Frontiers in Earth Sciences

Gilles Ramstein
Amaëlle Landais
Nathaelle Bouttes
Pierre Sepulchre
Aline Govin Editors

Paleoclimatology
Frontiers in Earth Sciences

Series Editors
J. P. Brun, Clermont-Ferrand, France
Onno Oncken, Potsdam, Brandenburg, Germany
Helmut Weissert, Zürich, Switzerland
Wolf-Christian Dullo, Paleoceanography, Helmholtzzentrum für Ozeanforschrung | G, Kiel,
Germany
More information about this series at http://www.springer.com/series/7066
Gilles Ramstein • Amaëlle Landais •
Nathaelle Bouttes • Pierre Sepulchre •
Aline Govin
Editors

Paleoclimatology

123
Editors
Gilles Ramstein Amaëlle Landais
LSCE/IPSL, CEA-CNRS-UVSQ LSCE/IPSL, CEA-CNRS-UVSQ
Université Paris-Saclay Université Paris-Saclay
Gif-sur-Yvette, Ariège, France Gif-sur-Yvette, Ariège, France

Nathaelle Bouttes Pierre Sepulchre

LSCE/IPSL, CEA-CNRS-UVSQ LSCE/IPSL, CEA-CNRS-UVSQ
Université Paris-Saclay Université Paris-Saclay
Gif-sur-Yvette, Ariège, France Gif-sur-Yvette, Ariège, France

Aline Govin
LSCE/IPSL, CEA-CNRS-UVSQ
Université Paris-Saclay
Gif-sur-Yvette, Ariège, France

ISSN 1863-4621 ISSN 1863-463X (electronic)

Frontiers in Earth Sciences
ISBN 978-3-030-24981-6 ISBN 978-3-030-24982-3 (eBook)
https://doi.org/10.1007/978-3-030-24982-3

Translated from the original French by Mary Minnock

© Springer Nature Switzerland AG 2021
This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part of the material is
concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction
on microfilms or in any other physical way, and transmission or information storage and retrieval, electronic
adaptation, computer software, or by similar or dissimilar methodology now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication does not
imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and
regulations and therefore free for general use.
The publisher, the authors and the editors are safe to assume that the advice and information in this book are believed
to be true and accurate at the date of publication. Neither the publisher nor the authors or the editors give a warranty,
expressed or implied, with respect to the material contained herein or for any errors or omissions that may have been
made. The publisher remains neutral with regard to jurisdictional claims in published maps and institutional
affiliations.

This Springer imprint is published by the registered company Springer Nature Switzerland AG
The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland
Foreword

A brief history of paleoclimates

Climate is undeniably a topical issue of utmost importance. It has been the focus of attention
for several decades now, during which the study of ancient climates (paleoclimatology) has
progressed and gained a solid reputation. Currently, this work has become fundamental to our
understanding of how the climate system functions and to validate the models used to establish
future projections. Thanks to the study of past climates, a database documenting a much
greater diversity of climate changes than during recent centuries, has been created. This
diversity makes it possible to test climate models in situations that are vastly different from
those we have known over the last 150 years and, in some cases, for climates that are closer to
those that await us in the future if we apply the conclusions of the Intergovernmental Panel for
Climate Change.
The Earth’s climate changes have always changed over time and will continue to change in
the future. While we are all aware of the weather phenomena that condition our daily lives, few
of us are aware of what climate really is. Climatology is the science that explores the great
variability in meteorological conditions over time and space, throughout history. This word
comes from the Greek word klima meaning inclination, in this case referring to that of the rays
of the Sun. Since the dawn of our civilization, therefore, we have linked the variations in climate
and in the energy received from the Sun in a relationship of cause and effect. For a long time, the
term ‘climate’ was reserved to describe the characteristics of air temperature and precipitation
particular to different parts of the globe. This description was based on meteorological mea-
surements and their averages conducted over a few decades. It is only recently understood that
climate also varies over much longer time scales and therefore concerns more than just the
atmosphere. At present, specialists studying climate and its variations analyze all of the fluid and
solid envelopes of the Earth. Along with the atmosphere, we associate the hydrosphere and the
cryosphere which, together, represent the systems where water exists in solid (snow, glaciers,
and ice sheets) and liquid form (rivers, lakes, and seas), the continents where plate tectonics and
volcanic activity occur, and finally at the surface, the whole living world (biosphere) that
influences nature, the properties of soil cover, and the biogeochemical cycles.
Climatology has evolved from being a descriptive discipline to become a multidisciplinary
science involving five complex systems and their various interactions. It is therefore not sur-
prising that the resulting climate studies vary on scales ranging from the season to millions of
years. Although it is only in the last few decades that this science has exploded, the first
discovery and study of climate change beyond the annual and decadal scales date back to the
eighteenth century. It was at this time that the presence of erratic boulders in the mountainous
landscape became associated for the first time with the massive extension of glaciers. In 1744,
the Grenoble geographer Pierre Martel (1706–1767) reported that the inhabitants of the Cha-
monix Valley in the Alps attributed the dispersion of roches moutonnées to the glaciers
themselves, which would have extended much further in the past. This was a revolutionary
idea, because until then, most scientists still referred to the myth of the Biblical Flood to explain
landscape structures. This was the case of Horace Benedicte de Chaussure (1740–1799) from
Geneva, the French paleontologist, Georges Cuvier (1769–1832) and the Scottish geologist,

v
vi Foreword

Charles Lyell (1797–1875), who continued to assume that these boulders were carried by the
strength of strongly flowing waters. However, the location and nature of these boulders and
other moraines led some scientists to admit that ice transport would provide a better explanation
for the various observations. The Scottish naturalist, James Hutton (1726–1797), was the first to
subscribe to this idea. Others followed his lead and detected the imprint of climatic changes in
the fluctuations of the extent of the glaciers. These pioneers were the Swiss engineer, Ignace
Venetz (1788–1859); the German forestry engineer, Albrecht Reinhart Benhardi (1797–1849);
the Swiss geologist, Jean de Charpentier (1786–1855); and the German botanist, Karl Friedrich
Schimper (1803–1867), who introduced the notion of ice ages. But it was the Danish-
Norwegian geologist, Jens Esmark (1763–1839), who, in pursuing his analysis of glacier
transport, proposed in 1824, for the first time, the notion that climate changes could be the cause
and that these could have been instigated by variations of Earth’s orbit.
It was the work of these pioneers that led the Swiss geologist, Louis Agassiz (1801–1873)
to make the address to the Swiss Society of Natural Sciences of Neufchatel in 1837 entitled
‘Upon glaciers, moraines and erratic blocks’. It was also at the beginning of the nineteenth
century that the Frenchman Joseph Adhémar (1797–1862), not content with studying the polar
ice caps, attempted to explain in his book, Révolutions de la Mer, Déluges Périodiques (1842),
the pattern of ice ages stemming from the precession of the equinoxes. The astronomical
theory of the paleoclimates was born and would be continued, thanks to the development of
celestial mechanics, by the Frenchmen, Jean le Rond d’Alembert (1717–1783), Jean-Baptiste
Joseph Delambre (1749–1822), Pierre-Simon Laplace (1749–1827), Louis Benjamin Fran-
coeur (1773–1849), and Urban Le Verrier (1811–1877). In parallel, other advances were made
with the first calculations of the long-term variations in the energy received from the Sun,
variations due to the astronomical characteristics of the eccentricity of the Earth’s orbit, the
precession of the equinoxes, and the obliquity of the ecliptic. This was demonstrated by the
work of John Frederick William Herschel (1792–1871), L.W. Meech (1821–1912), and Chr.
Wiener (1826–1896), supported by the work of the mathematicians André-Marie Legendre
(1751–1833) and Simon-Denis Poisson (1781–1840).
This sets the stage for James Croll (1821–1890) to develop a theory of ice ages based on the
combined effect of the three astronomical parameters, a theory according to which winter in
the northern hemisphere played a determining role. This theory was much appreciated by the
naturalist, Charles Robert Darwin (1809–1882), and was taken up by the Scottish geologist
brothers, Archibald (1835–1924) and James (1839–1914) Geikie, who introduced the notion
of the interglacial. It is also the basis for the classification of alpine glaciations by Albrecht
Penck (1858–1945) and Edward Brückner (1862–1927) and American glaciations by Thomas
Chowder Chamberlin (1843–1928). However, geologists became increasingly dissatisfied
with Croll’s theory and many critics of it emerged. Many refuted the astronomical theory and
preferred explanations that related to the Earth alone. The Scottish geologist, Charles Lyell
(1797–1875), claimed that the geographical distribution of land and seas explained the
alternation of hot and cold climates, while others turned to variations in the concentration of
certain gases in the atmosphere. Hence, the French physicist, Joseph Fourier (1786–1830),
expounded on the first notion of the theory of the greenhouse effect. He was followed by the
Irish chemist, John Tyndall (1820–1893), to whom we owe the first experiments on the
absorption of infrared radiation and the hypothesis of the fundamental role played by water
vapor in the greenhouse effect. Later, the Italian, Luigi de Marchi (1857–1937) and the
Swedish chemist, Svante Arrhenius (1859–1927) proposed, along with other scientists of their
time, that the ice ages were caused by decreases in atmospheric carbon dioxide concentration.
In 1895, Arrhenius suggested, in an article published by the Stockholm Physics Society, that a
40% reduction or increase in CO2 concentration in the atmosphere could lead to feedback
processes that would explain glacial advances or retreats.
A revival of the astronomical theory became, however, possible with advances in the
calculation of astronomical elements by the American astronomer John Nelson Stockwell
(1822–1920) and the Serbian astronomer Vojislava Protich Miskovitch (1892–1976) and of
Foreword vii

solar irradiation (1904) by the German mathematician, Ludwig Pilgrim (1879–1935). It was
Joseph John Murphy (1827–1894), however, who, as early as 1869, proposed that cool
summers of the northern hemisphere had instigated the ice ages. This original idea was taken
up in 1921 by the German paleoclimatologist Rudolf Spitaler (1859–1946), but was popu-
larized by the Serbian geophysicist engineer, Milutin Milankovich (1879–1958), mainly
through his books Mathematical Theory of Thermal Phenomena Produced by Solar Radiation
(1920) and Kanon der Erdbestrahlung und seine Anwendung auf des Eizeitenproblem (1941).
Milankovitch was a contemporary of the German geophysicist Alfred Wegener (1880–1930)
with whom he became acquainted through the Russsian-born climatologist Wladimir Köppen
(1846–1940), Wegener’s father-in-law (Thiede, 2017). The modern era of astronomical theory
was born, even if there remained much criticism related to the lack of reliable paleoclimatic
data and of a reliable timescale, both by geologists and meteorologists. It was not until the
1950s and 1960s that new techniques made it possible to date, measure, and interpret the
climate records contained in marine sediments, in ice and on land. In 1955, the American,
Cesare Emiliani (1922–1995), proposed a stratigraphy, which still applies today, based on the
succession of minima and maxima of the oxygen-18 / oxygen-16 isotopic ratio measured in
the foraminiferal shells found in sediments taken from the deep ocean. The interpretation of
this isotopic ratio in terms of salinity was made by Jean-Claude Duplessy (1970), and in terms
of temperature and volume of ice (1973) by Nicholas Shackleton (1937–2006) and Niels
Opdyke (1933–2019). Mathematical tools made it possible to establish transfer functions to
quantitatively interpret information collected in the oceans in 1974 by the American paleo-
ceanographers John Imbrie (1925–2016) and Nilva Kipp (1925–1989),) and in tree rings
(Harold Fritts, 1968). Efforts by the CLIMAP group (1976) resulted in the first seasonal
climate chart of the Last Glacial Maximum and the pivotal article by James Hays, John Imbrie,
and Nicholas Shackleton (1976). The arrival of big computers allowed the first climate sim-
ulations to be conducted using general circulation models (Fred Nelson Alyea, 1972), and
further astronomical calculations led to the establishment of a high-precision time scale ref-
erence, as well as the determination of the daily and seasonal irradiation essential for climate
modeling (André Berger, 1973 and Berger and Loutre, 1991).These calculations of the
astronomical parameters were based on the 1974 and 1988 developments of the orbital ele-
ments by the French astronomers Pierre Bretagnon (1942–2002) and Jacques Laskar,
respectively. These are valid over a few million years. The Laskar solution was extended over
a few tens of millions of years by Laskar et al. (2011) and over the whole Mesozoic with the
American paleobiologist Paul Olsen and colleagues (2019).
This evolution and the recent advances in paleoclimatology show the difficulties involved
in tackling the study of the climate system. Overcoming these difficulties requires high-quality
books to improve understanding and to update the range of disciplines involved. It is with this
perspective in mind that this book was written. Written originally in French, it unquestionably
fills a gap in the field of graduate and postgraduate third-level education that goes far beyond
its description. It provides an overview of the state of knowledge on a number of key topics by
outlining the information necessary to understand and appreciate the complexity of the dis-
ciplines discussed, making it a reference book on the subject. The first of the two volumes is
devoted to the methods used to reconstruct ancient climates, the second to the behavior of the
climate system in the past. Many of the thirty-one chapters are written by researchers from the
Laboratoire des Sciences du Climat et de l’Environnement and associated research labora-
tories each focusing on his or her area of expertise, which ensures a reliable document founded
on solid experience.
Understanding the evolving climate of the Earth and its many variations is not just an
academic challenge. It is also fundamental in order to better understand the future climate and
its possible impacts on the society of tomorrow. Jean-Claude Duplessy and Gilles Ramstein
have achieved this huge feat by bringing together fifty or so of the most highly reputed
researchers in the field.
viii Foreword

The book they have written is a whole, providing both the necessary bases on the recon-
struction techniques of ancient climates, their chronological framework, and the functioning
of the climate system in the past based on observations and models. This book will allow all
those who want to know more, to explore this science, which, although difficult, is hugely
exciting. It will also give them the essential information to establish an objective idea of the
climate and its past and future variations.
You may find most of the references and pioneering studies mentioned in this preface in
BERGER A. 2012. A brief history of the astronomical theories of paleoclimates. In: “Climate
change at the eve of the second decade of the century. Inferences from paleoclimates and regional
aspects”. Proceedings of Milankovitch 130th Anniversary Symposium, A. Berger, F. Mesinger,
D. Sijacki (eds). 107–129. Springer-Verlag/Wien.DOI 10.1007/978-3-7091-0973-1.

André Berger
Emeritus Professor at the Catholic University of Louvain
Louvain la Neuve, Belgium
Introduction

For a long time, geology books devoted only a few lines to the history of past climates of our
planet, mostly to establish the deposition framework for the sediments that geologists found on
the continents, the only area of enquiry available to them. Scientists soon realized that the
copious coal deposits of England, Belgium, Northern France, Germany, and Poland resulted
from the fossilization of abundant vegetation facilitated by a warm and humid equatorial
climate that reigned over Western Europe, some 350 million years ago (an illustrated insert in
Chap. 2 volume I provides a diagram of continental drift since 540 Ma). Fifty million years
later, the sediments of these same regions, red sandstone, poor in fossils and associated with
evaporites testify to the replacement of forests by desert areas, dotted with occasional highly
saline lakes, similar to what we currently find in Saharan Africa. Humidity gave way to intense
aridity and we had no idea why. It was not until the discovery of plate tectonics that we
realized that Europe had slowly drifted toward the tropics. This transformation of the face
of the Earth due to tectonics is illustrated through 16 maps in Chap. 2 volume I.
The discovery of glaciations was a revelation for the geologists of the nineteenth century.
A major polemic broke out at the Swiss Society of Natural Sciences in Neuchâtel when, in
1837, its president Louis Agassiz presented his explanation, incredible at the time, for the
presence of gigantic boulders that dot the Jura mountains. He daringly claimed that these
erratic boulders were not the remnants of the Biblical Flood, but rather enormous rocks
transported over long distances by gigantic glaciers which used to cover the high latitudes of
our hemisphere.
The controversy died down quickly, when European and American geologists discovered
traces of glaciers all over the Northern Hemisphere, just as Agassiz imagined. In Europe, as in
North America, mapping of the terminal moraines left behind by glaciers when they melted
showed proof of the presence of gigantic ice caps in a past that seemed distant. especially since
there was no idea how to date them.
As the idea of the Biblical Flood fell out of favor, a new theory, based on astronomical
phenomena, soon appeared. Scientists like Joseph Adhémar and James Croll realized that there
were small, quasi-periodic variations over time in the movement of the Earth around the Sun
and suggested that associated mechanisms could periodically cause glacial advances and
retreats. Finally, it was Milutin Milankovitch, a professor in Belgrade, who would lay the
foundations for a complete mathematical theory of glaciations, the legitimacy of which was
proven when paleoceanographers found the frequencies of orbital parameters reflected in the
isotopic analysis of marine cores. We now know that the last one of these glacial periods
culminated only 20,000 years ago and was preceded by many others.
The great contribution by Milankovitch was to plant a new idea within the scientific
community: Ancient climates are not only of immense curiosity to geologists; they obey the
same physical laws as those governing the current climate.
This intellectual revolution has had far-reaching consequences and has profoundly altered
the approach to the study of ancient climates making paleoclimatology a science with many
links to geology, geochemistry, oceanography, glaciology as well as the approach to the
physical and dynamic dimensions of the climate. The first part of this book describes the

ix
x Introduction

physical, chemical, and biological phenomena that govern the functioning of the climate
system and shows how it is possible to reconstruct the variations in the past at all timescales.
This is the work of paleoclimatologists. As soon as the means became available to them in
the second half of the twentieth century, they undertook to track down all traces of climate
change so as to establish a planetary vision. This led them to develop new methods of
sampling continental sediments, marine sediments in the context of major oceanographic
campaigns, and ice cores by carrying out large-scale drilling campaigns of mountain glaciers
and the ice sheets of Greenland and of Antarctica. The level of resources that needs to be
mobilized is such that the drilling campaigns of polar ice and of marine sediments from all the
world’s oceans could only be carried out in an international cooperative framework which
makes it possible to coordinate the efforts of the various teams.
This scientific investment has produced an abundant harvest of samples containing records
of past climates. On the continents, lake sediments; peat bogs; concretions in caves; and fossil
tree rings have provided many indicators of environmental conditions, especially of the
behavior of vegetation and the atmosphere. In the ocean, samples have been taken from all
of the large basins and cores are able to trace the history of the last tens of millions of years.
Finally, the large drillings in the ice sheets have provided information not only on polar
temperatures, but also on the composition of the atmosphere (dust and the concentrations of
greenhouse gases, such as carbon dioxide and methane).
Unfortunately, nature has no paleothermometer or paleopluviometer, and therefore, there is
no direct indicator of the changes in temperature or precipitation: Everything has had to be
built from scratch, not only to reconstruct the climates, but also to date them. Extracting a
reconstruction of the evolution of the climate from these samples has necessitated considerable
developments using the most innovative methods from the fields of geochemistry, biology,
and physics. Firstly, it was essential to establish a timeframe to know which period was
covered by each sample. Many methods were developed, and they are the subject of the
second part of this book. Radioactive decay, which is governed by strict physical laws, plays a
vital role. It has made it possible to obtain timescales converted into calendar years, and it has
provided clarification on stratigraphic geology. Other more stratigraphic approaches have been
implemented: identification of characteristic events that need to be dated elsewhere; counting
of annual layers; or modeling of ice flow. It has thus been possible to establish a chronological
framework, and paleoclimatologists are now trying to make it common to all data via an
on-going effort to make multiple correlations between the various recordings. Few climatol-
ogists rely on one indicator. The confidence that they have in reconstructing a climate change
at a given time is obtained by intersecting reconstructions from independent indicators but also
by confronting them with results from models. Methods of reconstructing the evolution of the
different components of the climate system from geological indicators then had to be devel-
oped. These are extremely varied, and their description constitutes the main and third part of
volume I. Many use the latest developments in paleomagnetism, geochemistry, and statistical
methods to empirically link the distribution of fossil plants and animals with environmental
parameters, primarily air and water temperature. Reconstructions achieved in this way have
now reached a level of reliability such that, for certain periods, not only qualitative variations
(in terms of hot/cold, dry/wet) can be obtained, but even quantitative ones with the associated
uncertainties also quantified. This is the level of climate reconstruction necessary to allow
comparison with climate models.
The use of climate models also gained momentum during the second half of the twentieth
century. First established to simulate atmospheric circulation, they have progressed by inte-
grating more and more efficiently the physics, processes, and parameterization of the radiative
budget and the hydrological cycle, in particular, by incorporating satellite data. However, the
atmosphere only represents the rapid component of the climate system.
The late 1990s dramatically demonstrated the need to link atmospheric models to global
patterns of the ocean and vegetation to reconstruct climate change. Indeed, teams from the
GISS in the USA and from Météo-France bolstered by their atmospheric models that had
Introduction xi

succeeded in reconstructing the current climate, independently tried to use the disruption of the
radiative budget calculated by Milankovitch to simulate the last entry into glaciation 115,000
years ago. In both cases, it was a total fiasco. The changes induced by the variation of the
orbital parameters in these models were far too small to generate perennial snow. The com-
ponents and feedbacks related to the ocean and terrestrial vegetation needed to be included.
Developing a model that couples all three of these components is what modelers have been
striving to achieve over the last 20 years, and these are the models that now contribute to the
international IPCC effort.
Today, the so-called Earth system models that incorporate aspects from the atmosphere–
ocean–terrestrial and marine biosphere, chemistry, and ice caps are used to explore the climate
of the future and the climates of the past. Spatially, they are increasingly precise, they involve
a very large number of processes and are run on the largest computers in the world. But, the
flip side of this complexity is that they can only explore a limited number of trajectories
because of the considerable computing time they require. Also, from the beginning, climate
modelers armed themselves with a whole range of models. From behemoths like the ‘general
circulation models’ to conceptual models, with models of intermediate complexity in-between.
From this toolbox, depending on the questions raised by the paleoclimatic data, they choose
the most appropriate tool or they develop it if it does not exist. With the simplest models, they
can explore the possible parameter variations and, by comparing them with the data, try to
establish the most plausible scenario. All of these modeling strategies are described in detail in
volume II, which constitutes the last and fourth part of this book.
This investigative approach at each step of the research work, dating, reconstruction,
modeling, and the back and forth between these stages allows us to develop and refine the
scenarios to understand the evolution of the past climates of the Earth. We are certain that this
approach also allows us, by improving our understanding of the phenomena that govern the
climate of our planet and through continuous improvement of the models, to better predict
future climate change. This comparison between models and data, which makes it possible to
validate numerical simulations of the more or less distant past, is an essential step toward the
development of climate projections for the centuries to come, which will, in any case, involve
an unprecedented transition.
Acknowledgement

This book would never have been possible without the very efficient help of Guigone Camus
and Sarah Amram. We are grateful to Mary Minnock for the translation of each chapter. We
also thank the LSCE for its financial support and Nabil Khelifi for his support from the origin
to the end. Last but not least, thank you to all our colleagues—more than 50—who patiently
contributed to this book.

xiii
Preface

Before taking this journey together into the Earth’s paleoclimates, it is important to know what
we will be facing. This exploration will bring us into the heart of the ‘Earth system’: a tangle
of interwoven components with very different characteristics and response times, a system in
constant interaction.
The first volume is dedicated (Chaps. 1 and 2) to an introduction to climate of the Earth.
Chapters 3–9 focus on different time measurement and datation technics. The most important
part of this first volume deals with the reconstructions of different climatic parameters from the
three major reservoirs (ocean, continent, cryosphere, Chaps. 10–21). The second volume is
devoted to modeling the Earth system to better understand and simulate its evolution (Chaps. 1
–9). Last but not least, the final chapter (Chap. 10) describes the future climate of the Earth
projection from next century to millennia.
The first part of this book (Chaps. 1 and 2) will equip the reader with a ‘climate kit’ before
delving into the study of paleoclimates. This quick overview shows the great diversity in the
systems involved. From the microphysics of the clouds that can be seen evolving over our
heads by the minute to the huge ice caps that take nearly 100,000 years to reach their peak, the
spatiotemporal differences are dizzying (Chap. 1). Yet, it is the same ‘Earth system’ that,
throughout the ages, undergoes various disturbances that we will address. Chapter 2 takes us
on a journey through the geological history of our planet. The distribution of continents,
oceans, and reliefs changes how energy and heat are transported at the Earth’s surface by the
ocean and the atmosphere.
The study of paleoclimates requires an understanding of two indispensable concepts in
order to describe the past climates of the Earth.
The first is the concept of time. Measuring time is fundamental to our research, and an
understanding of the diversity of temporalities particular to paleoclimatic records is essential.
The second part (Chaps. 3 to 9) of this book is devoted to the question of the measurement of
time. Different techniques may be implemented depending on the timescales considered in
Chap. 3. Thus, although carbon-14 (Chap. 4) provides us with reliable measurements going
back to 30,000–40,000 years ago, other radioactive disequilibria (Chaps. 5 and 6) need to be
used to access longer timescales. But it is not only the radioactivity-based methods that inform
us of the age of sediments; the use of magnetism (Chap. 7) is also a valuable way of placing
events occurring on the geological timescale into the context of climate. On shorter timescales,
the use of tree rings is also a valuable method (Chap. 8). Ice core dating techniques will also be
outlined (Chap. 9). This gamut of different methods shows how researchers have succeeded in
developing ‘paleo-chronometers’ which are essential to locate climate archives within a
temporal context, but also to establish the connections of cause and effect between the different
components of the Earth system during periods of climatic changes.
The second concept is that of climate reconstruction. Indeed, in the same way that there is
no single chronometer that allows us to go back in time, there is not one paleothermometer,
pluviometer, or anemometer. Just as it was necessary to invent paleo-chronometers based on
physical or biological grounds in order to attribute an age and an estimate of its uncertainty to
archives, the relevant climatic indicators had to be invented to quantify the variations in
temperature, hydrological cycle, and deepwater current. The third part of this book (Chaps. 10
xv
xvi Preface

–21) is devoted to the slow and complex work of reconstruction by applying this whole range
of indicators. Thus, we can reconstruct the climate of the major components of the climate
system: the atmosphere, the ocean, the cryosphere, and the biosphere. But we can also take
advantage of the specificities of temperate or tropical lakes, of caves and their concretions
(speleothems), of tree rings and even, more recently, of harvesting dates (Chap. 17). How can
paleo-winds or, to put it in more scientific jargon, the variations in atmospheric dynamics be
reconstructed? Based on the isotopic composition of precipitation (Chap. 10) or of the loess
(Chap. 13), not only can the evolution of the surface and deep ocean be reconstructed, but also
the geometry and dynamics of large water masses (Chap. 21). For land surfaces, palynology
and dendroclimatology enable us to retrace the evolution of vegetation and climate, respec-
tively (Chaps. 12 and 16). Finally, the cores taken from the ice caps of both hemispheres make
it possible to reconstruct the polar climate (Chap. 11).
In addition to these two main concepts, we also need to understand how fluctuations in the
hydrology of the tropics have caused variations in lakes (Chaps. 18 and 19) and glaciers
(Chap. 20); these factors also tell a part of the climate story. Other markers, such as spe-
leothems (Chap. 14) or lake ostracods (Chap. 15) reveal changes in climate in more temperate
areas.
Thus, a description of the global climate emerges from the local or regional climate
reconstructions. Through coupling these reconstructions with dating, our knowledge of cli-
mate evolution progresses constantly. Nevertheless, this image is both fragmentary, because
of the strong geographic and temporal disparity of our knowledge, and unclear, because of the
uncertainties in the reconstructions that the paleoclimatologist tries to reduce. There is still a
long way to go in terms of developing new indicators and improving those widely used in
order to complete and refine this description.
The second volume of this book (Chaps. 22–30) focuses on the major processes and
mechanisms explaining the evolution of past climate from geological to historical timescales,
whereas last Chap. 31 examines future climate projections. First of all, we address, in the very
long term, the interactions between tectonics and climate over the timescale of tens to hun-
dreds of millions of years (Chap. 22). Then, we deal with the biogeochemical cycles that
govern the concentrations of greenhouse gases in the atmosphere over the last million years
(Chap. 23). And finally, we consider the interactions with ice caps (Chap. 24).
We will continue our journey simulating the climate evolution through time from the
formation of the Earth (4.6 billion years ago) up to the future climates at scales from a few tens
of thousands to a hundred of years. On this journey, it becomes obvious that the dominant
processes, those that drive climate change, vary according to timescales: solar power, which
increases by about 7% every billion years places its stamp on very long-term evolution,
whereas at the scale of tens of millions of years, it is tectonics that sculpt the face of the Earth,
from the high mountain ranges to the bathymetry of the ocean floor. Finally, ‘the underlying
rhythm of Milankovitch,’ with a much faster tempo of a few tens of thousands of years can
produce, if the circumstances permit, the glacial–interglacial cycles described in the preceding
parts. On top of this interconnection of timescales, a broad range of processes and components
of the climate system is superimposed. Through these chapters, we would like to highlight the
need to model a complex system where different constituents interact at different timescales
(Chap. 25). With the development of these models, the scope of investigation is vast. Indeed,
ranging from recent Holocene climates (Chap. 30) to geological climates (Chaps. 26 and 27),
how they evolve is underpinned by very different processes: from plate tectonics (Chap. 22) to
orbital parameters (Chap. 28). The complexity of the system can also be seen in the abrupt
reorganizations of the ocean–atmosphere system (Chap. 29). The capacity acquired in recent
decades to replicate past climate changes using a hierarchy of models, and to compare these
results with different types of data, has demonstrated the relevance of this approach coupling
model simulations with data acquisition.
Preface xvii

Nevertheless, the field of investigation of the Earth’s past climates remains an important
area of research with many questions being raised about the causes of climate reorganizations
throughout the Earth’s history. Even though several chapters clearly show recent break-
throughs in our understanding of past climate changes, and the sensitivity of our models to
climate data has undeniably increased the extent to which we can rely on their outputs, we can
legitimately question what they contribute to future climate. Chapter 10 addresses these issues.
Will the ice caps, which have existed on Earth for only a short time relative to geological time,
withstand human disturbance? And can this disturbance, apart from its own duration, have an
impact on the rhythm of glacial–interglacial cycles?
At the end of these two volumes, you will have obtained the relevant perspective to project
into the Earth’s climates of the future. Indeed, by absorbing the most up-to-date knowledge of
paleoclimatology in this book, you will be provided with the necessary objectivity to critically
assess present and future climate changes. It will also give you the scientific bases to allow you
to exercise your critical judgment on the environmental and climatic issues that will be
fundamental in the years to come. Indeed, in the context of the Anthropocene, a period where
man’s influence has grown to become the major factor in climate change, the accumulated
knowledge of the climate history of our planet gathered here is precious.

Gif-sur-Yvette, France Gilles Ramstein

Amaëlle Landais
Nathaelle Bouttes
Pierre Sepulchre
Aline Govin
Contents

Volume 1

1 The Climate System: Its Functioning and History . . . . . . . . . . . . . . . . . . . . . 1

Sylvie Joussaume and Jean-Claude Duplessy
2 The Changing Face of the Earth Throughout the Ages . . . . . . . . . . . . . . . . . 23
Frédéric Fluteau and Pierre Sepulchre
3 Introduction to Geochronology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
Hervé Guillou
4 Carbon-14 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
Martine Paterne, Élisabeth Michel, and Christine Hatté et Jean-Claude Dutay
5 The 40K/40Ar and 40Ar/39Ar Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
Hervé Guillou, Sébastien Nomade, and Vincent Scao
6 Dating of Corals and Other Geological Samples via the Radioactive
Disequilibrium of Uranium and Thorium Isotopes . . . . . . . . . . . . . . . . . . . . . 89
Norbert Frank and Freya Hemsing
7 Magnetostratigraphy: From a Million to a Thousand Years . . . . . . . . . . . . . 101
Carlo Laj, James E. T. Channell, and Catherine Kissel
8 Dendrochronology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117
Frédéric Guibal and Joël Guiot
9 The Dating of Ice-Core Archives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
Frédéric Parrenin
10 Reconstructing the Physics and Circulation of the Atmosphere . . . . . . . . . . . 137
Valérie Masson-Delmotte and Joël Guiot
11 Air-Ice Interface: Polar Ice . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145
Valérie Masson-Delmotte and Jean Jouzel
12 Air-Vegetation Interface: Pollen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151
Joël Guiot
13 Ground-Air Interface: The Loess Sequences, Markers of Atmospheric
Circulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157
Denis-Didier Rousseau and Christine Hatté

xix
xx Contents

14 Air-Ground Interface: Reconstruction of Paleoclimates

Using Speleothems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 169
Dominique Genty and Ana Moreno
15 Air-Interface: d18O Records of Past Meteoric Water Using Benthic
Ostracods from Deep Lakes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179
Ulrich von Grafenstein and Inga Labuhn
16 Vegetation-Atmosphere Interface: Tree Rings . . . . . . . . . . . . . . . . . . . . . . . . 197
Joël Guiot and Valérie Daux
17 Air-Vegetation Interface: An Example of the Use of Historical Data
on Grape Harvests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 205
Valérie Daux
18 Air-Ground Interface: Sediment Tracers in Tropical Lakes . . . . . . . . . . . . . . 209
David Williamson
19 Air-water Interface: Tropical Lake Diatoms and Isotope Hydrology
Modeling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 213
Florence Sylvestre, Françoise Gasse, Françoise Vimeux,
and Benjamin Quesada
20 Air-Ice Interface: Tropical Glaciers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 219
Françoise Vimeux
21 Climate and the Evolution of the Ocean: The Paleoceanographic Data . . . . . 225
Thibaut Caley, Natalia Vázquez Riveiros, Laurent Labeyrie, Elsa Cortijo,
and Jean-Claude Duplessy

Volume 2

22 Climate Evolution on the Geological Timescale and the Role

of Paleogeographic Changes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 255
Frédéric Fluteau and Pierre Sepulchre
23 Biogeochemical Cycles and Aerosols Over the Last Million Years . . . . . . . . . 271
Nathaelle Bouttes, Laurent Bopp, Samuel Albani, Gilles Ramstein,
Tristan Vadsaria, and Emilie Capron
24 The Cryosphere and Sea Level . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 301
Catherine Ritz, Vincent Peyaud, Claire Waelbroeck,
and Florence Colleoni
25 Modeling and Paleoclimatology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 319
Masa Kageyama and Didier Paillard
26 The Precambrian Climate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 343
Yves Goddéris, Gilles Ramstein, and Guillaume Le Hir
27 The Phanerozoic Climate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 359
Yves Goddéris, Yannick Donnadieu, and Alexandre Pohl
28 Climate and Astronomical Cycles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 385
Didier Paillard
Contents xxi

29 Rapid Climate Variability: Description and Mechanisms . . . . . . . . . . . . . . . . 405

Masa Kageyama, Didier M. Roche, Nathalie Combourieu Nebout,
and Jorge Alvarez-Solas
30 An Introduction to the Holocene and Anthropic Disturbance . . . . . . . . . . . . 423
Pascale Braconnot and Pascal Yiou
31 From the Climates of the Past to the Climates of the Future . . . . . . . . . . . . . 443
Sylvie Charbit, Nathaelle Bouttes, Aurélien Quiquet, Laurent Bopp,
Gilles Ramstein, Jean-Louis Dufresne, and Julien Cattiaux
About the Editors

Gilles Ramstein is a director of research at Laboratoire des Sciences du Climat et de

l’Environnement (LSCE, France). His initial degree is in physics and since 1992 he has
specialized in climate modeling.
He has been responsible for many French and European research projects on the Pleis-
tocene, Cenozoic, and Precambrian eras. He has also been the advisor of many Ph.D. students
who have explored and expanded the frontiers of paleoclimate modeling.
As a climate modeler, he studies very different climate contexts from ‘Snowball Earth’
episodes (717–635 Ma) to more recent, and occasionally future, climate situations.
The main research topics he focuses on are
• Geological time from the Precambrian to the Cenozoic:
– Investigation of relationships between tectonics, the carbon cycle, and the climate with an
emphasis on the impact on the climate and the atmospheric CO2 cycle of major tectonic events
such as plate movements, shrinkage of epicontinental seas, mountain range uplift, and the
opening/closing of seaways.
– Leading international collaborations on projects on monsoon evolutions and the dispersal
of human ancestors during the Neogene periods.
• From the Pleistocene to future climate: In this framework, his major interests are
interactions between orbital forcing factors, CO2 and climate. More specifically, his focus is on
the response of the cryosphere, an important component of the climate system during these
periods, with an emphasis on the development of the Greenland ice sheet at the Pliocene/
Pleistocene boundary and abrupt climate changes driven by ice sheet variations.
He has also published several books and co-edited the French version of ‘Paleoclima-
tologie’ (CNRS Edition) and contributed to an online masters program devoted to educating
journalists on climate change (Understanding the interactions between climate, environment
and society ACCES).

Amaëlle Landais is a research director at Laboratoire des Sciences du Climat et de l’Envi-

ronnement (LSCE, France). Her initial degree is in physics and chemistry and, since her Ph.D.
in 2001, she has specialized in the study of ice cores.
She has been responsible for several French and European research projects on ice cores
working on data acquisition both in the laboratory and in the field, interacting extensively with
modelers. She has been the supervisor of ten Ph.D. students and is deeply committed to
supporting and training students in laboratory work.
Her main research interests are the reconstruction of climate variability over the Quaternary
and the links between climate and biogeochemical cycles. To improve our understanding
of these areas, she develops geochemical tracers in ice cores (mainly isotopes), performs
process studies using laboratory and field experiments, and analyzes shallow and deep ice
cores from polar regions (Greenland and Antarctica). Through numerous collaborations and
improvement of ice core dating methods, she tries to establish connections with other pale-
oclimatic archives of the Quaternary.

xxiii
xxiv About the Editors

Nathaelle Bouttes is a research scientist at the Laboratoire des sciences du climat et de

l’environnement (LSCE/IPSL). Following the completion of her Ph.D. in 2010 on the glacial
carbon cycle, she went to the University of Reading (UK) for a five-year postdoc on recent and
future sea-level changes. She then spent a year at Bordeaux (France) with a Marie–Curie
Fellowship on interglacials carbon cycle before joining the LSCE in 2016. Since then, she has
specialized in understanding glacial–interglacial carbon cycle changes using numerical models
and model–data comparison.
She is mostly using and developing coupled carbon-climate models to understand past
changes of the carbon cycle, in particular the evolution of the atmospheric CO2. She has been
focusing on the period covered by ice core records, i.e., the last 800,000 years. She uses
model–data comparison by directly simulating proxy data such as d13C to evaluate possible
mechanisms for the orbital and millennial changes. She has been involved in several projects
covering this topic as well as teaching and supervising Ph.D. students.

Pierre Sepulchre is a CNRS research scientist at the Laboratoire des sciences du climat et de
l’environnement (LSCE/IPSL). He completed a Ph.D. on the Miocene climate of Africa in
2007, then went to UC Santa Cruz (USA) for a two-year postdoctoral position working on the
links between the uplift of the Andes and atmospheric and oceanic dynamics. His lifelong
research project at CNRS is to evaluate the links between tectonics, climate, and evolution at
the geological timescales, focusing on the last 100 million years. Through the supervision of
Ph.D. students and his collaboration with geologists and evolutionary biologists, he also
worked at evaluating paleoaltimetry methods with the use of an isotope-enabled atmospheric
general circulation model, as well as linking continental surface deformation, climate, and
biodiversity in Africa and Indonesia. In recent years, he led the implementation and validation
of a fast version of the IPSL Earth system model that allows running long climate integrations
dedicated to paleoclimate studies.

Aline Govin is, since 2015, a research associate at the Laboratoire des Sciences du Climat et
de l’Environnement (LSCE, Gif sur Yvette, France). She studied Earth Sciences at the Ecole
Normale Supérieure of Paris (France) and obtained in 2008 a Ph.D. thesis in paleoclimatology
jointly issued by the University of Versailles Saint Quentin en Yvelines (France) and the
University of Bergen (Norway). Before joining the LSCE, she worked for five years as a
postdoctoral fellow at the Center for Marine Environmental Sciences (MARUM, University of
Bremen) in Germany.
Her research activities focus on the reconstruction of paleoclimatic and paleoceanographic
changes by applying various types of geochemical and sedimentological tracers on marine
sediment cores. She has mostly worked on the Earth’s climatic changes of the last 150,000
years and is an expert of the last interglacial climate, which is an excellent case study to
investigate the response of the Earth’s climate to past warming conditions that could be
encountered in the coming decades. Her research interests include the past variability of the
deep North Atlantic circulation, the responses, and drivers of tropical monsoon systems (e.g.,
South American Monsoon), the development and calibration of paleo-tracers, the development
of robust chronologies across archives, and the quantification of related uncertainties, as well
as the comparison of paleo-reconstructions to climate model simulations of past climates.
She has authored around 30 scientific publications and has been involved in many French,
German, and other international (e.g., Brazilian) projects.
 The Climate System: Its Functioning
and History 1
Sylvie Joussaume and Jean-Claude Duplessy

Climate plays an important role for mankind. It determines the average state of the atmosphere and its variability from
the conditions in which societies can develop as well as the this average. In practice, climate is defined by the average
resources available to them such as water and biological conditions over a thirty-year period. Although this working
inputs (agriculture, forests, livestock). However, climate is a definition makes sense while the weather is relatively stable,
complex system. It is the result of interactions not only it becomes more difficult to apply during a period of rapid
between the atmosphere, the oceans, landmasses and ice but change. This was the case in the twentieth century during
also the biosphere: the living world. It varies depending on which two phases of rapid increase in the average temper-
the timescale, and different mechanisms may come into play ature of our planet were detected by weather stations in the
at different scales. The aim of this book is to show how a WMO network, one from 1910 to 1940 and the other from
multi-disciplinary scientific community can now reconstruct, 1975 onwards (Fig. 1.1d). The period 1961–1990 is often
with increasing accuracy, the major features of past climates taken as a reference.
and discover how they are regulated by the geological
evolution, geochemistry, physics and biology of our living
planet, Earth. Climate Changes in the Past
Human living conditions are dependent on climate, but
human beings, in turn, influence the climate system. They Climate is essentially variable, regardless of the time scales
change the atmospheric concentration of greenhouse gases under consideration. Over past two millennia, historical
and aerosols, as well as the vegetation through deforestation chronicles and earliest instrumental measurements dating
and agriculture. For this reason, it is of major importance to back to the seventeenth century have shown the existence of
society to understand how climate works and how man may a very cold period in Europe from the sixteenth to the
be altering its course. This is a complex scientific problem nineteenth century (the Little Ice Age), preceded by the
because of the large number of feedbacks likely to occur and Medieval Warm Period and another warm period during
the study of past climates contributes to a better under- Roman times.
standing of them by analyzing major climate changes pro- Geological data also show large upheavals in climate. Of
voked by natural causes. course, this is over much longer periods than thirty years, but
geologists strive to define a stratigraphic framework and
precise geochronology to put these events into context
Climate Change within the history of our planet (Fig. 1.1). For example,
about seven hundred and fifty million years ago, the Earth
Definition of Climate went through an intense glaciation phase; glaciers flowed
down to sea level on every continent, even in low latitudes,
Climate is defined by the statistics of the physical charac- to such an extent that our planet could be described as a
teristics of the atmosphere. It differs from meteorology by snowball. Conversely, during the Mesozoic Era (25–
focusing on statistics over several decades, by calculating 65 million years), the conditions were hot, even at high
latitudes. During the Cenozoic Era (from 65 million years),
S. Joussaume (&)
J.-C. Duplessy the glaciers grew slowly, first on the Antarctic continent and
Laboratoire des Sciences du Climat et de l’Environnement, then on Greenland. For the past three million years or so, the
LSCE/IPSL, CEA-CNRS-UVSQ, Université Paris-Saclay, 91191
Gif-sur-Yvette, France Earth has experienced a succession of ice ages, marked by
e-mail: [email protected] glaciers advancing over land at high and middle latitudes of

© Springer Nature Switzerland AG 2021 1

G. Ramstein et al. (eds.), Paleoclimatology, Frontiers in Earth Sciences,
https://doi.org/10.1007/978-3-030-24982-3_1
2 S. Joussaume and J.-C. Duplessy

Fig. 1.1 a Variations in the 0

climate of Earth over the past Précambrien Ere
Paleozoic Ere
Mesozoic Ere
Cenozoic
Precambrian Primaire Secondaire Tertiaire
4 billion years, estimated from Era Era Era
geological data. b Changes in
mean climate estimated from hot
chaud
variations in the volume of
glaciers and ice sheets present on
the surface of continents for the present day
actuel
last million years. c Variations in
the average air temperature in the A
froid
cold
northern hemisphere from the
year 1000 reconstructed from 4600 3000 2000 1000 540
570 250
225 65 1,8
Age (Ma)
Time
paleoclimate data, including Niveau
Isotopic composition of oxygen (δ18O , ‰) Sea
historical data and the study of de lalevel
mer
2.5 (m)
tree rings. d Variations in the
3
average surface air temperature 0
calculated from the global 3.5
meteorological network data for 4
the period 1860–2010. Note that 4.5
the number of stations has varied 5 B -130
during this period and was sparse
5.5
throughout the late nineteenth
1000 900 800 700 600 500 400 300 200 100 0
century Age (ka)
Time
Ecart à la tempé
rature moyenne (°C)
0.8
0.6
0.4
0.2
0
-0.2
-0.4
-0.6 C
-0.8
1000 1100 1200 1300 1400 1500 1600 1700 1800 1900 2000
Age (AD)
Time
rature moyenne (°C)
0.8
0.6
0.4
0.2
0
-0.2
-0.4
-0.6 D
-0.8
1850 1870 1890 1910 1930 1950 1970 1990 2010
Age (AD)
Time

the northern hemisphere, separated by interglacial periods oscillations between glacial and interglacial periods marks
when the ice caps receded and remained confined to the intervals which are the result of small changes in the
Antarctic and Greenland. Earth’s orbit around the Sun. The glacial periods last
We have been in an interglacial period, called the Holo- almost ten times longer than the interglacial periods but
cene, for the past 11,000 years. The various aspects of the they are interspersed with rapid warmings which follow
evolution of climate will be expanded upon later in this the outbreaks of cold in the North Atlantic and neigh-
book. boring landmasses. All these major climate shifts do not
The last million years is the best understood geological occur as a result of chance and it is the work of clima-
period, because climate can be reconstructed from detailed tologists and paleoclimatologists to understand them by
information provided by polar ice and marine and conti- analyzing climate mechanisms and the causes of their
nental sediments. Over this period, a succession of variability.
1 The Climate System: Its Functioning and History 3

Climate Mechanisms amounts to 161 W/m2. It is offset by an infrared flux emitted

by the Earth and its atmosphere to space. In fact, the Earth
The Sun is the major driver of climate. Received solar behaves as a ‘black body’: it emits energy whose intensity is
energy plays a key role in establishing climate conditions on proportional to the fourth power of its absolute temperature
the surface of our planet. But these depend critically on the (287 K), in accordance with Stefan’s law. This radiation is
composition of the atmosphere and energy exchanges almost entirely concentrated in the infrared range between 4
between the surface of the planet and the atmosphere that and 100 lm (microns), with a maximum intensity centered
surrounds it. The Earth radiation balance compares, for each around 12 lm. Solar radiation also behaves like a ‘black
point on the Earth’s surface, the energy received from the body’ but at temperatures of around 6000 K, and covers a
Sun and that which is emitted back into space. Significant range of wavelengths from ultraviolet to near infrared, from
geographical differences drive wind and ocean currents 0.2 to 4 lm, and has a maximum intensity in the visible
which redistribute energy, influenced by the shape of the wavelengths of around 0.6 lm.
ocean basins and the relief of the land. In the absence of any greenhouse effect, i.e. if the
atmosphere were perfectly transparent to infrared radiation
emitted by the Earth, the temperature in equilibrium with an
The Radiation Balance of the Earth average absorbed flow of 161 W/m2, would be only −19 °C.
In reality, water vapor, liquid water in clouds, carbon
The Greenhouse Effect dioxide and other trace elements present in the air absorb a
A disk with a surface area of 1 m2, located equidistant large portion of infrared radiation emitted by the surface,
between the Earth and the Sun and intercepting solar radi- limiting the loss of energy towards space. Acting as ‘black
ation at a perpendicular angle, would receive an energy flow bodies’, all these constituents re-emit infrared energy in all
of 1368 W at the top of the Atmosphere (TOA). However, directions including towards the ground. This additional
the Earth is a sphere whose surface area is four times greater contribution means that the average surface temperature of
than that of a disk with the same diameter. This is why, on the Earth is 14 °C, not −19 °C. This greenhouse effect is a
average and over the course of a year, the solar flux inter- natural phenomenon due in large part to the presence of
cepted by a unit area is four times lower. It corresponds to a water vapor, which contributes about 55% of the total
power of 340 W/m2 TOA with a known accuracy of roughly greenhouse effect, to other greenhouse gases (carbon diox-
1 W/m2 (Fig. 1.2). Yet all this energy is not accessible to the ide, methane, nitrous oxide) which account for 28%, with the
Earth/atmosphere. A portion, about 36%, returns back to remainder caused by clouds. Throughout the geological
space after being reflected by the clouds, the suspended history of the Earth, the composition of the atmosphere has
aerosols in the air, the Earth’s surface and by the air mole- changed significantly and changes in the greenhouse effect
cules themselves. So, the real amount of energy absorbed have greatly contributed to past climate variations (see

Fig. 1.2 Radiation balance of

the Earth. The solar radiation
incident at ground level is fully
offset by infrared radiation
emitted towards space (Source
IPCC 2013)
4 S. Joussaume and J.-C. Duplessy

Chap. 2, Volume 1 as well was Chaps. 22, 26 and 27, the Sun since the formation of the solar system modify the
Volume 2). Since the beginning of the industrial era (about amount of energy it emits. In the second case, the slow
1850), human activities have significantly increased the variations of the movement of the Earth around the Sun
concentration of greenhouse gases already naturally present influence the seasonal and geographical distribution of
in the air and have introduced new ones, such as energy received in a given place on our planet.
chlorofluorocarbons (CFCs) which are active agents of the
greenhouse effect. Solar Cycles
In the mid-nineteenth century, the German astronomer,
The Water Cycle H. Schwabe, discovered spots on the Sun’s surface that
As solar radiation passes through the different layers of the appear and disappear over an eleven-year cycle. When solar
atmosphere, part of it is absorbed by ozone in the strato- activity is more intense, marked by a greater number of
sphere and by water vapor in the troposphere. About half of spots, the Sun emits more energy. Since the 1980s, satellite
the incident energy reaches the surface of Earth, where it is measurements allow the estimation of variations in intensity
partially compensated for by the loss of infrared radiation to of solar energy. These are around 0.1%, which corresponds
the atmosphere. An energy surplus of 104 W/m2 (Fig. 1.2) to a very small perturbation (0.24 W/m2) in the radiation
remains available at the surface. This energy warms up the balance of the Earth. Solar activity directly reflects changes
surrounding air and causes evaporation of water from the in the Sun’s magnetic field. The spots reappear in larger
surface of oceans and land, feeding into the water cycle on numbers when the magnetic field intensifies. Solar flares
our planet. The water vapor is then transported by winds then become stronger; they eject a larger number of particles
until it condenses as precipitation, releasing into the atmo- toward outer space and thus reinforce the solar wind. These
sphere the energy acquired at the surface during evaporation. electrically charged particles, mainly electrons and protons,
Thus, the cycle of evaporation and precipitation of water reach the Earth’s atmosphere where they cause magnetic
takes energy from the surface of the oceans and land and storms—strong disturbances in the magnetic field—as well
redistributes it in the atmosphere. This transfer of latent heat as magnificent auroras in the polar regions.
cools the surface and warms up the atmosphere, thus less- The influence of solar activity on climate has been
ening the differences in temperature between the upper and debated for many years. In the second half of the seventeenth
lower layers of the atmosphere, as well as between the century, documented observations indicate an almost total
equator and the poles. The water cycle thus plays a funda- disappearance of spots for a period of several decades,
mental role in the redistribution of energy between the sur- during the Little Ice Age (Fig. 1.3). At the end of the
face and the atmosphere. nineteenth century, the German astronomer H. Spörer and
Evaporation and condensation continuously renew the his English colleague W. Maunder linked these two phe-
store of water vapor in the atmosphere. However, the nomena, thus starting a controversy that persists today. The
amount of water vapor in the air at any given moment nature of the connection between the minimum solar activity
remains quite low. If it were completely condensed, the (called the Maunder minimum) and a decrease in the
liquid layer thus formed would cover the Earth’s surface in a intensity of solar radiation sufficient to induce a marked
layer 2.5 cm thick. Yet, on average, the water cycle involves cooling that coincided with that time, still needs to be
the evaporation and the precipitation of water which would explained.
correspond to a layer of about 80 cm per year. The recycling As the direct disruption in the solar radiation balance is
time of water in the atmosphere is therefore very fast and the too small to explain the phenomenon, it is believed that solar
water vapor is completely renewed in ten days. The water, activity may affect climate through circulation in the upper
most of which evaporates from the oceans (86%), returns atmosphere. Nevertheless, the link between variations in
there either by precipitation or through the flow of rivers and solar activity and the Earth’s climate remains a subject of
streams after runoff from land. Globally, on average, evap- research and a source of controversy given the absence of a
oration and rainfall balance each other exactly, thereby recognized physical mechanism. The relative role of external
maintaining a constant concentration of water vapor in air, as forcing (solar radiation) and internal/geological forcing
long as the average temperature of the air remains constant. (volcanism) in explaining the Little Ice Age still needs to be
assessed.
The Sun exhibits variations over longer periods. These
Sun-Related Variability can be seen not only in the number of sunspots, but also in
variations in solar diameter. This varies with a periodicity of
Variations in energy emitted by the Sun and the variations in 900 days, but this oscillation is influenced by solar activity.
the solar energy received by the Earth will affect the climate. It is minimal when the activity is at its maximum. Like
In the first case, the solar activity cycles and the evolution of sunspots, solar diameter measurements started in the
1 The Climate System: Its Functioning and History 5

Irradiance
Solar solaire
irradiance (W/m (W/m
2 )
)
2

1370

1365

1360
800 1000 1200 1400 1600 1800 2000
Irradiance
Solar solaire
irradiance (W/m ) (W/m
2 2
) Age (AD)
Time
1363
B

1362

1361

1360

1359

1358

1357
10/2/2003 24/6/2004 6/11/2005 21/3/2007 2/8/2008 15/12/2009
Date
Nombreofde
Number tâches
solar solaires
spots
200
C Minimum
180
de
Maunder
160
Maunder
minimum
140
120
100
80
60
40
20
0
1610 1660 1710 1760 1810 1860 1910 1960 2010
Age (AD)
Time
Fig. 1.3 a Variations in solar energy reconstructed from changes in values measured using satellites. b Variations in energy emitted by the
the beryllium content of polar ice and from the modeling of stellar Sun, NASA satellite measurements. c Changes in the number of
activity. It should be noted that the mean values calculated using sunspots observed by astronomers since 1610
stellar-activity modeling are higher, by about 5 W/m2, compared with
6 S. Joussaume and J.-C. Duplessy

seventeenth century. They led to the discovery of a cycle of Because of the elliptical nature of the Earth’s orbit, the
80–90 years, called the Gleisberg cycle, which modulates distance between the Earth and the Sun varies at different
the Schwabe cycle. times of the year. Currently, in the northern hemisphere, this
Solar wind plays only a minor role in the flow of charged distance is at its minimum in winter and at its maximum in
particles received by the Earth. Most comes from galactic summer, and the opposite is true for the southern hemi-
cosmic rays, which consists of electrons, protons, a particles sphere. In fact, the amount of solar radiation intercepted by
(ionized helium nuclei) and heavier ions in very small the Earth decreases as the distance increases. This causes
quantities. It is isotropic and comes from everywhere in milder winters and cooler summers in the northern hemi-
space. In periods of high solar activity, intense solar wind, sphere, while the seasonal contrasts are accentuated in the
through the magnetic field it creates, acts as a shield repel- southern hemisphere (although this impact is minor com-
ling the galactic cosmic radiation falling to Earth. This pared with the seasonal variations in high latitudes caused by
phenomenon inspired a geochemical method for determining obliquity).
variations in solar activity. Indeed, galactic cosmic rays, Over the millennia, the position of the solstices and
through spallation reaction on the atoms in the upper equinoxes slowly moves along the ellipse resulting in a
atmosphere, are responsible for the production of several variation in the solar energy received during each season.
cosmonucleides, the most well-known of which is This movement of precession of the equinoxes is caused by a
Carbon-14. Less Carbon-14 is produced during intense solar combination of two rotational movements. The first is the
activity. Measurements by geochemists on well-dated tree rotation of the Earth around an axis running through the
rings showed pseudo-periodic variations in the production of poles which is perpendicular to the elliptic plane. A gradual
this cosmonucleide. This is attributed to fluctuations in solar shift in the orientation of the axis of rotation is caused by the
activity, with periods of about 150–300 years (Suess cycles) attraction of the Sun and the Moon and traces out a circle
and 2300 years (Hallstattzeit cycles). The existence of these over the North Pole in a cycle of approximately
cycles has been confirmed by the measurement of other 26,000 years. The second is the elliptical orbit of the Earth
cosmonucleides, such as beryllium-10, in polar ice. These around the sun which is superimposed on the first. The
are trapped in ice in Greenland, whose location in time can combination of these two movements results in a periodicity
be determined simply by visually counting the annual layers of the precession of the equinoxes of about 22,000 years.
or, for earlier periods, through more complex methods More specifically, the Earth’s distance from the Sun fluctu-
described in Chap. 9. The paleoclimatologists are now ates, not only due to the precession movement of the equi-
investigating if these periodicities can be reflected in geo- noxes, but also due to variations in the eccentricity of its
logical records. orbit which varies according to a set of cyclical changes
occurring over two proximate periods, one of 19,000 years
Long-Term Variations in the Movement and the other 23,000 years. Thus, approximately
of the Earth Around the Sun 10,000 years ago, the Earth reached its closest point to the
The movement of the Earth around the Sun varies over time Sun at the time of the summer solstice and not at the boreal
under the influence of the gravitational attraction of other winter solstice as it does today. At that time, the northern
planets (see Chap. 28, Volume 2). The orbit traveled by the hemisphere received more solar energy in summer than it
Earth over a full year is almost exactly an ellipse with an does today and obviously less in winter.
eccentricity (the parameter which defines the degree of All of these modifications in the orbital parameters affect
flattening of the ellipse with respect to a circle) that can vary sunshine levels (still referred to as the insolation) at the dif-
over time. With periodicities close to 100,000 and to ferent bands of latitude on Earth, and particularly the intensity
400,000 years, the orbit goes from a circle with an eccen- of the seasonal cycle. Already, in 1924, the Serbian mathe-
tricity of zero to a slightly flattened ellipse with a maximum matician, Milutin Milankovitch, proposed that these slow
eccentricity of 6%. variations of the movement of the Earth around the Sun could
The tilt in the axis of the Earth relative to the ecliptic explain the glacial cycles. Indeed, as these slow variations
plane is known as its obliquity and it influences the amount induce a decrease in solar energy received in the summer at 60°
of sunshine received at different latitudes in different sea- N, snow, which has fallen in the winter, does not melt com-
sons. It is the reason for phenomena such as the polar night pletely. Furthermore, it strongly reflects solar radiation,
in winter and the midnight sun in summer at the highest facilitating the snow to persist. Gradually, the snow accumu-
latitudes. For this reason, the climate at high latitudes is lates and turns into an ice cap. This hypothesis has been
especially sensitive to variations in the obliquity. With a debated for many years and was strongly opposed until the
periodicity of around 41,000 years, the obliquity angle 1970s, when the cycles predicted by this theory were clearly
oscillates between 22° and 25°, the current value being close confirmed by paleoclimate records in marine sediments and
to 23° 26′. later in polar ice. We will see more precisely in Chap. 28 of
1 The Climate System: Its Functioning and History 7

Volume 2 the state of our knowledge about the Milankovitch periods, around 280 ppmv (280 cm3 of CO2 per m3 of air)
theory, or the ‘astronomical theory’ of paleoclimates. and only 200 ppmv during cold periods. Similarly, methane
ranged from *700 ppbv (mm3 per m3 of air) in warm
The Sun’s Evolution periods to less than 400 ppbv during cold ones, with a high
Since the formation of the solar system, the Sun, like all stars temporal variability (Fig. 1.4).
of the same type, slowly consumes its hydrogen to produce Polar ice is the only direct recording of the composition
helium, and the amount of heat it emits varies very slowly of the atmosphere. As ice in the ice caps flows very slowly
over long time scales. The standard stellar evolution models and is continuously renewed throughout geological time, it is
estimate that four billion years ago the luminosity of the Sun impossible to reconstruct a record of carbon dioxide levels
was 25–30% lower than it is today, and that it has increased before a million years ago. For earlier periods, it is therefore
more or less linearly over time. This model seems in necessary to use indirect empirical methods which have a
accordance with observations made by astronomers of much lower level of precision. These indirect “proxies” to
young stars. With the same Earth’s atmosphere as today four reconstruct atmospheric CO2 may be derived from stomata,
billion years ago, the average temperature of the Earth would boron isotope or alkenone (Chap. 27, Volume 2). For
be below 0 °C, oceans would be frozen, and life would be example, botanists observed that the stomata—pores through
impossible. Geological observations, however, indicate the which leaves absorb carbon dioxide from the air—are
presence of water in the liquid state and the first traces of life smaller and fewer when carbon dioxide is high. This plant
3.5 billion years ago. This is the ‘Pale Young Sun paradox’ characteristic has been used as a means of establishing CO2
which is solved by assuming that the atmosphere had a very levels for the past. However, the results were not clear-cut.
different chemical composition from that of today. Indeed, For one, the fossil species being studied must be the same as
the elimination of carbon dioxide by the young Earth, and the current species on which the empirical relationship
the low rates of weathering given the absence of continental between the levels of carbon dioxide and the number or
crust, meant that the atmosphere during these ancient periods diameter of the stomata is being established. Moreover, the
acquired an exceptionally high level of CO2 and hence was relationship, which can only be determined in the current
responsible for a strong greenhouse effect, further enhanced conditions, also depends on the availability of water to the
by the presence of methane produced by bacteria. This will plant, so it is not clear whether changes observed in fossil
be discussed in detail in the chapter on the Precambrian stomata are due to variations in CO2 or in moisture.
(Chap. 26, Volume 2). As the CO2 content of the air is governed by the partial
pressure of this gas in the surface waters of the ocean,
geochemists have tried to use carbon-13, a tracer of the
Reconstruction of the History of Atmospheric oceanic phase of the carbon cycle, as a tool to reconstruct
Composition changes in atmospheric CO2. One of the proposed markers is
the 13C/12C ratio in foraminifera, microscopic animals in the
Although the Sun is the source of energy for the Earth, the form of plankton living in the surface waters of the oceans.
energy made available depends essentially on the composi- These animals secrete a calcareous shell whose size is a few
tion of the atmosphere: greenhouse gases and particles. tenths of a millimeter and which are found in abundance in
Reconstructing the past history of the composition of the marine sediments. The 13C/12C ratio of planktonic for-
atmosphere is therefore an important element in under- aminifera therefore depends on the isotopic composition of
standing climate dynamics. dissolved CO2 in the surface waters, and indirectly on that of
Again, the last hundreds of thousands of years constitute the atmosphere.
the best documented period because of the fossil air bubbles This isotopic approach can be compared against the
contained in the polar ice caps. The snow falling on the polar independent records provided by the polar ice cores, so that
caps forms a porous firn, within which air circulates freely. the method can be evaluated over the last few hundred
Under the weight of accumulated snow, the pores gradually thousand years. The correlation is only proximate due to the
compact and the firn turns into ice that traps tiny air bubbles complexity of the oceanic carbon cycle which depends in
within it. This air keeps its original chemical composition, particular on the temperature of the sea water, on the primary
which allows the reconstitution of variations in the compo- production of the ocean, on the decomposition of organic
sition of the atmosphere over time, as long as we can find matter and on the circulation of the bodies of water. Biolo-
well-preserved ancient ice. The oldest ice is found in gists came up with another method when they noticed that
Antarctica, where a continuous recording of the greenhouse the fractionation of carbon isotopes during the absorption of
gas content (CO2, CH4) over the last 800,000 years has been carbon dioxide by seaweed depends on the dissolved carbon
established. These records show that the levels of carbon dioxide content and therefore the partial pressure of CO2 in
dioxide have not remained constant; they were high in warm the seawater. This led them to the hypothesis that variations
8 S. Joussaume and J.-C. Duplessy

Fig. 1.4 a Variations in CH4 (ppbv)

atmospheric methane 1000
concentrations inferred from
variations in methane 900
concentration in the air bubbles 800
trapped in the ice cores drilled at 700
Dome C (EPICA). b Variations in
600
atmospheric carbon dioxide
concentrations inferred from 500
variations in carbon dioxide 400
concentration in the air bubbles 300
trapped in the ice cores drilled at A
200
Dome C (EPICA). c Air
900 800 700 600 500 400 300 200 100 0
temperature variations in
Antarctica inferred from changes 350
in the isotopic composition of the CO2 (ppmv)
ice cores drilled at Dome C
(EPICA). d Changes in mean 300
climate of the Earth estimated
from variations in the isotopic 250
composition of oxygen in benthic
foraminifera, a proxy for
variations in the volume of 200
glaciers and ice caps on land B
surfaces over the last million 150
years 900 800 700 600 500 400 300 200 100 0
6
Anomalie de température
Temperature anomaly par
4 rapport à l'actuel (°C)
with respect to present day (°C)
2
0
-2
-4
-6
-8
-10 C
-12
900 800 700 600 500 400 300 200 100 0

Isotopicisotopique
Composition composition
des 2.5
foraminifères benthiques
of benthic (‰)
foraminifera (‰)
3

3.5

4.5

5
D
5.5
900 800 700 600 500 400 300 200 100 0
Age (ka)
Time

in the 13C/12C ratio in specific compounds formed during geochemical parameters, produce only rough estimates of
photosynthesis, such as in alkenones found in marine cores, the carbon dioxide composition of the air and its variations.
would reflect changes in the CO2 composition of the surface At long time scales (>106 years), the changes in the levels
waters of the ocean and of the air. However, the relationship of CO2 in the atmosphere are determined by the relative
obtained depends on the ratio of surface to volume of cells extent of degassing by volcanoes and mid-ocean ridges on
performing photosynthesis, which introduces a new uncer- the one hand, and the consumption of CO2 by chemical
tainty in the reconstructions. Finally, it is clear that even erosion of silicates on the other. This means that the key role
isotopic methods, which use accurately measurable is played by plate tectonics. Thus, a gradual reduction of
1 The Climate System: Its Functioning and History 9

CO2 in the air may be due to either a lower degassing rate or The Atmosphere
an increase in erosion of the surface of the continents. The
latter depends on a complex set of parameters, themselves The Main Features of Atmospheric Circulation
related to climate, such as air temperature, precipitation,
continental runoff and vegetation. Geochemists therefore try The net balance between the radiation received from the Sun
to reconstruct the changing partial pressure of atmospheric and that emitted into space does not have a uniform distri-
CO2 using models; CO2 emissions are estimated using bution. The net energy flux varies, depending on the latitude,
geological data on the speed of movement of the plates; geographic regions and season. Solar radiation decreases
consumption of the gas is taken into account in simplified significantly between the equator and the poles, but there is
models by coupling the carbon cycle to climate and by little difference in emitted infrared radiation. The result is a
considering the geographical context resulting from plate surplus of energy in the tropics and a deficit in the north and
tectonics. For example, the breaking-up of the arid super- south latitudes above 40°. Heated at the equator, cooled at
continent Rodinia, into a multitude of small humid conti- the poles, the atmosphere and the ocean are activated and
nental masses, 800–700 million years ago, led to the carry the excess energy from tropical regions to the deficient
creation of basaltic regions, easily erodible chemically. This higher latitudes. According to currently available measure-
resulted in a significant drop in carbon dioxide levels in the ments, the two fluids of the planet contribute with relatively
air which may explain the great glaciations of the period. similar amplitude to this transport (Fig. 1.5).
Airborne dust also plays an important role in the radiation A strong circulation in the atmosphere traveling from the
balance of the atmosphere, mainly by intercepting solar equator to the poles is established in order to ensure the
radiation and thereby reducing the amount of energy transport of energy necessary for the thermal balance of the
reaching ground level. Dust levels have varied considerably planet. The warmer air, and therefore lighter, rises above the
in the past, as is evidenced in polar ice. Falling snow brings equator, before diverging and heading at high altitudes
down atmospheric dust with it which then remains trapped in towards the poles. Above the polar regions, on the contrary,
the ice. The more the air is charged with dust, the more of it cold, dense air descends toward the surface, and travels
the snow absorbs. In this way, strong atmospheric dust levels toward the equator, which forms a large loop between the
during the glacial periods of the Quaternary have been equator and poles. This mechanism, described in 1735 by the
demonstrated. The dust came from continental erosion English scientist George Hadley, would happen if the Earth
which was then transported by winds. They gave rise to huge was turning very slowly. In reality, this large convection cell
accumulations of very fine particles. These created the loess remains confined between the equator and the subtropical
present in China, and in smaller quantities, in Western regions, where it forms the so-called ‘Hadley’ circulation.
Europe (Chap. 13, Volume 1). Associated with the Hadley circulation, low-pressure belts

Fig. 1.5 Average transport of Heat transport (PW)

energy by the atmosphere (thin
dotted) and ocean (dashed), and
total transport (solid line).
Positive transport towards the
north and negative towards the
south is recorded

South North
10 S. Joussaume and J.-C. Duplessy

predominate in the equatorial regions, while high pressures For example, during the Little Ice Age, variations in the
belts predominate in subtropical regions. A convergence of position of the winds were detected in the Pacific Ocean by
winds towards the equator, the trade winds, is thus observed analyzing the travelling time of galleons transporting goods
in the tropics at sea level. The trade winds are dry at the start between Manila (Philippines) and Acapulco (Mexico). The
of their journey, since they are powered by the descending General Archive of the Indies held in Seville relates that the
branch of the Hadley cell. Like the Harmattan over Africa, journey could take between less than three months and more
they maintain desert conditions on tropical continents. Over than four months. The routes were always the same:
the ocean, there is high evaporation of surface waters heated departing from Manila, the galleons went east, allowing
by solar radiation and the trade winds pick up this vapor and themselves to be carried by the stable westerly winds. For
carry it towards the low latitudes. the return, they headed west, catching the northeast trade
At the equator, with its ascending branch of warm moist winds, and the duration of the journey was determined, in
air, low-altitude convergence is manifested by strong con- the end, by the location of the opposing winds (from the
vective activity and heavy rainfall. These allow the devel- southwest) that they encountered as they approached Manila.
opment of a lush, tropical rainforest on land, while over the Historians were thus able to show the existence of a period
great ocean basins, convective activity is focused along a of about forty years in the middle of the seventeenth century
narrow longitudinal strip of one hundred kilometers wide: when the headwinds were very common due to a northward
the intertropical convergence zone. It is in this area that shift of the large depressions. Changes in the strength and
storms and rain are concentrated: the ‘doldrums’, dreaded by direction of winds are therefore an important manifestation
sailing vessels in the past, and now by solo sailors. of past climate changes.
Beyond 30° latitude, the flow of air, deflected eastward
by the Coriolis force, reaches such speeds that it becomes
unstable and breaks into eddies and meanders. Large Water Vapor, Clouds and Rainfall
meanders in this western circulation appear as vast oscilla-
tions, usually between three and six of them, which encircle Water in the atmosphere, in the form of vapor in the air, or as
the Earth. Depressions and anticyclones succeed each other a liquid or ice in clouds, plays an important role in climate
in the middle latitudes between 30 and 60° north and south, dynamics. Firstly, the amount of water vapor contained in
creating very variable weather conditions. This is the cause the air is a function of increasing temperature as defined by
of the ‘temperate’ climate prevailing in Western Europe. By the Clausius-Clapeyron relationship that links the saturation
mixing the hot subtropical air and cold polar air, these level of water vapor in air to temperature. When air tem-
vortices take over the transfer of the excess energy from the perature increases, its water vapor content increases. Fur-
tropics to the poles from the Hadley circulation. However, thermore, water vapor is the most important greenhouse gas
this circulation is affected by the contrast between land and in the atmosphere. Any increase in its concentration in the
oceans, and by the presence of mountains, both favoring the air in turn induces further warming of the atmosphere,
anchoring of global planetary waves whose intensity and engaging a positive feedback mechanism which amplifies
position change over time. These waves impact on the the original disturbance.
geographical distribution of climate and cause, for example, When the water vapor content exceeds the threshold of
a warmer climate on the west coasts than on the east coasts saturation, water vapor condenses, causing cloud formation.
of the continents of the northern hemisphere. The contrast Clouds have a particularly complex role, because of two
between the climate of Canada and that of France is a opposite effects. On the one hand, they reflect part of the
striking example. solar radiation, which has the effect of cooling the surface of
There have been very significant changes in the intensity the Earth. On the other hand, they have a greenhouse effect
and location of the winds in the past, particularly during that causes it to heat up. These two effects are not completely
glacial-interglacial oscillations. They are evidenced by the balanced. On average, with current climate conditions, the
presence, more or less marked, of pollen or desert dust reflectivity effect is more important that the greenhouse
transported to the ocean, sometimes very far from the coast, effect, and so, overall, clouds cool the Earth. But in the case
where they contribute to marine sedimentation. In the marine of climate change, will clouds play a moderating or ampli-
environment, when the wind blows parallel to the coast, it fying role? Low-level clouds and clouds in the upper layers
causes upwelling of deep cold water. Variations in their of the troposphere are very different. Low-level clouds are
intensity, reflecting that of the wind, result in variations in usually thicker, and reflect solar radiation, but being near the
the temperature of surface water that paleo-oceanographers surface, they have little impact on the greenhouse effect.
have managed to reconstruct (see Chap. 21). Conversely, high altitude clouds like cirrus, are much thinner
For recent periods, historical records provide information, and are very cold. They let solar radiation pass through, but
sometimes subtle, on the variability of the winds and storms. they contribute strongly to an increase in the greenhouse
1 The Climate System: Its Functioning and History 11

effect. It is not clear if global warming will be accompanied Quaternary. This has resulted in enormous variations in the
by more low-level clouds with a moderating effect, or volume of water flowing in the major rivers which are fed by
high-level clouds with an amplifying effect. The answer is the rains, such as the Niger and the Nile in Africa, or the
complicated further by the fact that it depends on changes in great rivers that drain the Himalayas and discharge into the
the general circulation in the atmosphere. Bay of Bengal. Variations in freshwater inputs to the ocean
Through coalescence, micro-droplets of water grow and have been so significant that they have resulted in large
turn into rain, snow or ice, depending on the temperature. fluctuations in salinity in areas close to the mouths of these
How precipitations are distributed reflects the main features rivers. Paleoceanographers are able to reconstruct these
of the general circulation of the atmosphere: ascendance and changes through isotopic and micropaleontological analysis
heavy rains at the equator, subsidence associated with a lack of marine sediments deposited near the river mouth
of rainfall in the sub-tropics, rainfall in the mid-latitudes (Fig. 1.6, Chap. 21, Volume 1).
linked to the passage of depressions. On average, excess in Land surfaces contain many other traces of major changes
the evaporation rate over the oceans is offset by a surplus of in the hydrological cycle in the past. During periods of great
rain over land. This transfer of water from oceans to land is aridity, dry winds have facilitated the creation of dunes which
particularly apparent during the seasonal phenomenon of the became established when the rains returned. The vegetation
‘monsoon’, well known in South East Asia but also in growing in different regions is as much determined by air
Africa. In summer, when the land warms up, low-pressure temperature as it is by precipitation and water availability.
heat expands and causes a convergence of ocean winds This is what causes the variation in the thickness and density
towards the mainland. Charged with humidity, the winds rise of the annual rings of trees. Pollens found in lake sediments,
and discharge a large amount of rain over land. peat bogs and in marine sediments close to the coast are used
by geologists to reconstruct the major vegetation types that
developed throughout the various geological periods allow-
Reconstructing Changes in Precipitation ing them to infer the temperature and humidity conditions
that then prevailed (Chap. 12, Volume 1).
There are many signs to be found on the surface of our
planet of how precipitation has changed over a given area.
The accumulation of ice in the polar ice caps is directly Modes of Variability of the Atmosphere
dependent on the supply of snow. Glaciologists have shown
that snowfall was half as abundant during glacial periods as Atmospheric circulation is very variable. Over short time-
it is today. At lower latitudes, fluctuations in rainfall are scales, variability is dominated by the duration of depres-
recorded in lake sediments. Deposits left above the current sions, usually a few days. Over longer time scales, the
water level are evidence of phases of intense rain, particu- circulation shows variability patterns over periods of up to
larly in the beginning of the interglacial periods. The levels several years. In Europe, the variability is dominated by
of the lakes drop when rains abate. For example, an aerial fluctuations in the system caused by low-pressure from
view of Lake Chad shows tracks of the various shorelines Iceland and high pressure from the Azores. This dipole
that record the regression of the lake over past millennia. oscillates between a ‘positive’ phase which is marked by a
About 6000–8000 years ago, it occupied an area of strengthening of the low and high pressures, stronger west-
340,000 km2 (equivalent to more than 2/3 of the area of erly winds bringing rain, and high temperatures in northern
France). By 2000, it was reduced to 1500 km2, or less than Europe, and the ‘negative’ phase where pressures and
1% of its maximum size. westerly winds subside, moving the rainy zone to the south
Lakes are not the only records of rainfall fluctuations on of Europe (Fig. 1.7). The ‘North Atlantic Oscillation’ (often
land. In limestone terrain, variations in the growth of con- designated by the acronym NAO) occurs at all time scales
cretions in caves are another indicator of fluctuations in the and explains about a third of the variability in weather
supply of groundwater by rain. Dating using geochrono- conditions in Western Europe, especially in winter. Positive
logical methods (see Chap. 14) can detect slowdowns or and negative phases tend to predominate for ten years or
arrested growth during dry periods, followed by recovery more, which makes the NAO particularly interesting in the
when the groundwater supply resumes in wet periods. study of the climate of Europe. This mode seems to be
Monsoons are a prime example of intense rainfall caused by the atmosphere alone, and yet it has an influence
affecting large areas, both in Africa and Asia. Rain falls over ocean circulation. The mechanisms that allow atmo-
during the summer months when the overheated land masses spheric circulation to present an oscillation over such a long
are the source of low pressures towards which the period are still not fully understood.
humidity-charged oceanic air masses converge. The intensity Did the North Atlantic Oscillation exist in the past?
of the monsoons has fluctuated considerably during the Measurements of atmospheric pressure, in particular ones
12 S. Joussaume and J.-C. Duplessy
 1 The Climate System: Its Functioning and History 13

b Fig. 1.6 Impact of variations in the intensity of the monsoon between the southeast of Northern America, but below average in
the last glacial maximum and current times on the hydrology of the Greenland, the Labrador Sea, in northwestern Africa and the
North Indian Ocean. a The current salinity of the Bay of Bengal where
the rivers draining the Himalayas flow is much lower than that of the Eastern Mediterranean. The opposite is true in the negative
Arabian Sea where evaporation is dominant. b The isotopic compo- phase of NAO.
sition of oxygen in planktonic foraminifera in recent marine sediments These teleconnections between different sites are used by
is a good record of these changes because there is little variation in the paleoclimatologists to reconstruct changes in the NAO in the
temperature of the northern Indian Ocean. c The isotopic composition
of oxygen in planktonic foraminifera in marine sediments deposited pre-instrumental period. Tree rings provide a good record of
during the last glacial maximum dated by 14C at around 18,000 years climate conditions, especially of periods of extreme heat or
BP shows a considerable decrease in the flow of rivers into the Bay of drought (Chap. 16, Volume 1). Recordings obtained from
Bengal and thus in the intensity of monsoon rains trees of North America and Europe show alternating phases,
carried out in Reykjavik and Gibraltar, allow scientists to some positive, some negative, since the beginning of the
reconstruct past trends as far back as 1850. Comparison eighteenth century, with short periodic elements (8 and
between winter air temperatures measured in weather sta- 2.1 years) but also multi-decadal ones (70 and 24 years).
tions shows that during the positive phase of the NAO, Snowfall and ice accumulation in the western part of
temperatures are higher than average in Western Europe and Greenland are abnormally low during the positive NAO

Fig. 1.7 The positive and

negative phases of the NAO and
their impacts on the distribution
of temperature, precipitation and
winds

Storm frequency
anomaly (%)

Temperature
anomaly (°C)

Precipitation
anomaly (mm)
14 S. Joussaume and J.-C. Duplessy

phase, and this has allowed the reconstruction of the NAO The atmosphere and the oceans exchange momentum
and its intensity over the last 350 years with, again, the through the friction exerted by winds at the air-sea interface.
detection of multi-year and multi decadal intervals. Tele- They are thus responsible for the great marine currents,
connections associated with the NAO are also recorded in well-known to ocean-going sailors. The atmosphere and the
marine sediments. Over the last ten thousand years, there has oceans also exchange energy and water. Energy exchanges,
been a tendency towards cooling of the surface waters of the through solar radiation, the infrared flow, turbulent eddies at
eastern North Atlantic Ocean while there has been a con- the surface, and sensitive and latent heat, impact on the
trasting warming of the subtropical western Atlantic and the temperature of surface waters of the ocean. Water exchan-
eastern Mediterranean. This trend is seen as a sign of a ges, through evaporation and precipitation, have an impact
continued weakening of the NAO during the Holocene. on salinity: evaporation increases the salinity of seawater
However, on these time scales, it becomes difficult to while rain, conversely, decreases salinity. These interactions,
distinguish between a change in the NAO expressed over which create variations in the temperature and salinity of
several millennia and a long-term climate trend, driven by seawater, ultimately determine its density. Density is indeed
the slow fluctuations in orbital parameters. Indeed, cyclic inversely proportional to temperature and directly propor-
variations in the precession (with a cycle of 21,000 years) tional to salinity. Density differences are then the cause of
were responsible for an increase in the winter incident solar the movement of large, deep-water masses in the world’s
radiation in the tropics 10,000 years ago, followed by its oceans.
progressive decrease accompanied by a drop in the differ- These physical interactions, shown schematically in
ence in atmospheric pressure between the tropics and the Fig. 1.8, are supplemented by exchanges of matter, such as,
high northern latitudes during the Holocene. This is an for example, of carbon dioxide or sulfur compounds, which
example of the changes in insolation changes predicted by interact with the biogeochemical cycles of the different
astronomical theory. elements. In this way, physics, chemistry and biology are
very closely linked in the ocean.

The Oceans
Oceanic Circulation
Main Characteristics of the Oceans
In the tropics, winds cause large ocean anticyclonic circu-
The oceans cover two thirds of the surface of the planet. lation, called ‘vortices’ or ‘gyres’, turning in a clockwise
With an average depth of 3900 m, they have a very high direction in the northern hemisphere and counterclockwise
thermal inertia, much greater than that of the atmosphere. in the southern hemisphere. However, there is a marked
A layer 3 m deep of ocean surface waters has the same heat asymmetry between the eastern and western sides of the
capacity as the 10 km troposphere. This feature explains ocean basins. For example, to the east of the North Atlantic,
why coastal regions have a much less contrasted climate than the Canary Current spans a much wider area and has a much
regions in the interior of large landmasses. It also plays an lower intensity than the current on the western side, the Gulf
important role in determining the response time of the Stream, which is very intense and is only a little more than
atmosphere-ocean system to a disturbance in the radiation 100 km wide. This strengthening of currents along the
balance. western edges of ocean basins is not specific to the Atlantic

Atmosphere

Winds

Temperature Ocean currents Salinity

Ocean

Fig. 1.8 Interactions between the atmosphere and the oceans: winds oceans condition the temperature and surface ocean salinity that change
drive the surface currents that transport temperature and salt in the the density of the water, thereby causing ocean currents
oceans. The exchange of energy and water between the atmosphere and
1 The Climate System: Its Functioning and History 15

Ocean. It is also observed in the warm current of Reconstructing Ocean Circulation in the Past
Kuro-Shivo in the North Pacific and in the Agulhas Current
which runs along the South African coast in the southern Paleoceanography reconstructs past ocean circulation by
hemisphere. This phenomenon is a complex result of the analyzing sediments that have settled in more or less regular
increasing strength of the Coriolis force with distance from layers on the ocean floor, the uppermost layers corre-
the equator. sponding to the most recent deposits. This discipline has
Differences in density also lead to large movements of the flourished thanks to several scientific and technical devel-
oceanic water masses. The densest waters are in the polar opments. One of these developments is in the methods of
regions where sea ice forms. Already dense because of their coring and drilling which allow cores to be brought to the
low temperature, the waters receive a further load from two laboratories which have relatively undisturbed sediments
salt inputs: one coming from the flow of currents from the reliably recording the conditions in the ocean at the time the
subtropics, where evaporation is intense and the other from sediment was deposited. Furthermore, in-depth analysis of
salt released during sea ice formation. These dense waters fauna (foraminifera) and flora (diatoms, coccoliths) fossils
tend to drop due to gravity below the warmer and less salty that lived in the illuminated area or in the water column
waters, then to spread out at the bottom of the ocean where permit the reconstruction of the temperatures of surface
the temperature, around 0–2 °C, varies little from the poles waters (see Chap. 21). Finally, new geochemical methods
to the equator. These dense water masses are the starting based on the analysis of stable and radioactive isotopes of
point of the great global circulation loop of the ocean, called elements present in marine sediments have provided dating
thermohaline, since denser waters tend to sink below less methods, stratigraphic markers and tracers of large marine
dense waters and the density of water masses is dependent currents (see Chaps. 4, 6 and 7, Volume 1).
only on temperature and salinity. This mechanism plays an It is now possible to draw up temperature maps of surface
important role in ocean circulation, as it contributes over waters of the ocean at critical times of the history of climate
75% to the formation of all the masses of deep waters of the on Earth, such as during the Last Glacial Maximum with a
14
world ocean. Paradoxically, this process of downwelling C age close to 18,000 years (in other words, a calendar age
deep water occurs only in a very small fraction of the surface of about 20,000 years, Chap. 2) or during interglacial peri-
of the oceans: in the Labrador Sea, the Norwegian and ods of the Quaternary. Paleoceanography also allows the
Greenland Seas and in some regions of North Atlantic as reconstruction of the movements of fronts separating surface
well as at the edge of the Antarctic continent, particularly in water masses with very different characteristics: a descent of
the Weddell Sea. In overall, the combination of the ther- polar waters in lower latitudes, driving out temperate waters,
mohaline circulation and the circulation caused by winds is accompanied by a strong cooling also felt by adjacent
makes up the ocean meridional overturning circulation. coastal areas. Conversely, their retreat is accompanied
Downwelling waters in the North Atlantic descend to a immediately by a significant warming.
depth of 2000–3000 m up to latitude of around 60° south, Marine sediments also contain markers for the conditions
where water masses undergo a slow movement of ascent to that prevailed at depths, at the water-sediment interface. The
the surface. Carried by the Antarctic Circumpolar Current most important of these are benthic foraminifera, micro-
that runs from west to east around the southern polar conti- scopic animals with a calcareous shell whose isotopic
nent, the deep waters from the North Atlantic then spread out composition is a particular reflection of the temperature and
in the South Pacific and the Indian Ocean. The return part of dissolved carbon dioxide content of the deep waters of the
this great circulation loop occurs through warm currents near ocean. Measurements by oceanographers show that at a
the surface. They pass between the Indonesian islands, cross certain depth in the ocean, the physical and chemical char-
the Indian Ocean, circumnavigate Africa by the Agulhas acteristics of the waters are almost constant over a distance
Current, and then up towards the North Atlantic with the Gulf of several dozen kilometers. By taking sediment cores at
Stream and the North Atlantic Drift. But, while the return by different depths in an ocean basin, it is possible to recon-
the warm currents takes a few decades or even up to a hun- struct the features of large deep water masses, to deduce the
dred years, it takes several hundred to a thousand years from main features of their circulation within a given period
the time the cold waters sink in the North Atlantic to their and to track their variations over time. Paleoceanography
arrival in the center of the Pacific, showing the slowness of therefore allows the reconstruction of the main features of
this gigantic mixing achieved by the deep circulation. changes in the ocean in three dimensions.
16 S. Joussaume and J.-C. Duplessy

El Niño, Interplay Between the Atmosphere El Niño events in the past are known. In the Andes, the
and the Oceans arrival of the warm waters on the Pacific coast brought
heavy, sometimes catastrophic, rainfall that caused floods or
Interactions between the atmosphere and the oceans can lead even huge landslides that geologists are able to date. Pre-
to climate variability modes. El Niño is a perfect example. historic sites also bear traces of these events, and archaeol-
Every two to ten years, an abnormal situation occurs in the ogists have established that certain Andean civilizations
Pacific, manifested by the appearance of unusually warm developed during periods when El Niño events were rare or
waters along the coast of Peru and major disruptions in the weak and regressed with the return of torrential rains.
tropical rainfall pattern. However, the most precise way to reconstruct the sequence
In ‘normal’ periods, the water temperature is around 28– of these events is by analyzing the geochemistry of the
29 °C in the western tropical Pacific, while it does not corals abundant in the waters of the equatorial Pacific Ocean
exceed 20–25 °C in the east. This strong asymmetry in and located at the heart of the phenomenon. For example,
temperature between the east and west maintains the atmo- Tarawa Atoll located near the International Date Line (180°
spheric circulation which, in turn, maintains the temperature meridian) usually has a dry, almost desertic climate. When
gradient. The warmer waters provide the heat and humidity an El Niño event occurs, the warm waters reach it, atmo-
necessary for the development of strong convective activity spheric convection becomes intense locally and heavy rains
over the western Pacific, which develops the air ascendance fall on the entire atoll and the coral reef that surrounds it.
associated with the Hadley circulation, while the air des- Corals, animals with a calcareous skeleton with recognizable
cends over the cold waters of the eastern Pacific. This annual bands, record these rainy passages. Analysis of them
asymmetry between the eastern and western Pacific is has allowed the number of El Niños in the last century to be
associated with a circulation called Walker. In turn, the trade counted and to show that their frequency has changed over
winds at the surface, which blow from east to west, maintain the last hundred years. Conversely, near Australia, which is
the east-west temperature gradient. usually in a region of warmer water, El Niño events are
In the eastern Pacific, they cause a surface current characterized by decreased rains and cooling, which were
deflected by the rotation of the Earth to the right in the reliably recorded by the coral reefs of New Guinea or Fiji.
northern hemisphere and to the left in the southern hemi- These recordings show that El Niño events have also existed
sphere, which drives out the surface water on both sides of during periods of glacial climate, and confirm that their
the equator, and causes an upwelling of cold water to frequency and intensity have varied in the past, the twentieth
compensate. The trade winds also propel surface waters century being a period during which they were particularly
towards the west, where the accumulation of water inhibits strong (Fig. 1.9).
the upwelling process; heated by the Sun, these waters reach
the highest ocean temperatures, thereby favoring intense
convective activity. This situation is called La Niña when the The Terrestrial and Marine Biosphere
differences between east-west are particularly strong.
During an El Niño event, the circulation of both the ocean The biosphere, defined as all living organisms, also inter-
and the atmosphere change simultaneously following a ser- venes in the operation of the climate system. Some theories
ies of mutual actions and reactions in which it is impossible even propose that, throughout geological time, the biosphere
to distinguish which of the ocean or the atmosphere triggers has contributed to the regulation of climate in order to create
the phenomenon. In particular, the water of the central conditions compatible with life.
Pacific is warmed to 28–29 °C, which has the effect of
moving the high convective activity towards the east. A de-
crease in the strength of the trade winds in the western The Geographical Distribution of the Biosphere
Pacific follows, and possibly even a reversal of their direc-
tion. With weaker trade winds, the surface current weakens On land, the biosphere is mainly made up of vegetation
and the warm waters of the western Pacific flow back which is distributed according to the critical climate char-
towards the east, a backlash that warms the central Pacific acteristics which are sunshine, temperature and precipitation.
and interrupts the upwelling of cold waters at the coasts of Rainforests can only develop if temperature and humidity
South America. But El Niño starts a series of waves in the conditions are favorable for the twelve months of the year.
ocean that eventually pans out and restores the so-called They are replaced by a dry forest or savannah if the soil
‘normal’ situation. water content decreases over several months. The savannah
Because of their well-recorded and varied consequences, itself becomes increasingly sparse as aridity increases
the existence of variations in the intensity and frequency of eventually becoming a desert. North of the tropics, seasonal
1 The Climate System: Its Functioning and History 17

Salinity

Temperature (°C)

El Niño event

Fig. 1.9 Changes in the temperature and salinity of surface waters in 1960–1995 for which instrumental measurements are available was
the archipelago of Fiji (Western Pacific) reconstructed from changes in used as a calibration period (Courtesy of Dr. Anne Juillet LSCE)
the isotopic composition of coral living in coastal areas. The period

changes in temperature and humidity dominate the climate. seem to have reinforced the intensification of monsoon rains
Vegetation takes the form of deciduous forests in humid during the mid-Holocene, 6000 years ago.
temperate regions, of drought-resistant flora in the During the 1980s, the discovery of past variations in
Mediterranean area or prairie grass in dry areas with strong carbon dioxide and methane in air trapped in Antarctic ice
seasonal contrasts. Even further north is the area of the has directed the spotlight onto the role of the biosphere,
boreal forest (taiga), consisting of birch and conifers, and hitherto neglected. Indeed, these changes that show disrup-
finally the tundra, where trees cannot grow. The mapping of tion of biogeochemical cycles cannot be explained solely by
current vegetation is therefore closely linked to major cli- the physical exchanges between the atmosphere and oceans.
mate zones. The biosphere must be taken into account. The land-based
In the oceans, the marine biosphere depends on the biosphere absorbs carbon dioxide by photosynthesis, but
temperature and salinity, but also on the amount of light and emits it by respiration during the life of the plant and also
nutrients available to enable the production of phytoplankton later, during its decomposition in the soil through the action
which forms the basis of the ocean food chain. However, the of bacteria. On average, as long as the climate remains
areas where the production of phytoplankton is abundant are constant, the absorption and emission of carbon dioxide by
very limited. Apart from some coastal fringes of tropical the land-based biosphere balances out and biological activity
regions, the Southern Ocean and the North Atlantic are the recycles atmospheric carbon.
only areas capable of producing enough nutrients to con- In the oceans, phytoplankton also absorbs carbon dioxide
tinuously feed a wide range of living matter. Most of the through photosynthesis. This carbon is then reused to form
ocean areas consist of essentially sterile water, hence the the tissues of other living organisms in the food chain as well
bright blue color of tropical waters. as organic waste of all kinds. In this way, phytoplankton is
responsible for a rapid recycling of carbon in the surface
waters of the oceans: the absorption carried out during
The Role of the Biosphere photosynthesis is offset by the constant emission of carbon
dioxide caused by the respiration of algae, zooplankton and
On land, vegetation alters the exchanges of energy, water and fish, as well as by the oxidation of waste. A fraction of this
momentum. Vegetation permits greater solar energy absorp- carbon, about 10%, is subtracted from this recycling. Fecal
tion than bare soil. Indeed, the albedo (reflection power) of pellets, dead tissue and other waste sink as a result of their
vegetation cover is 10–15% compared to 35% for bare soil. weight and take some of the carbon absorbed at the surface
Trees also enhance the evaporation from the surface, through of the oceans with them to the ocean floor. Most of this
transpiration by the foliage and through pumping water from ‘marine snow’ dissolves or decomposes through the action
the soil by the root system. Finally, a tree creates an obstacle at of bacteria before reaching the bottom of the ocean, releasing
the surface and increases turbulence close to the ground, hin- carbon organic matter which is added to the dissolved car-
dering the wind more efficiently than bare soil. Based on bon in the deep ocean waters. A tiny fraction of this ‘marine
numerical simulations, these physical effects of vegetation snow’, around 1% of the carbon taken from the surface, is
18 S. Joussaume and J.-C. Duplessy

deposited in the ocean abysses, forming sediment in which It is important to be aware that transfer functions are only
carbon is trapped for millions of years. valid insofar as the species discovered in ancient sediments,
both land and marine, are the same species as those that are
present today. Even if only a rough estimate of the condi-
Biosphere of the Past and Paleoclimates tions that prevailed in the geological past is required, it is
important to remember that species evolve and are likely to
For a long time, geologists have known that the land-based adapt to very different environments over millions of years.
and marine biospheres varied greatly in the past in response Thus, in the Jurassic era, 150–200 million years ago,
to the slow process of evolution (on the scale of millions of well-developed coral reefs in warm waters harbored a variety
years, Chap. 27, Volume 2) but also to changes in the global of mollusks such as Pholadomya, Tridacna and Astarte.
climate and in the general circulation of the atmosphere and Their distant descendants can be found today in very dif-
oceans, particularly at the pace dictated by the astronomical ferent environments: Pholadomya buried in the mud in
paleoclimate theory (Chaps. 28, 30, Volume 2). warm coastal waters, in a reef environment for Tridacna and
During their reproduction cycle, most plants produce tiny in polar waters for Astarte. This shows how the recon-
grains with a shape and decoration characteristic of their struction of paleoenvironments requires a cautious approach
species. These are the pollen (the male fertile element of the and the comparison of the various clues found in fossils.
flower) and spores (vegetative structure for propagation or
reproducing) which have a very hard outer shell. They are
transported by wind or streams and stored well if they land in The Cryosphere
a low-oxygen environment. They have been found in lake
sediments and bogs, where they provide a reasonably accu- Water in the form of ice or snow is the cryosphere. Glaciers
rate picture of the vegetation once covering the vicinity of the and ice caps cover approximately 11% of the surface of the
lake or bog. In this way, it has been discovered that, twenty Earth. Spread out over the oceans, the water they contain
thousand years ago, the vegetation of France was that of a would increase the sea level by about 77 m. Sea ice, which is
polar steppe, at the peak of the last glaciation. formed by freezing seawater, covers approximately 7% of
Land sediments contain other fossils which also provide the oceans, but is only a few meters thick. As it floats, its
information on the local climate: diatoms, mollusks, ostracods melting does not cause the sea level to rise. On land, snow
(tiny crustaceans with a calcareous shell) living in freshwater, cover varies greatly with seasons.
larger remains of plants (stem and leaf), charcoal remains from
large natural fires, fossil remains of larger animals, such as
bones of mammals found in archaeological sites. The Role of the Cryosphere
The link between the climate and plant fossils is so close
that statistical methods have been developed to quantify this The main property of the cryosphere is its albedo. It can
link, based on current observations. The resulting relation- reach 80–90% for fresh snow and barely drops below 50%
ships, called ‘transfer functions’ are used to estimate past when surface ice melts or when the snow covers trees. This
climate conditions if the same plant associations are found in property introduces the second major positive feedback loop
ancient sediments as those known today (see Chap. 12, of the climate system. Where snow or ice melts out, the
Volume 1). ground absorbs a larger fraction of the incident solar energy.
Leaving aside the abundant fossils present in coastal areas Containing more energy, it heats up, thus facilitating the
and in coral reefs, the marine biosphere leaves many traces melting of the remaining snow and ice. The process thus
of its diversity in marine sediments: fish otoliths and teeth, becomes amplified. In the Milankovitch theory of ice ages
aragonitic shells of pteropods (marine snails), calcareous (see Chap. 28, Volume 2), the decrease in incoming solar
shells of planktonic and benthic foraminifera, coccoliths radiation, due to the slow variations of the orbit of the Earth
(calcareous sheets secreted by microscopic algae, called around the Sun, triggers the formation of ice caps thanks to
coccolithophorids and living in warm or temperate waters), this positive feedback. Indeed, following the decrease of
siliceous skeletons of marine diatoms (algae living in cold sunshine, snow accumulated in the winter does not melt
waters rich in silica) and radiolarians (microscopic animals completely, in turn less energy is absorbed and the snow
living in deep water). The relationship between the tem- starts accumulating until ice caps are formed. As we shall
perature of the sea water and abundance of various species see later (Chap. 25, Volume 2) the mechanism is actually
of foraminifera fossils and diatoms is close, which helps to more complex, and involves feedback from the ocean and
establish transfer functions to estimate marine paleotem- boreal biosphere in response to changes in insolation, even if
peratures with an accuracy close to 1–2 °C (Chap. 21, this positive feedback does constitute the factor triggering
Volume 1). the start of an ice age.
1 The Climate System: Its Functioning and History 19

However, the role of the cryosphere is not limited to its On a geological scale, the variability of the cryosphere is
impact on the surface albedo. On the oceans, sea ice cuts off even greater. It is seen in the geological traces left by gla-
the ocean from the atmosphere, and blocks the air-sea ciers recording their passage and their broadest expansion
exchanges of water, salt and other chemicals. When it is (moraines, boulders streaked by friction marks as they were
forming, sea ice eliminates salt which makes the sur- transported by the glacier over the surface bedrock on which
rounding sea water denser causing it to sink to depths, thus it rested, vast continental shields like Canada eroded away).
feeding the ocean thermohaline circulation. Conversely, Geological observations have thus led us to believe that
when it melts in summer, the salinity of surface water some 750 million years ago, all continents were covered
decreases abruptly. On land, glaciers are not static. They with glaciers, and that the oceans were probably covered
flow under their own weight at a speed of up to several with perennial sea ice at the same time.
hundred meters per year. When they reach the coast, the ice About 450 million years ago, during the Ordovician, a
breaks into icebergs that are propelled by winds and cur- gigantic ice cap covered the Sahara where even today one
rents, and melt when they arrive in warmer water, causing a can see striated rocks, glacial valleys, remains of moraines
drop in salinity of surface waters of the ocean. Such changes and channels that collected the water from the melting ice in
are likely to bring about important feedbacks, such as summer and brought it to North Africa. Flying over western
stopping deep-water formation and the thermohaline circu- Mauritania, one can recognize the sandy bed of a great river
lation (see Chap. 29, Volume 2). that came into being beside an ice cap and traced out many
A climate change favoring the flow of glaciers could meanders before flowing further north, to the seas bordering
promote the destabilization of ice sheets, such as the one the glaciated African continent. These rivers were covered
covering West Antarctica. Indeed, it rests on a base that by icebergs that melted slowly and released stones they
would normally be covered by the sea, and the ice sheet is in carried. They can be found today in exposed terrains, in
contact with the ocean on all sides. Such a mass of ice is Morocco, Galicia and even in the Armorican massif, south
unstable. Because of this instability, gigantic floating ice of Caen.
platforms, several hundred meters thick, surround the entire About two hundred million years ago, a long, globally
part of the ice sheet in contact with the ocean. warm era commenced (Jurassic and Cretaceous), during
The sheets covering the Ross and Ronne Seas, each with which time glaciers appear to have been rare, if they had not
a surface area similar to that of France, calve huge icebergs, completely disappeared. The wide variety of flora and fauna
tens of kilometers in length, into the Weddell Sea. Without reflects a variety of environmental conditions, from tem-
any rocky terrain, they are in direct contact with the ice cap perate in Japan, Siberia and Australia to very hot in America,
and their speed often reaches several meters per day. What Africa and Eurasia. There are no known tracks of large
helps to safeguard the West Antarctic ice sheet is the pres- glaciers, even on the Antarctic continent, although, of
ence of several islands around it. These anchor points slow course, as the continent is now covered with a thick ice cap it
down the ice flow. Ice therefore progresses slowly until that is only accessible to geological observations at its periphery.
last barrier which, once crossed, leaves the field open to the The glaciations of the Quaternary are the culmination of a
calving of icebergs resulting from the fragmentation of the long process of cooling of the Earth that began more than
ice platform. The future of the West Antarctic ice cap in thirty million years ago, firstly with the development of an
response to ongoing climate change is a real concern for the ice sheet on Antarctica, and then on Greenland. For the last
centuries to come. million years, glacial-interglacial oscillations have domi-
nated the climate of our planet. While today it is the
Southern hemisphere which is the most glaciated with
The Cryosphere in the Past and Paleoclimates Antarctica containing about 28 million km3 of ice for only
1 million km3 of ice in Greenland, it was the Northern
The variability in the cryosphere is considerable, no matter hemisphere that was the most glaciated at the height of the
what time scale is considered. Over the span of a season, last glaciation 20,000 years ago, with 50 million km3 of ice
snow cover and sea ice show the biggest variations. Mea- over Canada (Laurentide Ice Sheet, 4 km thick) and northern
surements taken since the mid-twentieth century show that Europe.
the area of sea ice each year goes from 15 million km2 in Throughout the last ice age, the Laurentide Ice Sheet
early spring to just 6 million km2 at the end of the summer. extended beyond the American continent to reach the
However, since 1980, a clear decrease in this range has been Atlantic continental shelf and the Labrador Sea. This cap
observed and it has reduced to nearly 3 million km2 in 2012 could become unstable, suddenly releasing huge numbers of
summer. A similar trend in the retreat of mountain glaciers is icebergs which invaded the entire North Atlantic where they
observed during the twentieth century, whether in the Alps, melted. The sudden appearance of these armadas of icebergs,
or in the mountains of Africa or South America. designated as ‘Heinrich events’ caused a significant decrease
20 S. Joussaume and J.-C. Duplessy

in the salinity of the water at high latitudes, preventing amounts of dust and gas are ejected into the atmosphere, up
downwelling of dense water in winter, thus stopping the to several tens of kilometers in altitude. The emitted sulfur
thermohaline circulation and the transport of warm water by dioxide combines with the water vapor present in the
the Gulf Stream and North Atlantic Drift. The result was a stratosphere to form micro droplets of sulfuric acid, which is
freezing of Europe and a disruption of the climate over very effective in reflecting solar rays and causes a significant
almost the entire planet. The behavior of the cooling of the surface of the Earth, of around 0.5–1 °C on
ocean-atmosphere-cryosphere interactions during glacial average. In addition, these very small droplets can remain in
climate periods is complex. In addition to the Heinrich the stratosphere for several years before falling into the
events that occurred during the last glaciation at intervals of troposphere where they are eliminated by rain. For example,
eight to ten thousand years, the paleoclimate record inferred the most intense volcanic eruption of the last two centuries,
from the isotopic analysis of Greenland ice indicates a that of the Indonesian volcano Tambora in 1815, projected
sudden warming (>10 °C) between Heinrich events, over about 150 km3 of debris and gases into the atmosphere and
periods of less than a few centuries that end with a slower was followed by two exceptionally cold years. The year
cooling leading to a return to glacial conditions. These are 1816 was even deemed to Canada and New England to have
the Dansgaard-Oeschger events which occur every two to been ‘the year with no summer’. Recently, after the eruption
three thousand years and which are not explained by any one of the Philippine volcano Pinatubo in June 1991, one of the
unanimously accepted theory (instability of the European ice largest of this century, the weather stations recorded a sig-
sheets, internal oscillation in ocean circulation, amplification nificant lowering of temperature for several months.
of a weak solar forcing). The climate of the ice ages appears Although the influence of volcanic eruptions on climate is
to have been much more variable than the climate we have limited mostly to a cooling for a relatively short time, rarely
known for the past 10,000 years, but the conditions for exceeding a few years, geologists think that unusually
stability of the climate system remain a research topic that is intense eruptions, called trapps, could have contributed to
far from fully understood (Chap. 29, Volume 2). large-scale disruptions in climate and mass extinctions
through their emissions. For example, it is estimated that the
fissure eruptions that continued for several hundred thousand
The Lithosphere: Over Large Timescales years to form the gigantic Deccan plateau, could have been
of sufficient magnitude severally increased the carbon
The surface of the Earth, which makes up the lithosphere, dioxide content of the air and to warm it by 3–4 °C until the
also intervenes over long time scales of the order of a million weathering of rocks consumed the excess. Further in the
years or more. Plate tectonics determine the position of the past, 250 million years ago, huge fissure eruptions in Siberia
continents, the relief, the shape of the ocean basins, as well are often cited as one of the factors responsible for global
as the level of carbon dioxide emitted by volcanoes (Chap. 2, warming that marked the end of the Permian (Chap. 27,
Volume 1 and Chap. 22, Volume 2). For example, the Volume 2).
glaciation of the Sahara during the Ordovician is linked with
the movement of the African plate, which had a polar
location at that time. Pioneer studies suggested that the The Climate System
Antarctic ice cap was triggered 34 million years ago, when
the American and Australian continents had sufficiently Atmosphere, oceans, cryosphere, biosphere, and the litho-
drifted from the Antarctic to allow the establishment of the sphere, all contribute to climate changes through a complex
circumpolar current that isolated Antarctica from the set of actions and reactions, both physicochemical and bio-
mid-latitudes. More recently, it has been shown that the long logical (Fig. 1.10). The atmosphere and the ocean, the two
trend decrease of atmospheric CO2 was pivotal in the trig- fluids that transport the excess energy received in the tropics
gering of Antarctica ice sheet. The uplift of the Himalayas, to the poles, are central in climate dynamics, and this is true
through its impact on the weathering and erosion of surface for all periods of time ranging from a few years to millennia.
rocks, has likely contributed to increased consumption of Strongly coupled with the mechanical effect of wind on the
atmospheric carbon dioxide and to the global cooling surface of the sea and through the exchange of heat and
observed since the Eocene. The uplift of Tibetan Plateau as water, together they can cause fluctuations in natural climate
well as the Paratethys shrinkage amplified the Asian mon- phenomena, the scale of which can be glimpsed in El Niño
soon. The major tectonic phenomena therefore have a major for example. Their interactions can be strongly modified by
impact on global climate on a geological scale (Chap. 2 changes in the land surface (vegetation, snow, ice). By their
Volume 1 and Chaps. 22, 26 and 27, Volume 2). impact on the albedo and the water cycle, these changes
The lithosphere also interacts with the atmosphere modulate the exchange of heat and water between the two
through volcanism. During violent volcanic eruptions, large hemispheres, and between the oceans and land. Finally, past
1 The Climate System: Its Functioning and History 21

Fig. 1.10 The climate system

and the exchanges of energy, The climatic system
water and carbon that affect it

Atmosphere CO2, CH4, O3

1 day - 10 years 1 month - 109 years

H2O
H2O
Ice
1 month - 10 years

Ice sheet
103 - 106 years Terrestrial biosphere
1 month - 10 years
1 month - 100 years
Lithosphere Ocean
104 - 109 years
Pole 10 - 1000 years Equator

climates provide many examples of fundamental reorgani- cycles of a few dozen millennia. The recent discovery of
zations of the climate system. large iceberg armadas (Heinrich events), which can tip the
For periods extending beyond a few thousand years, climate into a glacial state in only a few decades shows that
changes in insolation due to slow changes in the Earth’s this is not the case. As soon as the calving of icebergs stops,
orbit around the Sun, the formation of ice sheets, the sinking the climate warms suddenly, a new rapid change that
of bedrock and changes in the composition of air also Neanderthals and Cro-Magnon witnessed. And between two
become drivers of climate change. Drivers which are how- Heinrich events, Dansgaard-Oeschger events are another
ever themselves dependent on other components of the cli- manifestation of quick, abrupt climate change. All these
mate system: the atmosphere, oceans and biosphere, through observations show that the climate system as a whole is
the underlying energy, water and carbon cycles. The for- unstable due to multiple feedbacks that can result from the
mation of ice sheets is in effect controlled by the temperature slightest disturbance.
of the air and by the accumulation or melting of snow. The We are not immune today to a brutal, unexpected change
concentration of carbon dioxide depends on the gas in climate, since human activity has reached a level that
exchanges between the atmosphere and oceans, themselves significantly disrupts the radiation balance of the atmo-
governed by winds and temperature; it also varies with the sphere. The role of paleoclimatology is to document climate
activity of marine phytoplankton, itself strongly controlled variability at all time scales and to help to highlight the
by ocean circulation. As a result, these interconnections can mechanisms that come into play so as to understand the
transform the slow variations of the movement of the Earth resulting changes. As such, paleoclimate information is
around the Sun into spectacular glaciations. Over millions of analyzed in the Intergovernmental Panel on Climate Change
years, plate tectonics become dominant, changing not only assessment reports to outlines our understanding of climate
the geography of the Earth’s surface, but also the composi- change and mechanisms as well as to evaluate climate
tion of the atmosphere. models. For example, the 5th Assessment Report reported
For a long time, scientists believed that climate changes how the current atmospheric concentration of carbon dioxide
of very large amplitude were phenomena that occurred is unprecedented in the past 800,000 years, and how the rate
slowly relative to the scale of human life, governed mainly of sea level rise since the mid-nineteenth century has been
by phenomena occurring over large time scales, such as the larger than the mean rate for the previous two millennia. It
evolution of the Sun measurable over hundreds of millions also informed on high sea levels during the last interglacial
of years, plate tectonics, the effects of which are felt grad- climate and proved the capability of models to reproduce
ually over millions of years, or the astronomical forcing with past warm and cold climates.
22 S. Joussaume and J.-C. Duplessy

Find out more IPCC (Ed.). Climate change (2013): The physical science basis; Contri-
References to studies mentioned in this chapter presenting bution of Working Group I to the fifth assessment report of the
Intergovernmental Panel on Climate Change. Cambridge, United
the climate system will be detailed and reported in the Kingdom and New-York, USA: Cambridge University Press.
various chapters of this book. We indicate below some Johnson, G. C., Mecking, S., Sloyan, B. M., & Wijffels, S. E. (2007).
general literature which readers may want to consult to Recent bottom water warming in the Pacific Ocean. Journal of
deepen their knowledge of the behavior and history of the Climate, 20, 5365–5375.
Joussaume, S. (2000). Climat, d’hier à demain (143 p). Paris: CNRS
climate system. éditions.
Mélières, M. A., & Maréchal, C. (2015). Climate change: Past, present
and future (416 p). Hoboken: Wiley.
References Neelin, D. (2011). Climate change and climate modeling (282 p). New
York: Cambridge University Press.
Philander, G. S. (1989). El Niño, La Niña, and the southern oscillation
Berger, A. (1978). Long-term variations of daily insolation and (293 p). London: Academic Press.
quaternary climatic changes. Journal of the Atmospheric Sciences, Rohling, E. J. (2017). The oceans: A deep history. Princeton: Princeton
35, 2362–2367. University Press.
Berger, A. (1992). Le Climat de la Terre, un passé pour quel avenir Ruddiman, W. F. (1997). Tectonic uplift and climate change. New
(479 p). Bruxelles: De Boeck Université. York and London: Plenum Press.
Broecker, W. S., & Peng, T. H. (1982). Tracers in the sea (690 p). Schneider, S., & Londer, R. (1984). The co-evolution of climate and life
Palisades, N.Y.: Eldigio Press. (563 p). San Francisco: Sierra Club Books.
Crowley, T. J., & North, G. R. (1991). Paleoclimatology (349 p). Steffen, W. Sanderson, R.A., Tyson, P. D., Jäger, J., Matson, P. A.,
Oxford: Oxford University Press. Moore III, B., Oldfield, F., Richardson, K., Schellnhuber, H. J.,
Duplessy, J. C., & Morel, P. (1990). Gros Temps sur la planète (296 p). Turner, B. L., Wasson, R. J. (2005). Global change and the earth
Paris: Éditions Odile Jacob. system, IGBP Series (336 p). Berlin: Springer.
Holland, W. R., Joussaume, S., & David, F. (1999). Modeling the Tomczak, M., & Godfrey, J. S. (2005). Regional oceanography: An
Earth’s climate and its variability. In École des Houches (565 p). introduction. PDF version available at: http://www.cmima.csic.es/
Amsterdam: Elsevier. mirror/mattom/regoc/pdfversion.html.
Hurrell, J. W. (2003). The North Atlantic oscillation: Climatic Wang, B. (2006). The Asian monsoon. Springer-Praxis books in
significance and environmental impact (279 p). Washington, D.C.: Environmental Sciences (779 p).
American Geophysical Union.
 The Changing Face of the Earth Throughout
the Ages 2
Frédéric Fluteau and Pierre Sepulchre

The face of the Earth has changed dramatically over the last we describe the techniques and tools used to retrieve such
4.5 billion years. The growth and emergence of the conti- information, as well as the uncertainties inherent in each
nental crust transformed a largely ocean-covered planet in its method.
early days into a planet with land masses. Under the action
of mantle dynamics, the first continental crusts merged with
island arcs to constitute the first continents at the end of the Continental Drift
Archean and the Paleoproterozoic (see Fig. 2.1 for a chart of
geological time). These continents gathered to form the first In 1915, the German meteorologist, Alfred Wegener, laid the
documented supercontinent (Bleeker 2003) around 1.5 Ga, foundations for continental drift in a book entitled Die
before breaking up and then coalescing again. Each phase Entstehung der Kontinente und Ozeane (The Origin of
has profoundly changed the Earth’s surface. Vast mountain Continents and Oceans). This theory is based on several
ranges got uplifted and ocean basins formed, then disap- arguments, in particular on the complementarity of the
peared, through erosion or later geological events for the continents bordering the Atlantic Ocean and the continuity
former, by posterious closure linked to tectonics for the of the terrain and their deformations. But Wegener was not
latter. The climatic upheavals that have marked the Earth’s the first to make these observations. A Dutch cartographer,
history are strongly linked to these paleogeographic events Abraham Ortelius, observed this complementarity of the Old
through direct or indirect coupling between the different and New Worlds in his atlas Theatrum Orbis Terrarum
solid, liquid and gaseous envelopes that are depicted in published in 1570 and questioned the cause of this rupture
volume 2. Here we provide an overview of the major geo- and this expansion towards the west in his book Thesaurus
logical stages of the early Earth before detailing the paleo- Geographicus published in 1596 (Romm 1994). Two hun-
geographic changes of the modern Earth in terms of dreds and fifty years later, the French geographer, Antonio
continental distribution and paleotopography. Snider-Pellegrini, published “The Creation and its Unveiled
Mysteries” (1858) in which he drew the position of the
continents after the closure of the Atlantic Ocean. However,
Paleogeographic Reconstructions the theories of Wegener were not limited to the comple-
mentarity of the continents bordering the Atlantic Ocean and
Paleogeographic reconstruction requires to quantify the geological continuity, he also postulated on the basis of the
changes in location of the continents, as well as their paleontological continuity of flora (Glossopteris) and fauna
coastlines and topography, through the geological ages. Here (Mesosaurus, Lystrosaurus, Cynognathus), and paleocli-
matic continuity. Glacial sediments from the Carboniferous
era were discovered at the end of the nineteenth century in
Africa, India and Australia. Wegener showed that by closing
F. Fluteau (&) the Atlantic and Indian oceans, all these outcrops would
Université de Paris, Institut de physique du globe de Paris, CNRS, form a coherent cluster close to what was the South Pole at
75005 Paris, France the time. All these arguments supported the idea of a
e-mail: [email protected]
supercontinent, the Pangea. This revolutionary theory of the
P. Sepulchre continental drift was strongly rejected by the Earth sciences
Laboratoire des Sciences du Climat et de l’Environnement,
LSCE/IPSL, CEA-CNRS-UVSQ, Université Paris-Saclay, 91191 community who preferred to believe in their fixity, and who
Gif-sur-Yvette, France criticized Wegener for the absence of mechanisms

© Springer Nature Switzerland AG 2021 23

G. Ramstein et al. (eds.), Paleoclimatology, Frontiers in Earth Sciences,
https://doi.org/10.1007/978-3-030-24982-3_2
24 F. Fluteau and P. Sepulchre

Fig. 2.1 Divisions of the geological time approved by the U.S. Geological Survey Geologic Names Committee (2018)
2 The Changing Face of the Earth Throughout the Ages 25

demonstrating his theory. The paleontologists of the early magnetic field can also be recorded in sedimentary rocks if
twentieth century clung to the concept of fixed geography, they contain magnetic particles (Chap. 7).
and invented the theory of “continental bridges,” land ties Sea floor spreading is the cornerstone of plate tectonics,
that would have linked the continents, to explain the but to complete this theory, temporal constraints needed to
migration of fauna and flora. Nevertheless, some geologists be integrated, enabling the understanding of the pace at
believed in continental drift and supported it, such as which ocean ridges opened. During this same period, the
Alexander Du Toit and Emile Argand. A decisive argument geophysicists Cox and Doell, and the geochemist Dalrymple
in favor of Wegener’s theory was established by Arthur established the first timetable of magnetic reversals for the
Holmes. He was the first to establish the basics of thermal last 4 million years (Chap. 7), using a new technique of
convection of the Earth’s mantle (Holmes 1929). But here isotopic dating with the potassium and argon elements (Cox
again, this argument did not receive the expected positive et al. 1964) (Chap. 5). By comparing the magnetization
response from the Earth science community, and it was not polarity and the age of basalt samples taken at sea, it is clear
until the 1960s that Holmes was seen as a pioneer of plate that the age of the oceanic crust increases with distance from
tectonics. the line of the ridge. From the 1960s onwards, the DSDP
From the 1940s, exploration of the ocean floor would (Deep-Sea Drilling Project) and ODP (Ocean Drilling Pro-
revolutionize Earth sciences. The mapping of the morphol- gram) scientific missions were launched with the aim of
ogy of the ocean floor, and the discovery of the Atlantic drilling for sediment deposits on this oceanic crust. These
Ridge under the initiative of Bruce Heezen and Marie Tharp sedimentary cores allow the sequences of magnetic polarities
(Heezen 1962; Heezen and Tharp 1965), led Harry Hess1to fixed by period to be established, thanks to the fossil content,
propose a theory of the expansion of the ocean floor (Hess and thus, to also date the marine magnetic anomalies. The
1962). Half a century after Wegener, Earth sciences were speed of opening of the ocean ridges can then be determined.
experiencing a revolution, following the discovery of mag- In 1968, Heirtzler and his group quantified the speed of
netic anomalies (Chap. 7). An American oceanographic opening of the South Atlantic Ocean by analyzing a marine
campaign mapped the magnetic field around the Juan de magnetic sequence dating back to the Pliocene (3.35 Ma). In
Fuca Ridge, off the North American coast, in the eastern the following years, the kinematic parameters of all the
Pacific Ocean. By subtracting the ambient magnetic field oceans were determined one after another, and the evolution
from the magnetic data, alternately positive and negative of the different ocean basins could then be traced. These
anomalies were found (Raff and Mason 1961). Based on parameters reflect the movements of one lithospheric plate
these observations, both Morley, and Vine and Matthews relative to another one that is arbitrarily fixed. These
(Vine and Matthews 1963), proposed the theory of the movements being defined on a quasi-spherical surface, they
renewal and expansion of the ocean floor in 1963. can be expressed by an angle of rotation about an axis
These marine magnetic anomalies result from the acqui- passing through the center of the Earth and defined by the
sition of a thermoremanent magnetization by the iron and longitude and latitude of its pole.
titanium oxide particles in the oceanic crust subjected to the Moving on from the oceans to the land, in the early
Earth’s magnetic field, after the oceanic crust (basaltic lava), 1950s, some scientists studied the natural remanent magne-
emitted at the ocean ridges at a temperature of about 1100 ° tization of rocks. At all points of the globe, the magnetic
C, begins to cool. When basalt reaches the Curie point (the field is defined by a vector collinear to the field lines. The
magnetic particle-dependent temperature, about 570 °C for magnetic field of the Earth originates from convective
the ferromagnetic magnetite crystals contained in the basalt), movements within the outer core, which consists of liquid
the direction of the field is fossilized by the magnetic carriers iron (about 90 wt%), nickel (about 4 wt%), along with some
of the oceanic crust. Above the Curie point, the material is lighter elements, such as silicon, sulfur and oxygen. The
paramagnetic, and each magnetic carrier behaves like a small movements within the conductive core induce electric cur-
compass that follows the direction of the magnetic field rents which, in turn, generate a magnetic field. The Earth’s
without storing it. As soon as the temperature of the rock magnetic field functions like a self-excited dynamo. To
passes below the Curie temperature, the magnetic carriers compensate for the energy losses associated with the elec-
aligned along the lines of force of the Earth’s magnetic field trical resistivity within the Earth’s core, energy input
are permanently frozen in this direction. The Earth’s obtained from the conversion of ohmic dissipation into heat,
from the gravitational energy and from the release of latent
heat during the crystallization of the inner core ensures the
thermal equilibrium of the milieu and permits the function-
ing of the geodynamo, as well as its continuity over geo-
1
When B. Heezen presented their findings to Princeton in 1957, Harry
Hess stood up and said: “Young man, you have shaken the foundations logical time.
of geology!” (Yount 2009).
26 F. Fluteau and P. Sepulchre

It is therefore important to check the time at which the (VGP), since the reference is a continental land mass whose
geodynamo was set up. To do this, it is necessary to find very past position is not known.
old rocks that would have registered and retained a primary The successive positions of the VGP over time track the
magnetization. Samples taken from the Matachewan Dikes path of the apparent drift of the magnetic poles, with the
dated at 2.5 Ga or basalts from the Fortescue Basin of the continent being studied fixed at its current position. The first
Pilbara Craton (Australia) dated at 2.7 Ga yielded a primary apparent polar wander path (APWP) over a period of
remanent magnetization which suggests the presence of a 600 Ma was established in 1954 by Creer and his research
dipolar magnetic field at the end of the Archean (Tarduno et al. group, based on magnetic measurements of samples taken
2014). Knowing this, it is technically possible to measure the from geological formations in Great Britain (Creer et al.
direction of magnetization from this time onwards. But what 1954). The virtual magnetic poles constituting the APWP
about rocks older than this? Archean formations are highly approximate, supposing a geocentric axial dipole, the suc-
likely to have undergone a complex geological history, and, in cessive paleopositions of the rotation axis with respect to the
particular, one or more episodes of metamorphism erasing the continent from which the paleomagnetic pole was deter-
primary magnetic signal in favor of a more recent secondary mined (Fig. 2.2). In the 1950s and 1960s, the existence of a
magnetization. To overcome this problem, a technique based central axial dipole was challenged, calling into question the
on the measurement of the magnetization carried by an isolated importance of the APWP. In 1964, however, Irving was able
mineral (single feldspar, quartz phenocrysts) was developed to to verify by means of climate indicators the hypothesis of a
show the presence of a magnetic field as far back as 3.5 Ga. central axial dipole. Indeed, the climate of the Earth is, first
Finally, zircons dated between 3.3 and 4.2 Ga discovered in and foremost, a function of insolation. The result is that the
the Jack Hills metaconglomerate showed a magnetic signal distribution of climate indicators (e.g. evaporites for sub-
carried by magnetite and considered to be primary. The pale- tropical zones, coral reefs for tropical areas, glacial sedi-
ointensity of the magnetic field (not its direction) was mea- ments at high latitudes), tends to be symmetrical on either
sured and it varies between 12 and 100% of the value of the side of the equator.
current field at the equator possibly suggesting the presence of To check if the distribution of the paleoclimate indicators
a terrestrial magnetic field as far back as the Hadean era discovered within geological formations changed over time,
(Tarduno et al. 2015). Irving replaced the continents (and the sites of the paleo-
The geomagnetic field H at any point on the surface of climatic indicators) in their paleopositions, using the virtual
the globe can be defined by two angles, the declination D magnetic poles, and applied the hypothesis of a geocentric
(angle between the magnetic north and the geographic north
(counted positive east of true north)) and the inclination I
(the angle between the horizontal plane and the direction of
the fieldH (counted positive if downward)). At the first
order, the present-day magnetic field can be represented by
the field produced by a magnetic dipole inclined to the
Earth’s axis of rotation by 11.5° and slightly off-centered by
about 500 km from the center of the Earth. Differences
between the current magnetic field and this theoretical field
exist locally, as evidenced by the contribution of non-dipolar
terms (quadrupole, octupole). In the even stronger hypoth-
esis of a perfectly dipolar, axial and centered magnetic field,
a simple mathematical relation connects the magnetic incli-
nation measured at a point of the Earth’s surface to the
latitude of this point. In other words, knowing the inclination
of the magnetization vector fossilized by the magnetic
minerals at a site at different times in the past allows us to
calculate the successive paleolatitudes of this site by
assuming that the geomagnetic pole has always coincided
with the geographical pole, itself supposed fixed. Con-
versely, from a set of samples distributed over a continental
land mass, we can calculate the position of the magnetic pole
Fig. 2.2 The apparent polar wander of the magnetic poles of India
associated with the continent or geological mass under (according to data from Besse and Courtillot 1991). The age of the
consideration, always assuming the axial and centered dipole magnetic pole (in Ma) is indicated. The ellipses represent the
magnetic field: this is therefore a “virtual” geomagnetic pole uncertainty (95%) on the magnetic pole position at a given age
2 The Changing Face of the Earth Throughout the Ages 27

axial dipole. Thus reconstructed, the latitudinal distribution rotation in response to the heterogeneities of masses in the
of the paleoclimate indicators is statistically identical to their mantle, modifying the tensor inertia of our planet (the
present distribution, thus supporting the hypothesis of the maximum axis of inertia is aligned with the axis of rotation
axial centered dipole. Other studies carried out in the 1960s of the Earth). In the paleomagnetic reference system, the
made it possible to validate this hypothesis, first for the APWP is thus caused by the movement of the plate due to
Plio-Pleistocene and then gradually as far back as the Pre- plate tectonics and to the overall movement of the
cambrian. In 1968, Le Pichon proposed a model of move- continents.
ment of rigid land masses relative to each other, thereby In 1972, Morgan proposed using hot spots associated
reconstructing seafloor spreading for all the Cenozoic (Le with convective plumes from the D″ transition zone as a
Pichon 1968). definitive frame of reference. Deep mantle convective
The trajectories of the APWP make it possible to place plumes are assumed to be fixed because of the sufficiently
the continents in their original position with respect to the slow movements in the lower mantle. The movements of
axis of rotation of the Earth, in other words, in terms of some lithospheric plates can thus be positioned within the
paleolatitude and orientation. However, paleomagnetism is “hot spots” reference frame. Unlike magnetic poles, paleo-
not enough, since it does not specify the paleolongitude of latitude and paleolongitude are constrained (by a frame of
the continents because of the spherical symmetry of the reference associated with the mantle), but this method can
magnetic field (assuming an axial and centered dipole only be used for the last 130 million years. Before this time,
magnetic field). Kinematic parameters of the oceans are used the traces of hot spots on the ocean floor cannot be traced,
to pin down the position of a continent relative to another. erased by the subduction zones, while the few remaining
However, this method can only be used when the ocean traces are not constraining enough.
separating two continents is bordered by passive margins The differences in latitude and in rotation of the continents
and especially when these kinematic parameters are known between the paleomagnetic and hotspot reference frames
(available for the last 170 million years only). If one of these make it possible to isolate the movement of the global drift of
oceanic margins is active, the positioning of one plate rela- the crust-mantle couple with respect to the axis of rotation or
tive to another is nevertheless possible by traveling across the true polar wander from the perspective of a continent. The
one or more other intermediate continents separated by amplitude of the global drift generally does not exceed 1° per
passive margins. Nevertheless, there are cases where the million years over the last 130 million years (Besse and
deformation of the continental land masses by tectonics (in a Courtillot 2002). Nevertheless, there are some rapid events
collision zone, for example), the presence of active margins, (occurring over a few million years) during which a global
or the age of reconstruction prevent the use of this method drift of about 10° of all the continental masses has been
combining paleomagnetism and ocean kinematics. In these observed, for example in the Paleocene (Moreau et al. 2007).
cases, paleogeographic reconstructions depend on paleo- Even with a lack of data in the hot spot reference, it is pos-
magnetism and the APWP. sible to isolate the true polar wander by determining common
These trajectories of the apparent polar wander are not all drifts in APWP of different continents. Events of great
of equivalent quality. To overcome this disadvantage, Besse amplitude have also been suggested, for example a 90°
and Courtillot (1991, 2002) and Torsvik et al. (2012) pro- movement between the Lower and Middle Cambrian
posed calculating artificial trajectories of APWP where the (Kirschvink et al. 1997) or during the Ediacaran (Robert et al.
poles of all the plates of a certain age are integrated into a 2017). These events result in a continental drift rate of the
single referential (i.e. transferred from one lithospheric plate order of 10° per Ma, therefore much higher than the maxi-
to another using the kinematic parameters of the oceans). mum continental drift rate caused by plate tectonics of about
This method is applicable to the last 320 million years 2° per Ma (Seton et al. 2012) but perfectly compatible with
(despite the absence of kinematic data for the oceans beyond the theoretical maximum speed of the true polar wander of
170 Ma since all the continents were then combined into a about 10° per Ma, taking into account a viscoelastic Earth
supercontinent, the Pangea). Before 320 million years, for (Greff-Lefftz and Besse 2014; Robert et al. 2017).
all continents and regardless of the period for those land
masses surrounded by active margins, reconstructions rely
solely on the trajectories of APWP, although geological The Paleomagnetic Tool, Tests
and/or paleontology arguments may provide constraints on and Uncertainties
the relative position of the masses. Of course, there is an
overall increase in uncertainties with age. To understand the difficulty of obtaining robust paleogeo-
The now well-known movements of large lithospheric graphic reconstructions, it is important to look at the paleo-
masses are part of the classical theory of plate tectonics. magnetic tool used (Chap. 7). The direction of the magnetic
However, the whole Earth can also tilt relative to the axis of field is measured in a laboratory using a magnetometer, an
28 F. Fluteau and P. Sepulchre

instrument capable of detecting natural remanent magneti- correspond to the average direction of the dipole. A second
zations of less than 10−12 A/m, on cylindrical samples of test consists of checking the reliability of the magnetic pole
approximately 10 cm3 taken from sedimentary or volcanic determined from sampling sites for which the dip in the
formations and as perfectly oriented in space as possible. This geological layers differs. Indeed, the geological formations
orientation in space of the sample collected in the field is can be subjected to deformations. To calculate the virtual
essential in order to determine the direction of the magneti- magnetic pole, the paleo-horizontality of the sampled sedi-
zation vector (represented by inclination and declination) and mentary formation that prevailed at the moment of the
then to calculate the position (longitude and latitude) of the acquisition of the magnetic signal must be restored. Taking
virtual magnetic pole associated with it. To obtain a reliable into account the tectonic (tilt) correction for each sampling
virtual magnetic pole, several criteria must be met. First of site must have the effect of clustering the set of magnetic
all, one needs to have a sufficient number of samples. The directions obtained for each site. If this structural correction
number of samples collected is usually eight per site, and the does not have the desired effect on the data, this means that
number of sampling sites within a geological formation is at the magnetization was acquired during or after deformation.
least six sites. As in the processing of the signal, increasing However, these tests are not sufficient to certify the
the number of sites and samples per site improves the robustness of the data that can be affected by various biases.
signal-to-noise ratio (thus reducing the sources of errors). The Regional tectonics in areas of active collisions such as in the
uncertainty on the virtual magnetic pole is of the order of a Alps or by rifting in Afar could lead to horizontal block
few degrees (1° = 111 km) in the best case, which still rep- rotation that is well recorded by paleomagnetism. These
resents a few hundred kilometers. The displacement of con- regional tectonics can mask the large drifts of plates or larger
tinent of less than several hundred kilometers is therefore masses. A second bias is related to the geometry of the
difficult to detect, which, taking into account the average magnetic field. The calculation of a virtual magnetic pole is
speed of plate drift (*4 cm/year), represents a time span of based on the assumption of a geocentric axial dipole. In the
about ten million years. The dating of the sampled geological case of samples collected from sedimentary formations, the
formations must also meet a standard of robustness in order magnetic measurements are carried out on cylinders of
to produce a reliable virtual magnetic pole. The primary approximately 10 cm3. Given the low accumulation rate of
character of the magnetization of the ferromagnetic carriers sediments and the effects of diagenesis (such as compaction),
must be verified. The magnetization signal must be syn- a cylinder of 10 cm3 may represent a time period of several
chronous or recorded in the millions of years (<5 Ma) after thousand years, and the measured magnetic direction is in
diagenesis (in the case of sedimentary formations). This is to some sense an “average” direction of this time period. For a
ensure that the magnetization age is essentially the same as sampling site, the magnetic direction is the average of the
the “stratigraphic” age of the formation. When these two ages magnetic directions of the samples taken from a unit of a
differ, a subsequent re-magnetization has occurred (in sedimentary formation. It is therefore possible that the time
response, for example, to the burial at several kilometers interval associated with the mean magnetic direction repre-
deep, which has the effect of imprinting a new magnetic sents several thousand years, or even more. The variations of
signal in the sample masking partially or wholly the primary the magnetic field are thus smoothed, and the magnetic
magnetization). direction measured is indeed that produced by a geocentric
Tests have been developed to evaluate the quality of the axial dipole field in the study area under consideration.
measured primary natural remnant magnetization (NRM). In the case of volcanic series, the thermoremanent magne-
The first is the reversal test. On average, the magnetic field tization (TRM) is acquired during the cooling of the lava flow.
experiences several reversals per million years (there are, It thus represents a (quasi-) instantaneous photography of the
nevertheless, periods of several tens of millions of years Earth’s magnetic field. To overcome the effect of the secular
during which no reversal is observed, Chap. 7). Over a time variation, the average direction of a large number of sites must
period of less than a few million years (<5 Ma), the calcu- be measured, so as to tend towards the direction which would
lated virtual magnetic pole must be statistically identical, be obtained with an axial geocentric dipole. However, the
regardless of the polarity of the magnetic field. This test presence of a persistent quadrupole term in the paleomagnetic
increases the confidence in the primary character of the data has been identified as a possible source of errors. The
NRM and ensures that the secular variation of the magnetic presence of quadrupole terms of about a few percent (the level
field resulting from the movements of molten iron in the generally observed) implies an error of a few degrees in latitude
outer core has indeed been averaged. The secular variation at the equator. This does not therefore significantly affect the
of the magnetic field gives rise to a rapid drift of the mag- virtual magnetic pole which is calculated assuming the axial
netic pole which can move the virtual pole by more than 10° geocentric dipole hypothesis. There is also a possibility that the
from its mean position. Thus, the direction provided by a magnetic field could include an octupole component affecting
single lava flow cooled in less than one year does not Asia during the Tertiary and Pangea during the Permian. If we
2 The Changing Face of the Earth Throughout the Ages 29

consider an octupole term equal to 10% (resp. 20%) of the Topography can also be controlled by dynamic pro-
dipolar component, the maximum error committed is *7° cesses2 related to movement of matter and heat transfer
(resp. 13°) in the mid-latitudes. Finally, a last source of error is within the viscous mantle. Upward or downward movements
related to the preservation of the magnetic signal in the sedi- of mantle matter or dipping of a lithospheric plate into the
mentary series. Indeed, during diagenesis, the magnetic carriers mantle create mass anomalies that can induce long-wave
can undergo a decrease in the inclination of the magnetic car- crustal deformations. This is known as dynamic topography
riers. The inclination measured is therefore less than the actual (Husson 2006) as opposed to isostatic topography. The uplift
inclination of the magnetic field that prevailed at the time this of the southern African plateau and the Colorado plateau
detrital remanent magnetization was acquired. This inclination during the Cenozoic is explained by movements of mass
flattening can be a source of error in paleoreconstructions anomalies in the mantle. Dynamic topography also helps to
(Cogné et al. 2013). explain the history of the Western Interior Seaway that
connected the Arctic Ocean to the Gulf of Mexico across the
North American continent during the Cretaceous, but also
The Topography of the Earth the flooding followed by the exondation of part of Australia
at the same time, or the flooding of the Sunda shelf in
Mountain ranges and high plateaus play an important role in Indonesia during the Pleistocene (Sarr et al. 2019).
atmospheric circulation. The uplift of the Himalayas and the Restoring the past topography of the Earth is undoubtedly
Tibetan plateau is undoubtedly the most characteristic the most difficult part of paleogeographic reconstructions. It
example of the relationship between topography and the involves determining the spatial expanse of the terrain, its
evolution of atmospheric circulation and climate. The hyp- age and altitude. The reliefs bring about deformations and/or
sometric curve of the Earth reveals that today, reliefs higher structural and petrological markers which, when fixed in
than 2 km represent about 10% of the land area. Still, this time, are used by geologists to constrain the tectonic event.
curve reflects the orogenic context of the Earth that has These markers are not always easy to detect because more
prevailed since the beginning of the Cenozoic only, and one recent events often mask earlier events. A phase of conti-
expects that this curve evolved through time. nental accretion caused the India-Asia collision in Southeast
The great areas of high-altitude terrain (*>2 km) typi- Asia during the Triassic period, but this orogeny remains
cally developed during the formation phases of a super- uncertain, partly because this collision erased part of the
continent (for example, in the Carboniferous, during the previous geological history. Determining the paleoaltitude of
formation of the supercontinent Pangea), while periods of mountains is therefore crucial in order to model
break-up of the supercontinents (the Mesozoic for example) pre-Quaternary paleoclimates. Several methods (flora, sedi-
are characterized by more modest reliefs (although low ment, oxygen isotopes, cosmogenic isotopes) have been
reliefs of less than 2 km in general, can form in case of developed, but only the most commonly used will be
continental rifting). In the oceans, seafloor reliefs are mostly depicted here. It is very important to differentiate between
dominated by ocean ridges and some high oceanic plateaus absolute methods, which makes it possible to estimate the
rising above the abyssal plains. paleoaltitude of a relief at a given time and methods esti-
The altitude of the continents, plains, collision mountain mating the vertical velocity of the rocks in these mountains.
ranges (Himalayas, Alps) and the zones of the East African The relative methods reflect a balance between the vertical
Dome result first of all from the (quasi-) isostatic equilibrium movements linked to a geodynamic event (a collision for
of the lithospheric column marked by density heterogeneities example) and those due to erosion which denudes the sur-
on the underlying asthenosphere. The bathymetry of the face, favoring the rise of deep-set rocks by isostatic
ocean floor is controlled by the cooling and by the pro- readjustment.
gressive thickening of the oceanic lithosphere formed at the To measure an altitude or to estimate a paleoaltitude, a
dorsal ridges in isostatic equilibrium on the asthenosphere. reference level is essential. The surface of the oceans pro-
However this scheme does not apply everywhere, sug- vides this reference, but this fluctuates over geological time.
gesting other mechanisms. Some reliefs are not at isostatic However, the amplitudes in these variations are less than the
equilibrium, as is the case, for example, of volcanic islands uncertainty obtained on paleoelevations, regardless of the
that develop on an oceanic crust (like Hawaii) or continental methods used. Nonetheless, eustatic variations are crucial
crusts previously subjected to the weight of an ice cap (like when examining the regions between the continental shelf (>
the Scandinavian region). The charge brings about a flexure −200 m) and the vast lowland plains (<200 m).
whose wavelength is related to the elastic rheology of the
crust. After the rapid melting of an ice cap, the deformation
gradually fades over time with a rate depending on the 2
Although dynamic, collision mountain ranges and intracontinental
viscosity of the underlying mantle. rifting are essentially isostatic equilibrium processes.
30 F. Fluteau and P. Sepulchre

The Evidence from Flora site located at sea level must be known. The uncertainty on
paleoaltitude is only slightly lower than for the preceding
The assemblage of plant fossils has been used as a marker method.
for paleoelevation for almost half a century. This method is The plant-based methods are also subject to uncertainties
based on the relationship between the vegetation type and because of the impact of vegetation on the climate that are
the average temperature at a given location. It is broken ignored. The vegetation cover affects the radiative balance of
down into several variants. The first method relies on finding the Earth through its albedo and the water balance through
an assemblage of modern plant taxa equivalent to that of the evapotranspiration (e.g. Otto-Bliesner and Upchurch 1997),
fossil site (or at least sharing the highest possible number of thereby could introduce a bias and cause paleo-altitudes to
taxa with it), but it assumes that there was no adaptation by be mis-estimated.
these taxa to climatic variations in the past (Su et al. 2019).
This method is particularly well adapted to the last 10 mil-
lion years. A second method is based on the physiognomy of The Evidence from Erosion Sediment
the leaves of plants (size, shape, thickness, type of leaf
margin), synonymous with the adaptation of the plant to a By restoring the mass of deposited sediments in a basin to
given climatic context. A relationship between the phys- the original relief, it is possible to calculate paleo-altitudes.
iognomy of current plant leaves and the mean annual tem- This presumes that the geometry of the sediment source is
perature has been established. This method assumes, known and that this has not changed over time, that there are
however, that the response of the leaf physiognomy to the no sediment losses due to subduction or to incorporation
climate has been constant over time. This tool is calibrated during more recent orogenic events and finally it assumes
on dicots and does not take into account all biogeographic that the relief has always been in isostatic equilibrium. These
provinces (e.g., Australia) or all taxa (such as conifers). considerations considerably limit the use of this method to a
Moreover, it is only applicable as far back as the Upper few endorheic basins that do not have subduction zones
Cretaceous, since flowering plants only appeared during the (intracontinental), such as the Tien Shan range in Asia
Cretaceous. However, the main advantage of this approach during the Miocene. Its application to older orogeny is
is that it avoids any systematic recognition of taxa because complex.
only the morphological characteristics count (which implies
optimal fossilization conditions). To determine altitude, it is
necessary to know the average temperature at sea level of a Stable Isotope Paleoaltimetry
site of the same age. The difference in temperature between
these two sites divided by the vertical gradient of the tem- Water oxygen is made of different stable isotopes (16O, 17O,
18
perature indicates the paleo-altitude. However, the temper- O, e.g. Chaps. 11, 14, 15, 16, 21). The ratio of heavy to
ature lapse rate varies from 4 to 10 °C/km. This depends on light isotopes, when compared to a global reference value
the latitude, the humidity of the air mass, the continentality (namely the Vienna standard mean ocean water, VSMOW),
of the site and the topography itself. The choice of this is noted d18O. It can be measured in various surface waters
parameter is therefore decisive. Uncertainty remains high (ice, ocean, lakes, rivers), carbonates from pedogenic or
(Peppe et al. 2010), around 700–1000 m for paleo-altitudes lacustrine sediments, or biogenic archives. At the global
estimated at 3–4 km. scale, is has been observed that d18O in rainfall decreases
To overcome the problem of the vertical gradient, a from low to high latitudes, and from coastal to inland areas.
method based on the preservation of moist static energy h in On continents, d18O measured directly in precipitation or in
the atmosphere has been developed (Forest et al. 1995). The rivers along different elevation transects also has been shown
parameter h is the sum of a thermodynamic parameter, the to scale with altitude, with d18O decreasing as elevation
enthalpy H of humidity and the potential energy gZ and has increases. These observations have been explained by the
the advantage of a distribution that is relatively zonal, Rayleigh-type distillation process that occurs theoretically in
especially in the mid-latitudes of the northern hemisphere. a cooling air parcel ascending along a mountain range: As
This method requires an atmospheric circulation where air rises and cools, water vapor condenses then precipitates.
horizontal movements predominate over vertical move- During these steps, heavy isotopes are more favorably
ments, to ensure the conservation of moist static energy. The removed from the air parcel, progressively depleting water
humid enthalpy H was calibrated on current leaf indices vapour along the way. Ultimately, rainfall is more and more
(similar to the mean temperature in the previous method). To depleted in heavy isotopes with elevation, i.e. d18O decrea-
determine the paleo-altitude of a site, its humid enthalpy ses (Fig. 2.3). This theoretical framework, together with
H and the humid enthalpy H0 of a contemporary reference regional measurements of d18O have led to determine
2 The Changing Face of the Earth Throughout the Ages 31

meteoric waters. The d18O of the pedogenic carbonates is

more a reflection of the averaged d18O of the drainage basin
than of the meteoric waters. For lake carbonates, d18O is
affected by evaporation on the surface of the lake.
Depending on the data location, these biases can add up to a
significant uncertainty, close to that observed with the
plant-based methods (see Mulch 2016 for a review).
As for the estimates of ocean paleo-temperatures
(Chap. 21), the clumped isotope (D47) measurement (Eiler
2007; Bonifacie et al. 2017) appears to be the next promising
tool in paleoaltimetry. The propensity to form molecules
containing bonds between heavy isotopes of carbon (13C)
and of oxygen (18O) in a carbonate (mainly in the form of
13 18 16
Fig. 2.3 Evolution of the isotopic ratio of oxygen (d18O) as a function C O O) is more probable, the lower the temperature is at
of altitude the formation of the pedogenic carbonate (Chap. 21). There
is therefore a relationship between this quantity and the
“isotopic lapse rates”, i.e. expected change in d18O as a formation temperature, which makes it possible to work
function of elevation (Mulch 2016). These lapse rates back to paleo-altitudes by establishing the vertical temper-
depend on many parameters, including initial relative ature lapse rate (which still may have varied in the past)
humidity and temperature of the air parcel. (Quade et al. 2007). Moreover, the measurements indepen-
The relation between d18O and elevation has been used to dent of the formation temperature through D47 and of the
estimate paleoelevations since the late 90s (e.g. Chamberlain d18O isotopic ratio of these carbonates, make it possible to
1999; Quade et al. 2007; Rowley and Garzione 2007; Mulch trace back to the d18O of the water and thus to isolate the
2016). Oxygen isotopic ratios can be measured in lacustrine effect of altitude from parasitic influences, such as climate,
or soil carbonates, and used to infer the paleo-rainfall d18O season, or latitude. Burial does not appear to affect the
using a relationship involving the temperature of carbonate measurement of D47 in general, an advantage in the case of
formation (Kim and O’Neil 1997) and assuming thermody- orogeny. This method makes it possible to obtain
namical equilibrium between the water and calcite. Recon- paleo-altitudes with a lower uncertainty than other methods.
structing rainfall d18O of dated records thus allowed to infer This isotopic tool was used to constrain the evolution of
past elevations and numerous studies have been carried out paleo-altitudes and the speed of uplift of the Altiplano, a
to reconstruct uplift history of mountain ranges worldwide, high plateau in the Andes in South America, during the
including the Andes, the north American cordillera and the Miocene (Garzione et al. 2014).
Tibetan Plateau.
Still, several processes can cause bias in the results. The
isotopic fractionation depends not only on temperature but Eustatic Variations and Ocean Gateways
also on physical processes that alter the simple model of air
parcel ascent and depletion. Large-scale climate conditions Ocean circulation plays an important role in the climate
(e.g. greenhouse climate) and dynamics can modify moisture system by transporting heat from low to high latitudes,
advection and can also lead to mixing of air masses with particularly through surface currents. However, this transport
different origins and isotopic signatures. Moreover, changes depends on the distribution of the continents and the shape
in atmospheric dynamics related to the peculiar conditions of of the ocean basins. At present, deep waters are formed
the warm climates of the Paleogene have been shown to alter mainly in the North Atlantic Ocean, specifically in the
the isotopic lapse rates. Such potential biases have been Norwegian Sea and the Labrador Sea. Therefore, the con-
highlighted for the reconstruction of the altitudes of the figuration of deep ocean circulation must have been different
Tibetan plateau (e.g. Botsyun et al. 2016, 2019; Li and in the past. At the end of the Cretaceous, the North Atlantic
Garzione 2017) and the Andes (Poulsen et al. 2010), with Ocean was not yet open and exchanges between the central
either extensive datasets of river d18O measurements or the Atlantic Ocean and the Arctic Ocean were very limited.
use of isotope-enabled general circulation models. Lastly, Deep water formation could therefore be transferred to the
the isotopic anomaly recorded in pedogenic carbonates is not Pacific Ocean, based on numerical models simulating the
from rainwater, but from runoff waters. In the case of the large-scale circulation of the ocean and atmosphere (e.g.
Himalayas, for example, measurements of d18O carried out Otto-Bliesner et al. 2002).
on various modern rivers can differ significantly from
32 F. Fluteau and P. Sepulchre

Changes in geographical configuration through conti- production are considered to be less variable. The determi-
nental drift, or more radically, through the opening or nation of sedimentary facies from core samples taken by
closing of ocean basins, are therefore likely to cause pro- drilling and the location of seismic reflectors on the conti-
found changes in ocean circulation. Either shallow or deep, nental margins and adjacent basins make it possible to
ocean gateways play a crucial role in water masses circula- reconstruct the arrangement of sedimentary deposits and to
tion and associated heat and salt fluxes (Ferreira et al. 2018). infer eustatic variations. Several curves have since been
During the Cenozoic, the position of the continents has not produced (Haq et al. 1987; Haq and Schutter 2008). The
drastically changed but some ocean gateways have been curves obtained (Fig. 2.4) show nested eustatic cycles. The
opened or closed thereby impacting ocean circulation. The largest eustatic variations, about 200 m, with a timescale of
Tasman seaway opened during the Late Eocene, the Drake tens of million years, are related to life cycle of supercon-
passage during the Late Eocene/Oligocene, the Fram Strait tinents (Wilson cycle), plate reorganization, dynamic
during the Oligocene. Conversely the East Tethys seaway topography and crustal production variations. Secondly,
closed during the Early/Middle Miocene, the Central significant variations (several tens of meters) but with a
American Seaway during the Late Miocene, the Indonesian lower timescale (from *2 to *10 Ma) are likely due to
throughflow during the Late Miocene/Pliocene. The timeline regional tectonic forcing. Thirdly, the sea level fluctuations
of these opening/closures are therefore crucial for paleocli- on the timescale of 0.5 to *2 Ma are not well understood, it
mate reconstruction purposes, and can be constrained could be related to the climate changes or ice volume vari-
through the use of paleoceanographic markers and sea level ations. Even more rapid variations on the timescale of tens to
records. hundred years are clearly related to glacioeustatism whereas
Sea level changes profoundly change the face of the postglacial rebound acts on timescale of thousands to hun-
Earth. Periods of high sea level, such as during the Ceno- dreds of thousands of years.
manian (*95 Ma) studied by Eduard Suess at the end of the However, this interpretation of the high frequency vari-
nineteenth century, caused the flooding of large continental ability found on an eustatic graph at the scale of geological
areas. Vast shelf seas, generally shallow (<200 m), covering time has been called into question. Moreover, the covering
up to 30% of the land area, were thus observed during the of coasts observed in sedimentary systems may not neces-
Phanerozoic period. In some cases the combination of a high sarily translate into eustatic variations.
sea level and a flexure of the lithosphere can result in the A different method from that of Haq et al. (1987) was
formation of shelf seas several hundred meters deep. In other proposed by Miller et al. (2005). It determines eustatic
cases, flooding or drying is controlled by dynamic variations using an inverse model by calculating the effects
topography. of the sediment load, compaction and the oceanic load
The dynamics of these epicontinental seas are still not necessary to simulate the deformation of a basin (subsidence
well understood because there is no analog of such large or uplift) located on a passive margin. The variations in sea
shallow water basins today. These epicontinental seas would level obtained by this method do not exceed 100 m, which is
have favored climates with little thermal contrast due to the half that proposed by Haq et al. (1987) for the Phanerozoic.
high calorific capacity of the water and their thermal inertia. Moreover, the eustatic variations obtained by these two
Conversely, periods with low sea levels would have favored methods may be out of phase with each other. Unlike the
more contrasted climates. Analysis of the sedimentary facies curve by Haq et al. (1987), which is based on numerous
makes it possible to locate the coastlines. These have records, Miller’s sea level variation curve is based on a small
changed with eustatic variations (change of sea level), but number of sites on the eastern margin of North America. The
also with continental uplift and subsidence, the flexure of the deformation undergone by this basin caused by internal
lithosphere in response to the build up of ice sheet or its dynamic processes originating in the mantle is not corrected.
melting or variations in the sedimentary fluxes. Changes in This basin may have recorded a “dynamic” topography with
coastlines inferred from the sedimentary facies do not long wavelength radiation (as opposed to the topography
therefore constitute a direct marker of sea level. linked to isostasy of a lithospheric column) partially skewing
In 1977, Peter Vail and his team, geologists with the the eustatic signal. Debates still exist regarding the exact
American oil company Exxon, produced a curve of sea level evolution of sea-level throughout the Phanerozoic, as a more
changes. Vail showed that the geometry and position of the recent study, based on a full geodynamical model, provided
sedimentary units deposited on the continental shelves and a large range of sea-level fluctuations for the last 500 Ma
in the basins had varied according to sea level, the subsi- (Vérard et al. 2015) (Fig. 2.4).
dence of the area of deposition, the sedimentary flow and the To understand these discrepancies, one must analyze the
carbonate production. The geometry and position of the mechanisms of sea level change. Variations in the position
sedimentary units are the main influencers of variations in of coastlines can come from several different sources: vari-
sea level, as subsidence, deposition and carbonate ation in the volume of water (the content) or variation in the
2 The Changing Face of the Earth Throughout the Ages 33

Fig. 2.4 Variations in sea level throughout the last 550 million-years, et al. (1987) and Haq and Schutter (2008); green curve is from Miller
according to various reconstructions. Red curve is from the EXXON et al. (2005), blue curve is from Vérard et al. (2015). The figure is
company, retrieved in Vérard et al. (2015); black curve is from Haq modified from Vérard et al. (2015)

capacity of ocean basins (the container). Changing the vol- Antarctic cap of today). Finally, water storage in lakes, rivers
ume of water in the ocean involves a process of contraction (*0.26 million km3) and in underground reservoirs
or thermal expansion, capture or release of this water by (*9 million km3) makes a very marginal contribution.
another reservoir (e.g., ice caps), subducted down with The most frequently cited mechanism to explain sea level
hydrated minerals in the mantle or released through degas- changes over the scale of geological time is oceanic crustal
sing at oceanic ridges. Although the volume of water present production rates and extruded oceanic plateaus. In the 1980s,
in the atmosphere in vapor and liquid form (*13,000 km3) the oceanic production rate was estimated for the last
is fundamental to the functioning of the climate system, it is 180 Ma. For the lower Cretaceous it was estimated to be
negligible compared to the volume of water contained in the twice as high as today, while a drop of about 50% over the
ocean basins (*1347 million km3). Variations in water last 50 million years was found. However these estimates
volumes in ocean basins on a time scale of between 1 and have been called into question following new estimates of
100 ka is due to the formation or melting of ice caps. The accretion rates and normalized fluxes which suggested
quantity of water held in the form of ice can lead to eustatic variations in the fluxes with an amplitude of 30% or less
variations of more than 100 m. At the geological time scale, around the present value (Cogné and Humler 2006). Such
the presence of ice at the poles is an episodic phenomenon, estimates could challenge the role of the ridges in sea level
because it implies the presence of a continent in a near-polar variations. The Wilson cycles, i.e. the construction followed
position and climate conditions allowing the formation of by the fragmentation of a supercontinent, have been identi-
permanent snow cover. For example, glaciation covered the fied as a possible mechanism to explain the first-order
southern part of the Gondwana continent (which corre- eustatic variations. Indeed, fragmentation periods produce
sponds to the southern part of South America, the southern young oceanic crust. The volume occupied by the ridges at
part of Africa, the southern part of the Arabian Peninsula, the expense of the old crust lost by subduction reduces the
parts of India and Australia and the whole of Antarctica, total volume of the oceans and raises their level, and con-
these areas having been united at one time) for about versely during the period of supercontinent construction, the
70 million years (335–265 Ma), marked by alternating total volume of the ridges decreases and the level of the
phases of growth and melting of the ice cap. The Gondwana oceans drops. In other words, the average age of the oceans
ice sheet at its climax may have sequestered a volume of reflects the first-order eustatic variations. It is thus a major
water of approximately 200 m (compared with 70 m for the challenge to develop plate motion models than can provide
34 F. Fluteau and P. Sepulchre

Fig. 2.5 Mean crustal age

through time according to
different models. Retrieved and
modified from Müller et al.
(2016)

estimates of the seafloor age evolution through time. Müller were formed by the amalgamation of modest-sized scraps of
et al. (2016) have provided a synthesis of the different model continental crust, island arcs, accretion prisms and oceanic
results since the pioneering works of Cogné et al. (2006), volcanic plateaus. The first major peak of continental crust
that shows how including new assumptions can make the production dates back to 2.7 Ga, initiated by an avalanche in
chronology of seafloor evolution change. Figure 2.5 depicts the mantle leading to the formation of a large number of
some of them. mantle plumes. This 2.7 Ga event resulted in an increase in
the number of cratons, that is to say, a permanent continental
crust, and thus a sharp increase in the continental surface
An Overview of the Changing Face of Earth (Hawkesworth et al. 2017). However, the existence of a
Through the Ages supercontinent at this period remains unlikely (Bleeker
2003).
To reconstruct the paleogeography of the past, the locations Reconstructing the paleogeography of the Archean is
of the large plates as defined by Morgan in 1968 and a very uncertain because paleomagnetic data are not sufficient
number of smaller continental masses need to be known. to constrain the position of all cratons in space and time.
This is a difficult task for the periods prior to 1.1 Ga However their geological histories can provide additional
(1100 Ma) due to the smaller number of paleomagnetic data information. The outlined scenarios are based on the simi-
available. The position of the paleoshorelines depends on the larity between the geological series preserved on each craton
available lithological information. Using available data, we (similarity in the lithological sequences, synchronism of
sketch the portrait of the Earth since the Archean. The main metamorphic and magma events affecting the cratons, con-
features of paleogeographic evolution are shown in the atlas tinuity of magmatic intrusions). However, this combination
(Fig. 2.6). does not always produce a unique scenario. For the period
Currently, the oldest observable geological units in a between 2.7 and 1.8 Ga, Bleeker (2003) counts no fewer
rocky outcrop are more than 3 billion year-old, but they are than 35 pieces of this paleogeographical puzzle. Internal
rare and do not allow to estimate the area of land present at heating twice as strong at this time is conducive to an
that time. However, the dating of a large number of zircons, organization of the plates more fragmented than today, and
an ubiquitous mineral in many rocks (igneous, metamorphic makes a vast and unique supercontinent unlikely. However,
and sedimentary), has allowed to retrace the main periods of it is not impossible that groupings of small cratons occurred
continental crust production. Zircons older than 4 Ga dis- during the Late Archean. Indeed, the presence of numerous
covered in Archean geological formations testify to the dykes dated between 2.4 and 2.1 Ga could testify for the
existence of a continental crust, even ephemeral, a few dislocation of these ephemeral land masses.
hundred million years after the formation of the Earth. The following period was marked by the formation of the
Around 3.5 Ga, the first stable continental land masses first supercontinent, Columbia (also referred to as “Nuna”),
appeared. Analysis of the oldest cratons reveals that these through the assembly of small cratons causing numerous
2 The Changing Face of the Earth Throughout the Ages 35

Fig. 2.6 Crustal aggregation states (supercratons and supercontinents) LAU = Laurussia; MAD = Madagascar; NUB = Nubia; QAN = Qan-
through time [adapted from Bleeker (2003)]. Position of the major tiang; RDP = Rio de Plata; SFCG = Sao Francisco-Congo; SIB =
continents and land masses in the past [according to Li et al. (2008) for Siberia, WAF = West Africa, WAN = West Antarctica. The thick
the Neoproterozoic, McElhinny et al. (2003) for the Paleozoic, Atlas lines represent the assumed location of the ridges (divergent litho-
Tethys and Peritethys, as well as reconstructions by the author for spheric plates); the thick black lines with triangles represent the
various studies on the Mesozoic, Fluteau et al. (2001, 2006)]. assumed location of the subduction zones (convergent lithospheric
AFR = Africa; NAM = North America; SAM = South America; plates). The shaded areas indicate the active orogenic zones for the
AMZ = Amazonia; ANT = Antarctic; ARA = Arabia; AUS = Aus- period being studied. The names of the main orogenies are indicated in
tralia; BAL = Baltic; NCH = Northern China; SCH = Southern China; italics. For the Cretaceous, the main shelf seas are represented in white
EAN = Eastern Antarctic; IND = Indochina; INDIA = India; IRA = (in a simplified way)
Iran; KAL = Kalahari; KAZ = Kazakhstan; LHA = Lhasa;
36 F. Fluteau and P. Sepulchre

Fig. 2.6 (continued)

2 The Changing Face of the Earth Throughout the Ages 37

Fig. 2.6 (continued)

38 F. Fluteau and P. Sepulchre

Fig. 2.6 (continued)

2 The Changing Face of the Earth Throughout the Ages 39

Fig. 2.6 (continued)

40 F. Fluteau and P. Sepulchre

Fig. 2.6 (continued)

2 The Changing Face of the Earth Throughout the Ages 41

Fig. 2.6 (continued)

42 F. Fluteau and P. Sepulchre

Fig. 2.6 (continued)

orogenies (Rogers and Santosh 2002; Zhao et al. 2002) and Rodinia supercontinent. The number of continents that
the structuring of several large cratons (Laurentia, Baltica, constituted Rodinia, the age of its formation or of its dislo-
Greenland, Siberia, Western Australia, India, cation, and even its very existence for some, remain open
Amazonia-West Africa, Congo-Sao Francisco) (Meert and questions. The reason for the lack of consensus is that the
Santosh 2017). The Laurentia craton is the result of the number and quality of the geological, geochronological and
assembly of several terranes (Superior, Rae, Slave, Hearne, paleomagnetic data does not lead to one single solution. For
Nain) between 1.95 and 1.8 Ga, and incorporated Wyoming a much more in-depth analysis of the paleomagnetic data
province by 1.7 Ga. The collisions of Volgo-Uralia with and the consequences in terms of configurations, the reader
Sarmatia by 2 Ga and with Fennoscandia by 1.7 Ga lead to is referred to the work of Li et al. (2008).
the formation of Baltica. Between 1.8 and 1.3 Ga, the After having drifted north, the supercontinent Rodinia
Columbia supercontinent experienced a period of continuous broke apart around 780–750 Ma. The dispersion of the
accretion along the active margins (subduction zones). These continents resulted in the opening up of ocean basins oriented
magmatic accretionary belts significantly increased the approximately north-south. Due to the spherical symmetry of
emerged land surface. A reconstruction of the Columbia the geocentric dipole magnetic field, the width of these ocean
supercontinent is made possible through paleomagnetic data basins is poorly constrained. At the end of the Neoprotero-
of some cratons at 1.5–1.4 Ga interval (Meert and Santosh zoic, around 600 Ma, a new supercontinent, Pannotia, could
2017). The development of continental rifts towards 1.5 Ga have formed. Made up of Laurentia and Gondwana domains,
marked the beginning of the fragmentation of the super- the existence of this short-lived supercontinent could only be
continent Columbia which lasted about 300 million years. linked to the uncertainties that impaired the age of break-up
However, very quickly, a new assemblage commenced. The of Laurentia, Amazonia and Baltica and the timing of
southern coast of Laurentia (which corresponds to the cur- Gondwana assembly (Li et al. 2008; Oriolo et al. 2017). This
rent east coast of the North American continent) collided is why the possibility of the Pannotia supercontinent has not
with the Amazon mass (a constituent block of South been retained in the reconstructions presented in this book.
America), and shortly after, the north Laurentian coast col- The period from the late Precambrian to the early Pale-
lided with Australia, East Antarctica and the South China. ozoic is marked by the gradual amalgamation of Gondwana
The unification of all existing cratons at that time ended from a mosaic of continents separated by oceans. These
around 950 Ma and the new supercontinent Rodinia was oceans constricted and closed, continents collided, causing
formed. These various collisions between continents brought orogenesis. These mountains have long disappeared, but the
about the existence of several large mountain ranges, whose present-day continents still bear relicts of them, such as
geological signature is found in metamorphic belts. This is metamorphic belts of high pressure, magmatism and/or
the case in North America, where the Grenville orogeny was deformations. These events, that have been dated by isotopic
dated to 1 Ga, and in South China with the Sibao orogeny. methods, have been extensively discussed in Cawood and
However, there is still no consensus on the history of the Buchan (2007).
2 The Changing Face of the Earth Throughout the Ages 43

The Nubie craton (NUB) and several small blocs were underneath Laurussia, inexorably bringing Laurussia and
amalgamated between 750 and 600 Ma. This event was the Gondwana closer together. The closure of this ocean con-
first phase of the East African orogeny dated to between 750 tinued until the Devonian (*360 Ma), after which period a
and 600 Ma. A little further east, the convergence of the generalized collision occurred along the suture between
Congo-Sao Francisco (CG-SF) and Amazon (AMZ) cratons Gondwana and Laurussia (Matte 1986). On the Laurussia
led to the closure of the Adamastor ocean, and their collision side, this resulted in the Alleghanian orogeny in North
around 650 Ma caused the Brazilian orogeny. This assembly America and the Hercynian (or Variscan) orogeny in Eur-
of cratons formed part of the future Gondwana continent. At ope. The Alleghanian orogeny affected the eastern margin of
the end of Precambrian, around 550 Ma, Laurentia and the North American continent (Canada and the United
Amazonia separated, marking the opening of the Iapetus States) and extended south through the state of Texas in the
ocean, while the cratons of Australia, Antarctica, West United States to Mexico with the Ouachita orogeny. The
Antarctica and India assembled to form another part of the Hercynian orogeny can also be observed on the Gondwana
future Gondwana continent. This event is reflected in the side through the Mauritanide orogeny in West Africa (Vil-
Kuunga and Pinjarra orogens. The closure of the ocean leneuve 2008). The dating of deformation and metamor-
separating the Kalahari and Congo-Sao Francisco cratons phism associated with this collision ranges from 340 to
leads to their collision shortly after 520 Ma, instigating the 270 Ma. The Hercynian orogeny affected much of western
Damara/Zambezi orogeny. This event occurred approxi- and central Europe. This event is dated to between 340 and
mately synchronously with the final amalgamation of the 290 Ma. At the end of the Carboniferous, these reliefs could
Gondwana continent around 530 Ma. This final stage in the have been as high as those of the Himalayas at the present
construction of the Gondwana continent brought about a time (Matte 1986). The amalgamation of Gondwana with
new orogenic phase in Eastern and Southern Africa (which Laurussia would go on to form the Pangea supercontinent,
was superimposed on the older one dated at 600 Ma). In the which would exist until 170 Ma. This supercontinent
final step in the construction of Gondwana, the passive reached its maximum size with the accretion of the
margin that bordered the western part of this continent Siberian-Kazakhstan plate in response to the closure of the
became an active margin (the oceanic crust was subducted Uralian ocean at the northeastern margin of Pangea during
under the Gondwana continent) over nearly 18,000 km, as the Permian, giving rise to the uplift of the Urals. The active
evidenced by the ages of magma activity between 550 and margin on the south of Gondwana experienced a new oro-
500 Ma. This active margin is linked with the genic cycle that occurred between the late Carboniferous
Ross-Delamerian orogeny in Australia and Antarctica, the (*310 Ma) and the Upper Triassic (*220 Ma). The
Saldanian orogeny in Southern Africa and Pampean orogeny Gondwanide orogeny affected Australia, southern Africa and
in South America. The final assembly of Gondwana and the southern regions of South America, already deformed by the
establishment of an active margin mark the end of the oro- Ross-Delamerian, Saldanian and Pampean orogenies during
gens along the sutures of the Gondwana mosaic. It is clear the Cambrian.
that the height of the reliefs remains very hypothetical, but The Lower Permian marked the beginning of the sepa-
the intensity of the collisions and geological evidence sug- ration of an assemblage of plates, called the Cimmerian
gest that the reliefs were high. plate, from the northeastern margin of Gondwana, leading to
The Gondwana continent would continue to exist for the opening of a new ocean, Neotethys (also known as Meso
nearly 400 million years until it ended during the Creta- Tethys) (Metcalfe 2002). The Cimmerian plate (southern
ceous. At the beginning of the Ordovician (*480 Ma), the China, Indochina, Lhasa, Qiangtang and others) drifted
Avallon plate, originally located close to the South Pole (its northward, closing the Paleo Tethys Ocean by subduction
remnants can be found in northeastern North America and under the northeastern margin of Pangea (the eastern part of
Western Europe), separated from Gondwana, causing the Laurussia and Kazakhstan) and under Tarim and North
birth of a new ocean, the Rheic Ocean (Nance et al. 2012). China, while to the south the Neotethys Ocean continued to
The Avallon plate migrated towards the north and collided open. This string of plates could have, for a while, isolated
with Baltica at the end of the Ordovician around 440 Ma. At the Paleo Tethys Ocean from the Panthalassa Ocean.
the end of the Silurian (*420 Ma), the collision between The configuration of the Pangea is full of uncertainties
Laurentia and Avalonia + Baltica completed the disappear- (this debate is not shown in the maps of Fig. 2.6). The
ance by subduction of the Iapetus Ocean and leads to the Pangea reconstruction is well constrained by geological and
Taconic orogeny in Laurentia. A new continent, Laurussia, geophysical data for the Late Triassic-Early Jurassic at the
was formed. This collision brought about the beginning of continental breakup. This is not the case for the
Caledonian/Acadian orogeny, which affected Scandinavia, pre-Late Triassic. Bullard et al. (1965) used to rotate the two
Greenland, Western Europe and the northeast part of North APWPs in a common frame permit to restore the paleopo-
America. The Rheic Ocean was gradually subducted sitions of two landmasses during these periods. Doing that, it
44 F. Fluteau and P. Sepulchre

results in a large continental overlap between Gondwana and and the connection with the central Atlantic Ocean only
Laurussia in excess of 1000 km. To solve this discrepancy, a occurred at the end of the Lower Cretaceous period, about
large disconnecting dextral fault operated during the Per- 30 million years later. At the beginning of the Lower Cre-
mian and Triassic periods was proposed (Irving 1977), taceous (*110 Ma), India began its drift northwards. The
resulting in a relative movement of Laurussia to the north Neo-Tethys Ocean was subducted under the southern margin
and of Gondwana to the south. Based on paleomagnetic of Eurasia, while in the south of India, the Indian ocean
data, the total dislocation length of this detachment could opened (McKenzie and Sclater 1971). Jagoutz et al. (2015)
have been several thousand kilometers, but could be much concluded that the exceptional rate of convergence exceed-
less if observations in the field of the cumulated dislocations ing 140 mm yr−1 is due to the existence of a double north-
of the known disconnecting faults in operation at that time ward dipping subduction zones between the Indian and
are considered. Moreover, the non-dipolarity of the magnetic Eurasian plates during the Cretaceous. Around 90 Ma,
field (strong influence from an octupolar field) during this Madagascar and India separated, the Carlsberg ridge is
period but more likely the lack of fidelity in the magnetic formed and marked the beginning of the opening of the
field recorded in the red sandstones caused by inclination northwestern part of the Indian Ocean.
shoaling may be “distorting” the paleomagnetic data The Cretaceous is characterized by a high sea level which
(Domeier et al. 2012). led to the formation of vast shelf seas starting at the Albian
At the beginning of the Jurassic, around 200 Ma, the (*100 Ma) to the Maastrichtian (*65 Ma). The flooding of
Pangea supercontinent broke into two vast continents, the continents reached its maximum at the beginning of the
Laurasia (Laurussia + Siberia + other smaller plates) to the Upper Cretaceous, around 95 Ma, when a large part of
north and Gondwana to the south. This separation marked Europe was inundated. At the height of the marine incursion,
the beginning of the opening of the central Atlantic Ocean. a shallow sea developed over North Africa across the current
The opening directed the drift of North America north- Sahara Desert, temporarily linking the Neo-Tethyan Ocean
westwards. This initiated the subduction of the Farallon and to the South Atlantic Ocean. In North America, during the
Kula plates (northeast region of the Pacific Ocean) under the Upper Cretaceous, a sea passage, the Western Interior Sea-
western margin of North America, resulting in the accretion way, was established between the Arctic Ocean and the Gulf
of small heterogeneous blocks and island arcs to the of Mexico, while a new orogenic phase affected the margin
North-American continent during the Jurassic. As this mar- of this continent (Laramide orogeny). A sea passage formed
gin deformed, this was the Sevier orogeny which lasted until in western Siberia linking the Arctic Ocean and the
the Lower Cretaceous. Neo-Tethyan Ocean, and only disappeared during the
Further east, the Paleo-Tethys Ocean became totally Eocene.
subducted around 200 Ma and the Cimmerian plates col- The beginning of the Cenozoic is marked in the northern
lided with the North China and Tarim plates. This conti- hemisphere by the opening of the third and last part of the
nental mosaic was separated from the southern margin of Atlantic Ocean, the northern part. North America and Eur-
Siberia by the Mongol-Okhotsk Ocean. Triangular in shape, asia separated. However the Arctic Ocean remains almost
this ocean closed like a scissors during the Jurassic and at the isolated from the rest of the oceans during the Early Ceno-
end of this period (*150 Ma), the small continent formed zoic favouring the deposits of black shales due to poorly
by the mosaic of small plates collided with the southern oxygenated water (Jakobsson et al. 2007) until the deepen-
margin of Siberia. The Eurasian continent was formed. The ing of the Fram Strait during the Late Eocene, *36 Myr
multiple collisions between the small plates more than likely (Poirier and Hillaire-Marcel 2011). In the southern hemi-
caused deformation, but probably did not create any sig- sphere, Australia and Antarctica are definitively separated
nificant reliefs. The opening of the central Atlantic Ocean during the Eocene. The Antarctic migrated to its polar
continued. The Panthalassa, Neotethys and Central Atlantic position. The circum-Antarctic basin was formed. In the
oceans all become connected in the subtropical zone of the western part of North America, subduction geometry
northern hemisphere. evolved, deformation progressed eastward, the Rocky
The Middle Jurassic (*170 Ma) marked the beginning Mountains lifted up, while some more coastal reliefs were
of the break-up of Gondwana. Madagascar, India, Australia lowered due to a change in the pattern of constraints in this
and Antarctica separated from the Africa-South America region. In South America, the uplift of the Andean Cordillera
duo. During the Lower Cretaceous (*130 Ma), South appears to have accelerated towards the end of the Cenozoic.
America and Africa started to split and the South Atlantic India collided with Asia at the beginning of the Cenozoic
Ocean opened between Patagonia and Southern Africa. This (*50 Ma). India’s drift to the north continued after the
event could mark the beginning of the deformation of the collision at a rate of 5–6 cm/year. Part of the crustal thick-
western central Andes margin in South America (Torsvik ening is accommodated by the play of large right-lateral
et al. 2009). The complete separation of the two continents deformations reactivating old sutures between the plaques
2 The Changing Face of the Earth Throughout the Ages 45

constituting the Asian mosaic, and by the uplift of the

Himalayas and other chains farther north. These great dis- Two contrasting opinions have been proposed to
connecting systems laterally eject plaques such as South explain the uplift of the Tibetan plateau: a “soft Tibet”
China and Indochina (Tapponnier et al. 2001). This defor- model and a “staggered model”. The “soft Tibet”
mation of the southern margin of Asia is in fact generalized model considers the uplift to be the result of an iso-
throughout the southern margin of Eurasia, in response to the static rebound caused by the “sinking” of the crustal
Africa-Europe overlap resulting in the Alpine orogeny. This root of a thickened Tibetan proto-plateau. The plateau
deformation contributed to the establishment of sea basins, reached its stable altitude of about 5000 m, and began
sometimes deep, in Eurasia. These basins were connected to to creep, as evidenced by the normal faults that line the
each other and extended from Western Europe to the foot of southern plateau. The other theory (Tapponnier et al.
the Tibetan zone (during the process of uplifting) to form a 2001) considers, on the contrary, that the uplift took
vast epicontinental sea, the Paratethys, connected to the place in successive stages. Part of the India-Asia col-
Atlantic Ocean and to the Tethys. The intense deformation lision is absorbed by the lateral extrusion of land-
of the southern margin of Eurasia during the Miocene masses (South China, Indochina), which caused old
gradually isolated the Paratethys from the rest of the oceans. sutures inherited from the accretion of small land-
Supplied only by runoff waters and subjected to the defor- masses during the Paleozoic and Mesozoic to pro-
mation of the substratum, this sea was gradually reabsorbed. trude, profoundly modifying the paleogeography of
Today, the Caspian Sea and the Black Sea are the only South-East Asia (resulting in the closure of the
descendants of the Paratethys. Indonesian passage). These large strike-slip faults,
The Arabia-Eurasia collision during the Early Miocene separating the landmasses, connect with small per-
lead to the closure of the Neo-Tethys sea and the uplift of the pendicular faults parallel to each other. The associated
Zagros collisional wedge (Pirouz et al. 2017). In the Middle reliefs isolate small endorheic sedimentary basins
Miocene, the northern part of the Arabian Peninsula became which gradually fill up with the output from the dis-
exposed, marking the birth of the Mediterranean Sea. This mantling of the chain “like a bathtub”. The uplift of
would dry up almost completely at the end of the Miocene the Tibetan plateau would have occurred in successive
(the Messinian crisis) for a few hundred thousand years, due stages since 50 Ma. The southern part of the Tibetan
to an upheaval in the Betic zone (Gibraltar region) combined plateau rose during the Eocene, the central part during
with a slight decrease in sea level. During the late Miocene, the Oligocene-Miocene and the northeastern part is
the northward drift of Australia initiated the closure of the currently uplifting.
Indonesian seaway, while the closure of the Central Amer-
ican Seaway linked to a tectonic uprising permanently iso-
lated the Atlantic and Pacific waters.
Box: Focus on the uplift of the Tibetan Plateau.
Conclusion

Since 50 Ma, the average convergence speed between The Earth moved from being entirely composed of an
India and Asia has remained around 5 cm per year. oceanic crust to the emergence of the first shreds of conti-
The Himalayan range occurred due to the deformation nental crust during the Archean, then to its episodic growth,
of the northern edge of the Indian sub-continent when mainly during the Precambrian. The paleogeography of this
it was subducted under the Asian continent. To the period is uncertain, given the small number of paleomag-
south of the Tsangpo suture is the Himalayan domain: netic data available, but it seems that the first consolidation
on Indian crust with remnants of the Tethyan sedi- of landmasses into a supercontinent (followed by a break-up
mentary cover and accretion prism associated with phase) dates back to that time. It is only from the end of the
subduction of the Tethyan Ocean before collision. Precambrian that we have global paleogeographic recon-
North of the suture, the Tibetan plateau has sedimen- structions, but there are no unique and definitive solutions
tary layers from the Asian paleomargin. These are two for much of the Paleozoic, as uncertainties remain on the
very different realms with different Cenozoic geolog- arrangement of some continents and the dimensions of the
ical histories, although related to the same event. oceans. Paleogeographic reconstructions become more reli-
During the millions of years after the collision, a able at the end of the Paleozoic, and data from ocean kine-
proto-Himalayan chain must have developed on the matics makes it possible to constrain the relative positions of
remains of the Andean chain associated with the some continents relative to others. The distribution of the
subduction of the Tethys, but the extent of this is not continents, either as they grouped together or dispersed,
known. The Tibetan plateau was still a low-lying area. represents an important climate forcing through direct and
46 F. Fluteau and P. Sepulchre

undirect effects, while a second trigger of long-term climate Cogné, J. P., & Humler, E. (2006). Trends and rhythms in global
change is mountain uplift. In that respect, the last seafloor generation rate. Geochemistry, Geophysics, Geosystems, 7,
Q03011. https://doi.org/10.1029/2005GC001148.
60 million-years have been crucial, since most of Cogné, J. P., Humler, E., & Courtillot, V. (2006). Mean age of oceanic
present-day mountain ranges have uplifted during this time crust drives eustatic sealevel change since Pangea breakup. Earth
interval. The Himalayan orogeny and the uplift of the and Planetary Science Letters, 245, 115–122.
Tibetan plateau are certainly the two most studied geological Cox, A., Doell, R. R., & Dalrymple, G. B. (1964). Reversals of the
earth’s magnetic field. Science, 144, 1537–1543.
events over the past three decades in terms of impact on Creer, K. M., Irving, E., & Runcorn, S. K. (1954). The direction of the
climate, but other mountain ranges such as the Andes, the geomagnetic field in remote epochs in Great Britain. Journal of
Rockies and the east African dome very likely have played Geomagnetism and Geoelectricity, 6, 163–168.
important roles on the Cenozoic climate change. Some clues Domeier, M., van der Voo, R., & Torsvik. T. H. (2012). Paleomag-
netism and Pangea: The road to reconciliation. Tectonophysics,
on this point are provided in Volume 2, Chap. 22. Still, the 514–517, 14–43.
paleoelevations of the reliefs remain shrouded in uncertainty, Eiler, J. M. (2007). “Clumped-isotope” geochemistry—The study of
and efforts both on the modeling and the paleoelevation naturally-occurring, multiply-substituted isotopologues. Earth and
indicator sides will need to be pursued to improve our Planetary Science Letters, 262, 309–327.
Ferreira, D., Cessi, P., Coxall, H. K., de Boer, A., Dijkstra, H. A.,
understanding of paleogeography evolution of Earth through Drijfhout, S. S., et al. (2018). Atlantic-Pacific asymmetry in deep
time, and ultimately better quantify its impact on long-term water formation. Annual Review of Earth and Planetary Sciences,
climate changes. 46, 327–352.
Fluteau, F., Besse, J., Broutin, J., & Ramstein, G. (2001). The Late
Permian climate. What can be inferred from climate modelling
concerning Pangea scenarios and Hercynian range altitude? Palaeo-
References geography, Palaeoclimatology, Palaeoecology, 167, 39–71.
Fluteau, F., Ramstein, G., Besse, J., Guiraud, R., & Masse, J. P. (2006).
The impacts of the paleogeography and sea level changes on the
Besse, J., & Courtillot, V. (1991). Revised and synthetic apparent polar
mid cretaceous climate. Palaeogeography, Palaeoclimatology,
wander path of the African, Eurasian, North American and India
Palaeoecology, 247, 357–381.
plates, and true polar wander since 200 Ma. Journal of Geophysical
Forest, C. E., Molnar, P., & Emanuel, K. A. (1995). Palaeoaltimetry
Research: Solid Earth, 95, 4029– 4050.
from energy conservation principles. Nature, 374, 347–350.
Besse, J., & Courtillot, V. (2002). Apparent and true polar wander and
Garzione, C. N., Auerbach, D. J., Jin-Sook Smith, J., Rosario, J. J.,
the geometry of the geomagnetic field over the last 200 Myr.
Passey, B. H., Jordan, T. E., et al. (2014). Clumped isotope
Journal of Geophysical Research: Solid Earth, 107, EPM 6-1–EPM
evidence for diachronous surface cooling of the Altiplano and
6-31.
pulsed surface uplift of the Central Andes. Earth and Planetary
Bleeker, W. (2003). The Late Archean record: A puzzle in ca.
Science Letters, 393, 173–181.
35 pieces. Lithos, 71, 99–134.
Greff-Lefftz, M., & Besse, J. (2014). Sensitivity experiments on true polar
Bonifacie, M., Calmels, D., Eiler, J. M., Horita, J., Chaduteau, C.,
wander. Geochemistry, Geophysics, Geosystems, 15, 4599–4616.
Vasconcelos, C., et al. (2017). Calibration of the dolomite clumped
Haq, B. U., Hardenbol, J. A. N., & Vail, P. R. (1987). Chronology of
isotope thermometer from 25 to 350 °C, and implications for a
fluctuating sea levels since the Triassic (250 million years ago to
universal calibration for all (Ca, Mg, Fe) CO3 carbonates.
present). Science, 235, 1156–1166.
Geochimica et Cosmochimica Acta, 200(1), 255–279.
Haq, B. U., & Schutter, S. R. (2008). A chronology of paleozoic
Botsyun, S., Sepulchre, P., Donnadieu, Y., Risi, C., Licht, A., & Caves
sea-level changes. Science, 322(5898), 64–68. https://doi.org/10.
Rugenstein, J. K. (2019). Revised paleoaltimetry data show low
1126/science.1161648.
Tibetan Plateau elevation during the Eocene. Science, 363. http://
Hawkesworth, C. J., Cawood, P. A., Dhuime, B., & Kemp, T. I. S.
doi.org/10.1126/science.aaq1436.
(2017). Earth’s continental lithosphere through time. Annual Review
Botsyun, S., Sepulchre, P., Risi, C., & Donnadieu, Y. (2016). Impacts
of Earth and Planetary Sciences, 45, 169–198.
of Tibetan Plateau uplift on atmospheric dynamics and associated
Heezen, B. C. (1962). The deep-sea floor. In S. I. Runcorn (Ed.),
precipitation d18O. Climate of the Past, 12, 1401–1420.
Continental drift (pp. 235–288). New York: Academic Press.
Bullard, E. C., Everett, J. E., & Smith, A. G. (1965). The fit of the
Heezen, B. C., & Tharp, M. (1965). Tectonic fabric of the Atlantic and
continents around the Atlantic. Philosophical Transactions of the
Indian Oceans and continental drift. In P. M. S. Blackett, E. Bullard,
Royal Society London, 258A, 41–51.
& S. K. Runcorn (Eds.), A Symposium on Continental Drift,
Cawood, P. A., & Buchan, C. (2007). Linking accretionary orogenesis
Philosophical Transactions of the Royal Society A, 1033, 90. https://
with supercontinent assembly. Earth-Science Reviews, 3–4, 217–
doi.org/10.1098/rsta.1965.0024.
256.
Heirtzler, J. R., Dickson, G. O., Herron, E. M., Pitman, W. C., &
Chamberlain, C. P., Poage, M. A., Craw, D., & Reynolds, R. C. (1999).
Pichon, X. L. (1968). Marine magnetic anomalies, geomagnetic
Topographic development of the Southern Alps recorded by
field reversals, and motions of the ocean floor and continents.
isotopic composition of authigenic clay minerals, South Island,
Journal Geophysical Research, 73, 2119–2136.
New Zealand. Chemical Geology, 155, 279–94.
Hess, H. H. (1962). History of ocean basins. In A. E. J. Engle, H.
Cogné, J.-P., Besse, J., Chen, Y., & Hankard, F. (2013). A new Late
L. James, & B. L. Leonard (Eds.), Petrologic studies: A volume in
Cretaceous to present APWP for Asia and its implications for
honor of A. F. Buddington (pp. 599–620). New York: Geological
paleomagnetic shallow inclinations in Central Asia and Cenozoic
Society of America.
Eurasian plate deformation. Geophysical Journal International,
Holmes, A. (1929). A review of the continental drift hypothesis. Mining
192, 1000–1024.
Magazine 1–15.
2 The Changing Face of the Earth Throughout the Ages 47

Husson, L. (2006). Dynamic topography above retreating subduction Nance, R. D., Gutierrez-Alonso, G., Duncan Keppie, J., Linnemann,
zones. Geology, 34(9), 741–744. U., Brendan Murphy, J., Quesada, C., et al. (2012). A brief history
Irving, E. (1964). Paleomagnetism and its application to geological of the Rheic Ocean. Geosciences Frontiers, 3(2), 125–135.
and geophysical problems. Hoboken: Wiley. Oriolo, S., Oyhantçabal, P., Wemmer, K., & Siegesmund, S. (2017).
Irving, E. (1977). Drift of the major continental blocks since the Contemporaneous assembly of Western Gondwana and final
Devonian. Nature, 270, 304–309. Rodinia break-up: Implications for the supercontinent cycle.
Jagoutz, O., Royden, L., Holt, A. F., & Becker, T. W. (2015). Geoscience Frontiers, 8, 1431–1445.
Anomalously fast convergence of India and Eurasia caused by Otto-Bliesner, B. L., & Upchurch, G. R. Jr. (1997). Vegetation-induced
double subduction. Nature Geoscience, 8, 475–479. warming of high-latitude regions during the Late Cretaceous period.
Jakobsson, M., Backman, J., Rudels, B., Nycander, J., Frank, M., Nature, 385, 804–807.
Mayer, L., et al. (2007). The early Miocene onset of a ventilated Otto-Bliesner, B. L., Brady, E. C., & Shields, C. (2002). Late
circulation regime in the Arctic Ocean. Nature, 447, 986–990. Cretaceous Ocean: Coupled simulations with the National Center
Kim, S. T., & O’Neil, J. R. (1997). Equilibrium and nonequilibrium for Atmospheric Research climate system model. Journal of
oxygen isotope effects in synthetic carbonates. Geochimica et Geophysical Research: Atmospheres, 107(D2), ACL 11-1–ACL
Cosmochimica Acta, 61, 3461–3475. 11-14.
Kirschvink, J. L., Ripperdan, R. L., & Evans, D. A. (1997). Evidence Peppe, D. J., Royer, D. L., Wilf, P., & Kowalski, E. A. (2010).
for a large-scale reorganization of early cambrian continental Quantification of large uncertainties in fossil leaf paleoaltimetry.
masses by inertial interchange true polar wander. Science, 277 Tectonics, 29(3). https://doi.org/10.1029/2009TC002549.
(5325), 541–545. Pirouz, M., Avouac, J. P., Hassanzadeh, J., Kirschvink, J. L., &
Le Pichon, X. (1968). Sea-floor spreading and continental drift. Journal Bahroudi, A. (2017). Early Neogene foreland of the Zagros,
Geophysical Research, 73, 3661–3697. implications for the initial closure of the Neo-Tethys and kinematics
Li, L., & Garzione, C. (2017). Spatial distribution and controlling of crustal shortening. Earth and Planetary Science Letters, 477,
factors of stable isotopes in meteoric waters on the Tibetan Plateau: 168–182.
Implications for paleoelevation reconstructions. Earth and Plane- Poirier, A., & Hillaire-Marcel, C. (2011). Improved Os-isotope
tary Science Letters, 460, 302–314. stratigraphy of the Arctic Ocean. Geophysical Research Letters,
Li, Z. X., Bogdanova, S. V., Collins, A. S., Davidson, A., De Waele, 38, L14607.
B., Ernst, R. E., et al. (2008). Assembly, configuration, and Poulsen, C. J., Ehlers, T. A., & Insel, N. (2010). Onset of convective
break-up history of Rodinia: A synthesis. Precambrian Research, rainfall during gradual Late Miocene rise of the Central Andes.
160, 179–210. Science, 328, 490–493.
Matte, P. (1986). Tectonics and plate tectonics model for the Variscan Quade, J., Garzione, C., & Eiler, J. (2007). Paleoelevation reconstruc-
Belt of Europe. Tectonophysics, 126, 329–374. tion using pedogenic carbonates. Reviews in Mineralogy and
McElhinny, M. W., Powell, C. M., & Pisarevsky, S. A. (2003). Geochemistry, 66, 53–87.
Paleozoic Terranes of Eastern Australia and the drift history of Raff, A. D., & Mason, R. G. (1961). Magnetic survey off the west coast
Gondwana. Tectonophysics, 362, 41–65. of North America, 40° n. latitude to 52° n. latitude. GSA Bulletin,
McKenzie, D., & Sclater, J. G. (1971). The evolution of the Indian 72, 1267–1270.
Ocean since the Late Cretaceous. Geophysical Journal Interna- Robert, B., Besse, J., Blein, O., Greff-Lefftz, M., Baudin, T., Lopes, F.,
tional, 24, 437–528. et al. (2017). Constraints on the Ediacaran inertial interchange true
Meert, J. G., & Santosh, M. (2017). The Columbia supercontinent polar wander hypothesis: A new paleomagnetic study in Morocco
revisited. Gondwana Research, 50, 67–83. (West African Craton). Precambrian Research, 295, 90–116.
Metcalfe, I. (2002). Permian tectonic framework and palaeogeography Rogers, J. J. W., & Santosh, M. (2002). Configuration of Columbia, a
of SE Asia. Journal of Asian Earth Sciences, 20, 551–566. Mesoproterozoic supercontinent. Gondwana Research, 5, 5–22.
Miller, K. G., Kominz, M. A., Browning, J. V., Wright, J. D., Romm, J. (1994). A new forerunner for continental drift. Nature
Mountain, G. S., Katz, M. E., et al. (2005). The Phanerozoic record 367, 407.
of global sea-level change. Science, 310, 1293–1298. Rowley, D. B., & Garzione, C. N. (2007). Stable isotope-based
Moreau, M.-G., Besse, J., Fluteau, F., & Greff-Lefftz, M. (2007). A new paleoaltimetry. Annual Review of Earth and Planetary Sciences, 35,
global Paleocene-Eocene apparent polar wandering path loop by 463–508.
“stacking” magnetostratigraphies: Correlations with high latitude Sarr, A. C., Husson, L., Sepulchre, P., Pastier, A. M., Pedoja, K., Elliot,
climatic data. Earth and Planetary Science Letters, 260, 152–165. M., et al. (2019). Subsiding Sundaland. Geology, 47(2), 119–122.
Morgan, W. J. (1968). Rises, trenches, great faults and crustal blocks. https://doi.org/10.1130/G45629.1.
Journal Geophysical Research, 73(308), 1959–1982. Seton, M., Müller, R. D., Zahirovic, S., Gaina, C., Torsvik, T.,
Morgan, W. J. (1972). Deep mantle convection plumes and plate Shephard, G., et al. (2012). Global continental and ocean basin
motions. AAPG Bulletin, 56, 203–213. reconstructions since 200 Ma. Earth Science Reviews, 113,
Mulch, A. (2016). Stable isotope paleoaltimetry and the evolution of 212–270.
landscapes and life. Earth and Planetary Science Letters, 433, 180– Su, T., Farnsworth, A., Spicer, R. A., Huang, J., Wu, F.-X., Liu, J.,
191. https://doi.org/10.1016/j.epsl.2015.10.034. et al. (2019). No high Tibetan Plateau until the Neogene. Science
Müller, R. D., Seton, M., Zahirovic, S., Williams, S. E., Matthews, K. Advances, 5(3), eaav2189. https://doi.org/10.1126/sciadv.aav2189.
J., Wright, N. M., et al. (2016). Ocean basin evolution and Tapponnier, P., Zhiqin, X., Roger, F., Meyer, B., Arnaud, N.,
global-scale plate reorganization events since Pangea Wittlinger, G., & Jingsui, Y. (2001). Oblique, stepwise rise and
breakup. Annual Review of Earth and Planetary Sciences, 44(1), growth of the Tibet Plateau. Science, 294, 1671–1677.
107–138.
48 F. Fluteau and P. Sepulchre

Tarduno, J. A., Blackman, E. G., & Mamajek, E. E. (2014). Detecting Vail, P. R. (1977). Seismic stratigraphy and global changes of sea level.
the oldest geodynamo and attendant shielding from the solar wind: Bulletin American Association of Petroleum Geologists Memoir, 26,
Implications for habitability. Physics of the Earth and Planetary 49–212.
Interiors, 233, 68–87. Vérard, C., Hochard, C., Baumgartner, P. O., Stampfli, G. M., & Liu,
Tarduno, J. A., Cottrell, R. D., Davis, W. J., Nimmo, F., & Bono, R. K. M. (2015). 3D palaeogeographic reconstructions of the Phanerozoic
(2015). A Hadean to Paleoarchean geodynamo recorded by single versus sea-level and Sr-ratio variations. Journal of Palaeogeogra-
zircon crystals. Science, 349, 521–524. phy, 4(1), 64–84.
Torsvik, T. H., Rousse, S., Smethurst, M. A., (2009). A new scheme for Villeneuve, M. (2008). Review of the orogenic belts on the western
the opening of the South Atlantic Ocean and the dissection of an side of the West African craton: The Bassarides, Rokelides and
Aptian salt basin. Geophysical Journal International, 183(1), Mauritanides. In N. Ennih & J.-P. Liégeois (Eds.), The boundaries
29–34. of the West African Craton (Vol. 297, pp. 169–201). Geological
Torsvik, T. H., Van der Voo, R., Preeden, U., Mac Niocaill, C., Society, London, Special Publications.
Steinberger, B., Doubrovine, P. V., et al. (2012). Phanerozoic polar Vine, F. J., & Matthews, D. H. (1963). Magnetic anomalies over
wander, palaeogeography and dynamics. Earth Science Reviews, oceanic ridges. Nature, 199, 947.
114, 325–368. Yount, L. (2009). Alfred Wegener: Creator of the continental drift
U.S. Geological Survey Geologic Names Committee. (2018). Divisions theory. New York: Infobase Publishing.
of geologic time—Major chronostratigraphic and geochronologic Zhao, G., Cawood, P. A., Wilde, S. A., & Sun, M. (2002). Review of
units. U.S. Geological Survey Fact Sheet 2018–3054, 2 p. https:// global 2.1-1.8 Ga orogens: Implications for a pre-Rodinia super-
doi.org/10.3133/fs20183054. continent. Earth-Science Reviews, 59, 125–162.
 Introduction to Geochronology
3
Hervé Guillou

Accurate knowledge of climate variations in the past is an the time scales for different types of records, both from the
essential preamble to any realistic modeling of future cli- land and ocean, isotopic stratigraphy plays an essential role
mate, and requires an understanding of the mechanisms that in providing a better understanding of the chronology and
govern the natural dynamics of the climate, and especially of dynamics of the mechanisms responsible for climate
its rapid changes. One of the current major concerns for this variations.
research is the quantification of the phase shift in climate Variations in climate are caused by many factors with
between different regions of the globe. This requires having characteristic durations ranging from hundreds of millions of
reliable, precise and comprehensive chrono-stratigraphic years for the evolution of the Sun to a few years for internal
tools in order to temporally locate and to synchronize the reorganization of the climate system. In addition, records of
various archives. climate signals that can potentially be dated are represented
Establishing a common time frame for all climate in very different substrates (sediments, ice, coral, cave con-
archives remains a major challenge. Research on long time cretions known as speleothems). The choice of the best
scales emphasizes the climate system’s response to external adapted geochronological tools to date them will depend on
forcing, but the study of rapid and abrupt changes in climate the nature of the records, their age, the time span of the
allows the internal variability of the climate system and the phenomena to be dated and the desired level of accuracy.
interactions between its various components to be investi- The dating methods most commonly used in paleocli-
gated. The last glacial period was characterized by a suc- matology are: dendrochronology, 14C, the Uranium/
cession of very rapid changes in the climate in the North Thorium relationship, Potassium-Argon (40K/40Ar) and its
Atlantic, which resulted in massive reorganization of the variant 40Ar/39Ar (isotopic methods), and magnetic stratig-
climate system on a global scale and was manifested in raphy (indirect method of dating). Often, in order to provide
particular by the massive discharge of icebergs into the an accurate geo-chronological framework, two or more of
ocean, known as Heinrich events. These events mainly these methods need to be compared.
occurred during glacial periods but some occurred as soon as In the following chapters, we will present the absolute
ice sheets developed on land in the northern hemisphere at dating methods implemented to provide a time scale inde-
the end of the last interglacial, and also at the beginning of pendent of astronomical parameters and to refine the strati-
the Holocene, even though interglacial periods seem to have graphic scales commonly used in paleoclimatology and in
been much more stable. In order to understand and model paleo-oceanography, very often based on variations in the
18 16
the mechanisms involved, it is important to know the precise O/ O relationship in ice and benthic foraminifera related
chronology of all these events. to the orbital signal. The principles of the methods men-
Another purpose of geochronological studies is to enable tioned above, their field of application, their implementation
the comparison of climate records on a common and abso- in the laboratory, their accuracy and limitations will also be
lute time scale. It is only through this approach that the phase presented. The scope of each method will be illustrated with
shifts between the hemispheres or between low and high a concrete example.
latitudes can be understood and explained. By harmonizing Geochronology plays an essential role for both geologists
and paleoclimatologists. For geologists, it has allowed
H. Guillou (&) traditional stratigraphy to be linked to a time scale covering
Laboratoire des Sciences du Climat et de l’Environnement, the full history of the Earth and to estimate the time con-
LSCE/IPSL, CEA-CNRS-UVSQ, Université Paris-Saclay, 91190
Gif-sur-Yvette, France stants of the great geological phenomena (plate tectonics,
e-mail: [email protected] uplift of mountains, renewal of ocean basins, long-term

© Springer Nature Switzerland AG 2021 49

G. Ramstein et al. (eds.), Paleoclimatology, Frontiers in Earth Sciences,
https://doi.org/10.1007/978-3-030-24982-3_3
50 H. Guillou

global climate changes). The latest version of the chrono- This has allowed the Milankovitch theory to be validated, a
stratigraphic scale is available on the website of the Inter- work that has been a focus of attention of geochemists over
national Commission on Stratigraphy: http://www. the last forty years. Currently, a major experimental effort is
stratigraphy.org/. underway to try to improve the dating accuracy and to detect
For paleoclimatologists, geochronology has provided a the phase shifts that accompany the response of various
time scale entirely independent of astronomical parameters. components of climate system to the insolation forcing.
 Carbon-14
4
Martine Paterne, Élisabeth Michel,
and Christine Hatté et Jean-Claude Dutay

Seventy years after its discovery by W.B.F. Libby and col- physicist Kurie (Kamen 1963), who studied neutron-induced
laborators (Arnold and Libby 1949; Libby 1952), the disintegration of light elements such as nitrogen (14N).
radiocarbon (14C) method of dating is still of great interest in During these experiments, Kurie observed infrequent and
many scientific fields in biology, earth science, climate, abnormal long thin traces in a cloud chamber filled with air.
environment and archeology. Libby received the Nobel Prize He attributed them to the emission of protons following the
in Chemistry in 1960 and the chairman of the Nobel Com- reaction 14N (n, 1H) 14C although other reactions such as 14N
mittee highlighted the importance of this discovery in these (n, 2H) 12C and 14N (n, 3H) 12C could also have been pos-
terms: “Seldom has a single discovery in chemistry had such sible. In 1936, Burcham and Goldhaber demonstrated that no
an impact on the thinking of so many fields of human a-particles were emitted in the slow neutron disintegration of
14
endeavor. Seldom has a single discovery generated such N and only the reaction 14N(n, 1H)14C was possible with
wide public interest”. The history of the 14C method is an proton emission and formation of 14C noted 14N(n, p)14C
excellent example of the fruitful exchanges among different (Kamen 1963).
scientific fields and of the complementarity between scien- The evidence of the chemical existence of 14C is due to
tific advances and technological innovations. Since its dis- Ruben, a chemist and Libby’s student, and to Kamen, a
covery, more than a hundred and fifty laboratories in the radiochemist of the Lawrence Livermore Radiation Labo-
world are now dedicated to 14C dating. In the 1980s, new ratory at Berkeley. They investigated the assimilation pro-
technologies, notably the accelerator mass spectrometry, cesses of CO2 during photosynthesis by incubating plant
allowed the use of samples of increasingly reduced sizes and species with the radioactive isotope 11C, which was pro-
a better precision of the 14C ages. Now, the physical and duced in the Livermore cyclotron (Ruben et al. 1949).
chemical processes involved in biological and environmental Labeled intermediate solutions were deposited on a blotting
changes may be analyzed at the molecular scale. paper, and, once dried, the paper was protected by a plastic
The method and techniques of 14C dating have been the film and wrapped inside a screen-wall counter. The use of
11
subject of several web and journal publications to which C in biology was however very difficult due to long sep-
readers may refer (Libby 1981; Taylor 1987; Taylor et al. aration phases of various photosynthetic pigments by ultra-
1992; Currie 2004). centrifugation and a half-life of 21 min. Furthermore, it was
not very competitive with the 13C labeling of plants. At the
request of Lawrence, who invented the cyclotron and
Principles of the Radiocarbon Method received the Nobel Prize in Physics in 1939, the existence or
not of long-lived radioactive isotopes was systematically
Discovery of the Method sought for each element of the first column of the periodic
table (H, C, N, O), and thus the search for the chemical
Kamen (1963) reported the history of the 14C discovery and existence of 14C (Kamen 1963). Kamen submitted a graphite
Libby’s meeting with 14C in 1939 at Berkeley. He attributed target to a deuteron beam in the cyclotron overnight. After
the physical prediction of the existence of this isotope to the burning the graphite, Ruben precipitated the CO2 into a
carbonate. This precipitate was furnished to their colleague
M. Paterne (&)
É. Michel
C. H. et Jean-Claude Dutay in the chemistry department, W.F. Libby, who developed
Laboratoire des Sciences du Climat et de l’Environnement, proportional counters to measure the radioactivity of ele-
LSCE/IPSL, CEA-CNRS-UVSQ, Université Paris-Saclay,
91190 Gif-sur-Yvette, France ments such as neodymium, samarium, rubidium and lute-
e-mail: [email protected] tium, to determine their period (Libby 1934). The detection

© Springer Nature Switzerland AG 2021 51

G. Ramstein et al. (eds.), Paleoclimatology, Frontiers in Earth Sciences,
https://doi.org/10.1007/978-3-030-24982-3_4
52 M. Paterne et al.

of a weak activity was the first proof of the existence of This explains the increase in neutron density at around
artificially created 14C; its half-life was estimated to be 16 km and the decrease towards the sea-level.
between 103 years and 105 years (Kamen 1963). During collisions, the neutrons slow down, and Korff
This research was interrupted in the early 1940s as Libby suggested that these secondary slow neutrons were captured
joined Harold Urey’s team in Chicago to develop techniques by nitrogen nuclei to form the cosmogenic isotope 14C fol-
of isotopic enrichment of uranium for nuclear weapons in the lowing the reaction 14N (n, p) 14C (Korff 1951). When
frame of the Manhattan Project. He would later use these receiving the Nobel Prize in 1960, Libby (1964) indicated
enrichment techniques in natural samples for 14C studies. that the idea of the 14C dating method was inspired by
While the 14C was artificially created, the existence of a Korff’s results. Later, Simpson (2000) showed that the
natural production of 14C still needed to be proven and thus density of neutrons varied with the latitude, as a function of
the existence of slow neutrons in the Earth’s atmosphere in the lines of the Earth’s magnetic field that deflect the
order to assess the feasibility of 14C dating. (electrically charged) particles of cosmic rays. The average
The method of 14C dating is linked to the discovery of production of 14C is in the range of 2.25 ± 0.1 atoms of
14
cosmic radiations, later termed cosmic rays, by Victor Hess C/cm2/s. It varies from one to six between the equator and
in 1912 by using electroscopes aboard a balloon (Libby the poles, and at the poles, it can vary by a factor of almost
1964; Rossi 1952). This discovery was the beginning of four depending on solar activity. Of minor importance, other
numerous studies, which investigated the composition, reactions on 16O, 17O, 13C, contribute also to the formation
intensity, origin, and effect of the cosmic rays on the Earth’s of atmospheric 14C.
atmosphere. Cosmic rays caused nuclear reactions in the
atmosphere, which were suspected by Grosse (Libby 1981).
14
These reactions were first evidenced by Blau in 1932, who Principle of the C Dating Method
pioneered the technique of photographic plates covered by
thick nuclear emulsions to separate the a-particles from Libby postulated in 1946 (Arnold and Libby 1949) that the
proton tracks. Such plates were exposed in the Austrian Alps production of 14C atoms and their decay as 14N by emitting a
revealing disintegration stars in the emulsion (Rossi 1952). b particle would be in equilibrium at steady-state conditions
Rumbaugh and Locher determined the nature of this radia- (Fig. 4.1). As the cosmic rays continuously bombard the
tion by sending photographic plates into the stratosphere to Earth and as the Earth age is much higher than the estimated
an altitude of about 20 km in the gondola of a balloon (Rossi period of 103–105 years, the distribution of 14C would be in
1952). Some plates were covered with different materials equilibrium within all the reservoirs of exchangeable carbon
about 1 cm thick and the others were free of materials. The (atmosphere, ocean, biosphere). Estimating the production of
control plates showed no traces, while those covered by neutrons per cm2 and per second based on the distribution of
paraffin, for example, showed four times more traces than neutrons observed by Korff (1951) and the amount of
those covered by lead or carbon. Due to the absence of traces exchangeable carbon between the reservoirs, Libby wrote
on the control plates, the traces could be only protons and that the specific activity of exchangeable carbon could be
not a particles, and these protons could have been emitted easily calculated taking into account the balance between
only from the different materials during collision of atoms production and decay:
with neutrons. Korff and colleagues then performed new
d14 C
experiments using proportional counters aboard balloon, ¼ Q  k14 C ¼ 0
some filled with boron trifluoride (boron is a neutron dt
absorber and emits a a particle upon collision with a neu- where Q is the production of a 14C atom per second and k,
tron), and others filled with a mixture of hydrogen, methane the decay constant, is equal to ln(2)/T1/2, where T1/2 is the
and carbon monoxide (sensitive to fast neutrons) (Rossi half-life (half of the radioactive atoms have decayed).
1952; Simpson 2000). These counters permitted precise

14 
measurement of the density of the neutrons and their energy d 12 CC  14

1 12 d C 14 d12 C
spectrum. These authors have thus shown that the density of ¼0¼ C  C
slow neutrons reached a maximum at an altitude between 12 dt ð12 CÞ2 dt dt
12
C  
and 16 km, and then decreased towards the sea level. When ¼ Q  k14 C
entering the atmosphere, the protons, which compose about ð12 C Þ2
90% of the cosmic rays, collide with atoms and molecules
(mainly nitrogen and oxygen). The products of their disin- because
tegration are protons and neutrons, which collide with other d 12 C
14
atoms while neutrons lose some energy on each collision. C ¼0
dt
4 Carbon-14 53

Fig. 4.1 Diagram of the 14C

formation and mixing in the
different terrestrial reservoirs
(Libby 1964)

The carbon stocks are expressed in Gt (Gigatons) and the isotopic ratio 14C/12C of highly enriched barium carbonates
number of neutrons per cm2 per second corresponds to [BaCO3] (up to 6%) produced by submitting to neutrons
pre-1950 estimates (modified according to 10): beams, solutions of ammonium nitrate in a cyclotron
(Engelkemeir et al. 1949). Proportional counters were then
14
C Q filled with the CO2 released by BaCO3 hydrolysis. By mea-
k 12 ¼ 12
fC g C suring the number of disintegrations per minute and per gram
of carbon (dN/dt) and knowing the number of atoms of 14C
The predicted activity should have been between 1 and 10
(N) in the samples, the half-life was estimated at
decays per minute per gram of carbon, given the uncertain-
5720 ± 47 years. Averaging all the published estimates,
ties, in all living matter. 14C activities of 10.5 disintegrations
Libby estimated the half-life at 5568 years. Redeterminations
per minute per gram were measured from isotopically enri-
have led to a value of 5730 ± 40 years (Godwin 1962). The
ched samples of biomethane in 1947, in good agreement with
latter value has recently been debated (Chiu et al. 2007).
the prediction. Libby and collaborators measured then the
Libby and collaborators then undertook the dating of
specific activities of natural tree samples from different
samples of known ages, mostly from the tombs of the
continents by reducing background counting with lead and
Egyptian kingdoms, and published them as the Curve of
iron shielding. They found worldwide homogeneous 14C
Knowns (Libby 1964).
activities at around 12.5 disintegrations/min/g of carbon1
(Libby 1981). A precise measurement of the 14C half-life,
and the specific activity of samples of known ages were
Principle of the Method
further undertaken to validate the 14C dating method.
Carbon-14 is formed in the upper atmosphere, where it is
14 rapidly oxidized to form 14CO2 molecules. All living matter
Estimation of the Half-Life and the First C
contains carbon, and, thus a very small proportion of 14C.
Dating
The 14C abundance is of some 1.2  10−10% (or
1.2  10−12 g of 14C per g of carbon), while those of the
The half-life of 14C was determined through many experi-
isotopes 13C and 12C are respectively 1.108% and 98.892%.
ments (Engelkemeir et al. 1949; Olsson et al. 1962). One of
The 14C exchanges between the living material and their
them consisted of measuring, by mass spectrometry, the
environment cease at the death of the animals or plants. The
time (t) since the death can be measured by comparing the
residual specific activity in dead organisms to that of the
1
The specific activity is now determined at 13.56 ± 0.07 atmosphere.
disintegrations/min/g of carbon.
54 M. Paterne et al.

The radiocarbon age (t) is calculated using the radioactive Calendar Age BP (years)
exponential decay: -450 -400 -350 -300 -250 -200 -150 -100 -50 0 50
500
14
14
C C
12 C
¼ 12 C
ekt Maunder Dalton
0
400
where (14C/12C)0 is the atmospheric ratio and k = ln(2)/T1/2

Raidocarbon Ages (years)

the decay constant. By convention, the Libby’s half-life at
5568 years is used to calculate the 14C ages. The mean 300
lifetime of the 14C atoms before decay is T1/2/ln(2).
The calculation of age becomes: Suess effect

14
! 200
1 C
12 C
t ¼ ln 14 C
k 12 C 0
100
The 14C dating method is based on the hypothesis of a
constant radioactive equilibrium between the 14C formation
and its disintegration in 14N. If we look at Libby’s diagram 0
1500 1600 1700 1800 1900 2000
(Fig. 4.1), we may observe that this is true if the production
Calendar AD (years)
of 14C, the size of the various reservoirs of carbon (atmo-
sphere, oceans, land and marine biosphere) and their carbon Fig. 4.2 Variations of the 14C ages as a function of calendar ages in
content remain constant over time, as well as the fluxes tree-rings. The time interval of the ‘Suess effect’, the Maunder and
between the various reservoirs. In addition, the physico- Dalton minima of solar magnetic activity are shown. The last two
chemical integrity of the dated fossils must have been pre- coincide with a rapid decrease of the atmospheric 14C ages due to lower
filtering of cosmic protons by the solar magnetic field. The ‘Suess
served after the death of organisms. For example, no isotopic effect’ corresponds to the increase of the 14C ages due to the dilution of
exchange or secondary crystallization should have occurred. atmospheric 14CO2 by the industrial and domestic emissions of CO2
Finally, the samples should have not moved from their into the atmosphere from the 14C-free fossil fuels
burying sites to date precisely any events.
Besides the anthropogenic changes of 14C, the natural
variations of 14C in the different carbon reservoirs were then
Validity of the Assumptions and Definition identified by comparing the 14C and dendrochronological
of a Reference Standard for the Atmosphere ages over the past millennia (Damon et al. 1978; Stuiver
et al. 1991). They are attributed to changes in 14C production
The first offsets between the 14C and known ages appeared in the upper atmosphere and to variations in the natural
very quickly, notably with the major contribution of den- carbon cycle linked to the size of the carbon reservoirs, their
drochronology, a counting method of the annual tree-ring composition and the carbon exchange fluxes.
growth. In 1955, Suess demonstrated (Taylor 1987; Damon De Vries observed rapid fluctuations (wiggles) in the
et al. 1978) that the 14C content in the atmosphere varied in atmospheric 14C from tree-ring analyses (Damon et al. 1978;
the last hundred years and decreased from 1890 AD to 1950 Stuiver et al. 1991). He attributed these fluctuations to
AD (Fig. 4.2). He suggested that the decrease was the result changes in climate and solar activity, as both the Little Ice
of the CO2 release into the atmosphere from the domestic Age (about 1560 AD to 1830 AD) and the Maunder and
and industrial combustion of 14C-depleted fossil fuels (coal, Dalton minima, two time intervals of few sunspots, occurred
oil). These annual emissions, approximately 150Gt of C as over this time.
CO2 until 1950 AD, were responsible for a 14C aging of the The charged particles released from the Sun, known as
atmosphere of about 160 years between 1890 AD and 1950 the solar wind, create a magnetic field in the interplanetary
AD, the so-called ‘Suess effect’. In 1957, Rafter and Fer- space. These eruptions are a manifestation of the magnetic
gusson observed a rapid atmospheric 14C increase that they activity of the Sun, with a minimum dipolar magnetic field
attributed to the 14C production during the aerial atomic corresponding to the maximum equatorial activity, and their
bomb tests. These peaked between 1960 and 1961 and intensity is reversed every 11 years. The solar magnetic field
doubled the 14C concentration in atmospheric CO2. This varies with a cycle of 22 years, reversing its polarity every
atmospheric 14C spike led later to life-size experiments to 11 years at every change of solar activity. The particles of
monitor carbon exchanges between the various earth the galactic cosmic rays are deflected by the solar wind. The
reservoirs. higher the solar magnetic activity, the fewer particles
4 Carbon-14 55

penetrate into the upper atmosphere. As a result, the 14C standard, i.e. a reference value of 14C for the atmosphere
production decreases as the solar activity increases, and vice (N0), and the calibration process which precisely quantifies
versa. Similarly, the cosmic ray intensity in the upper the difference between the true or ‘absolute’ ages and the 14C
atmosphere is modulated by the changes in the Earth’s ages.
magnetic field, which acts as a shield against the As requested by Arnold in 1956, the National Bureau of
electrically-charged cosmic protons. The greater the intensity Standards (Washington) prepared a standard (NBS-I) of
of the geomagnetic field, the fewer the number of cosmic 450 kg of oxalic acid (HOOC-COOH), an organic com-
protons arriving in the upper atmosphere, and therefore the pound extracted from a French crop of sugar beet in 1955
production of 14C is lower in the atmosphere (Damon et al. (Arnold and Libby 1949). It provided the reference activity
1978; Stuiver et al. 1991). Changes in the Earth’s magnetic of year zero, from which the 14C age of a sample is calcu-
field could explain about 50% of the 14C variation in the lated. The reference activity was taken at 95% that of NBS-I
atmosphere between the last glacial and the Holocene, and to account for the Suess and nuclear bomb effects and the
the 14C aging of the atmosphere during the last glacial reference year has been set arbitrarily at 1950. Since that
related to a high magnetic field intensity (Lal and Charles time, other standards were prepared (NBS-II, sucrose). The
14
2007). C ages are expressed in years BP (Before Present, present
The dendrochronological record of the atmospheric 14C being equal to 1950 AD). In archeology, the terms Anno
content over the last 10,000 years has, in turn, given the Domini (AD) and Before Christ (BC) are used. Lately
opportunity to evaluate the changes of solar activity beyond appears the term Common Era (CE) which is equivalent to
the first observations in the seventeenth century. Spectral AD and BCE to BC.
analyses of the atmospheric 14C emphasize long cycles of Before measuring the 14C activity in a sample and cal-
88 years, 208 years and 2050 years attributed to changes in culating its 14C age, the fractionation of the stable isotopes
solar activity (Damon and Peristykh 2000). During the last of carbon must be considered. Craig in 1953 demonstrated
70 years, the solar activity was at an exceptionally high level that the 13C/12C ratios (d13C) vary in contemporary materials
and many other periods of high activity, although shorter, as a function of the reservoir’s d13C in which they form. The
occurred in the past, such as, for example, at the beginning amount of enrichment/depletion of 14C due to biological and
of the Holocene. In addition, some authors have thought that physicochemical fractionation processes is approximately
the unusual solar activity may have contributed, in a small two times that measured by d13C in the same sample. The
part, to the recent climate change observed in the late d13C value, expressed in ‰ compared to the standard PDB
twentieth century (Muscheler et al. 2005). (Pee Dee Belemnite), is about −6.5‰ for the pre-industrial
Climate changes modify the size of the carbon reservoirs atmospheric CO2. It varies between −2‰ and +3‰ in the
and the CO2 fluxes between them and therefore the atmo- carbonates of seashells and between −20 and +3‰ for those
spheric 14CO2 as noted by de Vries (Stuiver et al. 1991). of lake shells. The d13C values of plants vary between −27
Libby (1952) had already estimated the impact of climate and −14‰ due to photosynthesis processes. Processes of
changes on the 14C ages. He assumed that the sea level assimilation of atmospheric CO2 by plants are carried out
lowering (*100 m) and the temperature decrease during the mainly according to two cycles of transformation of organic
last ice age reduced the oceanic carbon inventory, resulting compounds, called the Calvin cycle and the Hatch and Slack
in an increase of the specific activity of 14C (Fig. 4.1). cycle, or also as C3 and C4, the second with a lower isotope
Assuming that the 14C activity in the reservoir exchange discrimination than the first. As a result of the great vari-
changed by 10%, Libby calculated that the glacial 14C ages ability of the d13C values, the 14C activities have been
would be too young by some 800 years. We know now that normalized to a common reference of d13C set at −25‰ (see
the 14C ages are too young by about 2000 years with respect below). The complete procedure for calculation of age is
to ‘true’ ages during the last glacial maximum (Reimer et al. explained in Box 1.
2013). The ocean circulation and the carbon cycle were
deeply modified during the last ice age that modulated the
concentration of atmospheric CO2. Between the last glacial Box 1
maximum and the Holocene, the atmospheric CO2 concen- As in most definition of isotopes, the 14C content is
tration increased from 190 to 280 ppm as recorded in the expressed by a d in ‰, which defines the difference
Antarctic ice cores. between a sample and a standard, which may be the
The growing evidence of the atmospheric 14C variations NBS-I standard, the d13C of which is equal to −19‰
through time led to two consequences: the establishment of a relative to the PDB (Broecker and Olson 1959; Olsson
56 M. Paterne et al.


and Osadebe 1974; Stuiver and Polach 1977; Sten- 14 14 13 d14 C
ström et al. 2011). D C ð&Þ ¼ d C  2 d CS þ 25 1 þ
1000
The d14C may be written as follows:
ð4:6Þ

14 As 14
d C¼  1  1000 ð4:1Þ The calculation of the C age is then:
AOx 
1 1
With As and AOx, the respective activities of the tðyearsÞ ¼  ln 14
k 1 þ D C=1000
sample and of the standard NBS-I of oxalic acid.
" # In the case of precise measurements of activity
2  19 þ d13 COx performed in oceanography or to calibrate the 14C
AON ¼ 0:95  AOx 1  ð4:2Þ
1000 ages, it is necessary to consider the 14C decay between
the age of the sample (x) and its measurement (y) us-
The 14C activity in the atmosphere in 1950 (AON) is ing the period T of 5730 years. Equations 4.1 and 4.6
equal to 95% of the NBS-I standard activity, corrected then become:
for d13C (Eq. 4.2). The isotopic fractionation of d13C, 
which affects the abundance of the mass 14, is noted ASN ekðxyÞ
d14
n C ¼  1  1000
by the number 2. It indicates that the fractionation AOxN ekðy1950Þ
between the mass 14 and mass 12 is double that
between the masses 13 and 12. The negative sign since the 14C activity of the sample decreases at the
assigned to fractionation d13C means that the 14C same rate as the standard
activity measured in organisms with a negative d13C, 
ASN ekð1950xÞ
that is to say with less affinity for the mass 13 than the d14
n C ¼  1  1000
PDB standard, must increase to compensate for the AOxN
loss caused in 14C by fractionation (Broecker and and
Olson 1959; Olsson and Osadebe 1974; Stuiver and

Polach 1977; Stenström et al. 2011). d14 C
d14C is defined as: D14 C ð&Þ ¼ d14 C  2 d13 CS þ 25 1þ
1000

As As the activity of a sample in 1950 is the same than
d 14 C ¼  1  1000 ð4:3Þ
AON that of the standard, then D14C is zero in 1950 AD.
Now, the activity of a sample is expressed as
To take into account the variability in the frac-
fraction modern (F) that represents the 14C/12C ratio in
tionation of isotopic d13C measured in the samples,
a sample such that:
d14C is normalized to a common value, d13C, set at
−25‰ versus PDB, regardless of the sample type ASN
(carbonate, dissolved inorganic carbon or organic F 14 C ¼
AON
matter). This value was obtained by averaging the
measurements of the isotopic ratio d13C of several In oceanography, the 14C concentration is expres-
pieces of wood with an age less than 1890 AD sed as Δ14C or Δ in ‰:
(Broecker and Olson 1959; Olsson and Osadebe 
14 ASN
1974). D C¼  1  1000ðwithout age correctionÞ
AABS
We define:
 
14 ASN ASN ekðyxÞ
D C ð &Þ ¼  1  1000 ð4:4Þ D¼  1  1000ðwith age correctionÞ
AON AABS

with
with AABS is the absolute age, y the year of mea-
! surement, x the year of growth, and k = ln(2)/5730.
2  25 þ d13 CS
ASN ¼ AS 1  ð4:5Þ The notation Δ is often written as Δ14C-age corrected.
1000
4 Carbon-14 57

14
Calibration of the C Ages foraminifera in the Cariaco sediments in the Caribbean Sea
(Fig. 4.4). The sea surface 14C is not in equilibrium with that
It is now well-established that the 14C ages are not equal to of the atmosphere. The comparison of the atmospheric and
the absolute ages because (i) the 14C ages are calculated with marine 14C dating at a same calendar age between 0 and
the 5568 year half-life, (ii) the 14C concentration in the 13,900 cal BP allowed the atmosphere-ocean 14C difference
atmosphere varies as a function of time due to changes in the (the marine reservoir age-see below) to be quantified and the
production rate by cosmic rays which are modulated by the marine calibration to be anchored to the atmospheric record
solar and earth magnetic fields and (iii) of the changes in the by subtracting the reservoir age to the marine 14C ages. In
carbon cycle. The 14C calibration consists of the precise IntCal13, the variability of the reservoir ages during the
measurement of the difference between an absolute (calen- glacial period were taken into account by augmenting the
dar) age and a 14C age. An international group of scientists value by 200 years prior to 13,900 cal BP. Similarly, the 14C
led by Stuiver and Reimer joins efforts to iterate the cali- in speleothems is not equal to that of the atmosphere, and the
bration datasets. The last product is the IntCal13 calibration, difference (dead carbon fraction—see below) is measured
which extends over the past 50,000 years (Reimer et al. between the time interval 0–13,900 cal BP and then con-
2013). sidered constant prior to 13,900 cal BP.
As initiated with IntCal04, the mathematical approach
that defines the envelope of IntCal13 takes into account both
Methods and Results uncertainties on 14C dating and those associated with the
absolute ages. In addition, it also takes into account the
The calibration procedure consists in measuring the 14C age of diversity of records and the representativeness of atmo-
a sample while the absolute ages are determined by three spheric 14C by assigning different statistical weights to them
methods with the best possible accuracy. The first is based before returning the most likely fit between measured 14C
either on the counting of annual tree-ring growth (den- activity and absolute age. The IntCal13 error envelope rep-
drochronology) or on the counting of the annual laminae resents the best fit between the included datasets and it is
(varves) deposited in marine or lake sediments. The second much smoother than those of the previous calibration
consists in the precise U-Th dating of carbonates (warm water records. It does not take into account the scattering of the
corals or speleothems) by mass spectrometry (Chap. 6). The
14
C ages in the different datasets, which do not permit ‘real
14
third consists in synchronizing the variations of climate C variations’ to be discriminated from measurement noise.
proxies in marine sediments and in speleothems, the last two Those real variations are likely to be attributable to the
being dated by the U-Th method. complex history of each carbon reservoir in either the ter-
In the eighties, the 14C and U-Th dating benefitted from restrial (atmospheric) or marine environments over the past
new techniques in mass spectrometry coupled to an accel- 50,000 cal year due to climate and oceanic circulation
erator or by thermal ionization (TIMS), respectively—which fluctuations, and to production changes through the varia-
allowed their precision to be greatly improved and the size of tions of the Earth and solar magnetic field.
samples to be reduced by a factor of 1000 (from one gram to a
few milligrams). Bard had first proposed to extend the cali-
14
bration from 10,000 to 22,000 cal BP through paired 14C and Examples of Precise Calibration of C Ages
Th-U dating of corals from the Barbados islands (Fig. 4.3)
(Bard et al. 1990; Reimer et al. 2013). The calibration record The Dating of the Eruption of Santorini
IntCal13 is based on the 14C dating of the terrestrial and During the eruption of Santorini in Cyclades in the Aegean
marine samples referenced onto an ‘absolute’ timescale. It Sea, huge amounts of volcanic ash were emitted into the
extends up to 13,900 cal BP by means of dendrochronology atmosphere and spread eastward and southward, covering
and up to *50,000 cal BP from terrestrial plant species much of the Middle East. The ash bed is a useful chrono-
deposited in the varved sediments of Lake Suigetsu. It is logical marker for the whole region, allowing the chronology
completed by U-Th-dated samples and has corrected 14C ages of human connections to be refined throughout the eastern
(see below) of the Hulu speleothem and of Atlantic and basin of the Mediterranean Sea during the second millen-
Pacific corals. Finally, corrected 14C ages of foraminifera in nium BC. Archaeologists, for example, linked the apogee of
the varved and unvarved sediments from the Cariaco basin the Minoan civilization to that of the New Kingdom in Egypt
linked to the Hulu speleothem by coeval climatic fluctuations in the sixteenth century BC, on the basis of the elegance of
are also included. The marine calibration record similarly the decorations of objects found in the ash layer in Santorini.
covers the past 50,000 cal BP and it is based on the dating of The date of this eruption could not be defined to within
corals from the Atlantic and Pacific oceans and of planktonic 60 years, despite hundreds of 14C dating of charcoals found
58 M. Paterne et al.

Fig. 4.3 Variations between 14C 35000

and calendar ages (Reimer et al.
Dendrochronology
2013). The calendar ages are
based on tree ring counting back Barbados
to 11,950 cal BP and on the U-Th Mururoa
30000
dating of corals from Barbados
and Tahiti (in Reimer et al. 2013).
The 14C age plateau are shown in
the inset between 9000 cal BP
25000

Radiocarbon ages (years BP)

and 12,000 cal BP

11000
20000

Radiocarbon ages (years BP)

10500

10000
15000

9500

10000 9000

8500

5000 8000
9000 11000
Calendar Ages (years BP)

0
0 5000 10000 15000 20000 25000 30000 35000
Calendar Ages (years BP)

14
in the ash-bed (Friedrich et al. 2006). The discovery of an CO2 concentration is very sensitive to the rate of formation
olive branch with leaves buried in the ash bed enabled a very of deep water, because the CO2 in the atmosphere is trans-
precise dating of the eruption. Friedrich et al. (2006) dated ferred to the deeper layers of the ocean via the oceanic cir-
the annual tree-rings of a piece of branch (Fig. 4.5). The culation. This rate also regulates the heat transfer to the
variation (wiggle) of the 14C ages in the tree-ring matched northern high latitudes. During the cold event of the
precisely that of the calibration record, the number of cal- Younger Dryas, which lasted approximately 1200 calendar
endar years being equal to that of the tree-rings. years, the atmospheric Δ14C increased (younger 14C ages)
The calendar age of the eruption of Santorini was thus during the first 200 years and then decreased during the
estimated to be between 1627 and 1600 BC, with a confi- following 1000 years (Fig. 4.6a). The increase of Δ14C is in
dence level of 95%. The apogee of the Minoan civilization agreement with a reduction in the deepwater formation in the
would then be earlier than that of the New Kingdom, and northern North Atlantic, contributing to a cold climate, but
would be contemporary to the time when foreign kings— how may the further Δ14C decrease be explained when the
Hyksos—occupied the country (Friedrich et al. 2006). climate is still cold in the northern latitudes?
Glaciologists have shown that the abrupt cooling
The Bipolar Seesaw or the North-South Heat observed in the northern hemisphere was accompanied by a
Transfer warming in Antarctica, and vice versa (Blunier et al. 1998).
Rapid fluctuations of climate, referred to as Heinrich events, Comparing these two results, Broecker (1998) proposed a
Younger Dryas, and Dansgard-Oeschger events, punctuated seesaw pattern of the north-south heat transfer via the ocean
the last glacial period and the deglaciation. or the “thermal bipolar seesaw” (Fig. 4.6b): a reduction in
They were attributed to variations in the thermohaline deep water formation in the North Atlantic Ocean led to the
oceanic circulation (see Chap. 21). The mechanisms of these installation of a cold climate in the northern hemisphere and
rapid fluctuations were clarified by the analysis of the 14C an increase in the atmospheric 14C. Deep waters further form
variations in the varved deposits in the Cariaco Basin in the Southern Ocean contributing to the atmospheric Δ14C
(Hughen et al. 1998; Reimer et al. 2013). The atmospheric decrease.
4 Carbon-14 59

Fig. 4.5 14C and calendar ages of the Santorini eruption (modified
Fig. 4.4 Variations between 14C and calendar ages between 20,000 from Friedrich et al. 2006). Four groups of rings counted (grey
and 50,000 cal BP from the calibration record IntCal 13 (Reimer et al. rectangles) from an olive branch of an approximately ten centimeters
2013). The straight line represents the 1:1 relation length were dated. The total number of rings equals the number of
calendar years. The 14C ages match the 14C variations of the calibration
Apparent Ages record (Reimer et al. 2013). The calendar age of the eruption range
between 1621 and 1605 BC at the 1r confidence level (68.2%) and
between 1627 and 1600 BC at the 2r confidence level (95.4%)
Oceanic Environments: The Reservoir Ages
Carbon exists in several forms: carbon monoxide CO and
carbon dioxide CO2, methane CH4, many complex organic The 14C aging of deep waters is the result of the conveyor
molecules (cellulose, etc.). The dominant forms are the (thermohaline) circulation. In the modern ocean, wind-
bicarbonate and carbonate ions (CO32−, HCO3−, H2CO3) driven warm saline surface waters from the North Atlantic
dissolved in water, and calcium, barium and magnesium low latitudes flow poleward where they cool and sink to
carbonates (CaCO3, BaCO3, MgCO3). Most of the Earth’s form the North Atlantic Deep Water (NADW). NADW
living species directly assimilate the 14CO2 from the atmo- flows at depth to the Indian and Pacific Oceans where it
sphere. In the ocean, the organisms incorporate 14C, which is mixes and upwells then flowing back to the North Atlantic.
not in equilibrium with the atmospheric 14CO2. The benthic All the deep waters flow to the Southern Ocean where they
foraminifera are calcareous organisms living on the surface upwell and downwell to form a return flow to the north
of marine sediments. When collected alive at a water-depth through the intermediate waters and the Antarctic bottom
of 2000 m in the North Atlantic and Pacific oceans, they will waters. Once isolated from the atmosphere, the dissolved
14
have 14C ages of about 600 years and 1400 years, respec- CO2 in the intermediate and deep waters starts to decrease.
tively (before the nuclear bomb tests). These 14C ages are Although in contact with the atmosphere, the sea surface 14C
apparent ages that include the age of the organism and that is not equal to the atmospheric 14C, because of the mixing
of the water body in which the organisms formed their tests between the sea surface and the underlying sub-surface
from the dissolved ions according to: waters. The difference between the apparent 14C ages of
living organisms in the ocean and the atmospheric 14C ages
CO2 þ H2 O $ H2 CO3 $ H þ þ HCO
3 for the same absolute age is termed the reservoir age R. The
HCO þ 2
3 $ H þ CO3
sea surface values of R as well as those of the intermediate or
deep waters differ in each one of the ocean basins and
ðCa2 þ þ 2HCO
3 ! CaCO3 þ H2 O þ CO2 Þ: regionally within the basins.
60 M. Paterne et al.

Fig. 4.6 a Variations in the

atmospheric 14C content (Δ14C in
‰) (triangle) and grey scale (line)
in marine varved sediments from
the Cariaco Basin as a function of
calendar ages by varve counting
(Hughen et al. 1998; Reimer et al.
2013). b Variations of d18O
trapped in CO2 air bubbles in the
Greenland ice core GISP2 sites in
Greenland (line) and in the Byrd
ice core in Antarctica (diamond)
as a function of synchronized
time from analyses of CH4
concentrations in the two sites
(Blunier et al. 1998)

δ
δ

The modern values of R are quantified by the 14C dating marine organism from that of the atmospheric calibration
of carbonates (mollusk shells, corals) of known ages from record at the same absolute age (Reimer et al. 2013). The
the historic collections of museums (Bard 1988; Siani et al. mean value of R is about 300–400 years in the subtropical
2000; Tisnérat-Laborde et al. 2010) and, for older ages than sea surface, and it may be some 1100 years around the
those of these collections, by subtracting the 14C age of a Antarctic continent. Values of approximately 700 years were
4 Carbon-14 61

measured in oceanic regions close to upwelling of interme- tends to increase the 14C age in lake waters relatively to that
diate and deep waters in the North Indian and Pacific of the atmosphere.
Oceans. DR is the local departure from the global mean The speleothems are composed of calcium carbonates,
value of R. The latter is calculated from the modeled global formed by the dissolution of 14C-free geological carbonates
sea surface 14C age, using a global box-diffusion carbon by slightly acidified waters by CO2 from the atmosphere and
model, which accounts for the14C production and carbon from the degradation of the soil organic matter. In Fig. 4.7
cycle (Reimer et al. 2013). the procedure to estimate the fraction of dead carbon
In the past, the oceanic circulation changed as the reser- (DCF) in a Bahamas speleothem, located in the western
voir ages did. Changes of the differences between the marine North Atlantic, is shown (Beck et al. 2001). To an U-Th age
and atmospheric 14C ages may be estimated by dating the corresponds an atmospheric 14C age in the calibration
charcoals and foraminifera from widespread volcanic ash record. The DCF is calculated by subtracting the 14C age of
layers in both marine sediments and on land (Bard et al. the speleothem to that of the contemporaneous atmosphere.
1994; Siani et al. 2001; Austin et al. 2011; Thornalley et al. In this example, the 14C ages of the speleothem are older
2011; Siani et al. 2013; Sikes and Guilderson 2016), as well than those of the atmosphere at an average of about
as by dating paired charcoals and mollusk shells found in the 1450 years between 11,000 cal. BP and 15,000 cal. BP,
same deposits (Bondevik et al. 2006; Ascough et al. 2009). which corresponds to a DCF of 16%. This aging is not
These studies demonstrated that sea surface R changed constant as a function of time and varies between 1000 years
during the first step of the deglaciation from the modern and 2000 years. These variations appear to be closely cor-
value of 400 to 2000 years in the North Atlantic (Bard et al. related to the climatic fluctuations recorded either in the
1994; Bondevik et al. 2006; Austin et al. 2011; Thornalley Greenland ice-core GISP2 or in the marine varved sediments
et al. 2011) and to 800 years in the Mediterranean Sea (Siani of the Cariaco Basin. About 30% of the variability of the
et al. 2013). In the South Pacific, the subtropical R aug- DCF in this speleothem may be explained by such fluctua-
mented from *300 to *700 years (Sikes and Guilderson tions by the way of changes of the local temperature and
2016), while those of the sub Antarctic surface waters rainfall patterns.
increased from *800 to 1400 years and to 3200 years
(Siani et al. 2013; Sikes and Guilderson 2016).
14
Past increases in the sea ice extent at high latitudes during The C Exchanges in the Carbon Reservoirs
the glacial periods prevents the atmosphere-ocean 14CO2
exchanges contributing to the increase of the sea surface R. Radiocarbon is commonly used to test numerical simulations
In the IntCal13 calibration record, the modern value of R of the oceanic circulation in Ocean General Circulation
was subtracted to the 14C ages of the marine samples to be Models (O-GCM) (Toggweiler et al. 1989; Key et al. 2004).
compared to the atmospheric 14C ages. To account for The systematic measurements of the 14C content of the
changes in the oceanic circulation, the modern value of R dissolved inorganic carbon (DIC) in the different world
was augmented by 200 years from *14,000 cal BP and ocean basins started with the international oceanographic
then considered as constant down to 50,000 cal BP (Reimer campaigns GEOSECS (1972-1978), followed by WOCE
et al. 2013). (World Ocean Experiment, 1990–2002), and others
(Broecker et al. 1995; Key et al. 1996). In addition to 14C
Continental Environments: The Hard Water analyses, the physicochemical properties of the worldwide
and Dead Carbon Effects basin water masses were measured along depth profiles.
Modern lake vegetation and calcium carbonates in lakes in Both the natural and anthropogenic components of the 14C
calcareous regions exhibit older 14C ages than those of the concentration in the oceans offer the opportunity to validate
atmosphere. The dissolved inorganic carbon 14CO2 in lakes, the general ocean circulation simulated by numerical models
used during photosynthetic processes and during the pre- (Toggweiler et al. 1989; Key et al. 2004). The natural
cipitation of calcium carbonates originates from the disso- component tests the circulation of deep waters, while the
lution of 14C-free carbonates of geological age (dead carbon) anthropogenic component, resulting from the thermonuclear
and from the mineralization of old organic matter enclosed tests in the 1960s, allows the analysis of physical processes
in lake sediments. Impact of the later on the apparent 14C with time constants of a few decades, such as the formation
ages of lacustrine plants is much higher in lake surrounded of deep and intermediate waters and the ventilation of the
by peats and in artificial lakes implemented by soils flood- thermocline (transition zone between the cold intermediate
ing. The resulting aging of 14C ages is called ‘hard water’ and deep waters and the warm surface waters). Because of
effect. In addition, water stratification in lakes or the pres- radioactive decay, the 14C content of an ocean water mass
ence of an ice cover in high altitude and polar lakes tend to decreases during the oceanic transport, once the water mass
prevent the atmospheric 14CO2 input to the lake waters, that is isolated from exchanges with the atmosphere.
62 M. Paterne et al.

Fig. 4.7 Variations of the 14C 15000

ages as a function of U-Th A
calendar ages from a Bahamas
speleothem (Beck et al. 14000

Radiocarbon ages (years BP)

2001, Reimer et al. 2013). a The
black squares represent the 14C
ages from the speleothem and the 13000
white squares to the 14C ages of 1450 years
the speleothem adjusted onto the
atmospheric calibration record 12000
Dead Carbon
(grey squares). The mean dead
carbon age is 1450 years.
b Comparison of the variations of 11000
dead carbon (bold line) (Paterne,
unpublished) and of GISP 2 d 18O
(Blunier et al. 1998) 10000

9000
11000 12000 13000 14000 15000 16000

Calendar ages BP (years)

Calendar Ages BP (Years)

8150 9150 10150 11150 12150 13150 14150 15150 16150

B
-34
2200
Dead carbon (years)

GISP2 δ18O (‰)

-36

1700

-38

1200
-40

700 -42
7850 8850 9850 10850 11850 12850 13850 14850 15850

Calendar Ages BP (years)

The 14C provides independent and additional constraints of the ventilation of the deep ocean producing concentra-
for ocean circulation models validation beside conventional tions similar to observations. In particular, the ventilation
tracers of the water masses (temperature, salinity), by signature of the deep ocean associated with the Antarctic
allowing estimates of the time constants of the ocean deepwater formation coming from the Southern Ocean
ventilation. (Antarctic Bottom Water AABW) is reproduced, character-
ized by high values of Δ14C at the ocean floor.
The eastern Pacific is characterized by a thick water mass
Examples of Simulation of Modern Oceanic between 2000 m and 3500 m of almost homogeneous
Circulation 14
C-depleted values at around −200 and −240‰. These
deep waters correspond to the oldest deep waters of the
Figure 4.8 represents the 14C distribution simulated with the world ocean with 14C ages of 1790 years and 2200 years
NEMO global ocean circulation model. The simulation is respectively.
evaluated using the GEOSECS data (Broecker et al. 1995; The temporal changes of the sea surface 14C may be
Key et al. 1996). The model reproduces a realistic structure measured from the annual growth bands of recent corals over
4 Carbon-14 63

Natural 14C ORCA

0 0

-80 -80
1000 -120
1000
-120
Depth (m)

-140 -140
-160 2000 -160
2000
-180 -180
-200 -200
3000 -240 3000 -240

4000
4000

5000
5000
-40 -30 -20 -10 0 10 20 -40 -30 -20 -10 0 10 20

Latitude Latitude

Fig. 4.8 Observed and simulated 14C concentration (in ‰) in the Pacific Ocean from depths between 2000 and 3500 m. At about 40 °S,
eastern Pacific Ocean. The negative value indicates an aging of the the presence of a more recent water mass, between −180 and −200‰ at
water bodies compared with the age of the atmosphere. A decrease of 4000 m depth, formed on the edge of the Antarctic continent should be
10‰ is equivalent to an aging of about 80 years. A mass of nearly noted. It corresponds to the Antarctic Bottom Water (AABW)
homogeneous water between −200 and −240‰ occupies the eastern

200
B

100
(‰)
14C

-100
1950 1960 1970 1980 1990 2000
Year A.D.

Fig. 4.9 Variations in D14C in ‰ (circles) in a banded coral collected interannual OPA simulation (INT) (8.1) are represented. This simula-
near Bermuda (Druffel 1989), in the subtropical surface waters of the tion highlights the importance of the winds in the atmosphere-ocean
North Atlantic between 1950 and 1990. Δ14C increased rapidly exchanges in the subtropical North Atlantic (Tisnérat, Dutay, personal
between 1960 AD and 1965 AD due to the 14C input into the communication)
atmosphere during the aerial nuclear bomb tests. Results of the

several decades. These data provide additional validation of model simulation of 14C data measured in one coral from
general circulation models, which, in turn, allow a better Bermuda shore (Druffel 1989).
identification of the causes and mechanisms of the temporal Both the 14C coral data and the modeled 14C at the sea
variability of 14C. Figure 4.9 represents the ORCA/IPSL surface show coeval changes, indicating that the model
64 M. Paterne et al.

correctly reproduced the transfer of the tracer at the air-sea ten to hundreds of meters of the water column in which they
interface, as well as its penetration below the sea surface into lived. More constraints on the changes of the sea surface
the ocean. reservoir ages allowing robust calculation of Δ14C from
benthic foraminifera and more 14C analyses from deep sea
corals will be very useful for model simulations of the
Oceanic Paleocirculation oceanic circulation during the last glacial maximum and the
deglaciation using three-dimensional models (Tagliabue
In the meantime, from the systematic 14C measurements in et al. 2009). In the study by Zhao et al., (2018), the box
the world ocean layers, the idea quickly emerged to recover model resolution, which should at least take into account the
past changes in ocean circulation from 14C analyses of changing geometry of water masses in the last 25 kyr
marine biocarbonates. The 14C concentration of large vol- (Michel et al. 1995), is a limitation to estimate regional
ume samples such as corals and mollusks shells were first changes in paleoventilation.
measured by radioactive decay measurements (Stuiver et al.
1986; Druffel 1989). This could be achieved for small vol-
ume samples with the new AMS 14C technique. The Mineralization of Organic Matter in Soil
planktonic and poorly abundant benthic foraminifera con-
temporaneously deposited in deep-sea sediment cores from The contribution of soils and their role as sinks and sources
different water-depths could henceforth be easily dated in the global carbon cycle remain misunderstood until now.
allowing the paleo-ventilation of the paleocean layers to be The stock of soil organic matter is defined as a balance
recovered to the limit of the 14C dating (theoretically 10 between the input of organic matter through vegetation and
times the half-life). Meantime the changes of the Earth’s the loss through microbial decomposition. The balance can
orbital parameters, and thus changes of the insolation, the be disrupted by changes in agricultural practices and climate
atmospheric and oceanic circulation changes modulate the variations. For instance, a temperature increase may clearly
Earth’s climate. Considerable efforts, compiled in (Zhao increase the activity of soil microorganisms and the subse-
et al. 2018), have been made to measure the 14C differences quent soil organic matter mineralization. No consensus has
in the deep to surface ocean water masses in the past from however been reached on the relative importance of the
benthic-planktonic foraminifera. Benthic corals have pro- various climatic factors that affect soil organic matter
vided additional estimates of the deep Δ14C as a function of dynamics, such as temperature, aridity, land use. To better
calendar ages as they can be dated by both the U-Th and 14C evaluate the effect of these disturbances on the global carbon
methods (Adkins et al. 1998; Goldstein et al. 2001; Robin- cycle, it is essential not only to characterize soil carbon
son et al. 2005; Burke and Robinson 2012; Chen et al. stocks but also soil carbon dynamics. To do so, 14C is a
2015). Opposite results in the estimates of the powerful tool as it can be considered as a clock that registers
paleo-ventilation of the ocean during the deglaciation the carbon residence time in the soil organic mixture
emerged when using either the 14C dating of paired (Scharpenseel and Shiffmann 1971; Balesdent and Guillet
benthic-planktonic foraminifera or paired 14C/U-Th dating 1982).
of benthic corals (Adkins et al. 1998; Goldstein et al. 2001; Conceptual views of soil organic carbon dynamics have
Robinson et al. 2005; Burke and Robinson 2012; Chen et al. greatly evolved with time. Carbon sequestration was con-
2015) or the paired atmospheric-marine 14C dating from sidered to be related to the chemical structure of the com-
ash-layers (Sikes et al. 2000; Ikehara et al. 2011; Siani et al. ponents (lignin having a longer mean residence time than
2013; Ezat et al. 2017). Values of Δ14C are obtained from sugar), to the accessibility of organic matter in aggregates
the absolute (calendar) age and the 14C age (see Box 1). (the higher the pore, the more labile the organic matter), to
Thus the conflicting results are very likely related to the the affinity between organic matter and mineral surface (the
estimated Δ14C from paired benthic-planktonic foraminifera stronger the bond, the more refractory the components) (Six
due to (i) an incorrect estimate of the sea surface reservoir et al. 2002). The conceptual view is still developing (Kleber
age subtracted to marine 14C ages (in order to be referenced et al. 2007) and 14C brings powerful elements. 14C mea-
to calendar ages), (ii) the use of different atmospheric 14C surement is thus done on bulk organic matter, on density
calibration records, notably the much smoother variations of fractions, on granulometric fractions, on molecular fractions
14
C in IntCal13 than those estimated in previous records and on molecules according to the process or to the turnover
(Reimer et al. 2013), (iii) the use of 14C values of for- to be characterized.
aminifera picked in deep-sea sediment cores with a low Isotopic methods, such as dating by carbon-14, natural
sedimentation rate (see below: the bioturbation effects), and (percentage of plants in C3 and in C4) and artificial
(iv) few planktonic foraminifera spend their entire life at the carbon-13 labeling are very powerful tools in so far as they
sea surface, and their 14C content represents that of the upper make it possible to estimate the residence time of natural
4 Carbon-14 65

organic matter in the soil. Instrumented monitoring of the relative weights of the different fractions, can then be used to
13
C content in experimental soil plots can be used to estimate characterize the dynamics of carbon of the soil and to assess
the mean residence time from a few years to some decades. carbon stocks over time.
Many studies have been conducted since the beginning of It is important to keep in mind that whatever the fraction
the 14C method (Balesdent and Guillet 1982; Gaudinsky and since soil is the result of balance between input and
et al. 2000). At first, they were carried out by radioactive output, the dated sample is always a mixture of components
counting measurements (a few grams of carbon), and then by of different 14C ages. The resulting 14C is only a mean age of
mass spectrometry coupled to an accelerator (AMS), which all components of different 14C age. Likewise, it is important
did not allow the targeted molecular level to be reached, to remind that soil is alive and any molecule is recycled. It
because a few milligrams of carbon were still required. might reach the soil as vegetal sugar and as a source of
Nevertheless, their scope was large, and they identified energy for microbial life be bio-assimilated and metabolized
kinetic pools among the various elements of the soil’s into a microbial lipid. It will however keep the same 14C
organic matter, in other words, compartments that can be signature of the original vegetal sugar. That’s why a
defined by a specific carbon residence time. It has been microorganism molecule can give its old 14C age even if the
shown that the residence time of different carbon compart- microorganism is still alive. We definitively characterize the
ments in soil can range from one to several decades or even mean age of carbon and not the mean age of the molecule we
to a few thousand years for the stable fraction. analyzed.
Using the contamination of plant species by 14C nuclear
explosions in 1960 AD, Gaudinsky et al. (2000) modeled
14
over time the 14C activity of the different compartments of Treatment of Samples and Calculations of C
the soil receiving a constant organic input every year, Ages
according to their residence time (Fig. 4.10). The pool with a
residence time of 10 years reached a maximum activity of Radiocarbon dating is based on either decay-counting from
14
C in 1972, eight years after that recorded in the atmo- gas (CO2, C2H2, C6H6) in proportional or liquid scintillation
sphere. The one with a 50-year residence time recorded a counters or atom-counting of 14C, 13C and 12C from graphite
maximum activity in 1985. Using the results of this mod- targets by mass spectrometry (Accelerator mass spectrome-
eling to interpret the measurements conducted on the sepa- try: AMS). The greatest advantage of the AMS technique is
rate fractions from the same soil, the residence time of the the very small sample size required, nowadays as small as a
different fractions can be estimated. These, along with the few tens of micrograms. At the end of the seventies, the

Fig. 4.10 14C activity in a soil with a constant annual supply of different mean residence time (MRT). We note that the peak of bomb
carbon over time, expressed as D14C in ‰ (Gaudinsky et al. 2000). The detonations is clearly reflected in the more recent fractions (<10 years),
thin black line represents the atmospheric 14C activity in the northern while the effect is diluted in the fractions with MRT of a few decades. It
hemisphere; the bold line represents that of the soil fractions with is virtually nonexistent in older fractions
66 M. Paterne et al.

extracted ions from the graphite by a Cesium gun, were Cuzange et al. 2007). Clearly, such old ages cannot be due to
accelerated to a voltage difference of 2 MeV while the a contamination by the carbonates precipitate on cave walls,
acceleration voltage is nowadays reduced at 0.5 MeV. The as was frequently hypothesized. To increase the 14C age of a
reader is referred to various publications and web descrip- painting of 1 mg carbon and 15,000 years old, for example,
tions on the equipment and measurement methods (Taylor to an age of about 33,000 years would require some 900 µg
et al. 1992). of 14C-free contaminants, which is almost the entire sample
and this would have been seen through visual examination.
The internal contamination of samples, due to the
Physical-Chemical Treatment cleaning treatments, is carefully assessed by measuring the
14
C activity of an old (14C-free) sample (blank), cleaned in a
The samples are first cleaned by physicochemical treatments similar manner to samples of unknown ages. During a
14
to eliminate contaminants. Vegetation samples collected in C-AMS run of measurements, standard references, blank
soils may include old carbonates and living rootlets, and the samples and the samples to date are inserted. The “blank” or
14
C ages would thus increase or decrease respectively. After “background” activity is subtracted from the activity of the
a visual examination, the organic remains (seeds, coal, sample of unknown age. The smaller the sample, the greater
plants) undergo the classic Acid-Alkali-Acid chemical the effect of the internal contamination on the 14C age. This
treatment (noted AAA). The aim is to eliminate contami- contamination varies depending on the nature of the sam-
nants from bacterial decomposition of organic matter since ples. Even if the blank activity is very low, its variability
the burial of the sample. Finally, CO2 is obtained by burning determines the accuracy of the 14C age and the 14C age limit.
the sample into an evacuated quartz sealed tube filled with Consider two samples of carbonate and charcoal, with a
copper oxide at about 800 °C. Carbonate samples (for- 14
C/C activity of 6.88  10−15 (an age of 40,000 years BP)
aminifera, pteropods, corals, speleothems, earthworm gran- and an absolute error of 5%. The blank variability accounts
ules) are examined under the microscope to check their for 25% and the subtracted blank value for the carbonate and
homogeneity. Corals and mollusk shells are pre-cleaned by charcoal is 5  10−16 and 19  10−16, respectively. If the
sand blasting to eliminate secondary calcite precipitation, blank variability increases by a factor of 2, the 14C age
which may lower the 14C ages. All the carbonates are lea- uncertainties will increase by 100 years for the carbonate
ched in a weak acid to remove surface contaminants before and by 700 years for the charcoal, respectively. The preci-
hydrolysis in a vacuum device. For AMS 14C measurement, sion of a 14C age older than 30,000 years BP may be
the obtained CO2 is converted into graphite by metal catal- appreciably affected by a few thousand years when the blank
ysis (Tisnérat-Laborde et al. 2001; Hatté et al. 2003). activity is not well-estimated.

14
Determination of a C Age Some Examples of Post-depositional
Disturbances of the 14C Ages
The 14C age is obtained by comparing the activity of a
sample to that of a standard reference, representative of the The validity of the 14C dating of climatic, archaeological or
atmospheric 14C content in 1950 (NBS-I; NBS-II; sucrose) geological events depends, in many cases, of mechanical or
regardless of the dating techniques. Chemicals as well as biological disturbances of sedimentary deposits that may
vacuum lines and counting devices are contaminated by occur after the death and the burial of organisms. In
modern 14CO2 which has a 14C/12C ratio of 1.2  10−12 in archaeological sites, stratigraphic inversions of 14C ages may
1950 AD. The effect of such a contamination on the 14C ages be due to soils disturbances linked either to successive
of a 1 mg carbon sample may be estimated using a simple occupations by human and animals, or to wind and water
mixing equation. A 14C age of 40,000 years effects, or both. The increasingly small size of 14C-dated
(14C/C = 6.88  10−15) would be lowered by 4300 years to samples, which tend to migrate through the sedimentary
7200 years, due to a modern contamination of 5 µg to deposits, favors such 14C age inversions.
10 µg, respectively. This emphasizes the importance of a Archaeological sites contain numerous 14C datable
careful cleaning of the sample. The same amounts of remains, such as mollusk shells, charcoals and bones.
14
C-free contaminants will have little effect: about 40 years Researchers give priority to the 14C dating of charcoals,
and 80 years respectively regardless of the 14C age of the because the 14C dating of marine shells requires a correction
sample. of the reservoir ages to be compared with those of bone and
The first 14C dating of cave paintings revealed an age of vegetation samples. In a seaside Peruvian site, paired marine
about 31,000 years BP that deeply modified our under- mollusk shells and charcoals were associated in several
standing of how human art evolved (Valladas et al. 1992; sediment layers to allow the sea surface reservoirs ages
4 Carbon-14 67

Fig. 4.11 Variations in the d18O 0 2000

(bold line and diamonds) and in G. bulloides A
the number of foraminifera per

Number of foraminifera per gram

cm2 and per gram of sediment
(thin line): a for the subpolar
planktonic foraminifera 1 1500

vs%PDB)%

of sediment per cm2

Globigerina bulloides; b for the
polar planktonic foraminifera

d18O%(‰%
Neogloboquadrina pachyderma
sinistral as a function of depth in
a North Atlantic deep-sea 2 1000
sediment core. The 14C ages at 1
sigma (in BP years) at the
different depths (arrows) are
shown (modified from Bard et al. 11590 ± 150
1987) 3 500
11720 ± 170
11370 ± 170
11470 ± 140 15150 ± 270
10510 ± 140
4 0
70 90 110 130 150 170 190 210
Depth (cm)

2 5000
N. pachyderma s. B

Number of foraminifera per gram

4000
17800 ± 250
δ18O$(‰$vs$PDB)$

of sediment per cm2

3
3000

19300 ± 290
2000
13950 ± 270
4 10640 ± 160
11860 ± 160 15150 ± 220
1000
15790 ± 220

10950 ± 160
5 0
70 90 110 130 150 170 190 210
Depth (cm)

(R) during the Holocene (Kennett et al. 2002) to be quan- in the archaeological layers. Only statistical analyses of the
14
tified. While the modern value of R varies from 540 years to C ages of charcoals and shells would enable a conclusion
970 years depending of the intensity of the Peruvian to be reached. This Peruvian study clearly emphasizes that
upwelling (the rise to the surface of 14C-depleted interme- charcoals are not always the best candidates in radiocarbon
diate and deep waters), it was slightly lower at 430 years dating.
between 5000 and 5300 cal BP and much lower at only A second example is related to the 14C dating of two
30 years between 5600 and 5900 cal BP. The much lower species of planktonic foraminifera during the deglaciation in
value of R is very unlikely, as charcoals and associated a North Atlantic deep-sea sediment core (Bard et al. 1987).
mollusk shells would have the same 14C ages. This may be The species are characteristic of the cold polar waters
due to two phenomena: the charcoals originate from the (Neogloboquadrina pachyderma sinistral) and of the war-
burning of old trees at the time of burial (living over some mer subpolar waters (Globigerina bulloides). It can be seen
hundred years) or the small charcoals migrated downwards from Fig. 4.11 that the 14C ages of G. bulloides are constant
68 M. Paterne et al.

7000 22000
A
Initial
6000 h = 15 cm 21000

Number of grains per g of sediment

h = 10 cm
14C ages

Radiocarbon Age (years BP)

5000 20000

4000 19000

3000 18000

2000 17000

1000 16000

0 15000
100 110 120 130 140 150
Depth (cm)

7000 18500
B
Initial
h = 10 cm
6000 18000
Number of grains per gram of sediment

h = 15 cm
14C ages

5000 17500

Radiocarbon age (years BP)

4000 17000

3000 16500

2000 16000

1000 15500

0 15000
100 110 120 130 140 150

Depth (cm)

Fig. 4.12 Bioturbation effects on 14C age of an event, observed in a plotted by crosses, and the resulting deviations from a bioturbation
deep sea sediment core with sedimentation rates of 10 cm/1000 years depth of 10 cm (white triangles) and 15 cm (white squares). Note the
(a) and 30 cm/1000 years (b) using bioturbation depth (h) of 10 cm change of the 14C age scales in a and b
(black triangles) and 15 cm (black squares). The initial 14C ages are
4 Carbon-14 69

between 170 cm and 140 cm, while those of N. pachyderma d18O profiles, micropaleontological variations and accelerator mass
decreased. Moreover, the latter are much older than the spectrometric 14C dating. Climate Dynamics, 1, 101–112.
Bard, E., Arnold, M., Mangerud, J., Paterne, M., Labeyrie, L., Duprat,
subpolar species. The abundance of these species varies J., et al. (1994). The North Atlantic atmosphere-sea surface 14C
along core depth and peaks of abundance spreads over gradient during the Younger Dryas climatic event. Earth and
several tens of cm. The 14C age distribution of the two Planetary Science Letters, 126, 275–287.
species is due to bioturbation, a mixing process of sediment Bard, E., Hamelin, B., Fairbanks, R. G., & Zindler, A. (1990).
Calibration of the 14C timescale over the past 30,000 years using
grains by benthic organisms, which move up and down mass spectrometric U-Th ages from Barbados corals. Nature, 345,
through the sedimentary deposits. Applying a simple mixing 405–410.
filter of a constant thickness (h) over time, we can observe Beck, J. W., Richards, D. A., Edwards, R. L., Bernard, W., Silverman,
that the estimated abundance is lower than at the time of B. W., Smart, P. L., et al. (2001). Extremely large variations of
atmospheric C-14 concentration during the last glacial period.
deposition. The abundance peak is no longer at its initial Science, 292, 2453–2458.
position in the sediment, and its position depends of the Blunier, T., Chappellaz, J., Schwander, J., Dällenbach, A., Stauffer, B.,
mixing depth (h) (Fig. 4.12). In the examples, the biotur- Stocker, T., et al. (1998). Asynchrony of Antarctic and Greenland
bation effects on 14C ages led to lower the 14C age with climate change during the last glacial period. Nature, 394, 739–743.
Bondevik, S., Mangerud, J., Birks, H. H., Gulliksen, S., & Reimer,
respect to the ‘real’ 14C age at the onset of the event. The P. (2006). Changes in North Atlantic radiocarbon reservoir ages
effect is 900 and 1400 years depending on the h value with a during the Allerød and Younger Dryas. Science, 312, 1514–1517.
sedimentation rate of 10 cm per 1000 years. A much higher Broecker, W. B. F. (1998). Paleocean Circulation During the Last
sedimentation rate of 30 cm per 1000 years tends to lower Deglaciation: A Bipolar Seesaw. Paleoceanography, 13, 119–121.
Broecker, W. S., & Olson, E. A. (1959). Lamont radiocarbon
the bioturbation effects as the 14C ages decrease by measurements VI. American Journal of Science. Radiocarbon
270 years and 420 years, with h equal to 10 cm and 15 cm Supplement, 1, 111–132.
respectively. The 14C age of the end of the event increased Broecker, W. S., Sutherland, S., Smethie, W., Peng, T. S., & Ostlund,
by 600 years using a sedimentation rate of 10 cm per G. (1995). Oceanic radiocarbon: Separation of the natural and bomb
component. Global Biogeochemical Cycles, 9, 263–288.
1000 years. The higher the sedimentation rates, the lower the Burke, A., & Robinson, L. F. (2012). The Southern Ocean’s role in
impact of bioturbation on the 14C ages. carbon exchange during the last deglaciation. Science, 335, 557–
Accurate marine 14C ages are obtained from the 14C 561.
dating of foraminifera picked at the depth of maximum Chen, T., Robinson, L. F., Burke, A., Southon, J., Spooner, P., Morris,
P. J., et al. (2015). Synchronous centennial abrupt events in the
abundance in deep-sea cores with a high sedimentation rate. ocean and atmosphere during the last deglaciation. Science, 349
The example on Fig. 4.11 also emphasizes that the selected (6255), 1537–1541.
species must be in adequacy with the climatic period to be Chiu, T. C., Fairbanks, R. G., Cao, L., & Mortlock, R. A. (2007).
dated. The variability in the compiled deep and surface Δ14C Analysis of the atmospheric C-14 record spanning the past 50 000
years derived from high-precision Th-230/U-234/U-238,
values in Zhao et al. (2018) may be partly explained by Pa-231/U-235 and C-14 dates on fossil corals. Quaternary Science
bioturbation and inadequately dated species. Review, 26, 18–36.
Currie, L. A. (2004). The remarkable metrological history of radiocar-
bon dating. Journal of Research of the National Institute of
Standards and Technology, 109, 185–217.
References Cuzange, M. T., Delque-Kolic, E., Goslar, T., Grootes, P. M., Higham,
T., Kaltnecker, E., et al. (2007). Radiocarbon intercomparison
Adkins, J. F., Cheng, H., Boyle, E. A., Druffel, E. R., & Edwards, R. L. program for the chauvet cave. Radiocarbon, 49, 339–347.
(1998). Deep-sea coral evidence for rapid change in ventilation of Damon, P. E., Lerman, J. C., & Long, A. (1978). Temporal fluctuations
the deep North Atlantic 15,400 years ago. Science, 280, 725–728. of atmospheric 14C: Causal factors and implications. The Annual
Arnold, J. R., & Libby, W. F. (1949). Age determinations by Review of Earth and Planetary Sciences, 6, 457–494.
radiocarbon content: Checks with samples of known Age. Science, Damon, P. E., & Peristykh, A. N. (2000). Radiocarbon calibration and
110, 678–680. application to geophysics, solar physics, and astrophysics. Radio-
Ascough, P. L., Cook, G. T., & Dugmore, A. J. (2009). North Atlantic carbon, 42, 137–150.
marine 14C reservoir effects: Implications for late-Holocene chrono- Druffel, E. R. M. (1989). Decade time scale variability of ventilation in
logical studies Quaternary. Geochronology, 4, 171–180. the North Atlantic: High-precision measurements of bomb radio-
Austin, W. E. N., Telford, R. J., Ninnemann, U. S., Brown, L., Wilson, carbon in banded corals. Journal of Geophysical Research, 94,
L. J., Small, D. P., & Bryant, C. L. (2011). North Atlantic reservoir 3271–3285.
ages linked to high Younger Dryas atmospheric radiocarbon Engelkemeir, A., Hamill, W. H., Inghram, M. G., & Libby, W. F.
concentrations. Global and Planetary Change, 79, 226–233. (1949). The half-life of radiocarbon (14C). The Physical Review, 75,
Balesdent, J., & Guillet, B. (1982). Les datations par le 14C des 1825–1833.
matières organiques des sols. Science du sol, 2, 93–112. Ezat, M. M., Rasmussen, T. L., Thornalley, D. J. R., Olsen, J., Skinner,
Bard, E. (1988). Correction of accelerator mass spectrometry 14C ages L. C., Hönisch, B., et al. (2017). Ventilation history of Nordic Seas
measured in planktonic foraminifera: Paleoceanographic implica- over-flows during the last (de)glacial period revealed by
tions. Paleoceanography, 3, 635–645. species-specific benthic foraminiferal 14C dates. Paleoceanography,
Bard, E., Arnold, M., Duprat, J., Moyes, J., & Duplessy, J. C. (1987). 32, 172–181.
Reconstruction of the last deglaciation: Deconvolved records of
70 M. Paterne et al.

Friedrich, W. L., Kromer, B., Friedrich, M., Heinemeier, J., Pfeiffer, T., Olsson, I., Kaplen, I., Turnbull, A. H., & Prosser, N. J. D. (1962).
& Talamo, S. (2006). Santorini eruption radiocarbon dated to 1627– Determination of the half-life of 14C with a proportional counter.
1600 BC. Science, 312, 548. Arkiv för fysik, 14, 237–255.
Gaudinsky, J. B., Trumbore, S., Davidson, E. A., & Zheng, S. (2000). Olsson, I., & Osadebe, F. (1974). Carbon isotope variations and
Soil carbon cycling in a temperate forest: Radiocarbon-based fractionation corrections in 14C dating. Boreas, 3, 139–146.
estimates of residence times, sequestration rates and partitioning of Reimer, P. J., Bard, E., Bayliss, A., Beck, J. W., Blackwell, P. G.,
fluxes. Biogeochemistry, 51, 33–69. Bronk Ramsey, C., et al. (2013). IntCal13 and Marine13 radiocar-
Godwin, H. (1962). Half-life of radiocarbon. Nature, 195, 984. bon age calibration curves 0–50,000 years cal BP. Radiocarbon,
Goldstein, S. J., Lea, D. W., Chakraborty, S., Kashgarian, M., & 55, 1869–1887.
Murrell, M. T. (2001). Uranium-series and radiocarbon geochronol- Robinson, L. F., Adkins, J. F., Keigwin, L. D., Southon, J., Fernandez,
ogy of deep-sea corals: Implications for Southern Ocean ventilation D. P., Wang, S., et al. (2005). Radiocarbon variability in the western
rates and the oceanic carbon cycle. Earth and Planetary Science North Atlantic during the last deglaciation. Science, 310, 1469–
Letters, 193, 167–182. 1473.
Hatté, C., Poupeau, J. J., Tannau, J. F., & Paterne, M. (2003). Rossi, B. (1952). High Energy Particles (p. 268). Englewood cliffs, NJ:
Development of an automated system for preparation of organic Prentice Hall Physics series Inc.
samples. Radiocarbon, 45(3), 421–430. Ruben, S., Kamen, M. D., & Hassid, W. (1949). Photosynthesis with
Hughen, K. A., Overpeck, J. T., Lehman, S. J., Kashgarian, M., radioactive carbon: II. Chemical properties of the intermediates.
Southon, J., Peterson, L. C., et al. (1998). Deglacial changes in Journal of the American Chemical Society, 62, 3443–3450.
ocean circulation from an extended radiocarbon calibration. Nature, Scharpenseel, H. W., & Shiffmann, H. (1971). Radiocarbon dating of
391, 65–68. soils, a review. Zeitschrift für Planzenernähring, Düngung,
Ikehara, K., Danhara, T., Yamashita, T., Tanahashi, M., Morita, S., & Bodenkunde, 140, 159–174.
Ohkushia, K. (2011). Paleoceanographic control on a large marine Siani, G., Michel, E., De Pol-Holz, R., Tim DeVries, T., Lamy, F.,
reservoir effect offshore of Tokai, south of Japan, NW Pacific, Carel, M., et al. (2013). Carbon isotope records reveal precise
during the last glacial maximum-deglaciation. Quaternary Interna- timing of enhanced Southern Ocean upwelling during the last
tional, 246, 213–221. deglaciation. Nature Communications, 4, 2758. https://doi.org/10.
Kamen, M. D. (1963). Early history of carbon-14. Science, 140, 584– 1038/ncomms3758.
590. Siani, G., Paterne, M., Arnold, M., Bard, E., Métivier, B., Tisnerat, N.,
Kennett, D. J., Ingramm, L. B., Southon, J. R., & Wise, K. (2002). et al. (2000). Radiocarbon reservoir ages in the Mediterranean Sea
Differences in 14C age between stratigraphically associated charcoal and Black Sea. Radiocarbon, 42(2), 271–280.
and marine shell from the archaic period site of kilometer 4, Siani, G., Paterne, M., Michel, E., Sulpizio, R., Sbrana, A., Arnold, M.,
Southern Peru: Old wood or old water? Radiocarbon, 44, 53–58. & Haddad, G. (2001). Mediterranean Sea surface radiocarbon
Key, R. M., Kozyr, A., Sabine, C. L., Lee, K., Wanninkhof, R., reservoir age changes since the last glacial maximum. Science, 294,
Bullister, J. L., Feely, R. A., Millero, F. J., Mordy, C., & Peng, T.H. 1917–1920.
(2004). A global ocean carbon climatology: Results from Global Sikes, E. L., & Guilderson, T. P. (2016). Southwest Pacific Ocean
Data Analysis Project (GLODAP). Global Biogeochemical Cycles, surface reservoir ages since the last glaciation: Circulation insights
18, GB4031. https://doi.org/10.1029/2004gb002247. from multiple-core studies. Paleoceanography, 31, 298–310.
Key, R. M., Quay, P. D., Jones, G. A., McNichol, A. P., von Reden, K. Sikes, E. L., Samson, C. R., Guilderson, T. P., & Howard, W. R.
F., & Schneider, R. J. (1996). WOCE AMS radiocarbon I: Pacific (2000). Old radiocarbon ages in the southwest Pacific Ocean during
ocean results; P6, P16 & P17. Radiocarbon, 38, 425–518. last glacial period and deglaciation. Nature, 405, 555–559.
Kleber, M., Sollins, P., & Sutton, R. (2007). A conceptual model of Simpson, J. A. (2000). The cosmic ray nucleonic component: The
organo-mineral interactions in soils: self-assembly of organic invention and scientific uses of the neutron monitor. Space Science
molecular fragments into zonal structures on mineral surfaces. Reviews, 93, 11–32.
Biogeochemistry, 85, 9–24. Six, J., Conant, R. T., Paul, A., & Paustian, K. (2002). Stabilization
Korff, S. A. (1951). Cosmic-ray neutrons. American Journal of Physics, mechanisms of soil organic matter: Implications for C-saturation of
19, 226–229. soils. Plant and Soil, 241, 155–176.
Lal, D., & Charles, C. (2007). Deconvolution of the atmospheric Stenström, K. E., Skog, G., Georgiadou, E., Genberg, J., & Johansson,
radiocarbon record in the last 50,000 years. Earth and Planetary A. (2011). A guide to radiocarbon units and calculations. Interna-
Science Letters, 258, 550–560. tional Report LUNFD6 (NFFR-3111), Lund University, 1–17.
Libby, W. F. (1934). Radioactivity of neodymium and samarium. Stuiver, M., Braziunas, T. F., Becker, B., & Kromer, B. (1991).
Physical Review, 45, 196–204. Climatic, Solar, Oceanic and Geomagnetic Influences on
Libby, W. F. (1952). Radiocarbon dating (Vol. 37, 124 p.). Chicago, Late-Glacial and Holocene Atmospheric 14C/12C Change. Quatem-
IL: The University of Chicago Press. ary Research, 35, 1–24.
Libby, W. F. (1964). Radiocarbon dating. Nobel Lecture 12 December Stuiver, M., Pearson, G. W., & Braziunas, T. (1986). Radiocarbon age
1960. Nobel Lectures, Chemistry 1942–1962, Amsterdam: Elsevier calibration of marine samples back to 9000 cal yr bp. Radiocarbon,
Publishing Company. 28, 980–1021.
Libby, W. F. (1981). In R.Berger & L. M. Libby (Eds.), Radiocarbon Stuiver, M., & Polach, H. A. (1977). Reporting of 14C. Radiocarbon,
and tritium (Vol. I, 300 p.). Ocean Way Santa Monica, CA: 129 19, 355–363.
Geo Science Analytical Inc., 90402. Tagliabue, A., Bopp, L., Roche, D. M., Bouttes, N., Dutay, J.-C.,
Michel, E., Labeyrie, L. D., Duplessy, J.-C., Gorfti, N., Labracherie, Alkama, R., et al. (2009). Quantifying the roles of ocean circulation
M., & Turon, J.-L. (1995). Could deep Subantarctic convection feed and biogeochemistry in governing ocean carbon-13 and atmo-
the world deep basins during the Last Glacial Maximum? Paleo- spheric carbon dioxide at the last glacial maximum. Climate of the
ceanography, 10, 927–942. Past, 5, 695–706.
Muscheler, R., Joos, F., Müller, S. A., & Snowball, I. (2005). Climate: Taylor, R. E. (1987). Radiocarbon dating: An archeological perspec-
How unusual is today’s solar activity? Nature, 436, E3–E4. tive (212 p.). London: Academic Press Inc. Ltd.
4 Carbon-14 71

Taylor, R. E., Long, A., & Kra, R. S. (1992). Radiocarbon after four Toggweiler, J. R., Dixon, K., & Bryan, K. (1989). Simulations of
decades. An interdisciplinary perspective (596 p.). New York: radiocarbon in a coarse-resolution world ocean model. 1. Steady
Springer. state prebomb distributions. Journal of Geophysical Research, 94,
Thornalley, D. J. R., McCave, N., & Elderfield, H. (2011). Tephra in 8217–8242.
deglacial ocean sediments south of Iceland: Stratigraphy, geochem- Valladas, H., Cachier, H., Maurice, P., De Quirost, F. B., Clottes, J.,
istry and oceanic reservoir ages. Journal of Quaternary Science, 26, Valdes, V. C., et al. (1992). Direct radiocarbon dates for prehistoric
190–198. paintings at the Altamira, El Castillo and Niaux caves. Nature, 357,
Tisnérat-Laborde, N., Paterne, M., Métivier, B., Arnold, M., Yiou, P., 68–70.
Blamart, D., et al. (2010). Variability of the northeast Atlantic sea Zhao, N., Marchal, O., Keigwin, L., Amrhein, D., & Gebbie, G. (2018).
surface Δ14C and marine reservoir age and the North Atlantic A synthesis of deglacial deep-sea radiocarbon records and their
Oscillation (NAO). Quaternary Science Reviews, 29, 2633–2646. (in) consistency with modern ocean ventilation. Paleoceanography
Tisnérat-Laborde, N., Poupeau, J. J., Tannau, J. F., & Paterne, M. and Paleoclimatology, 33, 128–151.
(2001). Development of a semi-automated system for routine
preparation of carbonate sample. Radiocarbon, 43, 299–304.
 40
K/40Ar and 40
Ar/39Ar Methods
The
5
Hervé Guillou, Sébastien Nomade, and Vincent Scao

The 40K/40Ar method and its variant, 40Ar/39Ar, are based on 40

K became more accurate. Aldrich and Nier could therefore
the natural radioactive decay of 40K, one of the isotopes of see the potential of the 40K/40Ar pair for the dating of rocks.
potassium, in 40Ar, one of the isotopes of argon. 40K The K-Ar clock is based on the principles of radioactive
decreases in 40Ar* (the * symbol indicates that this is a decay and the accumulation of a daughter isotope. However,
radiogenic isotope) with a period of 1.25
109 years, studies subsequent to Aldrich and Nier’s work showed that
according to the law of radioactive decay N = N0 e−kt. In there were many causes of disturbance in the K/Ar clock.
other words, if we consider a closed system, containing at an Among these are the inability of certain rocks or minerals to
initial time (t0) N0 atoms of 40K, then N0/2 atoms of 40K will retain all of the radiogenic argon-40 (40Ar*), or the presence
remain in the system after 1.25
109 years. This gives us of ‘excess argon’ in some samples. Consequently, for a K-Ar
an indication of the geochronological application. If, in a age to be accepted as correct, the following must be true:
geological sample, both the number of parent atoms
remaining (40K) and the number of daughter atoms formed 1. when starting the clock (at time zero t0), the 40Ar/36Ar
(40Ar*) can be measured, then it is possible to calculate the ratio in the sample is the same as that of the atmosphere
age of formation of this sample. The relatively high abun- (298.56), in other words that 40Ar* = 0;
dance of the isotope 40K (K is the seventh most abundant 2. and between t0 and the moment the sample is dated, it
element on Earth), combined with a low decay rate, makes behaves as a closed system with regard to 40K and 40Ar.
the 40K/40Ar method and its variant 40Ar/39Ar two of the
most widely used geochronological tools in Earth Sciences. The conventional K/Ar method does not allow verifica-
They are applicable to various geological materials and tion of these two major assumptions. To remedy this, the
cover a wide range of ages, given the long period of 40K. 40
Ar/39Ar variant was developed by Wänke and König
Already in 1921, Aston, using a mass spectrograph, (1959) and Merrihue (1965), which showed that the
proved the existence of two isotopes of potassium (39K and 40 40
K/ Ar ages can be obtained by irradiating samples of
41
K). In 1935, Klemperer, and also Neuman and Walker, rocks or minerals. When a sample is subjected to a neutron
experimentally demonstrated the natural radioactive decay of flux in a reactor, some 39K becomes 39Ar. The measurement
40
K to 40Ca and 40Ar*. In 1948, Aldrich and Nier confirmed of the 39Ar content by a counting method calculates the
the radiogenic origin of argon 40Ar*. They experimentally number of parent atoms (40K) remaining in the sample, since
determined the 40Ar/36Ar ratio of several potassic minerals we know the relative abundances of different isotopes of
and compared it to that of the atmosphere, assuming it to be potassium (see below). Using the same counting method, the
constant and equal to 298.56 (this updated value, determined number of 40Ar isotopes present in the sample can also be
by Lee et al. (2006), replaces the previous one of 295.5 measured. However, this approach is unsatisfactory because
established by Steiger and Jäger (1977)). As the ratios it does not allow for necessary corrections related to the
obtained were superior to that of the atmosphere, the source process of irradiation to be made, as we shall see later, and
of argon 40Ar* by radioactive decay of 40K was demon- the precision achieved on the different concentrations of
strated. In parallel, understanding of the decay constant of argon isotopes is insufficient. The crucial work for the
40
Ar/39Ar method was carried out by Merrihue in 1965. He
H. Guillou (&)  S. Nomade  V. Scao showed that the argon 39Ar generated in a nuclear reactor
Laboratoire des Sciences du Climat et de l’Environnement, from the 39K of a sample can be measured precisely by mass
LSCE/IPSL, CEA-CNRS-UVSQ, Université Paris-Saclay,
91190 Gif-sur-Yvette, France spectrometry. This 39Ar, derived from the 39K, is annotated
e-mail: [email protected] as 39ArK. In addition, the other isotopes of argon, 40Ar and

© Springer Nature Switzerland AG 2021 73

G. Ramstein et al. (eds.), Paleoclimatology, Frontiers in Earth Sciences,
https://doi.org/10.1007/978-3-030-24982-3_5
74 H. Guillou et al.

36
Ar (essential for the atmospheric correction and 40Ar* expense of a proton. The 40Ar atom thus produced is in an
content calculation) can be measured in the same way. Thus, excited state. It then returns quickly to its ground state by
in a single measurement, one can calculate the proportions of emitting gamma radiation.
40
K and 40Ar* present in a sample, and from this, calculate
its age. In their 1966 article, Merrihue and Turner estab-
lished the fundamentals in terms of approach and concepts The Age Equation
for the 40Ar/39Ar method. In particular, they showed that the
relative proportions of radioactive parents and radiogenic As with the other isotopic clocks, the fundamental law of
daughters can be calculated accurately from a measurement radioactive decay applies:
by mass spectrometry.
In addition, since isotopic ratios can be measured more N ¼ N0 ekt ð5:1Þ
precisely than the concentrations of K and Ar, this method N: number of radioactive parent atoms ( K) at time t, N0: 40
improves the accuracy of the ages and can be used for dating number of radioactive parent atoms at t0, k: decay constant.
smaller samples than the 40K/40Ar method does. This same From Eq. (5.1), we can calculate the number of daughter
work laid the groundwork for the application of isochrones atoms (D = 40K + 40Ca) formed over time t:
and age spectra (concepts that will be discussed below) to
the 40Ar/39Ar method. No ¼ Nekt
The 40K/40Ar and 40Ar/39Ar methods became very pop-


ular in geology as they are applicable to different terrestrial D ¼ No  N ¼ Nekt  N ¼ N ekt  1 ð5:2Þ
geological materials, such as terrestrial magmatic rocks
(volcanic, plutonic, metamorphic) and extraterrestrial (me- The constants and isotopic abundances required for the
teorites, moon samples) ones. For some measurements, the age calculation are listed in Table 5.1.
40 40
K/ Ar method is also well suited to the dating of clay The age equation is established from Eq. (5.2):
minerals. The range of application of these isotopic age
40 k 40
kt 
dating methods has an upper limit of 3 billion years and a Ar ¼ K e 1 ð5:3Þ
k
lower limit of 10,000 years.
These two methods were used to date major events in the where 40Ar* is the isotope of argon produced from the in situ
history of the Earth (fauna and flora of the Mesozoic and decay of 40K, k the total radioactive decay constant of 40K
Cenozoic, mass extinctions, origin and evolution of homi- equal to ke + kb. The ratio of proportionality kk corresponds
nids, major volcanic eruptions, genesis and evolution of the to the fraction of the decay leading to the formation of 40Ar*
large mountain chains, etc.). They were, and still are, used to (and not of 40Ca*).
establish and calibrate the geological time scale, including From Eq. (5.3) we get:
the time scale of the reversals of Earth’s magnetic field, very 40  
useful tie-points in paleoclimatology. 1 Ar k
t ¼ ln 40 þ1 ð5:4Þ
In the following, we present the main principles and areas k K k
of application of these two methods. For further details, the
reader may refer to the works of Dalrymple and Lanphere with t expressed in years.
(1969) and McDougall and Harrison (1988).
40
K
E=0.05MEV.
c.e.

Principles of the K-Ar Method

40
Diagram of Radioactive Decay in K E=1.46MEV. E=1.51MEV. E=1.32MEV.
11% 0.16% 88.8%
γ c.e. E=0.49MEV. β-
The principle of the 40K/40Ar method is based on the natural 0.01%
β+
radioactive decay of 40K in 40Ar (Fig. 5.1). The decay of 40K
is complex. At 88.8%, the 40K decays to 40Ca by emitting
b−. At 11.2% it decays to 40Ar*, either by emitting b+ 40
Ca (ground state)
(0.01%), or by direct electronic capture (0.16%) or by
electron capture followed by a c emission (11%). This last 40
Ar (ground state)
mechanism is the most common. An electron from the atom
40
is captured, resulting in the formation of a neutron at the Fig. 5.1 Diagram of radioactive decay of K
 40
5 The K/40Ar and 40
Ar/39Ar Methods 75

Table 5.1 Decay constants of Constant Value

40
K and isotopic abundances of K
and Ar. The values attributed to ke = k40Ar (5.80 ± 0.014)
10−11 a−1
the constants were determined by kb = k40Ca (4.884 ± 0.099)
10−10 a−1
Min et al. (2000), and for isotopic
k = ke + kb (5.464 ± 0.107)
10−10 a−1
ratios by Garner et al. (1975) and
39
Nier (1950) K 93.2581%
40
K 0.01167%
41
K 6.7302%
40
Ar 99.600%
38
Ar 0.0632%
36
Ar 0.3364%

The half-life period (N = N0/2) T is calculated from (5.1): (40Ar/39Ar). 40K is deduced from the isotopic composi-
tion of K. This ratio has changed over time due to
ln 2
T¼ ¼ 1:25
109 years: radioactive decay, but this term is not included in the age
k equation. At a given t, this ratio is constant in all mate-
rials because these isotopes do not fractionate as a result
of the geological processes.
Operation of the Potassium-Argon Clock 3. We consider that at t = 0, the moment of formation of the
sample, it is devoid of radiogenic argon (40Ar* = 0);
The radioactive clock 40K/40Ar is based on the process of otherwise, ages obtained would be marred by an error of
accumulation. K is one of the components of magma. When excess argon. In geochronology, this is the same as
this is in liquid form, the argon 40Ar* formed from the decay assuming that at t = 0, the 40Ar/36Ar ratio of the sample,
of 40K escapes from the system. During a volcanic eruption, called the initial ratio, is considered to be equal to that of
the magma that reaches the surface cools very quickly. Thus, the atmosphere, or 298.56. There are some deviations
argon 40Ar* is trapped in the solidified lava and accumulates from this principle. These are cases of excess argon or
in the crystalline lattice. The radiogenic argon (40Ar*) thus inherited argon, which cannot be directly detected by the
trapped can only escape if the rock or mineral are either K-Ar method, but can be detected more easily by the
40
melted or recrystallized, or heated to temperatures generally Ar/39Ar method. These excesses of argon show up as
greater than or equal to 200 ◦C, in such a way that the argon an overestimation of the calculated ages and are an
can diffuse through the crystal lattice. Dalrymple and Lan- important limitation of the K-Ar method.
phere (1969) illustrated the operation of the 40K/40Ar clock 4. It is also necessary that the formation time of the system
(Fig. 5.2) in a diagram, taking the crystallization of magma be negligible compared to the age of the sample.
as an example. Ideally, there are three distinct stages. During Therefore, volcanic rocks that form by very rapid cooling
the first stage, at high temperatures, the phenomenon of provide the most suitable samples for this method of
diffusion prevails. 40Ar* is not retained in the lattice. The dating.
second stage corresponds to a start of cooling and partial 5. It is essential to assume that the sample evolved within a
accumulation of the argon 40Ar*. The last step corresponds closed system with regard to K and Ar ever since the
to the rapid cooling of the surface of the silicate or magmatic geological event to be dated. This condition involves
melt. At this point, 40Ar* is retained entirely within the rigorously selecting unaltered samples, in order to avoid
crystal lattice. any disruption (re-opening subsequent to formation) in
From this evolution came the basic assumptions for the the isotopic system.
application of the K-Ar clock which are detailed below.

1. The parent isotope 40K decays at a constant rate, inde-

pendently of the physical conditions of the system (P and Datable Materials and Age Ranges
T). The constants used are those in Table 5.1.
2. The 40K/Ktotal ratio is constant in natural materials. This The main materials suitable for testing by the K-Ar and
39
condition is important because it is not 40K that is Ar/40Ar methods as well as the age ranges are listed below
directly measured, but the total K (K-Ar) or 39ArK in Fig. 5.3:
76 H. Guillou et al.

Fig. 5.2 Principle of the

40 40
K/ Ar clock in the case of a
simple-story magmatic rock (from
Dalrymple and Lanphere 1969)

The Unspiked K-Ar Method

Selection and Preparation of the Samples

The stratigraphic and geographic location of the sample

needs to be noted and established precisely in the field. This
will enable external limits to be placed on the ages obtained.
Only unaltered samples are retained, mainly to comply with
the condition of evolution within a closed system, as dis-
cussed above. Freshness is controlled by macroscopic and
microscopic observations, and by chemical analysis which
measures the loss on ignition (L.O.I. almost equivalent to the
water content), a reliable marker of the degree of alteration
of the sample. For example, a basalt is considered unaltered
if its L.O.I. value is less than 1%. The phase of the sample
(mineral, glass, etc.) must be representative of the event to
be dated. In the case of volcanic rocks, the selected phase for
40
Fig. 5.3 Applicable age range of K-Ar and Ar/39Ar dating for lava is the microcrystalline groundmass because it is formed
various materials (from Renne 2000)
by rapid cooling when the magma arrives at the surface; for
tephra, phenocrystals (feldspar, mica and amphibole) syn-
• volcanic rocks (calibration of time scales, stratigraphic chronous with the eruption are selected. After sampling in
studies); minerals: sanidine (  2 ka) anorthoclase the field, the rocks are cut, ground and sieved to a 0.250–
(  5 ka), plagioclase (  200 ka), amphibole (  1 Ma), 0.125 mm size. The granulate obtained is washed in an
leucite, nepheline; ultrasonic bath of acetic acid (1 N) at 50 °C. The sample is
• rocks (lava and tephra): mesostasis (groundmass) of all then rinsed thoroughly with deionized water. Then, the
types of unaltered rock (  3 4 ka), non-hydrated glass separation of the mineralogical phases is done in the labo-
(  300 ka); ratory, by magnetic sorting and densitometry or picking
• plutonic rocks, metamorphic minerals; potassic feldspar, under microscope. In the case of tephra, the preparatory
plagioclase, biotite, amphibole, muscovite, phengite, procedure is identical. However, given the nature of the
alunite, adularia; deposit, the grinding phase is bypassed.
• sediments; neoformed clays, glauconite, evaporites,
detrital minerals rich in K, alunite/jarosite.
 40
5 The K/40Ar and 40
Ar/39Ar Methods 77

40
Determining Ar* titanium sublimation pump and a SAES 10 GP-MK3 Zr-Al
getter operated at 400 °C. This first step of gas clean-up (i.e.
To calculate an age from Eq. (5.4), only the measurement of elimination of active gases) generally lasts 30 min (Fig. 5.5)
two variables is required: the content of radiogenic argon 40 and is followed by three consecutive exposures of five
(40Ar*) and of potassium 40 (40K). The K-Ar method of minutes each of the gas to SAES 10 GP-MK3 Zr-Al getters
dating used by the LSCE is the unspiked (with no tracer) also operated at 400 °C. The remaining gas, mostly argon, is
K-Ar technique developed by Cassignol et al. (1978), Cas- then adsorbed 5 min by a second active charcoal maintained
signol and Gillot (1982). at liquid nitrogen temperature.
A schematic representation of the ultra-high vacuum line The argon is then freed from the active charcoal finger n°
connected to the K-Ar mass spectrometer is given in 2 by bringing it to room temperature. After a rapid
Fig. 5.4. Prior to the dosing of argon, several stages, such as cryo-pumping of the spectrometer, the argon is introduced
extraction and purification of the argon are required. into the mass spectrometer. The argon, an inert gas, is
Groundmass splits (0.5–2.0 g) of samples are wrapped into ionised in the mass spectrometer, under the effect of an
99.5% copper foil packets, loaded in the sample holder, electronic source. 40Ar becomes 40Ar+ and 36Ar becomes
36
which has been turbo-molecular pumped for about 20 h Ar+. The atoms thus charged are accelerated under the
(Fig. 5.4). During the last two hours of that stage, the influence of a difference in potential (about 620 volts). They
molybdemium (Mo) crucible is degassed at about 1500 °C then pass through a magnetic field. At this point, their tra-
until the pressure decreases to 10−9 Torr. The sample is then jectory becomes circular. In a chamber with a high vacuum,
dropped into the Mo crucible and becomes molten at full these ions with a mass of 40 and 36 and charged e, animated
power of the induction furnace. During the melting stage at speed due to a difference in potential (620 V), trace a
(i.e. 20 min), the extracted gas is adsorbed by the first active trajectory of radius R, as they pass through a magnetic field
charcoal finger at liquid nitrogen temperature. H (3600 Gauss), according to the equation:
After the melting, the gas is released by heating the
charcoal to 110 °C and purified via the mutual action of a 1439 hm i12
R¼ V
H e

Viewport

sample
Sample
holder
primary pump

turbo pump

Auto valve open

Auto valve closed
turbo pump
Mo
crucible Manual valve
Induction
heating
air tank
Ti-sub

gauge

getter
getter
gauge

V8
V4 V5 V6 mass
V9 spectrometer

V1 V2 V3 V7 V10 V12 V14

getter

getter

V11
getter
getter
getter
getter
bellow

Active Charcoal Active Charcoal

finger 1 finger 2
gauge

Cryogenic
pump

Fig. 5.4 Schematic representation of the ultra-high vacuum line connected to the K-Ar mass spectrometer. Ti-Sub: Titanium sublimation pump,
V1-14: Ultra-high vacuum all metal valves. NB: the diagram is at a different scale between the preparation line and the sample holder
78 H. Guillou et al.

Fig. 5.5 Evolution of pressure

versus time during the gas
clean-up by means of a Titanium
sublimation pump and a SAES 10
GP-MK3 Zr-Al getter. #01 and
#02 are the reference numbers for
the experiment

with R: radius in mm, H: magnetic field measured in Gauss, The third measurement of an aliquot of gas is performed
m: atomic mass of the ion, e: number of elemental charges for calibration, in other words, the conversion of an electrical
carried by the ion, V: difference in potential in volts. signal into a number of atoms. A known number of argon
40
Thus, for a spectrometer configured with H = 3600 Gauss Ar is fed from a calibrated cylinder into the mass spec-
and V = 620 volts, the following is obtained: trometer. The number of atoms is deduced from the mea-
R40 = 62.9 mm and R36 = 59.7 mm. surement of standard minerals of a known age. The
The 40Ar and 36Ar isotopes of the sample are measured procedure for establishing the equation of the calibration
simultaneously on a dual manifold comprised of two Fara- curve (Fig. 5.7) is described in Charbit et al., (1998). The
day cups, set at m/e = 40 and m/e = 36. The signals are three previous steps are shown in the diagram in Fig. 5.6.
integrated over a period of 100 s. Once the analysis of the The characteristics of the standard minerals commonly
gas is made, the sample is removed from the mass spec- used in K-Ar for calibration are shown in Table 5.2.
trometer by cryo-pumping. The reference atmospheric argon Potassium analysis is done by flame spectrometry (atomic
(at.) is taken by means of a double valve from a cylinder absorption and emission), on several selections taken inde-
containing air from the laboratory (designed as Ref.Atm.). It pendently to ensure sample homogeneity. As this method is
is then introduced into the mass spectrometer and measured a conventional one, we will not enter into the details here.
at the same pressure conditions as the sample. It allows the
direct comparison of the two aliquots of gas (the sample and
atmospheric reference) and the determination of the relative Example of a Calculation of Age
content of radiogenic argon (Fig. 5.6). This is done by
varying the analysis volume via a variable volume Calibration of the Mass Spectrometer
(VV) connected to the mass-spectrometer cell. Calibration requires an analysis of mineral standards of a
The content of 40Ar* is given by the equation: known age. First, the 40Ar and 36Ar content of the standard
is measured by mass spectrometry. The mineral standard
40 40

40  36 Ar
Ars
 Arat
36 A studied here is LP-6. Its K content is 8.37% and that of 40Ar*
Ar ¼ s
40 Ar
at
is 1.158
1015 at./g. In the following example, 0.04442 g
s
36 Ar
s
of this standard was molten. Then, three aliquots of air were
 40
5 The K/40Ar and 40
Ar/39Ar Methods 79

Fig. 5.6 Principle of the K-Ar

method with no tracer. Two
isotopes of argon (40Ar and 36Ar)
are measured by mass
spectrometer. The sample gas is
composed of atmospheric argon
40
Arat and 36Arat, as well as
radiogenic argon 40Ar*, resulting
from the decay of 40K.
s = sample; at = atmospheric,
cd = calibrated dose

Fig. 5.7 Calibration curve of the

mass-spectrometer on the 19th of
November 2001

Table 5.2 Values of mineral Standard K (%) Ar* (10−9 mol/g)

40
Age (Ma) References
standards
HD-B1 7.987 0.335 24.03 ± 0.41 Fuhrmann et al. (1987)
GL-O 6.56 1.093 93.60 ± 0.90 Charbit et al. (1998)
LP-6 8.37 1.923 127.8 ± 1.4 Odin (1982)
SB-3 7.483 2.213 162.9 ± 0.9 Lanphere and Dalrymple (2000)

taken up using a double valve system (pipette) from a cali- 40

Ar in volts 36
Ar in millivolts
bration canister, and the 40Ar and 36Ar contents are Sample (LP-6) 8.921 3.110
measured.
Atmospheric reference 7.637 28.252
Calibrated dose (3 doses) 7.637
The results of the analysis by mass spectrometry are shown
in the following table.
80 H. Guillou et al.

From these values, the equivalent 40Ar/36Ar ratios can be The age is obtained from the equation:
calculated as follows: (40Ar/36Ar)sample = R.I.s = 2 868.49; 40    
(40Ar/36Ar)tam = R.I.atm = 270.32; and the concentration of 1 Ar k
t ¼
ln 40 þ 1 ¼ 296000 years
40
Ar* as: 40Ar*, hence: (R.I.s − R.I.atm)/(R.I.s) = 90.58%. k K ke
We know the content of 40Ar* (1.158
1015 at/g) and the
melted weight (0.04442 g) of the standard. We can calculate the
number of argon 40Ar* atoms introduced in the mass spec-
(k = 5.543
10−10 and ke = 0.581
10−10).
trometer: Nat = 1.158
1015
0.044 42 = 5.144
1013 at.
Out of the 40Ar signal of the sample (8.921 V), 90.58%
corresponds to 40Ar* that is to say 8080 V. From this, we The 40
Ar/39Ar Method: General Principles
deduce that 8080 V corresponds to 5.144
1013 at. of 40Ar*.
Three aliquots of air produce a signal of 7.637 V, this
The Age Equation
therefore corresponds to (7.637
5.144
1013)/8.080 =
4.862
1013 at. of argon 40Ar*.
This method is a variant of the K-Ar method. Firstly, the
An aliquot of air taken from the calibration canister is
samples undergo neutron activation. This activation under
therefore equivalent to 4.862
1013/3 = 1.621
1013 at.
fast neutron flux within a nuclear reactor is intended to
of 40Ar. This canister can then be used to calibrate the mass
transform the isotope 39K to 39Ar. The amount of argon 39Ar
spectrometer for the measurement of ordinary samples.
thus generated is proportional to the number of 39K atoms
Obviously, for each calibration measurement, the content of
40 and therefore of 40K (parent atoms) present in the sample,
Ar decreases in the cylinder. This change is monitored by a
the 40K/39K ratio being (supposedly) constant in nature. To
regular measurement of the mineral standard. This is the
do this, the samples are placed with samples of known age
curve shown in Fig. 5.7.
(standards) in aluminum discs themselves stacked in an
aluminum tube (shuttle). This shuttle is then subjected to a
Measuring a Sample of Unknown Age
fast neutron flux, for a period of between a few minutes and
Experimental data: melted weight: 1.0669 g; potassium
24 h depending on the age and nature of the samples.
content of the analyzed rock K % = 0.643; 1 calibrated
The irradiation causes the formation of an artificial argon
dose = 1.608 2
1013 atoms.
isotope, 39Ar, according to the reaction 39 39
18 Kðn; pÞ18 Ar (cap-
ture band of 80 to 100 mbarn, Mitchell 1968; Roddick
Sample Atmospheric reference Calibrated dose
40
1983). 39Ar is radioactive. Its disintegration period is
Ar (V) 1.196 1.323 3.912
265 years. As the analysis by spectrometer is performed less
36
Ar (mV) 4.225 4.884 14.474 than one year after irradiation, the error margin on its esti-
mation is negligible. The advantage of producing argon 39Ar
in proportion to the parent element (40K) is that this trans-
The calculation of R.I. isotope ratios (40Ar/36Ar) gives: R. formation replaces the measurement of the 40K/40Ar ratio by
I.s = 283.08; R.I.atm = 270.88; R.I.cd = 270.28. two different methods (atomic absorption for 40K and mass
The level of radiogenic argon is calculated by: spectrometry for 40Ar) with the direct measurement of the
40
Ar/39Ar ratio (by mass spectrometry).
40 R:I:s  R:I:atm
Ar % ¼ ¼ 4:31% The precise knowledge of 39Ar production yield is
R:I:s
obtained by referring to known age standards. These stan-
The number of argon 40Ar atoms in the sample is calculated dards are irradiated in the same shuttles as the samples. The
using the calibration data. We know from the calibration curve radiation yield is calculated according to the equation
that 3.912 volts correspond to 1.608 2
1013 atoms. Therefore, established by Mitchell (1968).
1.196 V (40Ar sample) is equivalent to 0.492
1013 atoms. Z
39
The concentration of atoms per gram is this value divided by the Ars ¼ 39 KDT UE rE dE ¼ 39 KDTI ð5:5Þ
weight of the melted sample (1.0669 g) so 0.461
1013 at./g.
4.31% of the measured 40Ar is radiogenic. This gives 40Ar* at./ Z1
g = 0.461
1013
0.0431
1.987 = 1013. I¼ U E r E dE
The number of 40K atoms is calculated using:
0
 
40 K
0:01
K¼
0:0001167
6:023
1023 with: 39K being the number of atoms of 39K in the standard
39:098304 sample; 39Ars the number of atoms of 39Ar produced in the
¼ 1:156
1016 standard sample; UE, the energy flux; rE, zone of efficient
 40
5 The K/40Ar and 40
Ar/39Ar Methods 81

capture of the reaction 39K ! 39Ar at energy E; DT, the references may be consulted among others: Kuiper et al.
irradiation duration. (2008), Renne et al. (2010), Jicha et al. (2016), Niespolo
The amount of 40Ar* produced by the disintegration of et al. (2017).
40
K follows the equation:

40 ke 40
kts 
Ar ¼ K e 1 ð5:6Þ Corrections for Atmospheric Argon
k and Interference of Mass
with ts, known age of the standard.
Combining (5.5) and (5.6), we obtain: As for the 40K/40Ar method, the correction for atmospheric

 argon is essential in order to determine the proportion of
40
Ar 40 K ke 1 ekts  1
40
Ar*. This adjustment is done by repeated mass spectro-
39 Ar
¼ 39 ð5:7Þ
K k DT I metric measurements of aliquots of air. This defines the
instrumental 40Ar/36Ar atmospheric ratio. Most mass spec-
Equation (5.7) is simplified by defining the J parameter
trometers give values slightly different from 298.56. It is
which is the radiation flux actually received by the sample:
therefore through the repeated measurements of aliquots of
39
Kk air that the bias in the measuring apparatus can be calculated.
J ¼ 40 DTI ð5:8Þ As a first approximation and for samples without calcium, the
K ke
determination of the percentage of radiogenic argon can be
From (5.7), J becomes, done by simply comparing the 40Ar/36Ar ratio of the sample
and the instrumental 40Ar/36Ar ratio of the atmosphere.
ekts  1
J ¼ 40 ð5:9Þ However, during irradiation, secondary reactions occur
Ar =39 Ar from Ca, K and Cl isotopes which also produce artificial
It is then possible to resolve the age equation: isotopes of argon (Fig. 5.8):
 
1 ð40 Ar =39 ArÞe
kts  • 40
Ca(n, na) 36Ar
te ¼ ln 1 þ ð40  39 e 1 ð5:10Þ
k Ar = ArÞs • 42
Ca(n, a) 39Ar
• 40
K(n, p) 40Ar
with s = standard and e = sample. • 35
Cl(n, c) 36Cl − b ! 36Ar t1/2 = 300
103 years
The following table shows the age of the main standard • 37
Cl(n,c) 38Cl − b ! 38Ar t1/2 = 37.3 min
(or flux) minerals for 40K/40Ar and 40Ar/39Ar methods.
The correction for interference of the masses 40, 39 and
Name Mineral Age (Ma) References 36 due to Ca is possible because of an additional reaction:
Hb-3gr Hornblende 1072 ± 11 Turner et al.
(1971) • 40
Ca(n, a) 37
Ar t1/2 = 35.1 days
MMhb-1 Hornblende 520.4 ± 1.7 Samson and
Alexander
To know the 39ArCa and 36ArCa contents, it is necessary to
(1987)
irradiate a pure calcium salt, such as CaF2, and measure its
LP-6 Biotite 127.9 ± 1.1 Odin (1982)
39/37 and 40/37 ratios with the mass spectrometer. The
SB-2 Biotite 162.1 ± 2.0 Dalrymple initial value of argon 37Ar of the irradiated sample, when it is
et al. (1981)
taken from the reactor, also needs to be calculated. This is
GA-1550 Biotite 97.9 ± 0.9 McDougall
done by applying the law of radioactive decay:
and Roksandic
(1974) 37


Ar0 ¼ 37 Arm ek37 ti k37 ti = 1  ek37 ti
B4M Muscovite 18.6 ± 0.4 Flish (1982)
B4B Biotite 17.3 ± 0.2 Flish (1982) with 37Ar0 = amount of the isotope 37 produced at the end
FCTs Sanidine 28.187 ± 0.019 Phillips et al. of the irradiation; 37Arm = amount of the isotope 37 mea-
(2017) sured on the day of analysis; t = duration of the irradiation;
ACRs Sanidine 1.18404 ± 0.00068 Phillips et al. ti = time interval between irradiation and analysis;
(2017) k37 = 0.0197 4 j−1.
The correction factors (39Ar/37Ar)Ca and (36Ar/37Ar)Ca,
The age assigned to the standards may vary depending on which depend on the yield from the irradiation on the salts,
the authors. With regard to FCTs and ACRs, the following are thus well defined.
82 H. Guillou et al.

Fig. 5.8 Mass spectrum of an

irradiated sample, deciphering the
origin of the different isotopes (in
red) generated from the neutronic
activation

In order to know the correction factor associated with the

production of 40Ar from 40K, a pure salt of K (K2SO4 or KF)
is also irradiated. This allows the ratio (40Ar/39Ar)K to be
defined, which is then used to calculate the share of argon
40
Ar resulting from irradiation of 40K, which will be cut
back from the total argon 40Ar.
In summary, the masses corrected for interferences related
to radiation and to the atmospheric component can be written:
40


Ar ¼ 40 Arm  40 Arat þ 40 ArK
39
Ar ¼ 39 Arm  39 ArCa
36
Ar ¼ 36 Arat  36 ArCa

The Age Spectra

Fig. 5.9 Age spectra depicting 40Ar/39Ar experimental results from
The 40Ar/39Ar method allows for the collection of more duplicated measurements of sample G-04, a lava from the Gelso section
comprehensive information on the behavior of the of Vulcano (Aeolian Islands, Italy). Uncertainties are ±2r. In red, the
radioisotopic clock than the 40K/40Ar method. In the first step-heating experiment; in blue, the second step-heating
experiment
experimental approach known as ‘step-heating’ (Turner
et al. 1966), the sample is gradually heated in steps of
increasing temperature (for example, by steps of 60◦C). At steps. The result of such experiments will be a consistent ‘age
each step, the isotopic composition of argon in the extracted spectrum’ in horizontal form which defines the plateau age.
and purified gas is measured by mass spectrometry. An Several definitions have been proposed for a plateau age
apparent age can thus be calculated for each step. In the end, (Dalrymple and Lanphere 1974; Berger and York 1981;
this results in an age spectrum. The general appearance of McDougall and Harrison 1988). It is generally considered
these spectra shows whether the sample and, consequently, that a plateau is composed of at least three successive steps
the 40K/40Ar clock, were disrupted or not. with at least 60% of the 39Ark released and whose apparent
In the case of an undisturbed sample (Fig. 5.9), which ages are consistent within ±2 sigma (i.e. within the respec-
evolved in a closed system, the K is homogeneously dis- tive analytical error bars at 95.6% confidence interval).
tributed in the crystal lattice. This is also true for 40Ar* and Cases of inconsistent age spectra are relatively common
39
Ar. When a sample is subjected to degassing in increasing and the reasons for this discrepancy are varied. One of the most
temperature steps, the 40Ar* and 39Ar isotopes will be frequently cited reasons is related to the ‘recoil effect’. The
extracted at a constant ratio. As a consequence, a similar transformation of 39K into 39Ar by the reaction 39K(n, p)39Ar
apparent age, within error margins, will be obtained for all can be accompanied by a loss of 39Ar. In other words, the 39Ar
 40
5 The K/40Ar and 40
Ar/39Ar Methods 83

atom can be founder-distributed, due to this reaction, in a The Isochrones

different crystallographic site than that occupied by its parent,
39
K. The distance of this backwards movement is proportional, The 40Ar/39Ar method is particularly suited to data processing
firstly, to the energy employed during the neutron activation, by isochrones. In the inverse isochron diagram 40Ar/36Ar
and secondly, to the sample density. Thus, this recoil effect can versus 39Ar/36Ar (Fig. 5.11), the slope is equivalent to the
lead to a redistribution of 39ArK in the samples with grains of a 40
Ar*/39ArK ratio which is itself proportional to the age, and
size less than 5–10 lm. This redistribution of 39ArK can cause the intercept on the y axis corresponds to the (40Ar/36Ar)I ratio.
an over or underestimation of apparent ages at different tem- This last ratio indicates the proportion of 40Ar and of 36Ar at
perature steps. The loss of 39ArK, related to the recoil effect, is t = 0, in other words, at the moment the system closed. This
seen in an overestimation of the ages obtained. value is directly comparable to the 40Ar/36Ar atmospheric
Other natural phenomena are also the cause of inconsis- ratio. It is thus possible, from the inverse isochron diagram, to
tent age spectra. Alteration, metamorphism, hydrothermal- highlight the presence or absence of excess argon. This
ism are all processes that might disrupt the K/Ar clock. information is particularly important because it allows one of
These processes are the source of migration, loss or gain of the basic assumptions for application of the clock to be
argon and potassium isotopes. checked, namely that an age is considered correct if, at t = 0,
40
Ar* = 0 and (40Ar/36Ar)initial = (40Ar/36Ar)atmospheric.
This analysis by isochron is particularly useful for the
The Single Grain Method dating of tephra. Indeed, ideally, all the minerals from the
same layer of a tephra should be on the same isochron, as
Technological developments of recent decades (increasing they have, in principle, the same age. Furthermore, the value
sensitivity of mass spectrometers, laser fusion system) make (40Ar/36Ar)i must be equivalent to the atmospheric value. If
it possible to work on samples of smaller and smaller size. It some experimental points are not on this isochron, we can
is thus possible to get down to the level of the crystal. This deduce that the corresponding crystals are xenocrysts i.e.
approach is used in particular for dating tephra. Within a older crystals remobilized during the eruptive event at the
given tephra layer, several crystals of the same mineral type origin of the tephra (Fig. 5.11).
are selected. After irradiation, each crystal is individually
melted by laser, the gas is extracted and it is then analyzed
by mass spectrometer. An age is obtained for each con- Selection and Preparation of Samples
stituent crystal at the tephra level. It is thus possible to
establish spectra of age probability for a given tephra (Deino The procedure is the same as that followed for the K/Ar
and Potts 1992). The analysis of these spectra allows method.
homogeneity at the stratigraphic level to be estimated and
the most statistically probable age to be defined (Fig. 5.10).

Fig. 5.10 Probability diagram obtained for the stratigraphic level N°

2-2 (Notarchirico archaeological site, Basilicata, Italy). 15 crystals were Fig. 5.11 Inverse isochron diagram obtained for the stratigraphic level
analyzed. Experiments in blue (11 crystals) define an age of N° 2-2 (Notarchirico archaeological site, Basilicata, Italy). 15 crystals
661 ± 14 ka. Red boxes are xenocrysts (4 crystals out of the 15 were analyzed. Experiments in blue define an age of 660 ± 10 ka. Red
analyzed) and as a consequence, eliminated from the age calculation. boxes are xenocrysts, eliminated from the age calculation. Data are
Data are from Pereira et al. (2017) from Pereira et al. (2017)
84 H. Guillou et al.

Mass-Spectrometric Analysis Calculation of Age

• Stage 1: pumping—pre-degassing. The sample (crystal Determination of the J factor (Neutron flux received
or microcrystalline groundmass) is placed in a vacuum during irradiation)
chamber including a crucible in the case of fusion in a J is determined for each sample. See the following analysis
furnace, and a viewport in the case of fusion by Laser of a ACR-2 standard grain (1.194 Ma) subjected to a fast
CO2), then placed under secondary high vacuum by neutron flux for 30 min (Osiris reactor, CEA Saclay):
means of turbo-molecular pumps. The sample is then
heated to about 500 °C. This first gas extracted is elimi-
40 39 38 37 36
Ar Ar Ar Ar Ar
Measured 9.246
10−3 2.321
10−3 3.576
10−5 1.188
10−6 1.375
10−6
nated by pumping. (mV)
• Stage 2: fusion. The sample is melted either by a CO2 Blank (mV) 2.351
10−5 2.586
10−7 2.238
10−8 6.672
10−7 2.383
10−7
laser or in a double vacuum resistance furnace. In the case Corrected 2.312
10−3 7.967
10−4 1.234
10−5 3.161
10−7 2.981
10−6
measurement
of fusion by laser, it applies essentially to the analysis of
single crystals and to small groups of crystals (about 5–15
grains). The furnace is used for the analysis of micro-
crystalline groundmass. As seen before, this allows a step Using Eq. (5.9), J can be calculated by setting the
40
by step fusion, and hence a gradual degassing of the Ar*/39ArK ratio or Re as follows:
40 36  
sample required for the ‘step-heating’ method. ½40 Ar=39 Arm  Ar=36 Ar Ar=39 Ar m þ 40 Ar=36 Ar A 36 Ar=37 Ar ½37 Ar=39 Arm
Re ¼ A Ca
• Stage 3: purification. The gas extracted from the sample 40 
39 37 37 39
1  ½ Ar= ArCa ½ Ar= Arm
Ar
is purified by the combined effect of Getters pumps and a  39 Ar
K
titanium sublimation pump.
• Stage 4: measurement by mass spectrometer. After
purification, the gas is introduced into the mass spec-
trometer. The purified gas is measured using a with m: measured ratios (see table above).
high-sensitivity noble gas GV5400 instrument operated in
ion-counting mode. One analytical run consists of 20 [40Ar/36Ar]A = atmospheric reference ratio = 292.8 (for this
peak scans of each argon isotope with integration times of sample);
1 s (40Ar, 39Ar) or 10 s (36Ar, 37Ar, 38Ar, baseline), first
preceded by a peak centering routine on the five Ar iso- [36Ar/37Ar]Ca = (given by calcium salt) 5.60
10−4;
topes, upon admission of the sample into the
mass-spectrometer. The precision and accuracy of the [39Ar/37Ar]Ca = (given by calcium salt) 6.95 10−4;
mass discrimination correction is monitored by periodical
measurements of air argon. This monitoring is performed [40Ar/39Ar]K = (given by potassium salt) 3.52 10−3;
using a dedicated air-calibration system featuring a 6 L
tank filled with purified atmospheric argon. This tank is Re ¼
3:974  292:8
4:898
104 þ 292:8
5:60
104
5:09
102
1  6:95
104
5:09
102
connected to the mass spectrometer vacuum line via two
 3:52
103 ¼ 3:8355
pneumatically- actuated air pipettes of approximately 0.1
and 1.0 cc. This system allows for a 1 cc (e.g. 600,000 If ts = 1.194 Ma and k = 5.543
10−10; J is then cal-
counts s−1 (cps on 40Ar)) and a 0.1 cc (e.g. 70 000 cps on culated as follows:
40
Ar) atmospheric aliquots to be delivered into the mass

spectrometer and permits a careful monitoring of the mass J ¼ e1:194
0:0000000005543  1 =3:8355 ¼ 1:726
104
discrimination over a wide dynamic range. The mineral
The measurement is repeated on at least three grains to
standards used to calculate the flux are analyzed in the
quantify any possible external errors originating from the
same way as the ordinary samples.
heterogeneity of the age standard (*1%). A weighted
 40
5 The K/40Ar and 40
Ar/39Ar Methods 85

average is then calculated from these measurements and Application: Example of the Dating
used to calculate the sample value. of the Laschamp Event

Age Calculation of a Sample In the chapter on magnetic stratigraphy (Chap. 7), the
Here are the measurements obtained for 1 sanidine crystal importance of the dating of geomagnetic events is discussed.
irradiated for 90 min in the Osiris reactor (CEA Saclay): Here, we will show how the Laschamp excursion could be
correctly dated with a high degree of precision.
40 39 38 37 36
Ar Ar Ar Ar Ar
−03 −04 −05 −07
The dating of this excursion was obtained by a
2.491
10 7.969
10 1.241
10 4.612
10 3.136
10−06
geochronological study combining the 40K/40Ar and
Measured (mV)

Blank (mV) 1.791
10−04 1.515
10−07 7.158
10−08 1.451
10−07 1.554
10−07
40
Corrected 2.312
10−03 7.967
10−04 1.234
10−05 3.161
10−07 2.981
10−06 Ar/39Ar methods applied to two lavas from the Massif
measurement
Central (Guillou et al. 2004). Prior to this study, estimates of
ages were imprecise and inconsistent with the ages deduced
from other means of dating, such as astronomical calibration.
By combining the expression of J (Eq. 5.9) with Two lava flows have been subjected to paleomagnetic
Eq. 5.10, the age calculation gives: and geochronological study. One of these comes from the
  ‘Puy de Laschamp’ part of the Chaine des Puys, located in
1 40
Ar
te ¼ ln 1 þ J 39 the French Massif Central. The second, called ‘the Olby
k ArK
flow’ comes from the ‘Puy de Barme’ also part of the Chaine
with J = 6.530
10−4 (calculated for this sample); des Puys.
k = 5.543
10−10 (constant for total decay of 40K); For each sample, the K-Ar ages (unspiked method) are
Re = 1.790 8 (see calculation of Re above with calculated from two independent measures of potassium and
[40Ar/36Ar]A = 296.1); three, also independent, measures of argon. The ages
te = 1/(5.543
10−10)
ln(1 + 6.530
10−4
1.790 obtained for the Laschamps lava flow (41.5 ± 1.9 ka) and
8) = 2.108 Ma. that of Olby (41.4 ± 1.9 ka) are identical at the two sigma
level. The weighted average of these two values gives an age
of 41.4 ± 1.4 ka. As for 40Ar/39Ar ages (Fig. 5.12), seven
40
Advantages and Limitations of the K/40Ar experiments out of thirteen give consistent age spectra, for
and 40Ar/39Ar Methods which 100% of the extracted gas could be used to define a
plateau age. For the other six experiments, between 76 and
The table below summarizes the advantages and limitations 96% of the extracted gas was used to define a plateau age.
of both methods. Furthermore, the intercept values calculated from the inverse
isochron diagrams are equivalent to the atmospheric ratio.
Method 40
K/40Ar 40
Ar/39Ar This indicates that the age determinations are not marred by
Advantages • Rapid implementation • Basic assumptions can error due to either a loss or gain of argon. The weighted
• No need for prior be verified (age average for isochron ages for the two sampling sites for the
irradiation of samples spectrum, isochrons) Laschamps lava flow were 39.4 ± 2.6 ka and
• Precise measurement • Dating possible on 38.3 ± 2.6 ka. The Olby lava flow has an age of
of low amounts of very small sample
40
Ar* (well-suited to sizes (grain by grain 39.2 ± 4.9 ka. The combination of these three ages give a
young basalts of dating well suited to weighted average of 38.9 ± 1.7 ka.
mid-oceanic ridges) tephra) The K-Ar and 40Ar/39Ar measurements are compatible at
Limitations • The basic assumptions • Pre-irradiation leads to the two sigma level. As these two flows recorded the same
(initial
40
corrections paleomagnetic excursion, this is dated to 40.4 ± 1.1 ka.
Ar/36Ar = 298.56, (interference of
Note that if we take the uncertainty (2.4%) on the potassium
evolution in a closed masses)
system) for application • The recoil effect makes decay constant into account, the error in age goes from 1.1 to
of the clock are not dating of very 2.0 ka. Thus, the age retained for the excursion is
verified fine-grained samples 40.4 ± 2.0 ka. This age is comparable to that obtained by
• Large weight (>1 g) of or ones with a glassy
independent chronological methods (see Chap. 7). A new
sample required texture complicated
These two points study (Laj et al. 2014), combining K-Ar and 40Ar/39Ar
prohibit the dating of dating, associated with paleomagnetism, was applied to a
tephra by the K-Ar larger number of volcanoes from the Chaine des Puys, and
method
has since made it possible to narrow down the age of the
Laschamps excursion to 41.2 ± 1.6 ka.
86 H. Guillou et al.

Fig. 5.12 Age spectra and isochron diagrams for samples from the Laschamp-1, Laschamp-2 and Olby sites. Isochron ages and statistics are
weighted mean values from experiments on all the subsamples, (in Guillou et al. 2004). n = number of steps retained in the age calculation

This example highlights the potential of the K-Ar dating References

clock to date recent events of the Quaternary period. One of
the applications of these chronological data to calibrate time Berger, G. W., & York, D. (1981). Geothermometry from 40Ar/39Ar
scales, with various other useful applications in the earth dating experiments. Geochimica et Cosmochimica Acta, 45, 795–
sciences, archeology and paleoclimatology. 811.
 40
5 The K/40Ar and 40
Ar/39Ar Methods 87

Cassignol, C., Cornette, Y., David, B., & Gillot, P. Y. (1978). McDougall, I., & Roksandic, Z. (1974). Total fusion 40Ar/39Ar ages
Technologie potassium-argon. C.E.N., Saclay, Rapport CEA using HIFAR reactor. Geological Society of Australia, 21, 81–89.
R-4802, 37 p. Merrihue, C. (1965). Trace element determinations and potassium
Cassignol, C., & Gillot, P.-Y. (1982). Range and effectiveness of argon dating by mass spectroscopy of neutron irradiated samples.
unspiked potassium-argon dating: Experimental groundwork and Transactions American Geophysical Union, 46, 125.
examples. In G. S. Odin (Ed.), Numerical dating in stratigraphy Merrihue, C., & Turner, G. (1966). Potassium-argon dating by
(pp. 159–179). Chichester: Wiley. activation with fast neutrons. Journal of Geophysical Research,
Charbit, S., Guillou, H., & Turpin, L. (1998). Cross calibration of K-Ar 71, 2852–2857.
standard minerals using an unspiked Ar measurement technique. Min, K. W., Mundil, R., Renne, P. R., & Ludwig, K. R. (2000). A test
Chemical Geology, 150, 147–159. for systematic errors in 40Ar/39Ar geochronology through compar-
Dalrymple, G. B., Alexander, E. C. Lanphere, M., & Kraker, G. ison with U-Pb analysis of a 1.1 Ga rhyolite. Geochimica
P. (1981). Irradiation of samples for 40Ar/39Ar dating using the Cosmochimica Acta, 64, 73–98.
geological survey TRIGA reactor. U.S. Geological Survey Profes- Mitchell, J. G. (1968). The Argon-40/Argon-39 dating in
sional Paper, 1176. coesite-bearing and associated units of the Dora Maira Massif,
Dalrymple, G. B., & Lanphere, M. A. (1969). In J. Gilluly & A. Western Alps. European Journal of Mineralogy, 3, 239–262.
O. Woodford (Eds.), Potassium-argon dating. Principles, tech- Nier, A. O. (1950). A redetermination of the relative abundances of the
niques and applications to geochronology (251 p.). San Francisco, isotopes of carbon, nitrogen, oxygen, argon and potassium. Physical
CA: W. H. Freeman and Company. Review, 77, 789–793.
Dalrymple, G. B., & Lanphere, M. A. (1974). 40Ar/39Ar age spectra of Niespolo, E. M., Rutte, D., Deino, A. L., & Renne, P. R. (2017).
some undisturbed terrestrial samples. Geochimica et Cosmochimica Intercalibration and age of the Alder Creek sanidine 40Ar/39Ar
Acta, 38, 715–738. standard. Quaternary Geochronology, 39, 205–213.
Deino, A. L., & Potts, R. (1992). Age-probability spectra for Odin, G. S. (1982). Interlaboratory standards for dating purposes. In G.
examination of single crystal 40Ar/39Ar dating results: Examples S. Odin (Ed.), Numerical dating in stratigraphy (pp. 123–158).
from Olorgesailie, Southern Kenya Rift. Quaternary International, New York: Wiley.
13(14), 47–53. Pereira, A., Nomade, S., Bahain, J.-J., & Piperno, M. (2017). Datation
Flish, M. (1982). Potassium-argon analysis. In G. S. Odin (Ed.), par 40Ar/39Ar sur monocristaux de feldspaths potassiques: exemple
Numerical dating in stratigraphy (pp. 151–158). New York: Wiley. d’application sur le site pléistocène moyen ancien de Notarchirico
Fuhrmann, U., Lippolt, H., & Hess, C. J. (1987). HD-B1 biotite (Basilicate, Italie). Quaternaire, 28(2), 149–154.
reference material for K-Ar Chronometry. Chemical Geology, 66, Phillips, D., Matchan, E. L., Honda, M., & Kuiper, K. F. (2017).
41–51. Astronomical calibration of 40Ar/39Ar reference minerals using
Garner, E. L., Murphy, T. J., Gramlich, J. W., Paulsen, P. J., & Barnes, high-precision, multi-collector (ARGUS VI) mass spectrometry.
I. L. (1975). Absolute isotopic abundance ratios and the atomic Geochimica Cosmochimica Acta, 196, 351–369.
weight of a reference sample of potassium. Journal of Research of Renne, P. R. (2000). In J. S. Noller, J. M. Sowers, & W. R. Lettis
the National Bureau of Standards, 79A, 713–725. (Eds.), K-Ar and 40Ar/39Ar dating, in quaternary geochronology:
Guillou, H., Singer, B. S., Laj, C., Kissel, C., Scaillet, S., & Jicha, B. R. Methods and applications. Washington, D.C.: American Geophys-
(2004). On the age of the Laschamp geomagnetic excursion. Earth ical Union. https://doi.org/10.1029/rf004p0077.
and Planetary Science Letters, 227(3–4), 331–343. Renne, P. R., Mundil, R., Balco, G., Min, K., & Ludwig, K. R. (2010).
Jicha, B. R., Singer, B. S., & Sobol, P. (2016). Re-evaluation of the Joint determination of 40 K decay constants and 40Ar/40K for the
ages of 40Ar/39Ar sanidine standards and super-eruptions in the Fish Canyon sanidine standard and improved accuracy for
40
western US using a Noblesse multi-collector mass spectrometer. Ar/39Ar geochronology. Geochemica et Cosmochimica Acta, 74,
Chemical Geology, 431, 54–66. 5349–5367.
Kuiper, K. F., Deino, A., Hilgen, F. J., Krijgsman, W., Renne, P. R., & Roddick, J. C. (1983). High precision intercalibration of 40Ar-39Ar
Wijbrans, J. R. (2008). Synchronizing rock clocks of Earth history. standards. Geochemical and Cosmochimical Acta, 47, 887–898.
Science, 320, 500–505. Samson, S. D., & Alexander, E. C., Jr. (1987). Calibration of
Laj, C., Guillou, H., & Kissel, C. (2014). Dynamics of the earth interlaboratory 40Ar-39Ar Dating Standard, MMhb-1. Chemical
magnetic field in the 10-75 kyr period comprising the Laschamp Geology, 66, 27–34.
and Mono Lake excursions: New results from the French Chaîne Steiger, R. H., & Jäger, E. (1977). Subcommission on geochronology:
des Puys in a global perspective. Earth and Planetary Science Convention on the use of decay constants in geo- and cos-
Letters, 387, 184–197. mochronology. Earth and Planetary Science Letters, 5, 320–324.
Lanphere, M. A., & Dalrymple, G. B. (2000). First principles Turner, G., Huneke, J. C., Podosek, F. A., & Wasserburg, G. J. (1971).
calibration of 38Ar Tracers. U.S. Geological Survey Professional 40
Ar-39Ar ages and cosmic ray exposure rays of apollo 14 samples.
Paper, 1621. Earth and Planetary Science Letters, 12, 19–35.
Lee, J. Y., Marti, K., Severinghaus, K., Kawamura, K., Yoo, H. S., Lee, Turner, G., Miller, J. A., & Grasty, R. L. (1966). The thermal history of
J. B., et al. (2006). A redetermination of the isotopic abundances of the bruderheim meteorite. Earth and Planetary Science Letters, 1,
atmospheric Ar. Geochimica Cosmochimica Acta, 70, 4507–4512. 155–157.
McDougall, I., & Harrison, T. M. (1988). Geochronology and Wänke, H., & König, H. (1959). Eine neue Methode zür Kalium-Argon
thermochronology by the 40Ar/39Ar method (p. 212). New York: Alterbestimmung und ihre Anwendung auf Steinmeteorite. Zeits-
Oxford University Press. chrift für Naturforschung A, 14a, 860–866.
 Dating of Corals and Other Geological
Samples via the Radioactive Disequilibrium 6
of Uranium and Thorium Isotopes

Norbert Frank and Freya Hemsing

Abstract concept is that, in seawater, there is dissolved uranium but

very little thorium, an insoluble element. Therefore, each
U/Th dating methods have become cornerstone tools for
crystal of aragonite formed by the coral incorporates only
the determination of the age of climate change recorded in
uranium and not its first-generation daughters which are all
marine and continental carbonates. Here we describe the
thorium isotopes. Uranium concentration in seawater is very
theoretical principles and analytical methods along the
stable and homogeneous with 3.3 lg of U per liter of water.
example of U/Th dating the aragonite skeletons of
The activity ratio (234U/238U) is also very stable in the ocean
tropical corals. We demonstrate that a precision limiting
and is slightly above the radioactive equilibrium with a value
factor is built in the dating principle, known as U-series
of 1.1468 ± 0.0001 in the open ocean (Andersen et al. 2010).
open system behavior. Moreover, above all the quality of
This slight excess of 234U is due to preferential leaching of this
the samples is crucial for a successful and accurate age
isotope from rocks during weathering of the continental crust
determination. When using well preserved fossil coral
(Ivanovich and Harmon 1992). Henderson (2002) proposed
fragments ages provide measures of past sea level,
that the seawater ratio (234U/238U) has remained constant for at
contribute to the calibration of the radiocarbon time
least the past 800,000 years. However, moderate variations
scale, and allow for the reconstruction of reef accumu-
of ± 0.01 in this ratio are possible, due to changes in
lation rates, tectonic subsidence, or uplift. We finally
weathering of the continental crust, sea level changes, and
emphasize that U/Th dating also works for secondary
variations in freshwater runoff from rivers resulting from cli-
carbonates such as stalagmites, calcareous tuff and
mate changes (Esat and Yokoyama 2006). Very recently, it has
travertine, but the boundary conditions regarding the U
been suggested that this ratio may also be dependent on the
concentration, and the initial U and Th isotopic compo-
ocean circulation (Chen et al. 2016).
sition vary wildly and need careful consideration. When
Let us go back to the coral. If the skeleton of a coral
doing so, highest precision and accuracy can Is feasible.
remains a chemically closed system after its formation (in
other words, without any exchange of uranium or thorium
Dating methods based on the radioactive disequilibrium in the
with its sedimentary environment) the 230Th from the decay
uranium decay series were developed over the last fifty years.
of 234U accumulates progressively over time, while the
They can be applied to minerals that, at the time of their
excess of 234U decreases.
formation, incorporate uranium into their crystal lattice, but not
The state of this radioactive disequilibrium allows for a
thorium, whose isotope, 230Th, is a daughter of 234U (Fig. 6.1).
very precise determination of the coral age depending on the
This is the case for corals which form their aragonitic, cal-
measurement technique used. When this dating method was
careous skeletons from elements present in the seawater. They
developed, the isotopes of uranium and thorium were mea-
are used as an example throughout this chapter. The basic
sured by their radioactivity, either directly by a spectrometry
or indirectly by c spectrometry, enabling age determination
N. Frank ranging from a few thousand years to about 300,000 years.
Laboratoire des Sciences du Climat et de L’Environnement, Now, we are able to measure the abundance of these iso-
LSCE/IPSL, CEA-CNRS-UVSQ, Université Paris-Saclay,
Gif-Sur-Yvette, 91190, France topes by characterizing them according to their masses,
giving a much better accuracy. The age range measurable by
N. Frank (&)
F. Hemsing
Institute of Environmental Physics, Heidelberg University, thermal ionization mass spectrometry or even multi-collector
Im Neuenheimer Feld 229, 69120 Heidelberg, Germany inductively coupled plasma source mass spectrometry now
e-mail: [email protected]

© Springer Nature Switzerland AG 2021 89

G. Ramstein et al. (eds.), Paleoclimatology, Frontiers in Earth Sciences,
https://doi.org/10.1007/978-3-030-24982-3_6
90 N. Frank and F. Hemsing

travertine). In each case, the formation process of these min-

erals favors the incorporation of uranium over its radioactive
daughters. However, each mineral deposit has its own char-
acteristics with regard to the incorporation of uranium and
thorium and open system behavior Cheng et al. (2000a),
Mallick and Frank (2002). Here, we will focus on one par-
ticular archive: tropical corals. At the end of the chapter, the
possibility of dating other geological samples through uranium
238
series disequilibrium will be analyzed briefly.
Fig. 6.1 Start of the radioactive decay chain U

230
allows the dating range to be extended from a few years to a Methodology of Th/238U Dating
few hundred thousand years (>500,000 years). As a result,
230
U/Th dating has now become an indispensable tool for Principle of Th/238U Dating
precision geochronology and research on the environmental
and climate changes in the late Quaternary. This method is based on the radioactive decay chain of 238U
Undoubtedly, one of the greatest successes of this (Fig. 6.1).
chronometer is the precise reconstruction of sea levels The rates of decay in this radioactive chain are highly
throughout the climate cycles from growth series of corals variable, ranging from a few billion years to hours. The
(Thompson and Goldstein 2006). This geochronological radioactivity of 234Th and 234Pa decreases very rapidly with
technique also permits the determination of subsidence or half-lives (also called periods) of 24.1 days and 6.7 h
elevation of reefs caused by tectonic movements over time respectively. Therefore, these isotopes are insignificant rel-
(Frank et al. 2006). Significant progress has also been made ative to the periods of 234U and 230Th, key isotopes in
230
on the calibration of 14C ages by combined 230Th/U and 14C Th/238U dating. During the formation of the skeleton,
analysis (Reimer et al. 2013). only the U isotopes are incorporated, hence 230Th isotope is
Over the past decade, it has become clear that the mineral absent (230Th concentration is zero at t = 0). It is therefore
system of a coral is not a completely closed system but only by decay of 234U that 230Th accumulates over time in
sometimes uranium and thorium exchanges with its sedi- the coral structure.
mentary environment occur. This conclusion was reached To date a sample, it is essential to accurately measure the
from the observation that uranium 234U and 230Th are often activities of 238U, 234U and 230Th. Dating is performed by
in excess of the theoretical levels of the activity ratios using the radioactive decay equations to define the activity
(234U/238U and 230Th/238U) that would be expected from ratios of 234U/238U and 230Th/238U so that the time elapsed
radioactive decay. These inconsistencies are the result of since the formation of the coral skeleton can be calculated
recrystallization, early diagenesis, and radioactive decay of (Eqs. 6.1 and 2). The decay equation cannot be solved
uranium. This makes the dating less precise than the analysis analytically but ages are estimated by iteration.
might indicate and makes it necessary to apply corrections to
230 
Th d234 Um k230
the estimated ages. ¼ 1  ek230 t þ 
In light of these observations, correction models, cate-
238 U
 1000  k 230  k234
ðk230 k234 Þt
gorized as ‘open’ system models were developed which take  1e ð6:1Þ
account of the influence of such disturbances on tropical

234 
234 
corals. (Scholz et al. 2004, Szabo et al. 1994) U U  
¼ 238 initial  1  ek234 t ð6:2Þ
The focus of this chapter is to explore the dating of 238 U U
tropical corals and other carbonate climate archives using the
U/Th method including open-system models, but excluding with k230 and k234, the constants of decay of 230Th and 234U
microstructure and biomineralization. We will examine the respectively (Table 6.1), and t the time elapsed since the
methodology in detail including the relevance of the open system closed.
system and also some applications to geology and In Eq. (6.1), the activity ratio (234U/238U) is expressed as
paleoclimatology. ‰ compared to the radioactive equilibrium:
This principle of dating by radioactive disequilibrium in the 2234  3
U
decay chain of uranium applies to many sources of climate 238 U

records, such as deep-water corals, mollusk shells, and sec- d234 U ¼ 4 234 U   15  1000 ð6:3Þ
238 U
ondary carbonate precipitations on land (stalagmites and
6 Dating of Corals and Other Geological Samples … 91

Table 6.1 Decay of the T1/2 (years) k

daughters of 238U by Cheng et al.
(2000b) with T1/2 the half-life and
238
U 4,468,314,000 1.55125  10−10
k the decay constant 234
U 245,250 2.82629  10−06
230
Th 75,690 9.15771  10−06

230
Th). The mass spectrometric methods have the advantage
of directly measuring the number of atoms of an isotope,
instead of measuring its radioactivity. This results in a very
significant gain in accuracy because instead of measuring a
few alpha emissions per minute, tens to millions of ions per
second are detected. However, a very rigorous preliminary
chemical treatment is essential as the presence of any other
elements during ionization will decrease the quality of the
measurement or even lead to isobaric interferences. From the
number of atoms of each radionuclide measured by mass
spectrometry, its activity Ai can be calculated according to
the law of decay Ai = k  Ni; where Ni is the number of
atoms measured by mass spectrometry and k is the decay
constant of the radionuclide. To calculate the activities of the
daughters of 238U from isotopic measurements, we use the
Fig. 6.2 Temporal evolution of the activity ratios (230Th/238U) and decay rates shown in Table 6.1. These decay times have
d234U in a closed system. The points (◦) show the position of the coral recently been updated Cheng et al. (2013), but we have
every 50 thousand years opted to use the older values (Table 6.1) here because the
re-evaluated values by Cheng et al. (2013) have still not
For the method to be applicable, two basic conditions been independently confirmed and are identical to the pre-
need to hold: vious ones within a range of uncertainty.
In the following paragraphs, we will discuss the analytical
1. During the formation of the aragonitic skeleton, only aspect. However, we will firstly specify the sample selection
uranium is incorporated, and therefore 230Th is absent. criteria, before explaining the chemical process and finally
2. The system remains closed to any exchange of uranium the physical measurement (Fig. 6.3).
and thorium with the sedimentary environment after the
aragonite is formed.
230
Selecting a Coral for Th/238U Dating
If a coral satisfies these essential conditions, the activity
ratios of (234U/238U) and (230Th/238U) evolve over time In order to get good accuracy on the age of the corals, it is
(Fig. 6.2—assuming an activity ratio of (234U/238U) in sea- crucial to select the best-preserved samples possible. The
water equal to 1.1468 (d234U = 146.8‰) at the moment of purpose of the analysis is to obtain ratios between 238U, 234U
precipitation of the aragonite). and 230Th, produced exclusively by radioactive decay, and
The accuracy of U/Th dating depends mainly on the thus independent of any secondary addition or any depletion
accuracy with which the isotope ratios are measured, but the of uranium or thorium over the thousands of years the coral
quality of the sample also plays a critical role because it has spent as fossil in the reef environment. For this purpose,
determines whether the fundamental conditions for dating the selected coral fragments are physically and chemically
are met. cleaned of any metal oxide encrustations, traces of clay or
All of the radionuclides are alpha emitters. Historically, organic residue identified beforehand by binocular micro-
alpha spectrometry, i.e. the direct measurement of the alpha scope. This cleaning is carried out over several iterations of
emission of each radionuclide, was the preferred method. rinsing in an ultrasonic bath with ultrapure water, followed by
However, in the late 1980s, the use of mass spectrometry, diluted acid baths and oxidizing solutions. Finally, traces of
first with thermal-ionization, and nowadays with potential bioerosion, caused for example by foraging mol-
double-focusing inductively coupled plasma source mass lusks, are sought out, and the contaminated parts eliminated.
spectrometry, became feasible leading to very high precision To test the quality of the selection and cleaning processes,
measurements of 238U and its long-life daughters (234U, analysis by X-ray diffraction is conducted. This technique
92 N. Frank and F. Hemsing

Fig. 6.3 Schematics of the U/Th

disequilibrium methodology for
fossil tropical coral. Subsamples
of coral are taken from the
skeleton in order to check that it is
composed entirely of aragonite
(analysis by X-ray diffraction).
The scanning electron
microscopy (ESM) identifies
secondary aragonite fibers and
signs of coral dissolution.
Uranium and thorium are then
extracted, chemically purified
from the carbonate, and their
isotopes are measured by
inductively coupled plasma
source mass spectrometry
(MC-ICPMS)

identifies the abundance of various carbonate minerals in the directly estimated. Consequently, a tracer, known as a ‘spike’,
sample, such as aragonite and calcite, the latter being either which contains isotopes of uranium and thorium that do not
low or high in magnesium. Only samples identified with exist in the natural environment and have a well-established
more than 98% of aragonite qualify for U/Th dating. More concentration, is added to the solution. These artificial iso-
than 2% of calcite would indicate recrystallization and topes allow the calculation of the concentration of natural
therefore poor preservation of the skeleton. In a less sys- nuclides. Generally, any U/Th dating by mass spectrometry
tematic approach, scanning electron microscopy may be depends on spikes containing 233U, 236U and 229Th. Thus,
carried out to determine the presence of micro traces of measurements of the isotopic ratios 234U/238U, 236U/238U,
dissolution or precipitation of secondary aragonite fibers. 233 236
U/ U, 230Th/229Th and 232Th/229Th are needed to deter-
This selection process of the sample prior to dating is mine the concentration of 238U, 230Th and 232Th in the sam-
onerous, but often necessary, to ensure ages with the best ple, its isotopic ratios 234U/238U, 230Th/232Th and 230Th/238U,
accuracy and ‘precision’ possible. However, the samples are and the corresponding activity ratios.
of macroscopic size, varying from a few tens to a few Once chemical equilibrium between the spike and the
hundred milligrams, and a piece of coral is rarely perfectly sample in solution is reached, uranium and thorium are
preserved. The results of microanalysis on 1–5% aliquots of separated from the major, minor and trace elements by a
the sample, mean that the state of preservation was only column chemistry using an ion exchange resin. Several types
tested on part of the sample later used for U/Th dating. of resins are used to purify the uranium and thorium from the
sample. During the early days of Th/U dating most labora-
Chemical Procedure tories used successive series of anionic resin columns
The samples are placed in a strong acid solution (nitric acid or DOWEX 1X8. Nowadays, there are also separations for
hydrochloric acid), and undergo a chemical treatment which uranium and thorium using specifically designed resins such
involves several steps. Mass spectrometry measurements are as the UTEVA resin, which allows for faster and highly
of isotopic ratios, such as 234U/238U and 230Th/232Th. The effective purification. The sample dissolved in 3 N nitric
concentration or activity of the nuclides or the 230Th/238U acid is deposited on a column of UTEVA resin (0.5 ml),
ratio needed to calculate the age of the sample cannot be loaded in HNO3 3 N. The column is rinsed several times in
6 Dating of Corals and Other Geological Samples … 93

3 N nitric acid (10 column volumes), then uranium and tho- comparisons between standards and samples, and consider
rium are extracted with hydrochloric acid at different molar- the impact of background noise from the instruments and
ities (9–1 N) Douville et al (2010). This technique of from the chemical process (chemical blanks) on the mea-
chemical separation can be used to purify the uranium and surements. Uranium standards, HU1 or NBL 112, with
thorium of any other component of the coral with an known isotopic ratios, are generally used to determine the
approximated yield of 100%. The entire chemical procedure reproducibility of physical measurements.
can be done in approximately two days for about fifteen Overall, cutting edge mass spectrometry can achieve a
samples. For physical measurements, a similar time is dating with a higher precision than 100 years on corals of
required using the most effective tools, such as the around 100,000 years. This possibility remains, however,
MC-ICPMS described below. More recently it has become theoretical due to the fact that the error on the final age not
feasible to extract and purify uranium and thorium using only depends on the quality of the physical measurement,
automated extraction systems Wefing et al. (2017), but cur- but also, and above all, on the sample quality.
rently these tools remain the exception in the chemical Once the isotopic ratios are measured, Eqs. 6.1 and 6.2
preparation for U/Th dating. Lastly, the use of laser ablation shown above are used to estimate the age of the coral and its
systems has allowed for the direct extraction of material from initial 234U/238U ratio.
samples, which are carried with a gas stream to the source of a
mass spectrometer to detect the abundance of uranium and
thorium isotopes. However, those techniques allow for Limitations of the Method
rapid-age screening, but remain insufficiently precise for high
precision age determination (Spooner et al. 2016). Theoretically, the accuracy of U/Th dating is determined by
the precision of the measurements by mass spectrometry of
Physical Measurement by Mass Spectrometry the isotopic ratios 230Th/238U and 234U/238U. For this, it is
After chemical purification, the fractions of uranium and assumed that the coral being studied was perfectly preserved
thorium are deposited on a pure rhenium filament if analysis is and that the basic conditions mentioned above were
performed on a mass spectrometer with a thermal source. For respected. Therefore, the assumptions are that only uranium
inductively coupled plasma mass spectrometry, the final was incorporated into the aragonite skeleton at the time of its
solutions are diluted so that each sample has a similar con- formation, that the system remained closed to any exchange
centration. The uranium and thorium atoms are ionized in the of uranium and thorium with the sedimentary environment
source of a mass spectrometer. Formerly thermal ionization of after the formation of aragonite, and finally, that the coral
a solid sample from a heated filament was performed in skeleton contained no 230Th at the time of its formation. In
thermal ionization mass spectrometry (TIMS) at temperatures
of up to 1650 to 1800 °C. Most recent generations of mass
spectrometers (MC-ICPMS) have substantially improved the
ionization yield and hence the analytical accuracy due to their
detection of the ions by simultaneous multi-collection and due
to ionization temperatures in an Ar plasma of up to 8000 °C.
Ions are accelerated by a high voltage and subsequently
energy-filtered in an electrostatic filter and deflected by a
magnet according to their mass to charge ratio. The ions are
ultimately collected in a multi-collection system composed of
several faraday cups and electron multipliers. With the advent
of novel ultra-high resistances, faraday cups become available
for low ion intensities otherwise typically measured on elec-
tron multipliers. The MC-ICPMS technology now makes it
possible to determine isotopic ratios of uranium with an
accuracy of less than 1‰, and of thorium with an accuracy
of ± 1‰. This progress has resulted in a U/Th dating with a Fig. 6.4 Compilation of d234U activity ratios based on 230Th/238U
much higher analytical precision and accuracy than that ratios of fossil corals (black symbol: deep-water solitary corals; gray
obtained by TIMS. symbol: deep-water, reef-building corals; open symbol: tropical corals).
Measurement routines using these complex instruments The solid black line represents the variation in a closed system
compared to the d234U ratio of present-day seawater. The dots on the
vary and require rigorous data processing. Data processing right of the diagram in the ‘forbidden’ zone cannot be explained by the
must, among other things, correct for the effects of mass variation in a closed system. The straight dashed line indicates the
fractionation related to the measurement, perform distribution of isotopic ratios for co-genetic samples
94 N. Frank and F. Hemsing

addition, this dating method is based on the assumption that The Nuclear Recoil Effect and the ‘Open’ Dating
the activity ratio 234U/238U of sea water (equal to 1.1468) System
was constant over the last 500,000 years. For practical rea-
sons, this ratio is commonly expressed as the relative devi- During the radioactive decay of uranium isotopes, an a
ation (in ‰) from radioactive equilibrium and is denoted particle is ejected from the nucleus. The balance of kinetic
d234U (Eq. 6.3). energy requires that the nucleus produced (234Th or 230Th)
Thus, today’s d234U of seawater is 146.8 ± 0.1‰ recoils and therefore moves backward. In both cases, a
(Andersen et al. 2010). Given the conditions mentioned thorium isotope is produced. The 234Th in 234U decays
above, using Eq. 6.2, we would expect to find an initial ratio quickly (T1/2 = 24.1 days) by emitting an electron. This
of 234U/238U for the coral similar to that of seawater. process is not responsible for the movement of the nucleus,
However, it was observed that many corals more than because the recoil energy is too low. Therefore, the back-
80,000 years old had a wide range of values for this ratio, wards movement of the nucleus affects 234U (which occupies
and often values exceed the current value for seawater the position of 234Th) and 230Th. The mobile nucleus can
(Fig. 6.4). For example, a sample 125,000 years old, divided remain inside the crystal lattice but can also be ejected and
into several small pieces, can present age differences pass through the pore fluids, or even into another neigh-
between its various fragments of more than 10,000 years, boring crystal lattice (coral or sediment). This process,
even though the measurement accuracy is ± 1000 years for which occurs in both carbonates and sediments, or in organic
each subsample. Variability of age within the same sample matter present in a coral reef, is time-dependent: the more
can be ten times larger than the measurement precision. In time has elapsed, the more the nuclei will have moved.
addition, the sub-samples of the same specimen can also Obviously, this action is on a very small scale as dis-
show significant variability in their 234U/238U ratio over placement of the nucleus occurs over less than 20 nm.
time. This highlights the crucial role of the 234U/238U ratio Therefore, it cannot be measured directly, but observed
(or d234U) in the U/Th dating method for tropical corals. variations in concentration of the isotopes 234U (234Th) and
230
Henderson and Slowey (2000) and Gallup et al. (1994) Th likely reflect the overall redistribution of radionuclides
were the first to realize that the increase of 234U and of 230Th over the entire time elapsed since coral formation.
in ancient marine carbonates are often correlated (Fig. 6.4). Thompson et al. (2003) and Villemant and Feuillet (2003)
This proved to be a very important observation because were the first to incorporate this process into the equation for
diagenesis, such as the dissolution of the skeleton or pre- dating (Eq. 6.1). Here, we restrict ourselves to the theoretical
cipitation of secondary aragonitic fibers, cannot explain this approach of Thompson and his colleagues, because, on the
correlation. For example, the dissolution of the coral skele- purely mathematical level, the two models of radionuclide
ton during diagenesis will return uranium to pore waters, but redistribution are the same. The idea is simply to add a term
the thorium will be quickly re-precipitated because of its to the laws of decay (Eqs. 6.1 and 6.2), that takes the recoil
inability to remain in solution. Consequently, the 230Th/238U effect for 234U and 230Th into account. Equations 4 and 5 are
ratio of the coral will increase and the age calculated will be essentially the same as Eqs. 6.1 and 6.2:
overstated. This process does not involve a significant
230 
230   
Th Th
change in the isotopes of uranium, so it is expected that 238 U
¼ 238 U
initial  eðk230 tÞ þ f 230 f 234 1  eðk230 tÞ
sub-samples of the same coral will have variable 230Th/238U  
k230
ratios but a constant d234U. This equilibrium is reversed with þ f 230 eðk230 tÞ  eðk234 tÞ ðf 234  R0 Þ
k230  k234
the precipitation of secondary aragonitic fibers, as the fibers
ð6:4Þ
contain uranium but no thorium, leading to a reduction in the
230
Th/238U ratio of a coral and an underestimate of age.
234 

234  
U U
Since these secondary fibers are younger than the skeleton 238 U
 f234 ¼ 238 U
initial  f234 eðk234 tÞ ð6:5Þ
and they supply uranium taken from seawater, the d234U of
coral increases slightly but cannot exceed the d234U of R0 in Eq. 6.4 corresponds to the initial value of 234U/238U
seawater, i.e. 146.8‰ in total. It is clear that the disturbance in seawater fixed at 1.148 ± 0.010 (current value ± 10‰
in the dating system cannot be entirely attributed to these variability). f234 and f230 represent the proportions, expressed
processes. In particular, this does not explain the observed as activities, lost (f < 1) or gained (f > 1) following redis-
depletion of 234U and 230Th. tributions brought about by the nuclear recoil effect.
With these points in mind, Henderson and Slowey (2000) The redistribution factor f234 is estimated iteratively from
and Gallup et al. (1994) proposed that the nuclear recoil the difference between the 234U/238U ratio resulting from the
effect resulting from the radioactive decay of 238U and 234U temporal evolution in a closed system and the corresponding
is responsible for the disturbance and developed a method to evolution in an ‘open’ system.
take this into account.
6 Dating of Corals and Other Geological Samples … 95

The link between f230 and f234 (Eq. 6.6) is the difference of Instead of estimating f directly from the activity ratios
kinetic energy injected into the crystal lattice following the measured (Eq. 6.6), the values of f are fixed, but the excess
decay of 238U and 234U. The process of an a particle emission of 234U and redistribution slope m are estimated iteratively
is the same with only the kinetic energy being different. taking an external source into account (other corals or car-
bonates). Ultimately, we end up with equations that have the
f234 ¼ 1:157
ð1  f230 Þ ðVillemant and Feuillet 2003Þ same form as those derived from Villemant and Feuillet’s
ð6:6Þ model (2003), because the source considered in this model is
crucial and must have a composition similar to that of the
The definition of the redistribution factor f is purely
corals. However, here it is possible to vary the parameters in
mathematical, so gains and depletions in an open system for
the model and to find the slope m with the best fits for a set
uranium are expressed as f > 1 and f < 1. Up to this point,
of samples of the same age.
the two calculations of Villemant and Feuillet (2003), and of
These equations are used to correct for the nuclear recoil
Thompson et al. (2003) are identical. Assuming that the
phenomenon and to place the values of the activity ratios
measured ratios can be used to determine f as defined above,
measured, (230Th/238U) and (234U/238U), on a graph of
it is possible to estimate the age of the coral in an open
variations in these ratios within a closed system to calculate
system (Eq. 6.4). This approach assumes that only the ura-
the age of the corals.
nium and thorium in the coral are at the origin of the recoil
For example, Fig. 6.5 shows the activity ratios
process. However, this assumption is not always correct. In
(230Th/238U) and (234U/238U) measured in several corals of a
fact, the process of nuclear recoil, and thus the possible
coral reef on the Amedee Island, off the coast of New
ejection of nuclei over time, does not allow a gain of either
234 Caledonia (Frank et al. 2006). Ages calculated using
U or 230Th; the coral can only lose radionuclides.
Eqs. 6.1 and 6.2 and the measured activity ratios show a
Therefore, a value of f > 1 is not expected. But the reality is
wide dispersion, from 123,600 to 146,000 years. These
quite different, because most of the corals show an increase
samples also show a high variability in the initial d234U,
(f > 1), and only very few corals show a depletion of
ranging from 119.2 to 211‰. A linear correlation between
radionuclides (f < 1).
the ratios (230Th/238U) and (234U/238U) is obvious. This part
Thompson and his colleagues considered this obstacle to
of the reef very likely developed during the last interglacial
be theoretical and therefore introduced further complexity in
period (isotopic stage 5), corresponding to the last sea level
their approach. The gain of radionuclides can be explained,
maximum and dating back to about 125,000 years. The
either by direct exchange of nuclei between corals very close
results therefore demonstrate that this is an ‘open system’.
to each other, or by fluids circulating in the reef, in other
Two of the samples were in deficit and twelve had an excess
words, from an external source. Consequently, f is
of 230Th and 234U. None of the measured values thus reflect
always < 1 and the gain in 234U and 230Th is the sum of the
an evolution within a closed system. However, all samples
depletions over time and of the gain due to the retention of
were selected in a rigorous way and have 99% aragonite
radionuclides ejected and/or transported in the reef. To
with minor traces of dissolution and secondary aragonite
account for this phenomenon, Thompson and his colleagues
precipitation. Therefore, early diagenesis of carbonate can-
established the following equations derived for a simple
not explain these observations. By applying the models of
exchange model:

230 

234  
Th k230 t k230 U  k t 
measured ¼ 1  e þ initial  1 e 234  ek230 t
238 U k230  k234 238 U

234 

234  
1 U U k234 t
þ 238 U
measured  238 U
initial  1 e þ1 ð6:7Þ
m
 
ð1  f234 Þ 1  ek234 t
m¼    234   k t  ð6:8Þ
k234 t k230 t k230 t
ð1  f234 f230 Þ 1  k230kk
230
234
e þ k
k234
230 k 234
e þ ð 1  f 230 Þ k
k230
230 k 234
U
238 U initial e 234
 e
96 N. Frank and F. Hemsing

Figure 6.6 shows the ages of the different samples taken

from the reef according to depth. After correction using an
open system model, it is obvious that the points are now
organized in a linear trend, with older samples being located
deeper in the core than the corals sampled above. The
consistency in the linear relationship is such that it allows the
aggregation rate of the reef to be assessed. We can therefore
assume here that the process of nuclear recoil is the only
mechanism disrupting the dating system. In this example, we
applied the Thompson et al. model, with a value f234 of
0.975 ± 0.005 and an initial d234U of 148 ± 2‰, values
determined by Thompson and his colleagues for corals from
Barbados.

The Open System: Empirical Model

When dating a single coral, one option is to cut it into

Fig. 6.5 Activity ratios measured in corals from the Amedee Island in several sub-samples and to take them as independent sam-
New Caledonia. These corals came from a drilling site located between ples to which corrections for an open system could be
two zones of alteration, and show growth during the last stage of the
interglacial (MIS 5.5), 125,000 years ago applied. If the corrected values show a linear correlation
between the 230Th/238U and 234U/238U ratios indicating an
excess or depletion of 234U and 230Th, it is possible to cal-
culate the point where the regression line intersects the line
of evolution in a closed system and to deduce the age of the
coral.
In this way, a more accurate determination of age might
be arrived at without having to apply a complicated theory.
This idea was used by Scholz et al. (2004) to correct U/Th
ages of corals from the Mediterranean. Scholz and his
colleagues started with a result showing a variability of
activity ratios (230Th/238U) and (234U/238U) between
sub-samples of several corals that could not be explained by
the Thompson model. The open system model, subsequent
to the process of nuclear recoil, is logically limited to
samples which have not undergone any physicochemical
alteration. Once the coral skeleton is modified (dissolution,
recrystallization, or other), uranium and thorium can be
exchanged with the ambient environment. Hence, although
still present, the recoil effect is no longer the only and
predominant mechanism.
It is therefore obvious that theoretical models have strong
limitations and coral selection becomes critical. Moreover,
the open system models do not take small-scale variations in
uranium concentration within the coral into account, even
though these will also impact on the isotope redistribution
Fig. 6.6 ‘Raw’ results from dating of the coral shown in Fig. 6.5 process (Robinson et al. 2006).
( points) and adjusted by an open system model for the redistribution Thus, it is essential to adapt the interpretation of results
of U, subsequent to the recoil effect (□ points) according to the quality of the selected samples to obtain the
most accurate ages possible. It must be kept in mind that
Thompson et al. and Villemant and Feuillet, ages are cal- these ages, corrected by models, are approximations. In fact,
culated which are consistent with the evolution of the reef models are created to bring the ages as close as possible to
during the last interglacial period. an unknown reality, and the researcher cannot know if the
6 Dating of Corals and Other Geological Samples … 97

model chosen takes account of the real phenomena sustained complex. In most cases, tropical corals are located in vol-
by the sample during its geological history. canic areas that can be very tectonically active (subsidence
In conclusion, the analytical equipment available to us at or uplift). The sampled coral needs to have grown in situ and
present provides high precision and potentially accurate not moved by erosion. Also, corals can live at different
measures of U/Th ages of marine carbonates, with precisions depths depending on the species. On the continental mar-
on the permil magnitude for ages ranging from current day gins, the processes of eustasy (change in volume of ocean
to several hundreds of thousands of years old. The limits on water) and isostasy (variation in altitude of the continent) are
dating are not technical but primarily imposed by the quality also important. With changes in the volume of ice caps
of the samples. For recent samples, up to 15,000 years in during the climate cycles, the continent may well rise or fall
age, the U/Th dating system can be considered to be a closed depending on the weight it supports. Given all these vari-
system, since the physicochemical alteration of corals and ables, the reconstruction of sea level from tropical corals
the nuclear recoil effect have not yet had a significant impact requires the selection of corals from a habitat whose char-
altering the U-series age. acteristics are well known. Therefore, the tectonic processes,
For older corals, however, the impact of physicochemical which may have had an impact on their original position
alteration and especially that of the nuclear recoil displace- compared to where they were collected, need to be identi-
ment result in a system partially open to exchange of 234U fied. The parameters required can be defined as follows:
and 230Th, requiring subsequent correction to obtain rea-
sonable and accurate ages. The use of correction models • chronological age t;
reduces precision in dating, as not only the analytical error • the sampling height above present-day sea level h (+ or −);
must be taken into account, but also the parameters of the • the rate of uplift or subsidence Dh/Dt;
correction model used, such as the initial d234U and the • the average depth of the habitat of the species under
exchange factors f234 and f230. Selection of the study d;
best-preserved specimens possibly ensures the highest • the variability in the ecological habitat Dd.
quality of dating results that we can expect. If, however, the
sample shows signs of significant alteration, it is a prereq- The sea level over time m(t) can then be determined by
uisite to measure several subsamples to better identify the the following equation:
extent to which the system may have behaved open for
uranium. mðtÞ ¼ h þ d þ Dh=Dt  t:
The associated uncertainty is:

Estimating the Change in Sea Level DmðtÞ ¼ dh þ Dd þ t  dðDh=DtÞ þ Dh=Dt  dt:

from Tropical Corals
The reconstruction of sea level is highly dependent on
time and the Dh/Dt values. Moreover, it assumes that the rate
Changes in sea level come from the growth and melting of
of uplift or subsidence has remained constant over the time
polar ice caps when the climate of the Earth varies. Knowing
interval of the reconstruction.
the exact variations in sea level during climate cycles can help
The localization h and depth of habitat d can be well
to reconstruct a precise chronology of climate changes and to
constrained and are not, or only very little, influenced by
connect them with other parameters such as temperature
time. To minimize the impact of subsidence or uplift rates,
recorded in polar ice cores, the temperature and salinity of the
the ideal is to collect corals in the most stable places pos-
surface waters of the ocean, or even climate variations
sible, in terms of tectonics, for the sea level (m) to be almost
recorded in continental archives. A chronological framework
completely a function of t. However, most studies are
established in this way helps to better constrain the phase
located in places with strong uplift rates because these coral
relationships between the various components of the climate
terraces emerged and become easily accessible to sampling.
system and to determine the teleconnections between the
This sampling approach means that there is uncertainty
hemispheres. In addition, knowledge of sea level during cli-
about the uplift rate and therefore about the reconstructed sea
mate cycles is essential to study the variations in the conti-
level. Figure 6.7 shows reconstructions of sea levels
nental surfaces (expansion or reduction of the continental
obtained from U/Th dating of tropical corals, as they have
margins) and their erosion. We have seen that U/Th dating of
been published for various regions ranging from the Pacific
corals is a powerful tool to determine the evolution of coral
(Tahiti, the Marquesas, Vanuatu, Australia, New Caledonia)
reefs over time. But how can the ages of coral reef terraces
to the Atlantic (Barbados) (Bard et al. 1990; Cabioch et al.
contribute to a reconstruction of changes in sea level?
2003; Frank et al. 2006; Gallup et al. 1994; Hanebuth et al.
Linking the age (its time of formation) and location of a
2000; Thompson and Goldstein 2005). The graph also
coral taken in a particular part of the world to sea level is
98 N. Frank and F. Hemsing

Fig. 6.7 Reconstruction of

relative sea level (RSL) from
U/Th dating of tropical corals
(dots) and from oxygen isotopes
in benthic foraminifera (black and
gray lines)

shows a reconstruction of the variations in sea level and the global warming observed since the early twentieth century, a
associated uncertainty (black and gray lines) obtained from rise in sea level of over one meter per century is quite
isotopic measurements of oxygen in marine sediments conceivable if the ice sheets (Greenland, Antarctica) were to
(Waelbroeck et al. 2002). This reconstruction is based on the be destabilized.
assumption that the storage at the poles of a large quantity of The challenge for geochemists using U/Th dating of
freshwater depleted of 18O causes an increase in the oceanic tropical corals is not to predict future changes in sea level.
18
O content. Therefore, the 18O/16O ratio recorded in ‘deep Rather, they hope to find new coral reefs to reconstruct past
water’ foraminifera (less influenced by water temperature changes in sea level, and understand the pace and causes of
changes) can provide information on the size of ice sheets, these changes during climate transitions in more detail.
and, by consequence the sea level during the lifetime of the Using corals has the great advantage that they can be dated
foraminifera. With this approach, it is possible to reconstruct more precisely than marine sediments. Hence, episodes of
variations in sea level almost continuously. The various rapid climate change (Dansgaard-Oeschger cycle, Heinrich
reconstructions show very similar variations in sea level, events, described in Volume 2) are the subject of special
with a low level during cold climate periods and a high sea attention.
level during interglacial periods.
Over the past 140,000 years, the sea level has varied from
+5 m (relative sea level, RSL) during the last interglacial Other Geological Samples Datable
maximum to about -120 m during the last and penultimate by the U/Th Method
glacial maxima. A drop in sea level of this magnitude would
correspond to a volume of ice of about 47 million km3! By Other marine carbonates, such as deep-sea corals or shellfish
comparison, this is equivalent to an ice coverage over the shells can be dated by the U/Th method. In particular,
entire North American continent with an average thickness deep-sea corals behave very similarly to their tropical
of 2000 m. Knowing that an increase in sea level of one counterparts, with a slightly higher concentration of ura-
meter would have devastating impacts on much of the world nium. However, there are two major differences that are very
population living today in coastal areas, it should be high- important for U/Th dating. First, the aragonite skeleton is
lighted that natural changes in climate during the last tran- often more robust in deep-water corals, especially for the
sition (termination I) caused increases in sea level of 1.2 m species Lophelia pertusa and Desmophyllum dianthus.
every hundred years. It should further be noted that rapid Secondly, the physicochemical alteration is less important
changes in sea level are also evidenced during other climate because the skeleton remains underwater, from its formation
transitions between glacial and interglacial periods. During to its removal. As a result, the openness of the uranium
the last interglacial, about 125,000 years ago, sea level was system due to the processes of nuclear recoil and physico-
probably higher than the current level by about 5–9 m chemical alteration seems less important for these species.
(Dutton and Lambeck 2012). Hence, in the context of the However, in deep waters, an increase in 230Th resulting from
6 Dating of Corals and Other Geological Samples … 99

decay of dissolved uranium is observed. Deep-water corals carbonates, such as tuffs or travertines, several generations of
therefore incorporate small amounts of 230Th during the crystals can be found in a single layer of carbonate, Mallick
formation of their skeleton, and a significant correction is and Frank (2002). This means that U/Th dating of conti-
required to achieve precision and accuracy in equivalent nental carbonates is far more difficult than for corals. The
ages (Cheng et al. 2000a; Frank et al. 2004). isotopy of uranium cannot be used to test if the system is
Shells of mollusks have another distinctive feature. The open or closed, and the presence of an initial supply of
organism precludes incorporation of uranium during the thorium by groundwater or contaminants needs to be
formation of its aragonite shell through a still unknown investigated. To estimate the importance of contaminants,
biological process. A modern shell contains very little the isotope—232Th is used, Szabo et al. (1994). It is the most
uranium (in the range of a few tens of ng/g), whereas an abundant isotope of natural thorium. It is at the origin of a
aragonite coral skeleton contains a few micrograms/g. This decay chain called after 232Th. As clay contains lots of
would not be problematic for U/Th dating if the shell thorium (>5 lg/g), the appearance of 232Th in a carbonate
remained in a closed system during its preservation in sample is a sign of the presence of such contaminants
sediment. Unfortunately, it was observed that, after the (Ludwig and Titterington 1994).
death of the organism, the shell takes up uranium from its Therefore, as for corals, sampling is a crucial stage of
environment. The uranium concentration may increase by U/Th dating. It is essential to select a sample from a layer of
up to 10 mg/g, which is 100 times higher than at the time of carbonate containing a minimum of 232Th, indicator of
its formation. As it is impossible to identify the source of contaminants, and the sample must be representative of the
the excess uranium and how it has accumulated over time, original carbonate, i.e. the first generation deposited. Within
shells of mollusks are considered as open systems for these constraints, the U/Th dating of stalagmites, travertine
uranium. Thus, one of the two basic conditions of dating and tuff can be accurate, and nowadays allows for precise
mentioned above is violated (closed system), implying that determination of time frames for climate (Wang, et al. 2001),
based on our current knowledge, U/Th dating is not a and archaeological (McDermott et al. 1993), studies.
suitable dating method for shells and mollusks (Kaufman
et al. 1996).
For continental secondary carbonates, such as stalag- Conclusions
mites, travertine or tuff, dating by U/Th follows the same
‘analytical and theoretical’ principles presuming the same Dating by U/Th methodologies has become a very powerful
two assumptions and using the same basic Eqs. (6.2) and tool, widely used to obtain a fine chronology of the growth
(6.3) to determine the age. In groundwater a disequilibrium of coral reefs and many other carbonates. Although technical
between 238U, 234U and 230Th is created due to the fact that developments have been rapid in recent years, with precise
uranium is easily dissolved by the weathering of rocks, soils isotopic measurements of radionuclides from the decay of
and sediments, while the less soluble thorium is essentially uranium, the quality of the sample itself and the movement
absent. In fact, the precipitation of calcite in the form of of radionuclides induced by their own radioactivity cause
stalagmites, travertine or calcareous tuff leads to the problems for dating. Today, we no longer refer to a dating
co-precipitation of uranium in the carbonate without its system that is strictly closed, since the decay itself is partly
radioactive daughter, 230Th. Therefore, as in corals a the cause of exchanges of uranium and thorium with the
radioactive disequilibrium is created during the formation of coral environment. Correction models, known as ‘open
the carbonate. The concentration of uranium in these sec- system age models’ are emerging to incorporate this
ondary carbonates is related to the uranium content of important issue. However, prior to being studied, the sam-
groundwater and the type of mineral formed. It can vary pling process and characterization of the sample are crucial
from a few ng/g to hundreds of µg/g. However, the appli- as dating samples significantly altered by diagenesis is
cation of the dating method is more complex than for marine doomed to failure.
organisms. The isotopic composition of uranium in A precise geochronological framework established under
groundwater can be highly variable depending on the optimum analytical conditions and based on samples with as
weathering processes that come into play in an aquifer, good a level of preservation as possible, provides excep-
leading to activity ratio values for (234U/238U) that vary tional opportunities to determine geological parameters such
between 0.8 and more than 5, or in delta notation between as the uplift or subsidence of a coral reef or variations in sea
−200‰ and more than +5000‰ (Kaufman et al. 1996). level. The U/Th dating method has now become a key ele-
Another issue is that continental carbonates are often ment to place significant changes in major components of
contaminated with clay or even organic particles. These may the climate system within a precise temporal context. This
exhibit high uranium and thorium concentrations, without permits a direct comparison with astronomical forcings for
being in radioactive disequilibrium. Moreover, in porous the Quaternary.
100 N. Frank and F. Hemsing

References Ivanovich, M. & Harmon, R. S. (1992). Uranium-series disequilibrium:

Applications to earth, marine, and environmental sciences (2nd ed.,
p. 910), Oxford: Clarendon Press.
Andersen, M. B., et al. (2010). Precise determination of the open ocean Kaufman, A., Ghaleb, B., Wehmiller, J. F. & Hillaire-Marcel, C.
234U/238U composition. Geochemistry, Geophysics, Geosystems, (1996). Uranium concentration and isotope ratio profiles within
11(12), n/a–n/a. Mercenaria Shells: Geochronological implications. Geochimica et
Bard, E., Hamelin, B., & Fairbanks, R. G. (1990). U-Th ages obtained Cosmochimica Acta, 60(19), 3735–3746.
by mass spectrometry in corals from Barbados: Sea level during the Ludwig, K. R. & Titterington, D. M. (1994). Calculation of 230Th/U
past 130,000 years. Nature, 346, 456–458. Isochrons, ages, and errors. Geochimica et Cosmochimica Acta, 58
Cabioch, G., Banks-Cutler, K. A., Beck, J. W., Burr, G. S., Corrège, T., (22), 5031–5042.
Edwards, R. L. & Taylor, F. W. (2003). Continuous reef growth Mallick, R., & Frank, N. (2002). A new technique for precise
during the last 23 Kyr Cal BP in a Tectonically active zone uranium-series dating of travertine micro-samples. Geochimica et
(Vanuatu, South West Pacific). Quaternary Science Reviews, 22, Cosmochimica Acta, 66(24), 4261–4272.
1771–1786. McDermott, F., Grün, R., Stringer, C. B., & Hawkesworth, C. J. (1993).
Chen, T. Y., et al. (2016). Ocean mixing and ice-sheet control of Mass-spectrometric U-series dates for Israeli Neanderthal/early
seawater U-234/U-238 during the last deglaciation. Science, 354 modern hominid sites. Nature, 363, 252–255.
(6312), 626–629. Reimer, P. J., et al. (2013). Intcal13 and Marine13 radiocarbon age
Cheng, H., Adkins, J. F., Edwards, R. L. & Boyle, E. A. (2000a). U-Th calibration curves 0–50,000 Years Cal Bp. Radiocarbon, 55(4),
dating of deep-Sea Corals. Geochimica et Cosmochimica Acta, 64, 1869–1887.
2401–2416. Robinson, L., Adkins, J., Fernandez, D. P., Burnett, D. S., Wang, S. L.,
Cheng, H., Edwards, R. L., Hoff, J., Gallup, C. D., Richards, D. A., & Gagnon, A. C., et al. (2006). Primary U distribution in scleractinian
Asmeron, Y. (2000b). The half-lives of Uranium-234 and corals and its implications for U series dating. Geochemistry
Thorium-230. Chemical Geology, 169, 17–33. Geophysics Geosystem, 7, 1–20.
Cheng, H., et al. (2013). Improvements in 230Th dating, 230Th and Scholz, D., Mangini, A., & Felis, T. (2004). U-series dating of
234U half-life values, and U-Th isotopic measurements by diagenetically altered fossil reef corals. Earth and Planetary Science
multi-collector inductively coupled plasma mass spectrometry. Letters, 218, 163–178.
Earth and Planetary Science Letters, 371–372, 82–91. Spooner, P. T., Chen, T., Robinson, L. F., Coath, C. D. (2016). Rapid
Douville, E., Salle, E., Frank, N., Eisele, M., Pons-Branchu, E., & uranium-series age screening of carbonates by laser ablation mass
Ayrault, S. (2010). Rapid and precise 230Th/U dating of Ancient spectrometry.Quaternary Geochronology, 31, 28–39.
Carbonates using inductively coupled plasma-quadrupole mass Szabo, B. J., Ludwig, K. R., Muhs, D. R., & Simmons, K. R. (1994).
spectrometry. Chemical Geology, 272, 1–11. Thorium-230 ages of corals and duration of the last interglacial
Dutton, A., & Lambeck, K. (2012). Ice volume and sea level during the sea-level high stand on Oahu, Hawaii. Science, 266, 93–96.
last interglacial. Science, 337, 216–219. https://doi.org/10.1126/ Thompson, W. G., & Goldstein, S. J. (2005). Open-system coral ages
science.1205749. reveal persistent suborbital sea-level cycles. Science, 308, 401–404.
Esat, T. M. & Yokoyama, Y. (2006). Variability in the Uranium Thompson, W. G., Spiegelman, M. W., Goldstein, S. L., & Speed, R.
isotopic composition of the oceans over glacial-interglacial time- C. (2003). An open-system model for U-series age determinations
scales. Geochimica Cosmochimica Acta, 70, 4140–4150. of fossil corals. Earth and Planetary Science Letters, 210, 365–381.
Frank, N., Paterne, M., Ayliffe, L. K., Blamart, D., van Weering, T., & Thompson, W. G., & Goldstein, S. L. (2006). A radiometric calibration
Henriet, J. P. (2004). Eastern North Atlantic Deep-Sea Corals: of the SPECMAP timescale. Quaternary Science Reviews, 25(23–
Tracing upper intermediate water D14C during the Holocene. Earth 24), 3207–3215.
and Planetary Science Letters, 219, 297–309. Villemant, B., & Feuillet, N. (2003). Dating open systems by the
238
Frank, N., Turpin, L., Cabioch, G., Blamart, D., Tressens-Fedou, M., U-234U-230Th method: application to quaternary reef terraces.
Colin, C., et al. (2006). Open system U-series ages of corals from a Earth and Planetary Science Letters, 210, 105–118.
subsiding reef in New Caledonia: Implications for sea level Waelbroeck, C., Labeyrie, L., Michel, E., Duplessy, J. C., McManus,
changes, and subsidence rate. Earth and Planetary Science Letters, J. F., Lambeck, K., et al. (2002). Sea-level and deep water
249, 274–289. temperature changes derived from benthic foraminifera isotopic
Gallup, C. D., Edwards, L., & Johnson, R. G. (1994). The timing of records. Quaternary Science Reviews, 21, 295–305.
high sea levels over the past 200,000 years. Science, 263, 796–800. Wang, Y. J., Cheng, H., Edwards, R. L., An, Z. S., Wu, J. Y. Shen, C.-
Hanebuth, T., Stattegger, K., Grootes, P. (2000). Rapid flooding of the C. & Dorale, J. A. (2001). A high-resolution absolute dated late
sunda shelf: A late glacial sea level record. Science, 288, 1033– pleistocene monsoon record from hulu cave, China. Science, 294,
1035. 2345–2348.
Henderson, G. M. (2002). Seawater (234U/238U) during the last 800 Wefing, A. M., et al. (2017). High precision U-series dating of
thousand years. Earth and Planetary Science Letters, 199, 97–110. scleractinian cold-water corals using an automated chromatographic
Henderson, G. M., & Slowey, N. C. (2000). Evidence from U-Th U and Th extraction. Chemical Geology, 475, 140–148.
dating against northern hemisphere forcing of the penultimate
deglaciation. Nature, 404, 61–68.
 Magnetostratigraphy: From a Million
to a Thousand Years 7
Carlo Laj, James E. T. Channell, and Catherine Kissel

Since the publication in 1600 of the book De Magnete by Secular variation was therefore identified as the first
William Gilbert, and the measurements by magnetic obser- manifestation of the instability of the geomagnetic field. This
vatories progressively obtained from various parts of the was already surprising, but the biggest surprise was yet to
globe, we know that the Earth’s magnetic field is comparable come, when paleomagnetism methods were able to decipher
to one that would be created by a bar-magnet placed at the the history of the geomagnetic field over prehistoric periods.
center of the Earth and inclined by some 11° with respect to In 1906, Bernard Brunhes was the first to measure a
the axis of rotation (Fig. 7.1). direction of magnetization in rocks which was more or less
For each point on the Earth’s surface, the intensity and opposite to that of the present geomagnetic field. Brunhes
direction of the Earth’s magnetic field are defined in terms of measured this magnetization both in a Miocene lava flow
two components: the declination, which is the angle on the and in clays that had been baked when covered by this lava
horizontal plane between the magnetic north and the geo- flow, which he called “natural brick”. In doing so, Brunhes
graphic North, and the inclination, which is the angle used for the first time a test in the field, now called the
between the magnetic field vector and the horizontal plane. “baked contact test”, which is based on the fact that, when a
By convention, the declination is zero when the field vector lava flow settles on a sedimentary layer, it re-magnetizes it,
points to the North (it is 180° if the field vector points to the either partially or wholly, by heating it up.
South), and the inclination is positive when the field vector If the direction of the magnetic field changed between the
points downwards (which is the case today in the northern settling of the sediment and the arrival of the lava, the initial
hemisphere). magnetic direction of the sediment will be replaced by the
Measurements by observatories, which began in 1576 in magnetic direction of the overlying lava. (To quote Bernard
London and in 1617 in Paris, soon showed that declination Brunhes “If, in the banks of natural clay, we have a
and inclination were not stable throughout history: since well-defined magnetic direction and which differs from the
measurements began, the inclination in Paris has changed by direction of the current terrestrial field, it is reasonable to
about 10° and the declination by about 30°. assume that the magnetic direction is that of the Earth’s field
The intensity has decreased by about 5% per century. at the time when the lava flow transformed the clays into
This phenomenon is called the ‘secular variation’ of the “natural bricks”). Brunhes’s conclusion that “at the time of
geomagnetic field. Changes in declination are plotted on the Miocene, around Saint-Flour, the North Pole was
marine charts, essential tools for navigation by compass, pointing upwards: it is the Earth’s South Pole that was
before the development of GPS. closest to Central France” is the first suggestion that the
polarity of the magnetic field of the Earth could have
C. Laj (&) reversed in the geological past.
Department of Geosciences, Ecole Normale Supérieure, PSL Twenty years later, the Japanese scientist Motonori
Research University, 24 rue Lhomond, 75231 Paris Cédex, France Matuyama was the first to attribute reverse magnetization of
e-mail: [email protected] volcanic rocks in Japan and China to reversals of the Earth’s
J. E. T. Channell geomagnetic field and to differentiate Pleistocene lava from
Department of Geological Sciences, University of Florida, 241 Pliocene lava on the basis of the polarity of their magneti-
Williamson Hall, P.O. Box 112120 Gainesville, FL 32611, USA
zation. Matuyama was thus the first to use magnetic
C. Kissel stratigraphy as a way to order sequences of rocks in time.
Laboratoire des Sciences du Climat et de l’Environnement,
LSCE/IPSL, CEA-CNRS-UVSQ, Université Paris-Saclay, 91190 The modern era of magnetostratigraphy began in the
Gif-sur-Yvette, France 1950s in Iceland with the work of Hospers (1953). Hospers’

© Springer Nature Switzerland AG 2021 101

G. Ramstein et al. (eds.), Paleoclimatology, Frontiers in Earth Sciences,
https://doi.org/10.1007/978-3-030-24982-3_7
102 C. Laj et al.

Over the past two decades, significant technical develop-

ments have enabled the dating of recent volcanic formations
and sedimentary formations. In parallel, the development of
corers like the Hydraulic Piston Corer from the Ocean Dril-
ling Program (ODP) or the CALYPSO corer on board the
research vessel Marion Dufresne of the French Polar Institute
(IPEV) allowed the sampling of very long marine sedimen-
tary sequences, with very high sedimentation rates. Thus, it
was demonstrated that certain very brief changes in the
magnetic polarity of the sediment, initially thought to be due
to sampling artifacts, or to re-magnetization phenomena or to
various sedimentary processes, were, in reality, reflecting
coherent reversals in the polarity of the geomagnetic field,
Fig. 7.1 Diagram of the geocentric, inclined dipole which are observed simultaneously at different locations on
the surface of the globe. These “geomagnetic excursions”,
which occurred during periods previously considered
work, followed by Wensink (1966), led to a subdivision of “stable”, such as the Brunhes or Matuyama periods, provide
Icelandic Pliocene-Pleistocene lava in three polarity zones: an exceptional tool for correlations over long distances.
normal-inverse-normal, from most recent to oldest. After- Provided that these excursions are well dated (on volcanic
wards, the joint use of paleomagnetic measures and rocks with records of intermediary or reversed paleomagnetic
potassium/argon dating (K-Ar, Chap. 5) by Cox et al. (1963) directions), it is possible to establish a scale of geomagnetic
and McDougall and Tarling (1963a, b) marked the begin- instabilities (GITS = Geomagnetic Instabilities Time Scale),
ning of the development of the modern scale of magnetic allowing a better temporal resolution than the scale of mag-
polarities (Geomagnetic Polarity Time Scale, GPTS). netic polarities.
The first examples of magnetostratigraphic measures Finally, over the last decade, a new method of magnetic
obtained from sedimentary sequences were those by Creer correlation, based on the variations in intensity of the Earth’s
et al. (1954) and by Irving and Runcorn (1957), which magnetic field, has been proposed. The records of geomag-
demonstrated alternating positive and negative polarity in netic paleointensity obtained from marine sediments contain
thirteen sites of Torridonian sandstone in Scotland, and in a global signal, which is a global-scale correlation tool at an
Devonian and Triassic rocks. In parallel, Khramov (1960) exceptional time resolution. Although this new method is
published magnetostratigraphic results from more restrictive in its use (it requires a high level of
Pliocene-Pleistocene sediments in Turkmenistan, and pro- homogeneity of the magnetic mineralogy of the sediments),
ceeded to develop chromolithographic interpretations based it potentially allows cross-comparison with the stratigraphy
on the assumption that the durations of the different polarity of ice cores. As variations in geomagnetic paleointensity
periods were identical. We now know that this assumption is control the production of cosmogenic isotopes, like 10Be and
unfounded. Other pioneering studies of magnetostratigraphy 36
Cl in the upper atmosphere, their flux, measured in ice
were carried out on red Triassic sandstone from the Chug- cores from Greenland and Antarctica, is inversely correlated
water Formation, on the European Triassic Bundsanstein and with these variations. This makes it possible to envisage a
on the Moenkopi Formation from the Lower Triassic (see correlation between ice and marine sediments.
Opdyke and Channell 1996). All these studies were con- Below, we examine in turn the three major developments
ducted on sandstone and silt mainly of continental origin, in magnetostratigraphy, first the polarity scale, then the scale
largely devoid of fauna, and therefore, the correlations based of geomagnetic instabilities, and finally the development of
on the identification of polarity intervals were not supported the method of correlation by paleointensity. We adopt a
by biostratigraphic correlations. “historical” approach, retracing the different stages and the
The first studies of marine sediments from the main challenges encountered, and we will also attempt to
Pliocene-Pleistocene collected by coring in high southern highlight not only the advantages, but also the limitations of
latitudes (Opdyke et al. 1966) mark the beginning of modern each method. We will illustrate, through examples, the
magnetostratigraphy. These studies, combining magne- unique role that magnetostratigraphy can play, especially in
tostratigraphy and biostratigraphy, improved and expanded understanding the mechanisms of climate variability, by
the Geomagnetic Polarity Time-Scale (GPTS), which was allowing the evaluation and possible quantification of tem-
obtained by the paleomagnetic study of basaltic outcrops and poral phase differences between various regions and between
marine magnetic anomalies (MMA) (e.g. Heirtzler et al. different records (ocean, cryosphere, land).
1968).
7 Magnetostratigraphy: From a Million to a Thousand Years 103

Establishing the Scale of Magnetic Polarities

First Coupled Measurements: Magnetization

of Volcanic Rocks—K/Ar Dating; the McDougall
and Tarling Scale and the Mankinen
and Dalrymple Scale for the Plio-Pleistocene

In the 1960s, the first geomagnetic polarity time scales (GPTS)

were developed using studies coupling magnetic polarity with
radiometric dating (K/Ar, Chap. 5). These include the first
studies conducted by Cox et al. (1963) in California and by
McDougall and Tarling (1963a, b) on the Hawaiian Islands.
Identification of the different zones of polarity of the
Plio-Pleistocene was based solely on their radiometric age, as
none of the volcanic sequences studied (apart from the Ice-
landic sequences studied by Watkins and Walker) had the
continuity and duration to form a long, more or less continu- Fig. 7.2 The first scales of geomagnetic polarity for the
ous, stratigraphic sequence. The scale of polarities was Plio-Pleistocene period. After McDougall (1979). In black: normal
periods, in white: inverse periods
therefore constructed by combining results from locations
often very far from each other. So, there is no ‘stratotype’
deposited with a preferred orientation along the lines of
(typical locality), as is the case for sedimentary outcrops.
the geomagnetic field prevailing at the time of deposition.
The studies that marked this first phase of development of
In many cases, this orientation was preserved over time:
the scale were conducted from Jaramillo, the Cobb Moun-
just as for lava, sediments have a “magnetic” memory.
tain, in the Olduvai Gorge, Reunion, Mammoth, Cochiti and
Although the precise mechanisms by which sediments
Nunivak (see Opdyke and Channell 1996). All these studies
acquire magnetism are still poorly understood and this may
led to the gradual establishment of the magnetostratigraphic
limit their use in some applications, sediments, on the other
scale for the Plio-Pleistocene, shown in Fig. 7.2.
hand, have the enormous advantage of registering the
It was at this time that the custom of naming the long
direction of the geomagnetic field almost continuously
periods of polarity (polarity chrons) after the great geo-
unlike lava, where the continuity of the signal is condi-
physicists (Brunhes, Matuyama, Gilbert, Gauss) was intro-
tioned by the sporadic nature of volcanic eruptions.
duced, while the shorter periods (polarity subchrons) were
Figure 7.4 shows an example of magnetic polarity
named after the place where they were discovered (Jar-
recorded in the core RC12-65 collected by Opdyke et al.
amillo, Mammoth, etc.).
(1974), on the research vessel R. Conrad of the Lamont
The magnetic polarities scale by Mankinen and Dal-
Doherty Earth Observatory, Columbia University, USA.
rymple (1979) (Fig. 7.3), based on studies coupling mag-
While some details may differ, the sequence of polarity
netic polarity and K/Ar dating of volcanic rocks, is the last
periods from the Mankinen and Dalrymple scale can be seen
stage of this development phase of the magnetostratigraphic
in the record (from the Brunhes to the Gilbert period).
scale and remained the standard in the discipline covering
However, the sediment core goes back further in time; the
the last five million years, for over ten years.
lowest level is from around 10 million years ago, based on
Further back than 5 million years, however, analytical
paleontological estimates.
uncertainties in radiometric dating become of the order of
the duration of polarity periods, making their identification
unreliable. Other methods are required to extend this scale to
Magnetic Anomalies at Sea and the Heirtzler
more ancient times.
Scale of Magnetic Polarities

The study of magnetic anomalies at sea constituted the major

Magnetic Stratigraphy in Pliocene-Pleistocene
step towards establishing the sequence of magnetic polarities,
Sedimentary Series
from the middle of the Mesozoic era through to recent times.
The hypothesis by Vine and Mathews (1963) of the
It was the studies of magnetism of deep ocean sediments
spreading of the seafloor directly explains the formation of these
that first enabled this extension (Opdyke et al. 1974). The
magnetic anomalies on both sides of oceanic ridges (Fig. 7.5).
magnetic particles contained in the sediments were
104 C. Laj et al.

Fig. 7.4 Results obtained by Opdyke et al. (1974) on a marine

sediment core showing changes in declination of natural magnetization
Fig. 7.3 Magnetostratigraphic scale established for the last five illustrating the changes in magnetic polarity
million years. After Mankinen and Dalrymple (1979)

the ridges, exist over large areas of the North and South
During expansion, material from the upper mantle Pacific and the Atlantic and South Indian oceans. In addition,
reaches the surface at the axis of the ocean ridges, solidifies, assuming that the anomalies originate from a sequence of
and is then pushed laterally in a symmetrical manner on blocks magnetized in normal and reverse directions, they
either side of the ridge. The oceanic crust, composed of showed that these anomalies correspond to the same scale of
igneous rocks that contain magnetic minerals, acquires a magnetic polarities, the only difference being the rate of
thermoremanent magnetization during cooling and therefore expansion of the various ocean basins. Heirtzler et al. (1968)
registers the Earth’s magnetic field at the time of cooling. also calculated the speed of expansion of the South Atlantic
Alternating normal and reverse polarities are recorded by using the date (known from another source) of 3.35 Ma
“continuously” in rocks at the ocean bottom, which gives the for the Gauss/Gilbert boundary. Then, using various geo-
characteristic “pajama stripe” structure of the magnetic physical data and showing remarkable intuition, Heirtzler
anomalies distribution at sea. This structure can be detected et al. (1968) considered that the rate of expansion of the
by trailing a magnetometer behind a research vessel. South Atlantic remained constant for the last 80 million
This is exactly the type of work that was done by the years. With this assumption, they proposed a scale of mag-
Lamont Doherty Geological Observatory team led by Heirt- netic polarities going back to 80 Ma. With its new approach
zler, Pitman and Le Pichon. Their studies showed that and results, this scale was a gigantic step forward. It extended
magnetic anomalies, linear and symmetrical on both sides of the magnetostratigraphic scale from 5 Ma to 80 Ma and
7 Magnetostratigraphy: From a Million to a Thousand Years 105

Fig. 7.5 Measurement of

magnetic anomalies at sea on both
sides of the mid-ocean ridge and
comparison with the modeled
profile proposed by Heirtzler et al.
(1968)

proved to be very precise: in the thirty years since, studies The Cande and Kent Polarity Scale
have shown it to be out by only 5 Ma at 70 Ma!
In the scale of Heirtzler et al. (1968), the ages of the In a review of the Heirtzler scale, Cande and Kent (1992)
different polarity intervals were estimated solely assuming also considered the magnetic profile of the South Atlantic as
the rate of expansion of the seafloor. One of the objectives of a starting point, in which they replaced some intervals with
the international program for ocean drilling (Deep Sea equivalent, more detailed intervals obtained from ridges in
Drilling Project (DSDP), followed by ODP) was to link the the Pacific and Indian oceans with a higher speed of
scale of polarities inferred from the magnetic anomalies at expansion. However, they considered the rate of expansion
sea to biostratigraphic ages. Sediments accumulate on the of the South Atlantic to be continuous, but not constant as
ocean crust as soon as this latter is created at the ridges. Heirtzler et al. (1968) had thought in their first approxima-
Therefore, the age of the crust is given by the sediment just tion, and they adjusted the magnetic profile to match nine
above it. calibration points, whose ages had been estimated elsewhere
Over the past forty years, hundreds of cores have been (eight were high-precision 40Ar/39Ar ages, and the most
collected in the different oceans have allowed the Heirtzler recent age (2.60 Ma) was one proposed by Shackleton et al.
scale to be extended to older ages, and to link it with bios- (1990), at the Gauss/Matuyama boundary by astronomical
tratigraphic datings. calibration, see below).
In parallel, the magnetostratigraphic scale has been In a more complete version of this work, Cande and Kent
improved by studies of emerged marine sediments, which (1995) used astronomical calibration obtained by Shackleton
allow specific intervals to be analyzed with a better temporal et al. (1990), and Hilgen (1991a, b) for all the
resolution than was achieved by studies at sea. These Pliocene/Pleistocene polarity reversals. They modified the
“segments” were then inserted in the magnetostratigraphic Cretaceous-Tertiary boundary from 66 to 65 Ma. For a
scale, thanks to the identification of specific characteristics. detailed description of this new scale (called CK95), which
The study by Lowrie and Alvarez, published in 1981, on is the current benchmark for the Late Cretaceous and
the sediments of the Apennines in Umbria (Italy) is probably Cenozoic, we refer the reader to the original publication
the most important of these. Not only has it led to a better (Cande and Kent 1995).
understanding of the bio-magnetostratigraphic scale of the The magnetostratigraphic scale available today is very
last 100 million years, but it also led to the discovery of a reliable as far as the Late Jurassic. Traditionally, it is divided
thin level of iridium-rich clay, from the Cretaceous-Tertiary into two parts, the most recent corresponding to the Upper
boundary, which initiated the hypothesis of an extraterres- Cretaceous and Cenozoic eras, and the older corresponding
trial impact to explain the massive biological crisis that to the Lower Cretaceous and Upper Jurassic eras. Polarity
characterizes this boundary. sequences for these two major periods are called C and M
106 C. Laj et al.

sequences, respectively. On the ocean floor, they are sepa- astro-timeline, based on precession, was obtained at
rated by the normal, calm Cretaceous period, between 118 Site ODP 677, it became clear that the polarity intervals of
and 84 Ma, during which no geomagnetic reversal occurred the Mankinen and Dalrymple scale were not accurately
and therefore no magnetic anomaly is observed. dated.
Some important features of the geomagnetic field appear This study paved the way for a complete revision of the
evident when looking at the scale of magnetic polarities. On ages of the polarity intervals, especially during the Gauss and
the one hand, the total time for normal and reverse polarity Gilbert periods (Hilgen 1991a, b). The realization of the
intervals is essentially the same, with no tendency for the “youth” of the ages obtained by the K/Ar method for the
field to remain in one or the other of the polarities. On the Plio-Pleistocene era compared to those obtained by
other hand, during the Cenozoic, the rate of reversal cyclostratigraphy has led to the extensive use of the 40Ar/39Ar
increased: there was a reversal every million years or so at method (Chap. 5), which had recently been developed to test
the beginning of this period, but this reached four reversals the validity of cyclostratigraphic ages of this period. The age
per million years during the last 5 million years. The current of the Brunhes/Matuyama reversal (initially set at 0.73 Ma
normal period which has lasted for about 780,000 years by K/Ar) has been re-evaluated to 0.78 Ma due to a large
therefore seems abnormally long. number of independent 40Ar/39Ar measurements. The latter
Before the Upper Jurassic, the entire ocean floor corre- value is consistent with the cyclostratigraphic age. Good
sponding to the current tectonic phase was absorbed in ocean agreement is also observed for the Plio-Pleistocene era,
trenches. As a result, the polarity scale can only be extended where the new 40Ar/39Ar estimations coincide with the ages
in time through the study of continental rocks and is much given by Shackleton et al. (1990), and Hilgen (1991a, b).
less continuous and precise than for the Cenozoic and Although these new estimations, in general, confirmed
Mesozoic eras. The most obvious and probably most docu- the astronomical ages, the ages of geomagnetic reversals
mented characteristic is the long period of reverse polarity given by Cande and Kent (1992), based on magnetic
during the Permo-Carboniferous which lasted about 70 anomalies at sea, did not seem to be in agreement with
million years, and which is called the Kiaman interval. This astronomical dating. The authors had in fact adopted dura-
interval was preceded and followed by periods where the tions for the Plio-Pleistocene polarity intervals that have
field reversed frequently. proven inaccurate. As mentioned above, it is precisely the
astronomical ages for the Plio-Pleistocene that Cande and
Kent (1995) adopted in their new version of the scale. The
Astronomical Calibration of the Polarity Scale consistency they then obtained between 40Ar/39Ar estimates,
magnetic anomalies at sea and astronomical calculations
In a famous article published in 1976, Jim Hays, John Imbrie over the entire time period covered by the new scale can be
and Nicholas Shackleton were the first to show that some considered as a validation of this method. With this method,
indicators (proxies) of paleoclimate, such as isotopic records the age of a geomagnetic reversal can be estimated within
of oxygen, evolved over time depending on the orbital the duration of one precession cycle, in other words, less
cycles of the obliquity, eccentricity and precession. These than two or three times the length of a polarity reversal. This
cycles, initially calculated by Milutin Milankovitch, were very high resolution led Renne et al. (1994) to calibrate the
established much more precisely by André Berger, and gave age of one of the standards of the 40Ar/39Ar method (Fish
paleoclimatologists precise solutions for the last three mil- Canyon Tuff, FCTs) with the astronomical method, which
lion years. Hays et al. have “adjusted” their initial age model reduced the uncertainty to 0.6% for the calibrated ages
to “match” the obliquity cycle in the record to that given by compared with the standard.
astronomical calculation. In doing so, they established the
first ‘cyclostratigraphy’. This method was widely used in the
1980s to constrain the ages of isotopic records of oxygen Principle and Practice of Magnetostratigraphy
during the Brunhes period with a precision of a few thousand
years. The magnetic polarities scale shows that polarity reversals
Subsequently, the development of a hydraulic corer in the are largely random in time, which means that sequences of
DSDP program resulted in the acquisition of previously four or five successive reversals do not repeat themselves
inaccessible, very deep sediments, allowing the study of identically in time; therefore, they constitute a kind of
continuous sequences in even older sediments. Applied to “fingerprint” of specific geological periods. Magne-
the Matuyama period, astronomical calibration based on tostratigraphy is based on this characteristic: if a sequence
obliquity cycles did not initially show significant age dif- characteristic of the magnetostratigraphic scale can be
ferences between the limits of “astronomical” polarity and identified within a particular series, then a specific age can
those of the Mankinen and Dalrymple scale. But when an be assigned to this section.
7 Magnetostratigraphy: From a Million to a Thousand Years 107

Although simple in principle, matching of a particular unambiguous identification of the magnetic polarity at each
sequence to a segment of the magnetostratigraphic scale is far level of the studied sequence. This can sometimes be
from easy in practice. The ideal condition for a record of achieved if there is agreement between determination of
magnetic polarities to be perfectly continuous would be for polarity in two parallel and close sections. In general,
the “rain” of sedimentary particles to accumulate continu- however, this requires a complete paleomagnetic study
ously over time. But this is rarely the case. In general, the rate (determination of the magnetic mineralogy, field test, pro-
of accumulation of a sedimentary sequence is variable, with gressive demagnetizations either by heating or by applying
potential differences of an order of magnitude, depending, for alternating fields) to establish the stability of the magneti-
instance, on the changing climatic/environmental conditions zation and its acquisition at the time of deposition. For this
over time. Hiatuses can also occur. All these factors change purpose, Opdyke and Channell (1996) established ten cri-
the appearance of the sequence of polarities, making its teria to assess the quality of a study. The authors themselves
identification in GPTS complicated. However, in marine recognize that it is very difficult to simultaneously comply
sequences, close to ideal conditions can be found for periods with all ten for one section, but at least five should be present
of about a few million years. in a modern magnetostratigraphic study.
It is certainly a major advantage for magnetostratigraphic
studies to have thick sections (102–103 m). Indeed, the
thicker a section is, the more likely it is that it will contain A High-Quality Magnetostratigraphic Study: The
several polarity zones, and therefore can be more easily Siwalik Sequences in Pakistan
correlated with a defined segment of the reference magne-
tostratigraphic scale. Our colleague, Robert Butler, made an The deposits in the foothills of the Siwalik Basin in northern
amusing analogy between identification of a particular India and Pakistan are among the most studied fluvial sed-
sequence in the magnetic GPTS and fingerprinting in a iments. Indeed, in them, many fossils of mammals, including
police investigation. Usually, a full fingerprint can identify a primates, have been discovered. Their precise chronological
person, half a fingerprint leaves identification open to dis- study and the correlation of different outcropping sections is
cussion but a quarter fingerprint is unlikely to be accepted as of crucial importance for understanding the evolution and
irrefutable proof in a court of law. migration of these hominids and their land use.
In practice, even in favorable cases, the researcher is From a paleomagnetic point of view, the main problem
faced with a series of zones of normal and reverse polarity, with this study is that the deposits of Siwalik are “red beds”
which is often difficult to unambiguously correlate with the and magnetization is carried by hematite. The magnetization
GPTS scale. The assumption made at the outset is that the of this type of sediment is very complex, and it has been
rate of sedimentation of the sampled section is more or less shown in some cases that magnetization was acquired as a
constant. So, different ways of correlating the sequence to result of chemical reactions in the sediment, considerably
GPTS are tested. In most cases, however, it is necessary to after deposition. It is therefore essential for a magne-
have an independent marker in time (radiometric dating or tostratigraphic study of Siwalik to establish that magnetiza-
biostratigraphic data already independently correlated to the tion was acquired during deposition or immediately
GPTS) so that magnetostratigraphy can correlate the full afterwards so as to prevent invalidating correlation with the
section being studied to the GPTS, and thus, to specify its magnetic polarity scale (GPTS).
age, temporal thickness and rate of accumulation. The study of the magnetic mineralogy showed that the
Cross-correlation has been used by some authors to assess components of magnetization of these red beds fall under two
the agreement between the sequence being studied and a different categories of hematite: a red pigmentation phase and
particular segment of the GPTS, by calculating the correla- a specular hematite phase. The red pigmentation phase
tion coefficient corresponding to the various, visually eval- acquired its magnetization well after deposition (at least one
uated, solutions. A coefficient of maximum value indicates polarity interval later). However, the magnetization carried
the most likely correlation, while a value close to zero allows by specularite was acquired during deposition or immediately
rejection of that hypothesis. In practice, this method is only afterwards. These two components can be separated by
usable when the studied section contains a large number of thermal demagnetization, the magnetization of the specularite
polarity intervals. being isolated between 525 and 600 °C. This component can
Regardless of the method used to associate the studied then be used to establish the magnetic polarity of the sampled
sequence with a particular segment of the GPTS, the quality sections (Fig. 7.6) which is then correlated with the scale of
of a magnetostratigraphic study is primarily based on Mankinen and Dalrymple (1979) (Fig. 7.7).
108 C. Laj et al.

Fig. 7.6 Magnetostratigraphic

results obtained by Johnson et al.
(1985) on two sections from Chita
Parwala and Gabhir Kas. Some
levels of hard sandstone
(represented in gray) clearly
connect the two cuts. The
magnetic data is shown as the
virtual geomagnetic pole
(VGP) latitude, which is located
either in the southern hemisphere
(reverse polarity) or in the
northern hemisphere (normal
polarity). The filled-in circles
denote the most reliable
estimates, the open circles
represent the estimates of lower,
but acceptable, “quality”. The
column on the right shows the
sequence of polarities obtained
along both cuts and their
correlation

other hand, the discovery of rapid climate changes recorded

Geomagnetic Excursions and the Scale in ice cores encouraged paleoceanographers to collect mar-
of Magnetic Instabilities (GITS) ine sedimentary sequences with a very high sedimentation
rate, and with the longest possible series (IMAGES pro-
Discovery of Geomagnetic Excursions grams, ODP, IODP). This was made possible by the
development of new corers as previously mentioned. These
Throughout the process of establishing the scale of magnetic high rates of accumulation significantly increased the prob-
polarities, the presence of very short excursions of the field ability of recording rapid changes in magnetization, char-
polarity was observed in lavas and sedimentary sequences. acteristic of an excursion.
These excursions were often characterized by intermediate Despite this progress, detailed records of excursions
magnetization directions between direct and reverse polari- remain largely limited to the last two million years. The best
ties, and did not correlate to polarity intervals recognizable documented excursions occur as double changes in polarity
in the polarity scale well established on other sites. There- which define a very short polarity interval in the opposite
fore, these excursions were initially regarded with some direction to the initial and final polarities. The total duration
skepticism by paleomagnetic researchers and were some- of an excursion is very short: the most recent (and most
times attributed to episodes of extreme amplitude of the accurate) estimates converge around a duration of around
secular variation or even to artifacts due to sampling, 1500–2000 years (Laj et al. 2000, 2004, 2014; Laj and
re-magnetization or sedimentological processes. The short Channell 2007). In addition, directional changes related to
duration of these excursions also meant that their registration the best documented excursions like the Laschamp (LE) ex-
in volcanic or sedimentary sequences was often random, and cursion or the Iceland Basin (IBE) excursion, seem to have
until very recently, no excursion was observed in more than occurred more or less simultaneously in the various places
one site. where they were recorded. Therefore, geomagnetic excur-
A change of attitude in the community occurred about sions, as well as being of fundamental interest as a probe into
fifteen years ago, and can be largely attributed to a combi- the phenomena of magnetic instability in the Earth’s core,
nation of two main factors: firstly, considerable progress in also provide very precise points of correlation with a global
the dating methods of both volcanic rocks and sedimentary significance.
sequences allowed the unambiguous correlation of some of It is by now well established that excursions are accom-
these excursions recorded in these two media, and, on the panied by a decrease in the intensity of the geomagnetic
7 Magnetostratigraphy: From a Million to a Thousand Years 109

reversals. It is also important in paleoclimatology studies, where

an independent chronology of climatic/environmental phe-
nomena needs to be of higher resolution that the scale of
polarities to allow evaluation of the synchronicity and phase
shifts, either early or late, of climate events in different parts of
the globe (examples are described below).
Currently, the major obstacle to the widespread use of
this method is probably the difficulty of integrating sedi-
mentary data and volcanic data into a single unified scale of
geomagnetic instabilities. The brevity of the excursions is
both an advantage (giving very precise temporal tie points),
and paradoxically, it also represents an obstacle to the cre-
ation of this scale. In fact, a specific excursion is not sys-
tematically recorded in all sequences, including those with a
medium to high sedimentation rate. Studies to develop a
high-resolution chronology of geomagnetic instabilities,
especially of excursions, are often based on accurate dating,
using the K/Ar and 40Ar/39Ar methods, of as large a number
as possible of lava flows recording either an abnormal
direction, or a geomagnetic field with a very low intensity, or
both characteristics together. It is important to keep in mind
that 40Ar/39Ar datings are obtained by reference to standards
whose ages were defined based on astro-chronological cal-
ibrations. So, several ages have been proposed for the same
standard (e.g. Fish Canyon Sanidine, commonly used for
dating in the Quaternary). It sometimes appears that none of
them provides good agreement between the 40Ar/39Ar ages
and the glaciological or astronomical scales. Further work is
Fig. 7.7 Magnetic polarities observed in the Siwalik cut correlated therefore necessary to “reconcile” these various approaches.
with the Mankinen and Dalrymple (1979) polarities scale, the most
complete and reliable at the time
Today, at least seven geomagnetic excursions from the
Brunhes period have been inventoried in detail: the excur-
sions of Mono Lake (34 ka), Laschamp (41 ka), Blake
field. The field acts as a screen for cosmic radiation (con- (120 ka), Iceland Basin (188 ka) Pringle Falls (211 ka), Big
sisting essentially of charged particles), and this reduction Lost (560–580 ka) and Stage 17 (670 ka) (Laj and Channell
causes a significant increase in the amount of cosmonuclides 2007). Other excursions from the same period are being
(14C, 10Be, 26Al…) formed in the upper atmosphere through studied. Studies are also underway for earlier periods, such
the impact of cosmic radiation on the various particles air. as the Matuyama period during which at least eleven
As 10Be can be measured in polar ice cores, this opens up excursions seem to have occurred. All of these excursions
new opportunities for ice-sediment correlation. act as specific temporal tie points, that greatly increase the
temporal resolution of the magnetostratigraphic scale.

A Scale of Geomagnetic Instabilities?

Magnetostratigraphy Based on Variations
The use of geomagnetic excursions as precise temporal tie points in the Intensity of the Geomagnetic Field
could be particularly useful in at least two important areas of
Earth sciences. This is especially useful to understand the Introduction
operating mechanisms of the terrestrial dynamo which is the
origin of the field itself. Recent models, theoretical and numer- Over recent years, the stratigraphy of climate records has
ical, of the Earth’s dynamo (Glatzmaier and Roberts 1995) undergone a major change, particularly due to the discovery
indeed, give accurate assessments of time constants, frequencies of rapid and precise markers, both lithostratigraphic (Hein-
and geometries of the transitional field during excursions. It is of rich events in marine sediments in the North Atlantic) and
prime importance to characterize the role of the solid inner core climatic (Dansgaard-Oeschger events in Greenland ice
and the lower mantle in the mechanism of excursions and cores).
110 C. Laj et al.

Sedimentary records from the North Atlantic can be very distant from each other. In this method, the relative
correlated with a very good approximation to records changes in the paleointensity curve obtained from a specific
obtained in ice cores, by using anchor points common to sedimentary core is compared with a reference curve
both types of records, such as levels of volcanic ash or depicting the relative variations of the geomagnetic dipole
melting events (Bond et al. 1993). However, the correlation field. The latter is, in fact, the only component of the field
of ice cores with sedimentary records or sedimentary records which varies synchronously across the globe. After syn-
between themselves in very diverse parts of the world, is a chronization of two magnetic profiles with the reference
far more complex problem: a direct correlation, based on the curve, the phase shifts and/or synchronicity of the two
recognition of climate signals may not take into account the paleoclimate records from these two distant locations can be
possibility of a time difference between the occurrence of a evaluated.
particular climatic event in two distant regions or by to This method needs to be applied with some caution.
different media. And yet an accurate calculation of these Firstly, everywhere in the world, the local field is the
phase shifts (leads and lags) is of central concern to paleo- superposition of the dipole field on multipolar components,
climatologists, since it is essential to understand the mech- which are much more variable in space and have shorter
anisms of global climate change, and to the highlight of time constants. To plot the dipole field variations curve, the
chains of causality and the spatio-temporal spread of a cli- authors compiled records obtained from different parts of the
mate event. world. These compilations eliminate the non-dipolar com-
At the scale of a basin, the magnetic susceptibility mea- ponents which are averaged out in space. Compilations of
sured continuously in sediment cores, directly on-board relative paleointensity variations in the field for the last
research ocean vessels, usually allows the correlation of the 800,000 years (Sint 800) (Guyodo and Valet 1999) and the
different cores with a resolution of about one centimeter, last 1.5 Ma (PISO-1500) (Channell et al. 2009) show syn-
which is sufficient for most studies. This physical measure chronization of characteristic events globally and the wide-
allows splicing of records obtained from different cores, spread attenuation of the non-dipolar field. However, only
taken from the same site or in neighboring sites, to join characteristics with time constants of the order of 104–
together the different sections of cores obtained by the 105 years are apparent on these two compilations, as some of
hydraulic piston corer, aboard the Joides Resolution within the individual records have been obtained from sediment
the ODP international program. Magnetic susceptibility with accumulation rates of a few centimeters per thousand
reflects the ability of the sediment to acquire a magnetization years, insufficient to record short-term characteristics and
induced by a weak magnetic field. Measurement of magnetic rapid changes.
susceptibility is commonly used because it has the advantage These rapid characteristics are, however, clear in other
of being non-destructive, as the induced magnetization compilations characterized by a higher sediment accumula-
measured disappears as soon as the imposed field is switched tion rate. Initially proposed for the North Atlantic Ocean, the
off. It varies depending on the magnetic content of the NAPIS-75 compilation (North Atlantic Paleointensity Stack
sediment, and particularly on the concentration of magnetic for the last 75,000 years) (Laj et al. 2000), then SAPIS (South
particles, their nature and size. As changes in these physical Atlantic Paleointensity Stack) (Stoner et al. 2002) and finally
and chemical parameters are generally highly dependent on GLOPIS-75 (Global Paleointensity Stack) (Laj et al. 2004)
the paleo-environmental context, these are the same within have shown that rapid components of the variations in the
the same basin or water mass, and so the magnetic suscep- intensity of the geomagnetic dipole field can be recognized
tibility can be used as a local (or regional) correlation tool. globally in sedimentary records (Fig. 7.8). The outer limit of
Over long distances, oxygen isotopic ratios are the most temporal resolution appears to be of the order of 400 years,
commonly used stratigraphic tool. However, these ratios also which is the time constants of the dipolar field. GLOPIS-75 is
include regional climate components, precisely those that currently the reference for the last 75,000 years because it is
researchers are trying to fix in time relative to each other. precisely placed on the accurate ice age model (Cf. Sec-
Correlating these ratios between distant locations could tion “A Correlation Between Sediment and Polar Ice”). This
therefore mask the phase shifts which may really exist and new age scale for Greenland ice was developed by counting
that climatic studies attempt to quantify. Independent time annual levels (Greenland Ice Core Chronology or GICC05)
constraints of climate variations with a global value are (Andersen et al. 2006; Svensson et al. 2006). For earlier
required for this exercise. periods, up to 1.5 Ma, PISO-1500 is used.
Initially proposed a decade ago, a new method, based on Although, in principle, the method is simple, its imple-
relative variations in the intensity of the geomagnetic field in mentation is far from it. Firstly, a linear relationship between
the past recorded in sediments, is now being recognized as the intensity of the geomagnetic field existing at the time of
capable of revealing phase shifts over thousands of years at deposition and the magnetization of the sediment only exists if
most, in climate records obtained from sites geographically a single magnetic mineral carries the magnetization and the
7 Magnetostratigraphy: From a Million to a Thousand Years 111

Fig. 7.8 Changes in the intensity

of the geomagnetic dipole field.
a Over the last 75 thousand years:
global curve GLOPIS-75 (Laj
et al. 2004) here placed on the
most recent GICC05 ice age
model (Laj et al. 2014) and
compared to PISO-1500 on the
same period of time; b Over the
last 1.5 million years
(PISO-1500) (Channell et al.
2009)

size of magnetic grains is constant along the sedimentary the propagation time of the North Atlantic deep-water mass
column, in other words, if the magnetic mineralogy of the from north to south, and the other on the anti-correlation
sediment is uniform. All of these characteristics must be ver- between the intensities of the North Atlantic and Circum-
ified by a magneto-mineralogical study of the sediment (this Antarctic deep currents.
verification can lead to the rejection of a significant proportion
of cores, depending on the basin being studied), in order to
obtain a reliable record of the relative variations of the field. A Correlation Between Sediment and Polar Ice
This is obtained by dividing, at each stratigraphic level, the
intensity of the measured natural remanent magnetization by a The Laschamp geomagnetic excursion was the first to be
standardization parameter, which will take into account the discovered and is certainly the most studied excursion of the
variations in the concentration of magnetic grains. Of the three Brunhes period. Discovered in 1967 by Bonhommet and
magnetic parameters related linearly to concentration: sus- Babkine (1967), in lavas of the Puy de Laschamp, then in the
ceptibility; isothermal remanent magnetization (IRM) and Olby flow in the French Massif Central, it was originally
anhysteretic remanent magnetization (ARM),1 it is the latter dated between 20 and 8 ka. A series of studies conducted by
that is most often used, since it depends primarily on the same different research groups demonstrated the difficulty of dat-
magnetic grains as the ones carrying natural magnetization. ing lavas as recent as these. Recently the combined use of
Three recent examples of long-distance, multi-archive K/Ar and 40Ar/39Ar methods (Chap. 5) on basalt samples
correlations are described below showing the combined use collected at Laschamp and at Olby led to the establishment
of well-dated geomagnetic excursions and paleointensity of a critical requisite for reliable dating of these recent
records in paleoclimatology studies. The first one concerns a samples: the absence of excess argon in the initial compo-
sediment-ice correlation. Of the other two, one is based on sition of the lava, or of any other potential disruption of the
K/Ar system pre- or post eruption (Guillou et al. 2004).
Considering the uncertainty of 2.4% on the decay con-
1
Isothermal remanent magnetization (IRM) is the magnetization stant of 40K, Guillou et al. (2004) proposed a date of
acquired by a sample at a given temperature (most often room
temperature), after application of a constant magnetic field and
40.4 ± 2.0 ka, which was a considerable improvement, in
subsequent cancellation. Anhysteretic remanent magnetization terms of accuracy, over previous radiometric dating. Shown
(ARM) is obtained at room temperature through the combined action in Fig. 7.9 is a series of “snapshot data” obtained from lava
of a stationary field at a similar level to the Earth’s geomagnetic field flow from the Massif Central, the Canary Islands and New
and a strong alternative field in the same direction. The acquired
magnetization is measured after cancellation of the two fields.
Zealand superimposed on the GLOPIS-75 curve. The
112 C. Laj et al.

Fig. 7.9 a GLOPIS-75 curve

(red) (Laj et al. 2004) together
with Earth’s magnetic field
intensity data obtained from dated
lava flows from the Massif central
(black), the Canary Islands
(green) and New Zealand (blue)
(the grey dots are intensities
reported on the age given by the
geological map) (Laj et al. 2014).
b Zoom on the 20–48 ka interval
of the above diagram a with, in
addition, the Earth’s magnetic
field intensity reconstructed from
the 10Be data (blue) and 36Cl data
(green) obtained from the
Greenland ice core

absolute Earth’s magnetic field intensity values obtained accurately estimated and thus gives a measure of the dura-
from the lava flows are precisely dated using the coupled tion of the excursion as 1.5 ka, again in perfect agreement
K/Ar and 40Ar/39Ar methods. This figure also shows that the with sedimentary data.
coupled volcanic intensity/dating is consistent with the one If the Laschamp excursion (and also the more recent
obtained by GLOPIS-75, placed on the GICC05 age model Mono Lake excursion) constitutes a critical tie point for all
(Laj et al. 2014). the stratigraphies (in ice and sediment), it is also observed
It also shows that geomagnetic variations observed in the that, even outside these instability periods, the intensity
sediments used to build NAPIS-75 and then GOPIS-75 are profile and that of 10Be and 36Cl are extremely similar in
similar to those of the geomagnetic field, recalculated based time and amplitude for the 20–50 ka interval.
on the concentrations of cosmogenic isotopes registered in These studies show that it is possible to accurately
Greenland ice (Wagner et al. 2000). transfer the ice age model to sedimentary sequences, inde-
This results from the fact that the Earth’s magnetic field pendently of climate variations. The magneto-stratigraphic
shields our planet from cosmic and solar radiations and tool based on variations in the intensity of the geomagnetic
therefore modulates the production of cosmogenic isotopes field is therefore a very powerful tool, within the limits of its
(10Be, 36Cl, 14C) in the upper atmosphere. Assuming a application.
constant solar activity, a reduction (increase) in the intensity
of the geomagnetic field leads to an increase (decrease) in
this production. The concentration of these radionuclides, Paleo-Oceanographic Implications
especially 36Cl, recorded in Greenland ice, compared to the of High-Resolution Magnetic-Assisted
reference curve GLOPIS-75, has enabled a precise correla- Stratigraphy
tion, independent of climate, to be established between the
ice and sedimentary records. The last glacial period was characterized by very rapid cli-
Transferred to the GICC05 age model, the maximum mate changes which have been observed in the upper and
peak flow of 10Be corresponding to the intensity minima of middle northern latitudes. Heinrich (H) and Dansgaard-
the Laschamp excursion, is dated at 41.25 ka, in perfect Oeschger (DO) rapid climatic events, are associated with
agreement with the radiometric dating (41.2 ± 1.6 ka; Laj iceberg discharges into the ocean and large amplitude
et al. 2014). In addition, the width of this peak can be oscillations of air temperature over Greenland (Dansgaard
7 Magnetostratigraphy: From a Million to a Thousand Years 113

Fig. 7.10 Changes in the

intensity of the North Atlantic
Deep Water during the last glacial
period (below) reconstructed from
variations in the concentration of
magnetic particles along the path
of this deep-water mass. Top:
curve showing air temperature
changes over Greenland during
the same period. From Kissel
et al. (2008)

et al. 1993; Bond et al. 1993). The role of ocean circulation activity in the Deep Cape basin was therefore correlated by
in these rapid climate changes remained to be defined. the authors with warm events over Greenland.
Consequently, the evolution of the deep North Atlantic water However, this long-range correlation of climate parame-
mass was investigated, in particular by the study of the ters is entirely based on the assumption of synchronization
magnetic properties of a number of sediment cores collected between hemispheres. A new correlation between these
along the path of the Nort Atlantic Deep Water (NADW) paleoceanographic records from the north and south of the
(Kissel et al. 2008). Atlantic was proposed, using global variations in the inten-
By this approach, changes in the strength of the North sity of the Earth’s magnetic field as a long-distance,
Atlantic deep current (NADW) in the North Atlantic have climate-independent correlation tool, measured for each of
been shown to be completely synchronous with changes in the studied sedimentary sequences (Fig. 7.11). This showed,
atmospheric temperatures. This shows that the dynamics of for each warm event over Greenland, that when the NADW
deep ocean can change at a very high speed. This is shown becomes stronger in the north, it takes about
in Fig. 7.10, which illustrates the amount of small magnetic 860 ± 220 years to spread to the depths in the south,
grains, mobilized by the overflow water at the Faroe-Iceland “chasing out” and replacing the AABW (Kissel et al. 2008)
and Iceland-Greenland sills and transported by the deep (Fig. 7.11). This has been the only experimental attempt to
current, before being gradually deposited at each sampling quantify this time lag.
site. Its oscillations therefore reflect the relative changes in Compared with data obtained in the north from the
the intensity of the convection in northern seas and of the Greenland ice sheet, those obtained from ice cores in central
deep current, as well as its ability to transport the magnetic Antarctica appear to show a slightly different story, with less
particles from their basaltic source in the north (Iceland– abrupt variations than in the north.
Faroe) towards the south. More important, after synchronization with the Greenland
In the South Atlantic, during the same period, analysis of cores, using variations in the abundance of atmospheric
the carbon isotopic ratio of the benthic foraminifera shells methane trapped in the ice (Blunier and Brook 2001), gradual
(living on top of the sediment) and of the neodymium iso- warming, called type A events, began around 1500 years
topic ratio showed that the NADW also varied in intensity, before the main warm events (interstadials) in the northern
and was replaced by Antarctic deep water (AABW) as soon hemisphere. A bipolar seesaw mechanism has been proposed,
as it weakened (Charles et al. 1996; Piotrowski et al. 2005). according to which the southern hemisphere warms up when
These two water masses have indeed a different isotopic the northern hemisphere cools (Broecker 1998). More
signature. These variations were supposed to match with the recently, the EPICA community obtained a new record of the
variations in air temperature over Greenland. The NADW changes in air temperature from an ice core from the
114 C. Laj et al.

Fig. 7.11 a Top: Earth’s magnetic field intensity of the core from the (orange) in the north; b top: adjustment of the magnetic field intensity
Deep Cape Basin (green) reported on its own age model and compared from the Cape Basin (green) onto the reference curve GLOPIS-75 (red).
to GLOPIS-75 (red). Bottom: The age model of the core from the Deep The R2 is the correlation coefficient. Bottom: The lag between the two
Cape basin is based on the correlation between tracers of variations in paleoceanographic curves after this adjustment is of the order of
relative intensity of deep waters of the North Atlantic based on the eNd 880 years on average (underlined in yellow). From Kissel et al. (2008)
in purple in the south, and variations in the quantity of magnetic grains

Antarctic Dronning Maud Land site (EDML). This core,

because of its location, recorded the climate of the South
Atlantic, with a resolution comparable to that of the Green-
land ice cores. After synchronization, once again based on
methane, it appears that all the warm events perfectly reflect
the Dansgaard-Oeschger events in Greenland: the seesaw
mechanism therefore also works on short time scales.
A study of marine cores from the South Indian ocean
showed how the interhemispheric seesaw mechanism also
concerns the Antarctic Circum Current (Mazaud et al. 2007).
The record of relative paleointensity of core MD94-103
taken from the Kerguelen Plateau was precisely correlated Fig. 7.12 Normalized relative intensity obtained from core
with the reference curve GLOPIS-75, between 30 and 45 ka, MD94-103 (turquoise) compared and plotted on the same time scale
thanks, in particular, to the identification of the minima of as the reference curve GLOPIS-75 (Laj et al. 2004). Modified from
Mazaud et al. (2007)
the Laschamp and Mono Lake excursions (Fig. 7.12). This
correlation allowed all the paleoclimatic and paleoenviron-
mental records obtained from that core and other nearby concentration observed during the study period and accu-
cores to be transferred to the time scale defined for the rately fixed in time, correspond to increases in intensity of
Greenland ice cores (GISP2 at the time of the publication). the ACC. These maxima are found to be in phase with
Similarly, to what is observed for cores located along the periods of warming over Antarctica, events of type A, and in
path of the NADW, the concentration of fine magnetic grains antiphase with those of Greenland. Similarly, these periods
in the sediment shows, at these sites, located east of the of intense activity of the ACC are in opposition to the
volcanic islands of Kerguelen, the ability of the current to periods of intense activity of the NADW (Fig. 7.13)
remobilize particles coming from erosion of basaltic series demonstrating that the inter-hemispheric ocean seesaw
rich in magnetite and to transport them downstream, i.e. mechanism, suggested by orbital scale models, was also
towards the east. This is the deep Antarctic Circumpolar functioning at the millennial scale at least during the last
Current (ACC). As in the north, the maxima of magnetic glacial period (Mazaud et al. 2007).
7 Magnetostratigraphy: From a Million to a Thousand Years 115

The scale of geomagnetic polarity reversals resolves the

time intervals of the order of several hundreds of thousands
of years and the short instabilities like excursions allow a
resolution of a few hundreds of centuries to be achieved.
Furthermore, the significant decrease in the intensity of the
total geomagnetic field, documented in sediments during
excursions, permitted a first connection with other natural
archives such as ice cores from Greenland or Antarctica. The
reduction in the field led to an increase in the production and
therefore flux of cosmogenic isotopes such as 10Be, 36Cl and
14
C, arriving at the surface after their formation in the upper
atmosphere, whose concentration is accurately measured in
ice.
Finally, in recent years, identification of paleointensity
profiles in sedimentary sequences provides an extremely
accurate means of correlation. This opened up a new phase,
still in development, of high-resolution magnetic stratigra-
phy. Since the understanding of mechanisms of climate
change through the analysis of phase shifts between hemi-
spheres and/or between low and high latitudes is of critical
importance, it is clearly essential to dispose of a correlation
tool independent of climate with a resolution of about a few
hundred years.
By facilitating the unification of time scales for different
types of records, both continental and oceanic, high resolu-
Fig. 7.13 a Past variations in the intensity of the NADW between 30 tion magnetic stratigraphy, based on changes in the relative
and 50 ka based on the magnetic concentration in North Atlantic intensity of the field contributes to a better understanding of
sediments; b oxygen isotope record obtained from the GISP2 ice core
in Greenland (H is for Heinrich events); c variations in the concen- the chronology and dynamics of mechanisms responsible for
tration of magnetic minerals observed east of the Kerguelen Plateau climate and geomagnetic variations at the scale of the recent
(compilation of three separate records); d oxygen Isotopic curve from Quaternary, between 0 and 500 ka.
Byrd ice (Antarctica) showing the A1 to A2 events. In this figure, all
records are placed on the GISP2 age scale (following the correlation by
Blunier and Brooks (2001) for Greenland and Antarctica and that
shown in Fig. 7.12 for North Atlantic and Indian Ocean). From Mazaud References
et al. (2007)
Andersen, K. K., et al. (2006). The greenland ice core chronology 2005,
15–42 ka. Part 1: Constructing the time scale. Quaternary Science
Conclusions Reviews, 25, 3246–3257.
Blunier, T., & Brook, E. J. (2001). Timing of millennial-scale climate
During the years 1960–1980, the main objective of paleo- change in antarctica and greenland during the last glacial period.
Science, 291, 109–112.
magnetic studies of marine sediments and of the magnetization Bond, G., et al. (1993). Correlation between climate records from North
of the oceanic crust was to establish the sequence of magnetic Atlantic sediments and Greenland ice. Nature, 365, 143–147.
polarities. The resultant magnetostratigraphic scale, whose Bonhommet, N., & Babkine, J. (1967). Sur la présence d’aimantation
basic features were well established by the late 1970s, consti- inverse dans la Chaîne des Puys. Comptes rendus des séances de
l’Académie des sciences, série B, 264, 92–94.
tutes a huge step forward in an impressive number of strati- Broecker, W. S. (1998). Paleocean circulation during the last
graphic studies as well as tectonic studies focused in specific deglaciation: A bipolar see-saw? Paleoceanography, 13, 119–121.
geographic areas or more globally at the scale of plate tectonics. Brunhes, B. (1906). Recherches sur la direction de l’aimantation des
In its original form, magnetic stratigraphy, based on roches volcaniques. Journal de physique, V, 705–724.
Cande, S. C., & Kent, D. V. (1992). A new geomagnetic polarity
identifying polarity reversals of the geomagnetic field, typ- timescale for the late cretaceous and cenozoic. Journal of
ically had a million years as its unit of time. Geophysical Research, 97, 13917–13951.
The discovery, followed by the accurate documentation Cande, S. C., & Kent, D. V. (1995). Revised calibration of the
of geomagnetic excursions, was an important first develop- geomagnetic polarity timescale for the late cretaceous and cenozoic.
Journal of Geophysical Research, 100, 6093–6095.
ment, by providing tie points of well-defined age and very
short duration.
116 C. Laj et al.

Channell, J. E. T., et al. (2009). Stacking paleointensity and oxygen Laj, C., et al. (2004). High-resolution global paleointensity stack since
isotope data for the last 1.5 Myr (PISO-1500). Earth and Planetary 75 kyrs (GLOPIS-75) calibrated to absolute values. Timescales of
Science Letters, 283, 14–23. the Geomagnetic Field (American Geophysical Union, Washington,
Charles, C. D., et al. (1996). Climate connections between the C, 2004) Geophysical Monograph, 145, 255–265.
hemisphere revealed by deep sea sediment core/ice core correla- Laj, C., et al. (2014). Dynamics of the earth magnetic field in the 10–75
tions. Earth and Planetary Science Letters, 142, 19–27. kyr period comprising the Laschamp and Mono Lake excursions:
Cox, A., et al. (1963). Geomagnetic polarity epochs and pleistocene New results from the French Chaîne des Puys in a global
geochronometry. Nature, 198, 1049–1051. perspective. Earth and Planetary Science Letters, 387, 184–197.
Creer, K. M., et al. (1954). The direction of the geomagnetic field in Lowrie, W., & Alvarez, W. (1981). One hundred million years of
remote epochs in Great Britain. Journal of Geomagnetism and geomagnetic polarity history. Geology, 9, 392–397.
Geoelectricity, 6, 164–168. Mankinen, E. A., & Dalrymple, G. B. (1979). Revised geomagnetic
Dansgaard, W., et al. (1993). Evidence for general instability of past polarity time scale for the interval 0–5 m.y.b.p. Journal of
climate from a 250-kyr ice-core record. Nature, 364, 218–220. Geophysical Research, 84, 615–626.
Glatzmaier, G. A., & Roberts, P. H. (1995). A three-dimensional Mazaud, A., et al. (2007). Variations of the ACC-CDW during MIS3
self-consistent computer simulation of a geomagnetic field reversal. traced by magnetic grain deposition in Midlatitude South Indian
Nature, 377, 203–209. Ocean Cores: Connections with the Northern Hemisphere and with
Guillou, H., et al. (2004). On the age of the laschamp geomagnetic Central Antarctica. Geochemistry, Geophysics, Geosystems, 8,
event. Earth and Planetary Science Letters, 227, 331–343. Q05012. https://doi.org/10.1029/2006GC001532.
Guyodo, Y., & Valet, J.-P. (1999). Global changes in intensity of the McDougall, I. (1979). The present status of the geomagnetic polarity
Earth’s magnetic field during the past 800 Kyr. Nature, 399, 249– time scale. In M. W. McElhinny (Ed.), The earth: its origin,
252. structure and evolution (pp. 543–566). London: Academic Press.
Hays, J. D., et al. (1976). Variations in the Earth’s orbit: Pacemaker of McDougall, I., & Tarling, D. H. (1963a). Dating of reversals of the
the ice ages. Science, 194, 1121–1132. Earth’s magnetic field. Nature, 198, 1012–1013.
Heirtzler, J. R., et al. (1968). Marine magnetic anomalies, geomagnetic McDougall, I., & Tarling, D. H. (1963b). Dating of polarity zones in
field reversal and motions of the ocean floor and continents. Journal the Hawaiian Islands. Nature, 200, 54–56.
of Geophysical Research, 73, 2119–2136. Opdyke, N. D., & Channell, J. E. T. (1996). Magnetic stratigraphy
Hilgen, F. J. (1991a). Extension of the astronomically calibrated (346 p.). San Diego, CA: Academic Press.
(polarity) time scale to the miocene/pliocene boundary. Earth and Opdyke, N. D., et al. (1966). Paleomagnetic Study of Antarctic
Planetary Science Letters, 107, 349–368. Deep-Sea Cores. Science, 154, 349–357.
Hilgen, F. J. (1991b). Astronomical calibration of Gauss to Matuyama Opdyke, N. D., et al. (1974). The extension of the magnetic time scale
sapropels in the Mediterranean and implications for the geomag- in sediments of the Central Pacific Ocean. Earth and Planetary
netic polarity time scale. Earth and Planetary Science Letters, 104, Science Letters, 22, 300–306.
226–244. Piotrowski, A. M., et al. (2005). Temporal relationships of carbon
Hospers, J. (1953). Reversals of the main geomagnetic Field I, II, and cycling and ocean circulation at glacial boundaries. Science, 307,
III. Proceedings of the Koninklijke Nederlandse Akademie van 1933–1938.
Wetenschappen B, 56, 467–491. Renne, P. R., et al. (1994). Intercalibration of astronomical and
Irving, E., & Runcorn, S. K. (1957). Analysis of the palaeomagnetism radioisotopic time. Geology, 22, 783–786.
of the torridonian sandstone series of North-West Scotland. Shackleton, N. J., et al. (1990). An alternative astronomical calibration
Philosophical Transactions of Royal Society, London, A250, 83–99. of the lower pleistocene timescale based on ODP Site 677. Earth
Johnson, N. M., et al. (1985). Paleomagnetic chronology, fluvial and Environmental Science Transactions of the Royal Society of
processes, and tectonic implications of the Siwalik deposits near Edinburgh, 81, 251–261.
Chinji Village, Pakistan. The Journal of Geology, 93, 27–40. Stoner, J., et al. (2002). South Atlantic and North Atlantic geomagnetic
Khramov, A. N. (1960). Palaeomagnetism and stratigraphic correla- paleointensity stacks (0–80 ka): Implications for inter-hemispheric
tion. Gostoptechjzdat (218 p.), Leningrad. Geophys. Dept., A.N.U., correlation. Quaternary Science Reviews, 21, 1141–1151.
Canberra. Svensson, A., et al. (2006). The Greenland ice core chronology 2005,
Kissel, C., et al. (2008). Millennial-scale propagation of atlantic deep 15–42 ka. Part 2: Comparison to other records. Quaternary Science
waters to the glacial southern ocean. Paleocanography, 23, Reviews, 25, 3258–3267.
PA2102. https://doi.org/10.1029/2008pa001624. Vine, F. J., & Mathews, D. H. (1963). Magnetic anomalies over
Laj, C., & Channell, J. E. T. (2007). Geomagnetic excursions. In M. oceanic ridges. Nature, 199, 947–949.
Kono (Ed.), Treatise on geophysics (Vol. 5, pp. 373–416). Wagner, G., et al. (2000). Chlorine-36 evidence for the mono lake
Laj, C., et al. (2000). North atlantic paleointensity stack since 75 ka event in the summit GRIP ice core. Earth and Planetary Science
(NAPIS-75) and the duration of the laschamp event. Philosophical Letters, 181, 1–6.
Transactions of the Royal Society of London. Series A: Mathemat- Wensink, H. (1966). Paleomagnetic stratigraphy of younger basalts and
ical, Physical and Engineering Sciences, 358, 1009–1025. intercalated Plio-Pleistocene tillites in Iceland. Geologische Rund-
schau, 54, 364–384.
 Dendrochronology
8
Frédéric Guibal and Joël Guiot

Dendrochronology differs from other absolute dating meth- inside, responsible for, among other functions, the upward
ods in that the age assignment is not based on a simple, flow of the sap, and phloem, on the outside, responsible for
automatic count of annual deposits i.e. the rings, but on a set the downward movement of the elaborated sap (Fig. 8.1).
of intercomparisons of a large number of chronologies so as Year after year, the previously formed tissue is pushed
to ensure the annual status of each tree-ring (also known as inwards for wood and outwards for phloem.
growth rings), after eliminating the potential pitfall of In areas with a temperate climate, fluctuations in the
anomalies in the anatomy of rings which may result, some physical aspects of the atmosphere (temperature, humidity,
years, in the absence of a ring or the formation of double sunshine) mean that vegetation has a period of activity and a
rings (also known as false rings). period of rest within the same calendar year. Cambial activity
To understand the principle of this method, some fun- is discontinuous in time: in deciduous oaks on the plains in
damental aspects related to the formation of growth rings in western France, cambial activity lasts from April to Septem-
trees in temperate regions should be recalled. Because of the ber; for larches which grow in the internal Alps above 1500 m,
marked climatic seasonal contrast of temperate regions an cambial activity extends from mid-June to mid-August.
annual status can be assigned to each ring (with the excep- A ring is made up of two parts: the earlywood which
tion of some accidents in growth). develops at the beginning of the growing season, and the
final latewood which develops later in the growing season.
These differ in terms of the cells that compose them, their
A Bit of Botany and Ecology dimensions, their disposition and the thickness of their walls.
Variations in the thickness of the cell walls have conse-
The annual growth of woody plants is composed of an axial quences for the density of the wood, in the form of
component which leads to the lengthening of branches intra-annual and inter-annual variations more or less linked
(primary growth) and a radial component which leads to the to changes in climate conditions. These conditions act
formation of a ring (secondary growth). The radial growth of according to the principle of limiting factors. Growth cannot
the trunk, branches and roots results from a layer of proceed faster than is allowed by the most limiting factor.
actively-dividing cells, the cambium, immediately beneath This limiting effect may be continuous, variable or sporadic
the bark. This gives rise to vascular tissues: wood, on the depending on the case. The action of climatic factors is
attenuated or amplified by other factors, both abiotic (soil,
topography) and biotic (age, competition, pest attacks,
phenology).

F. Guibal (&)
Institut Méditerranéen de Biodiversité et d’Ecologie marine et
continentale, UMR7 263 CNRS/Aix-Marseille Université/ Crossdating
IRD/Univ Avignon, Europôle de l’Arbois, BP 80, 13545
Aix-en-Provence Cedex 04, France Aristotle, Buffon and Leonardo mentioned the existence of
e-mail: [email protected] annual tree rings. Leonardo da Vinci, in particular, observed
J. Guiot a relationship between ring widths and the weather condi-
European Centre for Research and Teaching in Environmental tions of the year. However, it is the American astronomer
Geosciences CEREGE, Aix-Marseille University, CNRS, IRD,
INRAE, Collège de France, BP 80, 13545 Aix-en-Provence Cedex Andrew E. Douglass (1867–1962) who, in laying out the
04, France methodological foundations, is considered the father of

© Springer Nature Switzerland AG 2021 117

G. Ramstein et al. (eds.), Paleoclimatology, Frontiers in Earth Sciences,
https://doi.org/10.1007/978-3-030-24982-3_8
118 F. Guibal and J. Guiot

over a wide enough geographical area to cause ring widths to

vary the same way in many trees. This principle is important
because ring widths can be crossdated only if one environ-
mental factor becomes critically limiting.
Crossdating, or synchronization, is essential to check the
accuracy of the ring count and the presence of any growth
abnormalities. Abnormalities may appear as double rings
(false rings) in the same calendar year or as missing rings.
Indeed, some years, after a cold winter, possibly followed by a
late spring or preceded by severe defoliation in the previous
year, the ring may be partially or totally absent. In other years,
as a result of the early onset of a summer drought (as in the
Mediterranean region), the cambium may develop latewood
Fig. 8.1 Cross-section of deciduous oak elements and, then, thanks to improved weather conditions
through the summer, may start producing earlywood elements
dendrochronology (dendron: tree, chronos: time, logos: again before producing latewood at the end of the normal
study). Dendrochronology remained fairly discreet until it growing season. In this case, the “first” ring is identified as
gained in notoriety about 1929, when Douglass succeeded supernumerary or false. Crossdating remains largely subjec-
for the first time in dating beams from ruins of Indian tive and various methods have been developed to describe the
buildings in the US state of New Mexico, qualifying it as a observed similarities more objectively (McCarthy 2004).
dating method with an annual resolution (Robinson 1976). In living trees of the same species, with a confirmed
Despite this, it was not until the arrival of the computer in contemporaneity between the samples, synchronization is
the 1960s, that dendrochronology truly took off, with a established by identifying sequences of similar rings over
proliferation of research laboratories. several series under a microscope (in cores or cross sections
In the regions where the first dendrochronological studies of trunk), firstly from the same tree, and then, between series
were conducted (semi-arid regions of the southwestern USA from different trees. The operator compares the series from
and cold regions), the existence of a single limiting climate the bark inwards, records the rings and counts the sequences
factor (rainfall in semi-arid regions, summer temperatures in of narrow rings or ones with a distinguishing feature (color
cold regions) was instrumental in creating a series of rings or width of the final wood, presence of any traumatic scars or
whose width varied from one year to the next. However, in ducts etc.). This stage allows the identification of each ring
regions with a temperate climate, where growth depends on in terms of its vintage, after any anatomical abnormalities
several factors, one factor can compensate for another, and the such as missing rings or double rings have been detected.
annual ring width series are less variable, making den- After synchronization of the series has been established, ring
drochronological studies more complicated. In fact, if weather width series are measured (1/100–1/1 000 mm).
conditions are adequate, year after year, to meet the ecological On samples of unknown date, i.e. samples of wood from
requirements of the tree, the rings form a temporal series of trees felled at an unknown date, the measurement of ring
constant width, and do not provide any chronological infor- width series allows the establishment of digitized series from
mation, since it is extreme variations that serve as landmarks. which graphs are drawn to compare the temporal variations
Dating with dendrochronology is based on a fundamental in ring widths. The graphs are then compared in pairs and
stage, called crossdating, which is relative dating assuring sequences of similar ring patterns are sought; maximum
the proper placement in time of each ring. Crossdating is similarity between two curves is obtained when contempo-
established by intercomparison of different pieces of wood rary years are superimposed: maximum values, minimum
on which sequences exhibiting similar ring patterns are values and the number of times two series show the same
identified. For this, similar sequences of coinciding narrow upward or downward trend in relation to the preceding year.
and wide rings, that is, separated by the same number of This analysis allows us to date the year of formation of
rings, need to be identified. This is possible only with two each ring and identifies the felling date of the tree. The
conditions. Firstly, the limiting factors for radial growth felling date of the tree is ensured when the outermost ring is
must vary in intensity from one year to another, with an still present; for species where the anatomical difference
unrepeatable series over time, so that the succession of ring between sapwood (functional wood nearest the bark) and
widths are also variable in such a way as to be irrepro- heartwood persists over time (oak, ash, elm, larch, etc.), the
ducible. Secondly, the limiting factors must act in a similar felling date of the tree is estimated based on the date
way on trees with the same environmental requirements and assigned to the last ring.
8 Dendrochronology 119

Fig. 8.2 Diagram of the theoretical construction of a master chronology

The dating of samples from trees which died at an etc.) or having experienced more or less different stresses in
unknown date requires synchronization between the ana- the form of local disturbances. This summation results in a
lyzed series and a chronology of previously dated ring master chronology that is the average of annual wood layers
widths, called a master chronology. This master chronology over time of a given species in a region exposed to the same
must be composed of several series of rings from the same macro-climate, over a more or less extensive area.
tree species as the one to be dated and from trees exposed to
the same climatic factors, i.e. geographically close. Building
a master chronology involves assembling several mean Temporal and Spatial Extension
chronologies, homogeneous in their ecological and geo-
graphical origin, partially synchronous, based on the pres- For the Holocene period, the longest chronologies, covering
ence of ring patterns common to these timelines, and one of several thousand years, come from North America (Fergu-
which is constructed from living trees for which the year of son 1969; Ferguson and Graybill 1983), the British Isles
formation of the outermost ring is known. This permits each (Pilcher et al. 1984; Baillie and Brown 1988), Central Eur-
ring in the master to be assigned the year of its formation ope (Leuschner 1992; Krapiec 1998; Schaub et al. 2008;
(Fig. 8.2). The representativity of a master chronology is Kaiser et al. 2011), North-West Europe (Eronen et al. 2002;
related to its sample depth, i.e. the number of series included Grudd et al. 2002) and Siberia (Naurzbaev and Vaganov
in the calculation of the mean ring width value. Even in 1999; Rashit et al. 2002). In the southern hemisphere, sev-
climatically homogeneous regions, the geographic area eral groups have built thousand-year chronologies in
covered inevitably leads to the inclusion of trees from forest Argentina and Tasmania (Barbetti et al. 1995; Roig et al.
stands subject to a variety of local climates, due to differ- 1996; Cook et al. 2000).
ences in altitude, exposure, continental character or even In the same way that the representativeness of a master
having grown in different site conditions (bedrock, exposure, chronology is related to the quality of the climate signal
phyto-ecological communities, degree of clearing of the site evidenced by a high frequency of pointer years, the
120 F. Guibal and J. Guiot

in the early 1980s. In a second example, master chronologies

representative of the South West of England have allowed
dating of sites located in the Loire valley, the Penthièvre and
the Rennes basin by the City of London Polytechnic and
Queen’s University Belfast.
We should also mention that before starting to crossdate a
piece of wood, it is often necessary to start with an
approximate age of the piece, provided by 14C dating which
crossdating will then refine until accuracy to the year is
achieved. It should also be noted that although den-
drochronology is an absolute dating process accurate to a
Fig. 8.3 Mean chronology (thick line) built up from multiple single year, this does not prevent occasional dating failures,
synchronous ring-width series of variable length (in line with the
custom in dendrochronology, the y-axis is shown on a logarithmic scale especially when master chronologies for the species and/or
to make the thinner rings more distinct) region are lacking.

14
chronological representativeness of the sample to be dated Contribution of C to Calibration
depends on the number of pointer years it contains. For this
reason, trying to date a chronology composed of too few Extremely valuable for its ability to date wood vestiges by
rings is usually an exercise doomed to failure. For a given establishing, under the conditions detailed above, the year of
site, a multiplicity of samples is essential to acquire a rep- formation of each ring, even the year of death of the tree,
resentative mean chronology for the site in which individual dendrochronology has the undeniable advantage of con-
variances are minimized; in the field, this means sampling at tributing to the calibration of radiocarbon dates by con-
least a dozen cases presumed to be contemporaneous, in verting 14C age to the true calendar age.
order to achieve, whenever possible, a mean chronology of In the 1950s, when the first radiocarbon datings were
at least 80 years (Fig. 8.3). This methodological requirement obtained on objects from past human societies, the match
explains the negative outcome of repeated attempts to date between the 14C dates and the calendar dates was considered
isolated pieces of wood, regardless of the context of their adequate. However, the archaeological material used was not
discovery, even if, in certain exceptional conditions (very very well dated or very old, and the ranges of uncertainty
long ring series, particularly well-documented period), stat- were so great that they masked potential minor deviations.
ues and dugout canoes may have been dated by den- According as the accuracy of 14C dating improved and the
drochronology (Arnold 1996; Eckstein 2006)! body of datings grew, it quickly became obvious that the 14C
Synchronization between different tree species, called dates obtained were more recent than the dates obtained
heteroconnexion, although discouraged because of differ- independently, in particular those obtained from remains
ences in climate response and ecological requirements from ancient Egypt. Given that any uncertainty inherent in
between species, is sometimes carried out between species relative dates obtained on such material could not be ruled
with very similar ecological requirements. For example, out, and that in a living tree, only the outermost ring has a
14
comparisons are commonly made between oak and chestnut, C content in balance with that of the atmosphere, a pro-
oak and elm, larch and spruce. gram of 14C dating of tree rings dated to the year by den-
Teleconnection or comparison of tree ring series over drochronology was initiated on long-living Methuselah pine
long distances, from areas subjected to different climate (Pinus aristata) from the slopes of the White Mountains in
conditions, although theoretically just as frowned upon as California (Fig. 8.4). The results confirmed the disparity
the previous exercise is nevertheless, often done if the study between 14C dates and calendar dates (de Vries 1958).
is initiated in an area previously never investigated and for Irregular fluctuations were noted in the dates obtained, and
which there is no knowledge base. led Suess (1965) to establish a calibration procedure for the
In this way, the first master chronologies of oak repre- 14
C dating of tree rings, in order to express the raw dates
senting the North East of France and Brittany were initiated. (conventional dates) in chronometrically calendar dates
In a first example, master chronologies already in place for (calibrated dates). Measurements carried out on sequences of
the South West of Germany were used to calibrate the first five or ten consecutive rings collected on the California pines
samples in Franche-Comté and Burgundy, which were have shown that the gap between the two calendars, 14C
analyzed by the Laboratory of Chrono-Ecology in Besançon, years and actual years, remained low for the last 2500 years,
8 Dendrochronology 121

of rings from American pines, oaks and European Scots

pines. After publication in 1993, these curves, called radio-
carbon calibration curves, established by thousands of
measurements, constituted the basis on which corrections are
now possible for the entire Holocene period, and currently,
for the past 12,400 years (Stuiver et al. 1998). Extended into
the past through the dating of tropical reef corals, varved
sediments and speleothems (see Chap. 4), the calibration
curves obtained from tree ring data are now being extended
using series of dated tree rings series from Bølling Allerød
and from Younger Dryas in Germany, the area around
Zürich, Northwest Italy and watersheds of tributaries in the
mid Durance valley (France).
The correction curve for 14C dates in calendar years
shows that the actual time seems compressed by about 15%
before the sixth millennium BC, that there are plateaus along
the curve (for example, around 500 BC. or during the ninth
and tenth millennia BC.), and that multiple small fluctuations
can, for some periods, affect the rectilinear shape of the
curve. The practical consequences of these types of varia-
tions, showing that variations in concentration of atmo-
spheric 14C have been erratic, are very different: small
Fig. 8.4 View of a several thousand year old Methuselah pine (Pinus fluctuations, after correction, can result in particularly inac-
aristata) from the White Mountains (California) curate dates; but in some cases, very precise dates can be
achieved for those periods particularly affected by fluctua-
but that from 500 BC, the gap increased sharply to almost tions in the atmospheric content of 14C.
800 years at 5500 years BC. Examined more closely, the radiocarbon calibration curve
A systematic study of this phenomenon was then carried with its very twisted appearance reflects a stochastic process
out in the 1980s and 1990s by several laboratories on blocks which constitutes, at certain times, a particularly valuable

Wiggle-matching of post 69 : plot of the match with the calibration curve.

3050

3000

2950
1-18
20-28 82-100 2900
Radiocarbon Age BP

37-5355-6667-80

2850

2800

2750

2700

2650

2600
-1250 -1200 -1150 -1100 -1050 -1000 -950 -900 -850 -800
cal Age BC

Fig. 8.5 Dating by wiggle-matching of the late Bronze age oak post y-axis corresponds to radiocarbon years. Vertical lines = standard
n◦ 69 from the submerged coastal habitat of Montpenèdre, Hérault deviation of the measure
(Oberlin et al. 2004). The x-axis corresponds to the calendar years; the
122 F. Guibal and J. Guiot

chronological marker. Indeed, the matching of these kinks Kaiser, K. F., Friedrich, M., Miramont, C., Kromer, B., Sgier, M.,
(known as wiggles), based on a multiplicity of 14C dates Schaub, M., et al. (2011). Challenging process to make the late
glacial tree-ring chronologies from Europe Absolute—An Inven-
obtained from blocks of tree rings, separated by a known tory. Quaternary Science Reviews, 13 p. https://doi.org/10.1016/j.
number of calendar years, allows, by the method of quascirev.2010.07.009.
“wiggle-matching” (Pearson 1986), a greatly improved Krapiec, M. (1998). Oak dendrochronology of the neoholocene in
accuracy of radiocarbon dating, as illustrated by the example Poland. Folia Quaternaria, 69, 5–133.
Leuschner, H.-H. (1992). Subfossil trees. In T. S. dans Bartholin, B.
of the post n◦ 69 (Fig. 8.5) from the submerged coastal E. Berglund, D. Eckstein, & F. H. Schweingruber (Eds.), Tree rings
habitat of Montpenèdre (Hérault, France). and environment. Proceedings of the International Dendrochrono-
logical Symposium, Ystad, South Sweden, 3–9 September 1990
(pp. 193–197). Lund: Lund University, Department of Quaternary
Geology.
References McCarthy, B. C. (2004). Introduction to dendrochronology, Ohio
University, World Wide Web homepage. http://www.plantbio.
Arnold, B. (1996). Pirogues monoxyles d’Europe centrale. Construc- ohiou.edu/epb/instruct/ecology/dendro.htm.
tion, typologie, évolution (Vol. 2). Coll. Archéologie aujourd’hui, Naurzbaev, M. M., & Vaganov, E. A. (1999). 1957-year chronology for
Archéologie Neuchâteloise. Eastern Taimir. Siberian Journal of Ecology, 6, 67–78.
Baillie, M. G. L., & Brown, D. M. (1988). An overview of oak Oberlin, C., Leroy, F., & Guibal, F. (2004). High precision 14C dating
chronologies. In E. A. Slates & J. O. Tate (Eds.), Science and of a bronze age tree-ring chronology from the pile-dwelling
archaeology (196, pp. 543–548), Glasgow 1987, Brit. Arch. settlement of Montpenèdre, Hérault, Southern France. In Proceed-
Rep. Brit. ings of the IVth Int. Symp. Radiocarbon and Archaeology, Oxford,
Barbetti, M., Bird, T., Dolezal, G., Taylor, G., Francey, R. J., Cook, E., 9–14/04/2002, Oxford University School of Archaeology Mono-
et al. (1995). Radiocarbon variations from tasmanian conifers: graph (Vol. 62, pp. 193–200).
Results from three early Holocene logs. Radiocarbon, 37(2), 361– Pearson, G. W. (1986). Precise calendrical dating of known
369. growth-period samples using a ‘curve fitting’ technique. Radiocar-
Cook, E. R., Buckley, B. M., D’Arrigo, R. D., & Peterson, M. bon, 28(2A), 292–299.
J. (2000). Warm-season temperatures since 1600 BC reconstructed Pilcher, J. R., Baillie, M. G. L., Schmid, B., & Becker, B. (1984).
from Tasmanian tree rings and their relationship to large scale A 7,272-year tree-ring chronology for Western Europe. Nature,
sea-surface temperature anomalies. Climate Dynamics, 16, 79–91. 312, 150–152.
de Vries, H. (1958). Variation of the concentration of radiocarbon with Rashit, M., Hantemirov, M., & Shiyatov, S. G. (2002). A continuous
time and location on Earth. Proceedings of the Koninklijke multimillenial ring-width chronology in Yamal, Northwestern
Nederlandse Akademie van Wetenschappen: Proceedings Siberia. The Holocene, 12(6), 717–726.
Series B, 61, 267–281. Robinson, W. J. (1976). Tree-ring dating and archaeology in the
Eckstein, D. (2006). Human time in tree-rings. Dendrochronologia, 24 American South-West. Tree-Ring Bull, 36, 9–20.
(2–3), 53–60. Roig, F., Jr., Roig, C., Rabassa, J., & Boninsegna, J. (1996). Fuegan
Eronen, M., Zetterberg, P., Briffa, K. R., Lindholm, M., Meriläinen, J., floating tree-ring chronology from subfossil Nothofagus Wood. The
& Timonen, M. (2002). The supra-long scots pine tree-ring record Holocene, 6(4), 469–476.
for Northern Finnish Lapland. Chronology construction and initial Schaub, M., Kaiser, K. F., Frank, D. C., Buentgen, U., Kromer, B., &
inferences. The Holocene, 12(6), 673–680. Talamo, T. (2008). Environmental change during the Allerød and
Ferguson, C. W. (1969). A 7404-year annual tree-ring chronology for Younger Dryas reconstructed from tree-ring data. Boreas, 37, 74–
bristlecone pine, pinus aristata, from the White Mountains, 86.
California. Tree-Ring Bull, 29(3–4), 1–29. Stuiver, M., Reimer, P. J., Bard, E., Beck, J. W., Burr, G. S., Hughen,
Ferguson, C. W., & Graybill, D. A. (1983). Dendrochronology of K. A., et al. (1998). IntCal98 Radiocarbon Age Calibration,
bristlecone pine: A progress report. Radiocarbon, 25(2), 287–288. 24,000-0 cal BP. Radiocarbon, 40(3), 1041–1083.
Grudd, H., Briffa, K. R., Karlén, W., Bartholin, T. S., Jones, P. D., & Suess, H. E. (1965). Secular variations in the cosmic ray produced
Kromer, B. (2002). A 7 400-year tree-ring chronology in Northern carbon-14 in the atmosphere and their interpretation. Journal of
Swedish Lapland: Natural climate variability expressed on annual to Geophysical Research, 70, 5937–5952.
millennial time scales. The Holocene, 12(6), 657–665.
 The Dating of Ice-Core Archives
9
Frédéric Parrenin

The wealth of testimony about past variations in our climate tens of thousands of years because the period of radioactive
and environment found in deep ice cores in Antarctica and decay of carbon-14 is 5730 years.
Greenland is acknowledged well beyond the limits of Datings developed by glaciologists are then based on
glaciological research. Uniquely, both local climate varia- complementary methods such as counting annual layers
tions and global atmospheric composition can be recon- (Section “The Counting of Annual Layers”), comparison
structed from a single archive: the ice. Effective use of the with other dated records (Section “Identification of dated
information provided by the glacial archives requires dating horizons”) and with variations in insolation (Section “Orbital
as precisely as possible of these various records. To do this, Tuning and Indicators of Local Insolation”), and glaciolog-
the specific characteristics of ice need to be considered. ical modeling (modeling of the accumulation of snow and
The first characteristic results from the compaction of the flow of ice, Section “Flow Modeling”). After these
snow layers under their own weight. At the surface, the snow methods have been presented, we will describe, in Sec-
is not very dense (0.3–0.4 g/cm3): air circulates freely in the tion “The Inverse Method: A Collective Approach”, a sta-
first meters of this porous milieu, the firn, and then with tistical technique, known as the ‘inverse method’, which
more difficulty as the density increases and the porosity consists of collecting these different sources of chronological
decreases. When the density is greater than about 0.83 g/cm3 information to achieve an optimum date and to assess its
(below about 100 meters in the center of Antarctica), air is confidence interval.
trapped in bubbles in the ice and insulated from the atmo-
sphere. In the depths, under the effect of pressure, the bub-
bles become compressed and are then transformed into Ice-Air Age Difference
clathrates, i.e. the gas molecules become incorporated into
the crystalline structure of the ice. This means that the air is Introduction
younger than the ice that imprisons it. Therefore, to date ice
core archives, which have some signals recorded in the ice The firn is the porous upper area of the ice caps. It marks the
and others recorded in the air bubbles, two distinct transition from snow on the surface to the ice below.
chronologies are required. The evaluation of the age differ- Depending on its location, its thickness can vary from
ence between gas and ice is discussed in Section “Ice-Air roughly 50 m (Greenland) to 120 m (central Antarctica). Its
Age Difference”. density varies from the surface density (typically 0.4 g/cm3)
Moreover, ice does not lend itself to the use of radioactive to the density at the close off depth, i.e. the depth at which
methods. Carbon-14 dating can only be used in exceptional the pores close (typically 0.83 g/cm3). At this depth, air is
cases, for example, on plant debris or when sufficient trapped in isolated bubbles and no longer circulates.
amounts of carbon dioxide are extracted. Although the The study of transport of air in the firn has led to the
quantities of ice necessary for carbon-14 dating have development of a simple model (Sowers et al. 1992) from
decreased since the advent of accelerator mass spectrometry, which we can distinguish four zones in the firn (Fig. 9.1).
the dates obtained are only averages over a few meters of
ice. In addition, this method is not applicable beyond a few • The convective zone is located just below the surface.
Convection in this zone is caused partly by the thermal
F. Parrenin (&) gradient and partly by surface winds. The depth of this
Institut des Géosciences de l’Environnement,
St Martin d’Hères, France zone varies from one site to another, and may reach 20 m
e-mail: [email protected]

© Springer Nature Switzerland AG 2021 123

G. Ramstein et al. (eds.), Paleoclimatology, Frontiers in Earth Sciences,
https://doi.org/10.1007/978-3-030-24982-3_9
124 F. Parrenin

than the ice surrounding it. The difference between the age
of the ice and the age of the gas bubbles is denoted as Dage.
In other words, the gas of the same age as the ice is found
lower down, and this difference in depth is denoted Ddepth.
In reality, it is not possible to attribute an exact age to the
gas at a given depth. As the air travels through the diffusive
column, it mixes the gases from atmospheres of different
periods, with a typical average time of a few decades.
Moreover, as the close-off boundary is not attributed to a
specific depth, but extends over several meters, gas trapped
at the same depth may have become imprisoned at slightly
different times. Therefore, the signal produced may be dif-
fuse, all the more so if the accumulation of snow is low.

Modeling the Densification of the Firn

Fig. 9.1 Diagram showing the different parts of the firn. Adapted To evaluate the Dage, one must evaluate:
from Sowers (1992)

• The density at the close-off which is often calculated from

in sites with high winds and rugged relief (dunes) at the the surface temperature, using observations carried out at
surface. different sites (Salamatin et al. 2009); it can also be
• The diffusive zone corresponds to an area where the air deduced from the concentration of air in the ice;
column is static (no convection). Movement happens at • The age of the gas at the close-off which is often ignored
the molecular level (diffusion), and in this zone, ele- in the case of Antarctic cores, as it is very tiny compared
mentary and isotopic fractionation is observed. For with the age of the ice; it can be assessed using a gas
example, the heaviest molecules are located, in priority, diffusion model for the firn in the case of Greenland
in the cold deeper part, while the lighter molecules are cores, where the age of the ice at close-off is only a few
more numerous in warmer, higher areas. hundred years (Schwander et al. 1997);
• The non-diffusive zone is the zone where the pores are • The density profile in the firn, which is derived from a
almost closed and where molecular diffusion is negligi- mechanical model; various mechanical models have been
ble. From this point, the composition of the bubbles no published (see, for example, Salamatin et al. 2009); they
longer changes, even though the total pressure can still generally take into account the slippage of snow grains
evolve. relative to each other, a dominant process at the surface,
• The close-off zone is located at the base of the firn. The and the deformation of the grains which becomes domi-
bubbles of air are closed off, and the air is trapped inside. nant at greater depths.
This close off zone has a well-defined density of about
0.83 g/cm3. This close-off depth will increase as the As shown in Fig. 9.2, the calculated depth of the close-off
surface temperature decreases and will vary according to increases when accumulation increases (vertical advection
the rate of accumulation on the surface of the firn. As a increases) or as the temperature decreases (densification
result, the depth of the close-off zone will vary during happens more slowly). These models were validated using
glacial and interglacial periods. current data (especially density profiles) from sites with very
varied average temperature and accumulation conditions,
The top of the non-diffusive zone is a depth of critical both in Antarctica and in Greenland (Fig. 9.2. See Salamatin
importance, above which the air contained in the pore space et al. 2007). However, it is worth noting that no site included
of the firn is still in contact with the surface atmosphere. corresponds to the conditions of the last ice age in Antarc-
Although the snow at this depth fell hundreds or even tica, which had very cold temperatures and very low accu-
thousands of years earlier, the gas is still at age ‘zero’, i.e. at mulation. Also, these validations only pertain to the present,
the age of the most recent snow. Consequently, at any depth with current orbital parameters and thus with very specific
of an ice core, the gas in the trapped air bubbles is younger daily and seasonal insolation distributions.
9 The Dating of Ice-Core Archives 125

Fig. 9.2 Depth of the close-off

of bubbles based on the rate of
accumulation and the surface
temperature, assumed to be at a
steady state, as calculated by the
model by Arnaud et al. (2000).
The conditions in different polar
sites are indicated by crosses.
Adapted from Landais et al.
(2006)

15 40 
a 
Application of the N and Ar Isotopes T
in the Bubbles dg ¼ 1  1000 ð9:2Þ
T0

Nitrogen and argon have isotopic compositions (in 15N and where T and T0 are the temperatures at either end of the
40
Ar) in the atmosphere almost constant over the timescales diffusive column and a is the thermal diffusion coefficient,
studied in ice cores. But the isotopic composition of the air which depends in a complex way on the temperature.
bubbles varies due to a process occurring in the firn.
Because of the mixing, no fractionation occurs in the Nitrogen-15 and argon-40 can thus be used in two dif-
convective zone. In the diffusive zone, two types of frac- ferent ways to constrain the age differences between ice and
tionation take place: gas.
Firstly, abrupt changes in temperature can be identified in
• Gravitational fractionation, under the influence of grav- both the ice (where it is recorded in the variations in the
ity, draws the heavy isotopes towards the bottom of the isotopic composition of oxygen and of hydrogen in the H2O
firn, according to the equation: molecule, see §11.3, Chap. 11), and in the air (in the isotopic

 
  composition anomaly due to thermal fractionation). Thus, an
T Dmgz estimate of Ddepth may be deduced. This method was used
dg ¼ exp  1  1000 ð9:1Þ
T0 RT to validate firn models in Greenland during major rapid
changes in temperature, called Dansgaard-Oeschger events.
where Dm is the difference in mass between the two iso-
In Antarctica, temperature variations are less abrupt, and so
topes, g is the gravitational acceleration, z is the height of the
detection of the temperature anomaly remains ambiguous.
diffusive column, R is the constant of perfect gases and T is
Secondly, assuming that the convective column is known
the temperature, expressed in Kelvin. This fractionation will
and that no fractionation takes place during the pore closing
therefore depend primarily on the height of the diffusive
process, we can calculate the thickness of the diffusive
column, and to a lesser extent, on the temperature of the firn.
column. This technique also served to validate the firn
All things being equal, the gravitational fractionation is
models in Greenland. For sites on the Antarctic plateau, the
proportional to the difference in mass between the two iso-
situation is more complex, because nitrogen-15 and
topes under consideration. So, it is four times higher for
argon-40 suggest a decrease in the diffusive column during
argon (40Ar and 36Ar) than it is for nitrogen (15N and 14N).
the glacial periods, although the firn models calculate an
• Thermal fractionation draws the heaviest types towards
increase in the thickness of the firn (Landais et al. 2006).
the cold extremity. Thermal fractionation in equilibrium
Three hypotheses may explain this discrepancy: (1) the
may be written as:
height of the convective zone increased during glacial
126 F. Parrenin

periods; (2) firn models do not apply to glacial conditions of and currently extends over the last 60,000 years (Svensson
the Antarctic plateau; (3) another fractionation process et al. 2008). More recently, the WAIS (West Antarctic Ice
occurs at the closure of the pores. This issue remains Sheet) Divide ice core has been counted back to
undecided at present. 31,000 years (Sigl et al. 2016).
Glaciologists use various records to identify annual lay-
ers. Where possible, the isotopic variations in the ice (d18O
Synchronization of Two Ice Cores and dD), which are dependent on the temperature at the
moment of the precipitation, provide the most reliable
As explained above, the ice/gas differential in cores with a recording of the changing of the seasons. However, water
low rate of accumulation from the Eastern Antarctic shelf molecules diffuse in the form of vapor through the firn, then
and during the ice ages is still poorly constrained. An more slowly through the ice. This diffusion smooths out the
alternative way to obtain an estimate, for both ice and gas seasonal isotopic signal until it disappears at a certain depth,
recordings, is synchronization with a core with a higher rate even more rapidly when accumulation is low and the tem-
of accumulation, wherein the ice/air differential is better perature is high. Thus, the seasonal cycle of isotopes is
constrained. hardly recognizable on the NorthGRIP core, which has a low
Loulergue et al. (2007) and Parrenin et al. (2012) have accumulation; it is quite muted in the GRIP core. The
applied this method to constrain the Dage of the EDC longest sequence on which the seasonal cycle oxygen-18
(EPICA Dome C) from the EDML (EPICA Dronning Maud was used was obtained from the Dye-3 core in Greenland:
Land) and the TALDICE (Talos Dome Ice Core) cores. Gas 67,000 isotopic analyses allowed the dating of the core year
synchronization is based on the rapid variations in methane, by year over the last 7900 years. Beyond that, the thickness
and ice synchronization uses volcanic signatures. So, this of annual layers is insufficient and the isotopic diffusion
study shows that the firn model, forced with temperature and through the ice makes counting inaccurate.
accumulation scenarios as for dating ice (Parrenin et al. Other data taken from the content of impurities in Con-
2007b), overestimates the Dage at EDC by 500–1000 years, tinuous Flow Analysis (CFA), from the Electrical Conduc-
during the last glacial period. Consequently, the densifica- tivity Measurement (ECM), from the insoluble dust content,
tion mechanism during glacial periods at EDC is poorly and from Visual Stratigraphy (VS) complete the isotopic
understood and the models need to be improved. information when this is available (Fig. 9.3). The CFA
The Dage, during glacial periods, for Antarctic plateau allows the various soluble compounds, such as Na+, Ca2+,
sites with low accumulation is therefore an open question. H2O2, NH4þ , NO 2
3 and SO4 to be separated. The ECM is a
Further studies are needed to clarify this issue. non-destructive measurement, conducted continuously in the
field, but it only provides information on an amalgamation of
these different soluble compounds. The VS uses the fact that
The Counting of Annual Layers impurities diffuse the light in the ice. However, this
recording generally shows several peaks in a year and is
On the polar ice caps and glaciers, many of the properties of therefore not easy to interpret. In Greenland and during the
snow differ depending on whether it accumulates in summer Holocene, a typical year is characterized by Na+ (dominated
or winter. For example, in summer, dust is more abundant in mainly by marine inputs) showing a peak in late winter.
the snow, because during this season, the winds are more Spring has a high dust content, high Ca2+ and low H2O2.
conducive to dust transport towards the poles. The annual Summer is characterized by high concentrations of NH4þ ,
layers can therefore be identified, either visually, or by NO 2
3 , and sometimes SO4 . This method, based on data
chemical analysis or by isotopic analysis. Counting annual from CFA, ECM and VS from GRIP and NorthGRIP, was
layers is a simple method of dating, provided that the the one principally used to establish GICC05 in the period
accumulation of snow is sufficient, so that the stratigraphy is between 7900 and 14,800 years b2k (this notation means
not destroyed by winds mixing the layers near the surface. ‘years before 2000’) (Rasmussen et al. 2006). In the older
For this reason, the counting of layers is impossible in the part (14,800–60,000 Years b2k), the method is the same, but
central regions of the Antarctic plateau where the deep only NorthGRIP data were used (Svensson et al. 2008).
drilling of Vostok, Dome C and Dome F are located, but it is In summary, none of the individual indicators is perfect,
possible over Greenland and the coastal regions of but combined, they permit an annual dating, as long as the
Antarctica. thickness of the layers, which thin out as they sink into the
A large project for systematic counting called Greenland ice cap, remain sufficient. For GICC05, counting was carried
Ice Core Chronology 2005 (GICC05) has been undertaken out by different people and on different cores, and the
by a Danish team at the Niels Bohr Institute in Copenhagen. independently obtained results were compared so as to
It is based on the cores of DYE-3, GRIP and NorthGRIP,
9 The Dating of Ice-Core Archives 127

Fig. 9.3 Example of a 1.2 m section from the GRIP core of about H2O2, Ca2+, NH4+ and d18O. For this last indicator, the thick line
8.8 ka with the annual layers marked by the gray vertical bars. From represents the raw data and the fine line represents data after correction
top to bottom, the records used to identify annual layers are: ECM, for the diffusion effect. Adapted from Rasmussen et al. (2006)

minimize subjective errors. Moreover, each uncertain layer Volcanic Horizons

was listed so as to obtain a confidence interval on the final
chronology. The listed errors are few: under 2% as far back Volcanic horizons can be identified in the cores, both in
as the last deglaciation, and around 5% before for older ages Antarctica and in Greenland. Some events are sufficiently
(Fig. 9.4). intense or occur sufficiently close to the core to deposit
volcanic dust (ash) visible to the naked eye. However, most
events only deposit fine aerosols. These can then be identi-
Identification of Dated Horizons fied with chemical analyses performed on the ice: sulfate, in
particular, has several peaks corresponding to volcanic
Even if ice cores cannot be dated directly by the conven- inputs that are easily identified because they far exceed the
tional radiochronological methods, events that have been usual levels. These volcanic horizons also change the
dated elsewhere can be identified in these cores: this is what dielectric properties of the ice, which enables them to be
we call the identification of dated horizons. In the following, identified by measuring its conductivity (this is a
we will detail the principal types of horizons used to date the non-destructive measurement, generally carried out in the
polar cores. field, immediately after the cores have been brought to the
surface).

Fig. 9.4 Depth/age relationship

on the NorthGRIP core according
to the GICC05 scale (solid line)
with associated uncertainty
(dotted line). The notation ‘b2k’
means before 2000
128 F. Parrenin

For the last few thousand years, a large number of these

volcanic events have been dated by different methods
(Zielinski et al. 1994): either by counting the layers in sites
with a sufficiently high accumulation or through historical
writings, or by dating volcanic material near the volcano in
question, especially by the carbon-14 method applied to
biological debris (e.g., trees caught in the lava). Starting with
an approximate glaciological dating, these events can gen-
erally be identified by their intensity in the cores. They can
also allow two ice cores to be stratigraphically connected
(see Fig. 9.5).
Before the Holocene, only a few layers of ash (Fig. 9.6)
can be dated accurately. Chemical analysis of these ashes
enables unambiguous identification of the relevant volcanic
eruption. This signature can sometimes be connected to that
of the volcanic material located close to a volcano. This Fig. 9.6 Ash layer present at about 239 m in the core drilled at Talos
volcanic material, in larger quantity, can be dated by con- Dome (East Antarctica). Copyright F. Parrenin (frederic.par-
[email protected])
ventional radiochronological methods.
In this way, a visible ash layer in the ice cores of Dome C
and Dome Fuji could be dated by an American team using changes in temperature during the last glacial period (see
the argon/argon method (see Chap. 5) to date ash found near §3.2, Chap. 4). Synchronous variations (within a few dec-
the volcano (Mt Berlin, Antarctica; Narcisi et al. 2006). For ades) were also observed in the atmospheric content of
Greenland, notable volcanic horizons were identified that are methane (Severinghaus et al. 1998) measured in air bubbles
referred as: Saksunarvatn, Vedde, Fugloyarbanki, ‘33 ka from both Antarctic and Greenland ice. These events can
14
C’ and Z2. These horizons were dated by the carbon-14 therefore be dated by the Greenland cores which are rela-
method or by the argon/argon method (see Svensson et al. tively precisely dated through counting of the annual layers
2008 and included references for further detail). (See Section “The Counting of Annual Layers”). This dating
can then be transferred to the Antarctic cores thanks to the
methane records.
Dansgaard-Oeschger Events The changes in climate associated with the
Dansgaard-Oeschger events, although their maximum
Dansgaard-Oeschger events (D-O) were identified for the
impact was probably at the level of the North Atlantic, are
first time in Greenland ice cores and correspond to abrupt
visible in many locations on the planet, in other climate

Fig. 9.5 Sulfate profiles for the first tens of meters in two ice cores Land). Several well-known eruptions can be identified. Moreover, these
from Eastern Antarctica: EDC (EPICA Dome C, bottom) and B32 volcanic profiles can be used to synchronize the ice cores between each
(top), the latter located near the EDML site (EPICA Dronning Maud other. Adapted from Severi et al. (2007)
9 The Dating of Ice-Core Archives 129

archives. They are found in the isotopic composition of Beryllium-10 can be accurately measured in ice cores,
speleothems (stalactites and stalagmites found in caves) in both from Antarctica (see Raisbeck et al. 2007 and included
Europe and Asia, in South America and the Indian Ocean. references) and Greenland (see Beer et al. 2006 and included
These speleothems can be dated with a precision of a few references). This allows the dating achieved by counting the
hundred years to one or two thousand years, using the annual layers in Greenland to be transferred to Antarctic
uranium/thorium method, which provides a specific age for cores for the Holocene (Ruth et al. 2007) and for the
the marked transitions of the D-O events. Figures 9.7 and Laschamp anomaly in the geomagnetic field, which occurred
9.8 summarize these recordings with their respective about 41 ka ago (Raisbeck et al. 2007).
dating. Carbon-14, meanwhile, is measured in tree rings, which
are very accurately dated for the last 12.4 ka using den-
drochronology (see Chap. 8). We can then import this dating
Variations in the Magnetic Field and in Solar to the ice cores when the variations in solar activity are
Activity significant enough so that beryllium-10 and carbon-14 can
be synchronized. This method has been used to date the
Beryllium-10 and carbon-14 are both produced in the upper Holocene part of the Antarctic ice cores, where counting of
atmosphere by the flux of cosmic particles. This flux is the layers is not possible (Ruth et al. 2007).
modulated, partly by the magnetic field of the solar wind Finally, significant anomalies in the geomagnetic field
which deflects the charged particles, and partly by the ter- can be identified in other paleoenvironmental archives such
restrial magnetic field. Unlike beryllium-10 whose deposi- as volcanic lava, which can be dated by the argon/argon or
tion on the surface of the Earth is almost directly related to potassium/argon methods. The Laschamp anomaly is thus
its production in the upper atmosphere, the composition of dated with a relatively good accuracy (Guillou et al. 2004),
carbon-14 in the atmosphere is also influenced by the while the older Bruhnes-Matuyama transition is only very
exchanges between the different carbon reservoirs on Earth. crudely dated (Raisbeck et al. 2006). Chap 7 provides more
But the major changes in these two indicators (carbon-14 detail on magnetic stratigraphy.
and beryllium-10) are simultaneous.

Age (yr)

Fig. 9.7 The Dansgaard-Oeschger events identified in the North-GRIP core (GICC05 dating), and in records from the Kleegruben (Spötl et al.
2006) and Moomi (Burns et al. 2003) caves, the latter being dated by a uranium/thorium method. Adapted from Svensson et al. (2008)
130 F. Parrenin

H1 H4 H5 H6

Age (yr)

Fig. 9.8 The Dansgaard Oeschger events identified in the NorthGRIP their error bar). The shaded areas represent Heinrich events as identified
core (GICC05 dating) and records from the Hulu cave (Wang et al. in speleothems in Brazil (Wang et al. 2004). These points dated by the
2001). The points dated in the records from Hulu Cave by the uranium/thorium method are also shown at the top of the figure.
uranium/thorium method are marked at the bottom of the figure (with Adapted from Svensson et al. (2008)

terrestrial biosphere. This is the Dole effect which depends

Orbital Tuning and Indicators of Local in a complex way on the reactions of photosynthesis and
Insolation respiration. The d18Oatm signal shows variations mainly
related to the precession, so it is very easy to use it to
The evolution of orbital parameters of the Earth over the last ‘count’ these cycles, at least for the periods during which
million years is known with great precision (Laskar et al. they are sufficiently important variations in the insolation
2004), and these variations leave a signature in most climate signal. However, there is no reason to assume that the
records. Therefore, the use of variations in insolation to date phase difference between d18Oatm and insolation remained
polar cores is a natural choice. Uncertainty in this dating is constant over time.
due, in part, to the assumption that there is a constant phase To avoid this limitation in the ‘traditional’ methods of
difference between orbital and climatic variations, and, orbital alignment, more direct indicators of local insolation
secondly, to the evaluation of this phase difference. The have recently been proposed. Local insolation in summer
advantage of this dating is that it has an almost constant alters the structure of the snow at the surface, and this sig-
uncertainty of a few thousand years along the length of the nature remains present down to the close-off zone, regardless
core, especially for the deeper parts, as long as we are able to of the densification process. These structural parameters
recognize the orbital cycles. have an impact on the volume of the pores at close-off (and
Several parameters recorded in the ice cores show strong thus the air content), and also on the molecular fractionation
variations in the orbital frequencies. These were then used processes between O2 and N2 when the pores close.
to align the ice cores with orbital cycles: these are the D/H Bender (2002), was the first to suggest that the O2/N2
ratio in the ice, an indicator of local temperature (for ratio in the air bubbles analyzed in the Vostok core was
example, Parrenin et al. 2004) and the 18O/16O (d18Oatm) dependent on the local insolation at the summer solstice
ratio in the air bubbles (for example, Dreyfus et al. 2007). (Fig. 9.9). Recently, this link has been confirmed in the first
The variations in d18Oatm are a reflection of two aspects of core drilled at Dome Fuji (Kawamura et al. 2007) (Fig. 9.9)
the environment (Landais et al. 2010). The first one is that and recent measurements in the second core, with an
the variations in the d18O of the ocean, directly related to improved analytical process, show an almost perfect
the volume of land ice, impact fully on the d18O of the correlation.
atmospheric oxygen through the process of photosynthesis Raynaud et al. (2007) also studied in more detail the air
by seaweed. The second one is that a portion of the vari- content in the EDC core and suggested that it depended on
ations in d18Oatm is determined by the behavior of the local insolation averaged over a period centered on the
9 The Dating of Ice-Core Archives 131

Fig. 9.9 a Isotopic composition

of the ice and temperature at
Vostok and Dome Fuji. b O2/N2
ratio measured in the air bubbles
at Dome Fuji and Vostok. The
dots are the raw data; the thick
lines represent the filtered data.
c Age markers deduced from the
alignment with local insolation,
with 2r error bars. d Insolation at
the summer solstice at 77° S used
as an alignment target. e 2r
uncertainty of the O2/N2 dating.
Adapted from Kawamura et al.
(2007)

Age (kyr)

summer solstice. However, for this second indicator, it is The parameter a is generally calculated from indicators
important to correct for variations in the altitude at which the measured in the ice core, while the parameter T is obtained
bubbles formed, as this will have an impact on atmospheric from a flow model. These two steps are detailed below.
pressure and thus the air content of these bubbles.
Although the specific physical link between these indi-
cators and insolation is still subject to debate and research, Evaluation of Accumulation on the Surface
we can make two observations. Firstly, no signal with a 100
000 year periodicity is present in the O2/N2 record, so that For the top few hundreds of meters, the thinning of the
the O2/N2 proxy does not seem to be dependent on climate. layers of snow and ice is minimal (T close to 1) and
Secondly, although these indicators are measured in the gas well-assessed by modelling. So, accumulation at the surface
bubbles, they are caused by modifications in the structure of may be determined from well-dated horizons such as layers
the snow at the surface, and therefore provide a dating of the of volcanic ash (described in Section “Volcanic Horizons”)
ice (and not of the bubbles!). This avoids uncertainty asso- using the formula (9.3). Below this depth, the isotopic
ciated with Dage. composition of the ice (D/H or 18O/16O) is generally used.
As for the surface temperature, the field measurements in
Antarctica and Greenland show a good correlation between
Flow Modeling isotopic composition and the surface accumulation of snow.
In a review of measurements in Antarctica, Masson-
The ice has an enormous advantage over other archives in Delmotte et al. (2008) derived a relationship as follows:
that it can be dated using physical models that take the
variations in the rate of accumulation of snow and the flow a ¼ a0 expðbðdD  dD0 ÞÞ ð9:4Þ
of ice into account. The age in the ice core at an altitude of where a and dD are respectively, a reference accumulation
z can be written as: and a reference isotopic composition, and where b = 0.0152.
Z This relationship is derived from the saturated vapor pres-
ðzÞ d ðz0 Þ
v ¼ dz0 ð9:3Þ sure of the ice and can be calculated from a simple model of
T ðz0 Þaðz0 Þ
precipitations of an air mass. Note, however, that it does not
where v is the age of ice, d is the relative density of ice take into account the phenomenon of re-deposition of the
(compared with pure ice), a is the initial accumulation of snow by the wind, which modifies the accumulation without
snow (expressed in cm equivalent to pure ice per year, altering the isotopic composition of the snow. On the other
denoted as cm-i.e./year) and T is the thinning function, i.e. hand, when we extrapolate this relationship to temporal
the thickness of an annual layer relative to its initial thick- variations in accumulation, it is important to consider the
ness at the time it fell. d can be measured from the ice core. temperature variations and isotopic composition at the
132 F. Parrenin

source of the air masses which also alter the isotopic com- linearly from 0 at the ice base interface to 1 at the surface,
position of the ice (Parrenin et al. 2007b). because the deformation is concentrated at the base of gla-
cier. For certain domes, the Raymond effect causes more
deformation at the top of the ice and therefore a less linear x
Ice Flow Models profile.
In a non-stationary case, variations in the thickness of the
Ice has a solid exhibiting viscoplastic behavior where the ice (related to climatic variations) cause bumps in the thin-
relationship between stress and strain can be determined ning function (Fig. 9.10). Moreover, for ice core drilled
experimentally and theoretically. It is thus possible to sim- along the flow line, like Vostok, the ice comes from
ulate the trajectory followed by a particle of ice within the upstream and more complex deformation effects exist. The
glacier over time in order to establish a chronology. parameter that most influences the thinning function is the
Modeling of behavior of ice within an ice sheet requires not thickness of ice at the place of origin of the ice: if this
only a good knowledge of the viscoplastic properties of the thickness is large compared to the thickness at the drill site,
material, but of the conditions at the boundaries of the then the column of ice has become compressed overall,
cap. These boundary conditions are: (1) the temperature and resulting in a strong thinning (that is, a low thinning func-
surface accumulation over time; (2) basal conditions, such as tion). And reciprocally.
the geothermal flow and the rate of friction on the bedrock;
(3) the lateral conditions for the area under consideration,
since local models are used for dating purposes. These lat- The Limitations of Modeling
eral conditions generally result from global simulations of
the polar cap over time (Ritz et al. 2001). In this way, the Unfortunately, dating using models becomes increasingly
thinning function adapted to the ice core drilling site is inaccurate as it approaches the base of the ice cap, for var-
obtained. ious reasons. Firstly, the mechanical properties of the ice are
Below is a qualitative description of how this function not perfectly understood. They depend not only on pressure
varies. For a stationary dome, the thinning function can be and temperature conditions, but also on the size and orien-
written: tation of the crystals that make up the ice. Secondly, the
conditions at the base of the bedrock cannot be measured
1  ma
T¼ ð9:5Þ directly in situ. Finally, the lateral conditions throughout the
x þ ma past, the outcome of a large-scale model, may also be tainted
by a significant error. These lateral conditions determine the
with l = m/a the ratio of basal fusion to surface accumula-
position of the domes and dividing lines in the domain, and
tion and with x the standard vertical profile of horizontal
therefore the direction of the ice particles.
flow (see Parrenin et al. 2007b, for details). x varies almost

Fig. 9.10 Thinning functions for ice cores from Dome C (Parrenin et al. 2007b), Dome Fuji (Parrenin et al. 2007a) and Vostok (Parrenin et al.
2004)
9 The Dating of Ice-Core Archives 133

The Inverse Method: A Collective Approach framework was developed (Parrenin et al. 2001). In the
context of dated horizons with a certain error bar, it facili-
All of absolute dating methods described in the preceding tates the determination of the probability density called ‘a
sections have advantages and disadvantages. The counting posteriori’ both for the uncertain parameters in the flow
of layers and flow modeling are precise methods in terms of models and for the final dating. Concretely, this probability
the duration of events (at least for relatively recent periods) density not only provides an optimum dating, but also a
because they are based on an assessment of the thickness of confidence interval. This probabilistic method, based on the
the annual layers. However, errors accumulate and inaccu- Metropolis-Hastings algorithm, was applied to the Vostok
racy in the absolute age increases rapidly with depth. (Parrenin et al. 2004), Dome C (EDC3 dating, Parrenin et al
Orbital tuning is generally applicable over the entire 2007a, b) and Dome Fuji (Parrenin et al. 2007b) ice cores.
length of a core, as long as the stratigraphy is maintained. In This inverse approach, however, has several limitations.
addition, accuracy does not diminish with depth, and so far, Firstly, it only takes into account the errors related to the
this is the most accurate method for dating the lower part of lack of understanding of poorly known parameters. In other
polar cores. Unfortunately, this method is not very accurate words, the model is considered perfect once an optimal
in terms of duration of events, and the accuracy in terms of estimation of these parameters is achieved. This is the same
absolute ages is limited by the assumption that there is a as saying that the model is able to describe all the relevant
constant phase difference between the record being studied flow mechanisms and is therefore, in agreement with all sets
and the insolation (and obviously does not allow the varia- of markers whose age errors are correctly estimated. In
tions in this phase to be reconstructed). Research into indi- reality, however, many physical phenomena influencing
cators of local insolation paves the way for a significant flow are not taken into account in the model, either because
improvement in methods of orbital alignment, with a pos- they are not properly understood, or because the level of
sibility of uncertainties below 1000 years. However, we still complexity necessary to describe them is incompatible with
lack sufficient distance to assess the real accuracy of these inverse modeling (direct model is too costly in computing
methods. In addition, other methods will always be more time with too many parameters to inverse). This limitation of
accurate for very recent times. the models appears clearly at the base of the EDC core
Volcanic eruptions provide important dated horizons. where the model, even after optimization of its parameters, is
This is particularly true for the last millennium, but beyond unable to reproduce the age markers obtained by orbital
this, only a few have an absolute age that is sufficiently alignment (Dreyfus et al. 2007). A second limitation of this
precise. Comparison with other dated records is particularly inverse method is that it can be applied to only one core at a
useful for the dating of Dansgaard-Oeschger events, which time, and the optimal dating obtained is different for each
can be precisely localized using variations in the isotopic core, making it difficult to compare climate and environ-
composition of ice (in Greenland cores only), or variations in mental signals. In reality, these cores can be synchronized
the methane content in air bubbles. Most of these events more accurately in the ice phase (for example, by volcanic
have been dated accurately for the last ice age, especially horizons), as well as in the gas phase (for example, CH4 and
from speleothems and the U/Th method (Chap. 6). Other d18Oatm).
studies are underway to improve this accuracy and to study As a result, a new method of optimization was developed
previous ice ages. In any case, these dated horizons do not (Lemieux-Dudon et al. 2010; Parrenin et al. 2015). This
provide continuous dating and are mainly relevant to recent considers the information gleaned from modeling to be
periods. weakly constrained (the model is not deemed perfect) and
As these different sources of chronological information applies to several cores simultaneously, taking into account
are complementary, it is clear that to obtain an optimum the stratigraphic links between these cores, both in the ice
dating from the ice records, it is essential to combine them. and in the air. This method can thus provide an optimum
This is what glaciologists have attempted to do with dating common to the different cores from Antarctica and
Antarctica ice cores. Initially, the poorly known parameters Greenland.
in flow models (such as melting and sliding at the base of the
glacier) were adjusted by trial and error to obtain a good
agreement with dated horizons. But this approach quickly Conclusion
becomes difficult when there are several free parameters and
when the different error bars for the dated horizons must be The dating of glacial archives is a complex problem which,
taken into account. in the absence of the radioactive methods, is based on sev-
In the early 2000s, an inverse method to formalize and eral complementary techniques. For the Holocene and in
systematize this optimization of dating in a probabilistic high accumulation sites, dating by counting layers is
134 F. Parrenin

accurate to about 1%. This dating was confirmed by com- insolation quantities of the earth. Astronomy & Astrophysics, 428,
parison with dendrochronology using beryllium-10, and with 261–285.
Lemieux-Dudon, B., Blayo, Petit, J. R. E., Waelbroeck, C., Svensson,
volcanic horizons dated by carbon-14. The last millennium A., et al. (2010). Consistent dating for Antarctica and Greenland ice
can also be dated to within a few years through identification cores’, Quaternary Science Reviews, 29(1–2), 8–20.
of known volcanic eruptions. For the rest of the Holocene, Masson-Delmotte, V., Hou, S., Ekaykin, A., Jouzel, J., Aristarain, A.,
synchronization with the dendrochronological scale using Bernardo, R. T., et al. (2008). A review of Antarctic surface snow
isotopic composition: Observations, atmospheric circulation, and
beryllium-10 gives us dating accuracy to within a few isotopic modeling. Journal of Climate, 21(13), 3359–3387.
decades. Narcisi, B., Petit, J.-R. & Tiepolo, M. (2006). A volcanic marker
For the last glacial period, the accuracy of counted time (92 ka) for dating deep east Antarctic ice cores. Quaternary Science
scales decreases to about 5% up to 60 ka. This dating pro- Reviews, 25, 2682–2687.
Parrenin, F., Barker, S., Blunier, T., Chappellaz, J., Jouzel, J., Landais,
vides ages for the Dansgaard-Oeschger events which are A., et al. (2012). On the gas-ice depth difference (Ddepth) along the
confirmed by uranium/thorium dating of speleothems in EPICA Dome C ice core. Climate of the Past, 8(4), 1239–1255
Europe and Asia. It is also compatible with some volcanic https://doi.org/10.5194/cp-8-1239-2012.
horizons dated by carbon-14 and argon/argon. The main Parrenin, F., Barnola, J.-M., Beer, J., Blunier, T., Castellano, E.,
Chappellaz, J., et al. (2007a). The EDC3 chronology for the EPICA
methods of dating for the last 60 ka therefore now seem to dome C ice core. Climate of the Past, 3, 485–497.
be in agreement to within a few hundred years rather than a Parrenin, F., Bazin, L., Capron, E., Landais, A., Lemieux-Dudon, B., &
few thousand years as was the case until recently. Masson-Delmotte, V. (2015). IceChrono1: A probabilistic model to
Further back than the last glacial period, counting of compute a common and optimal chronology for several ice cores.
Geoscientific Model Development, 8(5), 1473–1492. https://doi.org/
layers is no longer possible, and the chronologies are mainly 10.5194/gmd-8-1473-2015.
based on speleothems, volcanic horizons and orbital align- Parrenin, F., Dreyfus, G., Durand, G., Fujita, S., Gagliardini, O., Gillet,
ment. Orbital tuning has a precision of about 5 ka. Local F., et al. (2007b). Ice flow modelling at EPICA dome C and dome
insolation indicators (O2/N2 ratio, air concentrations) could Fuji, East Antarctica. Climate of the Past, 3, 243–259.
Parrenin, F., Rémy, F., Ritz, C., Siegert, M., & Jouzel, J. (2004). New
lead to an improved accuracy of 1–2 ka, but this must be modelling of the Vostok ice flow line and implication for the
confirmed by independent methods. glaciological chronology of the Vostok ice core. Journal Geophys-
ical Research, 109, D20102.
Raisbeck, G. M., Yiou, F., Cattani, O., & Jouzel, J. (2006). 10Be
Evidence for the Matuyama-Brunhes geomagnetic reversal in the
References EPICA dome C ice core. Nature, 444(7115), 82–84.
Raisbeck, G. M., Yiou, F., Jouzel, J., & Stocker, T. F. (2007). Direct
Arnaud, L., Barnola, J.-M., & Duval, P. (2000). Physical modeling of north-south synchronization of abrupt climate change record in ice
the densification of snow/firn and ice in the upper part of polar ice cores using beryllium 10. Climate of the Past, 3(3), 541–547.
sheets. In T. Hondoh (Ed.), Physics of Ice CoreRecords (pp. 285– Rasmussen, S. O., Andersen, K. K., Svensson, A. M., Steffensen, J. P.,
305). Japan.: Hokkaido University Press, Sapporo. Vinther, B. M., Clausen, H. B., et al. (2006). A new greenland ice
Beer, J., Vonmoos, M., & Muscheler, R. (2006). Solar variability over core chronology for the last glacial termination. Journal Geophys-
the past several millennia. Space Science Reviews, 125(1), 67–79. ical Research, 111, D06102.
Bender, M. L. (2002). Orbital tuning chronology for the Vostok climate Raynaud, D., Lipenkov, V., Lemieux-Dudon, B., Duval, P., Loutre,
record supported by trapped gas composition. Science Letters, 204, M.-F. & Lhomme, N. (2007). The local insolation signature of air
275–289. content in Antarctic ice. A newstep toward an absolute dating of
Burns, S. J., Fleitmann, D., Matter, A., Kramers, J., & Al-Subbary, A. ice records. Earth and Planetary Science Letters, 261(3–4), 337–
(2003). Indian Ocean climate and absolute chronology over 349.
dansgaard/oeschger events 9–13. Science, 301, 1365–1367. Ritz, C., Rommelaere, V., & Dumas, C. (2001). Modeling the evolution
Dreyfus, G. B., Parrenin, F., Lemieux-Dudon, B., Durand, G., of antarctic ice sheet over the last 420,000 years: Implications for
Masson-Delmotte, V., Jouzel, J., et al. (2007). Anomalous flow altitude changes in the Vostok Region. Journal Geophysical
below 2700 m in the EPICA Dome C ice core detected using d18O Research, 106(D23), 31943–31964.
of atmospheric oxygen measurements. Climate of the Past Discus- Ruth, U., Barnola, J.-M., Beer, J., Bigler, M., Blunier, T., et al.
sions, 3(2), 341–353. https://doi.org/10.5194/cp-3-341-2007. ‘EDML1: A chronology for the EDML ice core, Antarctica, over
Guillou, H., Singer, B. S., Laj, C., Kissel, C., Scaillet, S., & Jicha, B. R. the last 150 000 Years. Climate of the Past, 3, 475–484 (2007).
(2004). On the age of the Laschamp geomagnetic excursion. Earth Salamatin, A. N., Lipenkov, V. Y., Barnola, J. M., Hori, A., Duval, P.,
and Planetary Science Letters, 227, 331–343. & Hondoh, T. (2009). Snow-Firn Densification in Polar Ice Sheets.
Kawamura, K., Parrenin, F., Uemura, R., Vimeux, F., Severinghaus, In T. Hondoh (Ed.), Physics of Ice Core Records-2. Sapporo:
J. P., Matsumoto, K., et al. (2007). Northern hemisphere forcing of Hokkaido University Press.
climatic cycles over the past 360,000 years implied by absolute Schwander, J., Sowers, T., Barnola, J.-M., Blunier, T., Fuchs, A., &
dating of antarctic ice cores. Nature, 448, 912–917. Malaizé, B. (1997). Age scale of the air in the summit ice:
Landais, A., Barnola, J., Kawamura, K., Caillon, N., Delmotte, M., Implication for the glacial-interglacial temperature change. Journal
Ommen, T. V., et al. (2006). Firn-Air d15N in modern polar sites Geophysical Research, 102, 19483–19493.
and glacial-interglacial ice: A model-data mismatch during glacial Severi, M., Castellano, E., Morganti, A., Udisti, R., Ruth, U., Fischer,
periods in antarctica? Quaternary Science Reviews, 25(1–2), 49–62. H., et al. (2007). Synchronisation of the EDML1 and EDC3
Laskar, J., Robutel, P., Joutel, F., Gastineau, M., Correia, A. C. M., & timescales for the Last 52 Kyr by volcanic signature matching.
Levrard, B. (2004). A Long-term numerical solution for the Climate of the Past, 3, 367–374.
9 The Dating of Ice-Core Archives 135

Severinghaus, J., Sowers, T., Brook, E., Alley, R., & Bender, M. Svensson, A., Andersen, K. K., Bigler, M., Clausen, H. B.,
(1998). Timing of abrupt climate change at the end of the Younger Dahl-Jensen, D., Davies, S. M., et al. (2008). A 60 000 year
Dryas interval from thermally fractionated gases in polar ice. greenland stratigraphic ice core chronology. Climate of the Past, 4
Nature, 391, 141–146. (1), 47–57.
Sigl, M., Fudge, T. J., Winstrup, M., Cole-Dai, J., Ferris, D., Wang, X., Auler, A. S., Edwards, L., Cheng, H., Cristalli, P. S., Smart,
McConnell, J. R., et al. (2016). The WAIS Divide deep ice core P. L., et al. (2004). Wet periods in Northeastern Brazil over the past
WD2014 chronology—Part 2: Annual-layer counting (0–31 ka BP). 210 Kyr linked to distant climate anomalies. Nature, 432, 740–743.
Climate of the Past, 12(3), 769–786. https://doi.org/10.5194/cp-12- Wang, Y. J., Cheng, H., Edwards, R. L., An, Z. S., Wu, J. Y., Shen, C.
769-2016. C., & Dorale, J. A. (2001). A high-resolution absolute-dated late
Sowers, T. A., Bender, M., Raynaud, D. & Korotkevich, Y. L. (1992). pleistocene monsoon record from Hulu Cave, China. Science, 294
‘The d15N of O2 in air trapped in Polar ice: A tracer of gas transport (5550), 2345–2348.
in the firn and a possible constraint on ice age-gas age differences. Zielinski, G. A., Mayewski, P. A., Meeker, L. D., Whitlow, S.,
Journal of Geophysical Research, 97(15), 15683–15697. Twickler, M. S., Morrison, M., et al. (1994). Record of volcanism
Spötl, C., Mangini, A., & Richard, D. A. (2006). Chronology and since 7000 B.C. from the GISP2 greenland ice core and implications
Paleoenvironment of marine isotope stage 3 from two for the volcano-climate system. Science, 264(5161), 948–952.
high-elevation speleothems, Austrian Alps. Quaternary Science Parrenin, F., Jouzel, J., Waelbroeck, C., Ritz, C. and Barnola, J.-M
Reviews, 25(9–10), 1127–1136. (2001) Dating the Vostok ice core by an inverse method. Journal of
Geophysical Research: Atmospheres 106, (D23), 31831–837851.
 Reconstructing the Physics and Circulation
of the Atmosphere 10
Valérie Masson-Delmotte and Joël Guiot

The variability and evolution of the physical parameters of generally only available from 1860, except for the most
atmospheric circulation are currently monitored in real time inaccessible areas, such as Antarctica, where systematic
and on a global scale thanks to a dense network of weather meteorological monitoring did not start until the Interna-
stations (over 13,000 measurement sites on land and sea), tional Geophysical Year 1957–1958. The ‘instrumental
and to satellite observations of the Earth. This ‘instrumental’ period’ is therefore very short compared to the time frame of
period, during which the physical parameters of the atmo- the climate system and does not permit an understanding of
sphere were directly monitored, began in the mid seven- the natural climate variability on a global scale for the period
teenth century following the invention and use of prior to when human activities affected the composition of
thermometers, barometers, rain gauges etc. However, stan- the atmosphere.
dardization of measurement tools and their wide-scale use In order to characterize natural climate evolution and to
took a long time and was due to a continuous effort by the place the climate change of recent decades within a broader
meteorological services. The outputs of ancient instruments context, continental paleoclimatology has established meth-
from before 1950 must be homogenized to modern obser- ods of quantifying ancient climates by taking advantage of a
vation standards, and gaps exist in regional temperature large number of natural archives, in soils, lakes, vegetation,
information due to changes in the spatial monitoring net- continental and polar ice. These archives have allowed qual-
work. The oldest meteorological series of data available are itative or quantitative indices of the main parameters
in Europe, where temperature series for the center of Eng- describing climate to be defined. These indices are often
land start in 1659. Intensive work was carried out on the referred to as proxies. Below, we briefly review all the climate
weather records of the Alpine region, providing access to parameters reconstructed from these continental archives.
accurate measurements, from 1780, of average monthly The atmospheric parameters most commonly determined
temperatures and cumulative monthly precipitations. from continental paleoclimate archives are surface air tem-
Work is underway to extend the use of these historical perature (or surface lake water temperature) and parameters
measurements to study monthly variability in temperature related to soil hydrology (e.g. precipitation, drought indices,
and precipitation, and to assess other parameters (pressure, etc.). In some cases, these parameters can be estimated over
sunshine etc.). Use of these old measurements involves a season, when a resolution of less than a year can be
working on documents of the time, computerizing the data, detected in the archives (tree rings, ice cores in sites with a
and statistical analysis of regional databases. With the high level of snow accumulation) or when the archive is
exception of Europe, where instrumented measurements particularly sensitive to seasonal effects (temperature of the
were conducted particularly early, weather information is coldest month, temperature of the growing season for veg-
etation, etc.). Most continental proxies do not directly record
the amount of precipitation, but reflect the local water bal-
V. Masson-Delmotte (&)
Laboratoire des Sciences du Climat et de l’Environnement, ance (precipitation minus land-based evaporation, runoff into
LSCE/IPSL, CEA-CNRS-UVSQ, Université Paris-Saclay, 91190 lakes, net accumulation of snow on the glaciers at the dril-
Gif-sur-Yvette, France ling sites). Quantification of these climate parameters from
e-mail: [email protected] the records is often made difficult by the discontinuous
J. Guiot nature of geological recording, for example, the process of
Centre for Research and Teaching in Environmental Geoscience sedimentation in lakes. Some records, such as tree rings,
CEREGE, Aix-Marseille University, CNRS, IRD, INRAE,
Collège de France, BP 80 13545 Aix-en-Provence Cedex 04, function as threshold systems and identify an atmospheric
France signal once the threshold is reached (low temperature, dry

© Springer Nature Switzerland AG 2021 137

G. Ramstein et al. (eds.), Paleoclimatology, Frontiers in Earth Sciences,
https://doi.org/10.1007/978-3-030-24982-3_10
138 V. Masson-Delmotte and J. Guiot

season). Moreover, many proxies are not sensitive to a single archives and some proxies particularly well suited to each of
atmospheric variable but to the combination of effects related the interfaces under consideration. We have considered high
to temperature and hydrology. The combined use of multiple latitudes and low latitudes separately. Several archives (la-
markers within a single medium or multiple archives from custrine cores, speleothems) and proxies (pollen, diatoms)
the same site allows these effects to be separated out. Finally, are present in both temperate and tropical regions, but their
comparison between proxies, paleoclimate reconstructions, interpretation is specific to the particular geographical area.
modeling of climate and proxies all improve our under- In the high latitudes of both hemispheres, polar ice is of
standing of how climate dynamics and proxies operate. paramount importance in paleoclimatology because it
The dynamics of the atmosphere can also be estimated records both climate forcings and some local and global
from the continental paleoclimate records. During the climatic variations. Sedimentary archives (from lakes and
instrumental period, it is possible to determine how certain bogs) with their pollen records have long been studied in
modes of atmospheric circulation such as the El Niño-- paleoclimatology and cover almost all of the continents.
Southern Oscillation (ENSO), the North Atlantic Oscillation They were initially understood to mostly reflect a local
(NAO), the Southern Annular Mode (SAM) or the signal. However, the comparison with ice and marine
Pacific-North American oscillation (PNA) modulate the records revealed the broad geographical spread of many
spatial response of proxies, just as they affect the spatial events known to palynologists, such as the Younger Dryas
distribution of temperature and rainfall. Provided that the cold episode which lasted for close to a millennium and was
spatial distribution of proxy records is sufficient in key areas, felt throughout the Atlantic area of the northern hemisphere.
and depending on the stability of these tele-connections The loess covers a significant area of the continents; their
through time, this fingerprint can then be used to estimate sequences are excellent indicators of atmospheric circula-
past inter-annual variations in pressure indices characteristic tion. On a shorter timescale, some archives provide infor-
of patterns of atmospheric circulation in recent centuries mation with annual or near annual resolution. Among these,
(PNA, NAO, ENSO, SAM). tree rings provide a wealth of information on temperate
Over large time scales, loess deposits and dunes reflect regions. They are supplemented by multi-centennial archives
the prevailing wind direction. Similarly, concentrations of from historical written documents, such as wine harvesting
marine and continental aerosols, the size distribution of dates, in Europe, or cherry blossom dates, in Japan.
continental dust particles in polar ice, reflect changes in the At low latitudes, sedimentological tracers collected in
aridity of the regions of origin as well as changes in the tropical lakes, complemented by biological proxies, such as
efficiency of transport of aerosols in the atmosphere. How- the abundance of diatoms or pollen, contribute to a better
ever, quantitative estimates of the intensity of surface winds understanding of the functioning of the major inter-tropical
remain a challenge. Past changes in other atmospheric climate systems and give an insight into sometimes dis-
parameters such as cloudiness are difficult to determine from continuous records. Diatoms are good indicators of the
proxy records in natural archives. characteristics of lake water and, complemented by adequate
Continental paleoclimatology can also help to character- hydrological modeling, they allow an assessment of water
ize the frequency and intensity of ‘extreme’ events. The resources in watersheds. At high altitudes, tropical glaciers
intensity and amplitude of past droughts have been estimated are very sensitive to climate variations in the long and
using dendrochronological databases in North America and medium term.
Europe. Sedimentary and geochemical markers from lake Cave records, for instance from speleothems, offer
sediments are used to determine the intensity and occurrence well-dated, albeit discontinuous, records particularly sensi-
of flooding by the great rivers. High-resolution analysis of tive to changes in the hydroclimate and vegetation cover
lagoon sediments and the isotopic composition of tree rings above the caves, sometimes with very high temporal
or speleothems are currently being used in an attempt to resolution.
characterize past variations in the activity (trajectories,
intensity, frequency) of tropical cyclones and extratropical
storms. Interpretation of Records, Limitations
We have briefly presented the atmospheric variables that and Uncertainties
can be estimated from continental paleoclimate records. In
the following chapters, we will explain the reconstruction Paleoclimatology draws its information mainly from two
methods used; the assumptions upon which they are based; types of approaches with their advantages and limitations.
their limitations and uncertainties; and finally, we will pre- The first approach is the most basic and consists of using
sent several specific examples reflecting the diversity of simple equations to interpret a climate signal from a uni-
continental archives (lakes, vegetation, ice) whose dating variate series. This is the preferred approach for geochem-
techniques were described in Part II. We will present some istry which often uses scaling to transform an isotopic signal
10 Reconstructing the Physics and Circulation of the Atmosphere 139

into a temperature curve (in the case of polar and tropical observers. This provides series with a high resolution but
glaciers). This approach is also possible with data from over short periods. This review of climate records shows that
historical documents, which, when they do not provide no single proxy is perfect, and that without multi-proxy
direct climatic information, recount events related to a cli- comparison, errors of interpretation could easily occur.
mate variable (floods, droughts, freezing) that may be stan-
dardized semi-quantitatively (Pfister 1980).
Another approach is required for analysis of the living
Uncertainties Associated with Geochemical
world. The climate signal recorded by microorganisms is
Indicators: The Specific Case of the Isotopic
complex and is a reaction to a combination of several cli-
Composition of Precipitations
matic variables; temperature, salinity, and nutrients in the
ocean; temperature and precipitation on land. It is therefore
Over the past forty years, quantitative reconstructions of
not possible to decode this signal with a simple equation and
temperature changes have been obtained from the estimation
a multivariate approach is needed to interpret the changes in
of past changes in the isotopic composition of precipitation.
sets of pollen and diatoms in a continental environment, and
These are measured in various continental archives which
of diatoms and foraminifera in a marine setting (Chap. 21).
hold ancient precipitation directly (glaciers, ice caps,
On a smaller timescale, the thickness or density of a tree
groundwaters) or indirectly. Indeed, past changes in the
ring, also influenced by a complex environment, can seldom
isotopic composition of precipitation can leave a fingerprint
be interpreted with a simple calibration equation. Several
in the isotopic composition of molecules formed using this
series from the same region need to be used to get a clear
water such as the calcite fossil skeletons of lake microor-
climate signal.
ganisms, the calcite of speleothems, or the cellulose of tree
rings.
Different stable isotopic forms of the water molecule are
Uncertainties on the Temporal Scale
present on Earth. Their abundance is expressed by reference
to the international standard SMOW (Standard Mean Ocean
Before discussing the uncertainties and limitations associ-
Water), which has 0.038% of H217 O, 0.310% of HDO,
ated with the interpretation of series of proxies, we will
review the uncertainties related to their temporal resolution. 0.2005% of H218 O and 99.762% of the principal form, H216 O.
Figure 10.1 shows the temporal characteristics of climate The different isotopic molecules are characterized by a dif-
forcings (internal and external) and the different types of ferent number of neutrons, and therefore, different masses;
supports of commonly studied proxies. Many forcings and different vapor saturation pressures; as well as differences in
components of the climate system have characteristic times symmetry. During each change of phase (condensation,
of less than a year, while most proxies have longer charac- evaporation), the water molecules undergo isotopic frac-
teristic times. In addition, with the uncertainty of dating (see tionation, which includes processes at equilibrium (ex-
Part II), it is clear that time is an important factor of error in changes between infinite reservoirs) and kinetic processes:
the study of the interactions between climate and proxy. during evaporation on the surface of the ocean or
Among proxies providing a seasonal resolution, tree rings re-evaporation of rain drops during their precipitation, or
are prominent, but this is at the expense of the robustness of during the formation of ice crystals in the clouds, the pro-
the long term signal. Although it is possible in theory to cesses are faster than the diffusion time of water molecules
reach 10,000 years, trees have a much shorter lifespan and and cause kinetic effects associated with the molecular dif-
the long extended data series are achieved by splicing many fusivity of the different isotopic forms.
short series together. The behavior of trees in the low fre- Since 1958, the International Atomic Energy Agency
quency range is not exclusively due to climate, and this can coordinates a network of observations and a database of the
lead to significant disturbances in reconstructions. Glaciers, isotopic compositions of precipitations in modern times
another paleoclimate indicator, often have a high resolution (Fig. 10.2a). Since the 1960s, the measurements have revealed
for recent periods, due to compaction of the ice, but this a close relationship between the isotopic composition of pre-
reduces progressively as we go back in time. The same goes cipitations and air temperature (the ‘isotopic thermometer’,
for marine and lake cores. In some cases, the deposition of Fig. 10.2b). This relationship at a local level has been used
their sediments may have an annual resolution (varved intensively to quantify the changes in past temperatures.
sediments), but bioturbation (disturbance of the sediment by The use of natural archives to estimate variations in past
small aquatic animals) often prevents this level being climates from the isotopic composition of oxygen or of
achieved in practice. deuterium nevertheless poses many problems which are
Historical records are often very accurate, but they have sources of uncertainty for the quantification of climate
strong differences related to changes in instruments or reconstructions:
140 V. Masson-Delmotte and J. Guiot

Fig. 10.1 Temporal

characteristics of climate
components and proxy supports
used to reconstruct them. For
climate components (above), the
dark gray boxes represent the
temporal range of the signal;
these are extended by light gray
boxes when the component
becomes minor or occupies only
part of the time slot. For proxies,
black boxes indicate the
maximum scope possible of the
resolution, and gray boxes, the
maximum range of time covered

• uncertainty in the measurement of the isotopic content; precipitation. Each archive has specific biases described
this is generally low, with an impact of around a tenth of in the following sections, which are not necessarily
degree on estimates of temperatures; constant over time: these uncertainties can only be esti-
• uncertainty in the processes able to modify the relation- mated through proper understanding and modeling of
ship between the isotopic composition of precipitations transfer processes between the water from precipitation
and the isotopic composition of the archive (ice, calcite, and that of the archive (hydro-isotopic models of the
cellulose…). The physicochemical and biological pro- functioning of trees, and of lake systems). Recent work
cesses controlling the transfer of information between the has, for instance, shown that the isotopic composition of
precipitation and the archive must be understood and Greenland or Antarctic surface snow can evolve between
quantified in order to assess the uncertainty in the snowfall events, possibly due to the interplay of snow
reconstruction of the initial isotopic composition of the metamorphism with changes in surface water vapor
10 Reconstructing the Physics and Circulation of the Atmosphere 141

Fig. 10.2 a Global network

monitoring the isotopic
composition of precipitations,
stations with several years of
measurements (International
Atomic Energy Agency, IAEA,
www.iaea.org/water). b The
relationship between the content
of 18O in precipitations and the air
temperature at ground level,
based on network measurements
by IAEA (filled-in circles) and
from a synthesis of surface snow
measurements in Antarctica (open
circles). The slope of the
isotope-temperature relationship
(in ‰ by °C) is presented for
Antarctica, for IAEA data with
annual temperatures below 20 °C,
and IAEA data with annual
temperatures above 20 °C

-10

-20
δ O ( /oo)
o

Slope: 0.580±0.015
18

-30

-40

GNIP
-50
Antarctica
Slope: 0.795±0.015

-60
-50 -40 -30 -20 -10 0 10 20
Surface air temperature (°C)
b

isotopic composition associated with different air mass condensation temperature is the key factor governing the
trajectories. The exact effect of these uncertainties on isotopic distillation of an air mass, certain factors like the
temperature reconstructions remains difficult to quantify; evaporation conditions, changes in the source of moisture
• uncertainty concerning the stability over time of the or the vapor trajectory (including convective processes or
relationships between the isotopic abundances in precip- not), continental recycling (precipitation/evaporation
itations and weather factors: even with an almost perfect ratio), can affect the isotopic composition of the precipi-
archive of the isotopic composition of precipitations, how tations. Several critical aspects, such as the seasonality of
does this translate in terms of climate? Although the precipitations or the relationship between the
142 V. Masson-Delmotte and J. Guiot

temperature of surface air and the temperature of con- Uncertainties Associated with Biological
densation, can also affect the quality of reconstruction of Indicators
paleoclimates. For Antarctica, for example, the uncer-
tainty in the reconstruction of temperature is estimated to The commonly used methods to reconstruct climate from
be of the order of 20–30% between glacial and inter- biological assemblages are known under the term transfer
glacial periods, and recent work has produced evidence of function (Imbrie and Kipp 1971). Their principle is based on
spatio-temporal variations in this relationship over the last the expression of the relationship between the climate vari-
decades; able and the relative abundances of each taxon considered,
• Finally, it should be noted that in the tropics, it is not the as if the climate were dependent on the assemblage. This is
temperature of surface air that determines the isotopic an inverse approach, since the reality is that the assemblage
composition of rainfall. Indeed, both spatially and tem- depends on climatic conditions. The direct problem is called
porally (seasonal and inter-annual), the isotopic compo- the response function. A few equations suffice to show the
sition of rainfall is mainly related to its intensity (‘mass drawback of such an approach. Note X, the assemblage, C,
effect’), because the intensity of the isotopic distillation all climate factors combined, D, all non-climatic factors that
depends on atmospheric convective activity, irrespective may also influence X (e.g. soil), and R, the response function
of the air temperature at the surface. The interpretation of of the assemblage to C and X = R(C,D). If climate dominates
isotopic signals of tropical glaciers is therefore funda- over non-climatic factors, the relationship can be approxi-
mentally different from that of polar ice. Note that at time mated as follows: X = Rc(C). The transfer function may be
^ ¼R^ 1
c ð X Þ. But in general, the
scales greater than a decade, it is still possible that trop- obtained by inversion: C
ical sea surface temperatures have a leading role on number of variables included in the vector C is far fewer
tropical precipitation isotopic composition through their than the number of X variables, and in this case, only a least
impacts on atmospheric dynamics and convective activity squares method can solve the system of equations by mini-
(see Chap. 20). mizing the deviation between C and its estimation which
expresses C as a function of X: C ^ ¼ T^ ð X Þ where T^ denotes
Assessing the stability of the relationships between the the transfer function.
isotopes of precipitation and meteorological parameters The ‘transfer function’ (TF) approach is based on several
under different climate conditions requires the exploration of assumptions which should be kept in mind:
the processes associated with the three-dimensional atmo-
spheric circulation and water cycle. This can be achieved 1. climate conditions are the ultimate cause of any changes
using general or regional atmospheric circulation models observed in the data; human action which often modifies
implemented with the representation of the different isotopic the landscape is assumed to be negligible;
forms of the water molecule and the associated fractionation 2. the ecological properties of the studied species have not
processes. These modeling tools have been successfully used changed between the period of analysis and the present:
to explore the processes affecting the isotope-temperature the relationships between species and climate are con-
relationship at glacial-interglacial scales. Current challenges stant through time;
are related to the ability to perform long (multi-centennial or 3. current observations contain all the information neces-
longer) simulations using coupled ocean-atmosphere models sary to interpret fossil data: so, it is necessary that the
equipped with water stable isotopes to quantify the climatic vegetation of the past, for example, survived somewhere
drivers of precipitation isotopic composition in different in the world and that we have the corresponding infor-
regions and over different time scales (seasonal, inter-annual, mation. This third hypothesis, added to the second, may
decennial, centennial etc.). Recently, new understanding has be translated as the principle of uniformitarianism (the
emerged from in situ and remote sensing monitoring of present is the key to the past), which is implicit in any
water vapor isotopic composition, which provides more paleontological approach.
continuous insights than the sampling of precipitation. These
data are used to better understand the climatic drivers of It is clear that these three assumptions are quite strong.
water vapor isotopic composition, at the scale of weather The differences found between the various approaches often
events, but also to benchmark the ability of atmospheric stem from the fact that these assumptions are not always
models to correctly simulate the origin of atmospheric entirely verifiable.
moisture.
10 Reconstructing the Physics and Circulation of the Atmosphere 143

The analog method (AM) does not operate by calculating exist in the current data. The closest analog of A is Ao
a statistical relationship between climate and assemblages, whose climate is C(Ao). This shows that the AM is unable
but it is nevertheless based on the same assumptions, making to provide a climate different from that which exists in the
it subject to the same biases when these assumptions are not data.
met. However, this approach has its own peculiarities, • The fossil assemblage B also falls in an area without
because it is not based on a statistical calibration but on a current data. While the TF provides an estimate of TB
calculation of similarity. The fossil pollen spectrum (or any completely outside the realm of current data—which may
other assemblage of fossils) for which we would like to not be realistic, the AM provides the climate C(Bo) that
know the climatic conditions is compared to all current may underestimate the reality, but which has the advan-
spectra, and a measurement of each fit (‘distance’) is per- tage of being realistic.
formed (see Chap. 12). The few current spectra with the • The assemblage C has a very close analog (Co) which is
lowest distance from the fossil one are considered as the best isolated from the other points. The AM will naturally take
analogs. The reconstructed climatic conditions arise from the climate C(Co) as an estimate, but the TF will provide an
climatic conditions corresponding to these analogs, weighted estimate Tc which is very far from reality. The TF
according to the inverse of the distance from each analog to therefore follows the dominant gradient of the data and is
the fossil assemblage. unable to provide a reliable estimate for rare assemblages.
Figure 10.3 illustrates three marginal cases where the two
approaches (transfer function and analog) behave quite dif- This illustration shows that there is probably no perfect
ferently. In this figure, the horizontal axis represents the method and that the most effective way to validate the results
climate space (this space has several dimensions but here it (apart from comparing them to reconstructions from other
is simplified into one). Similarly, the vertical axis represents proxies) is to try several methods, such as advocated by
the space of the assemblages (in reality one axis per taxon). Kucera et al. (2005). About ten techniques can be identified
The gray circles represent the current data and the empty in the literature, with the two TF and AM families (Guiot
circles represent the fossil data. The line represents the and de Vernal 2007). It is therefore possible to select a few
transfer function (TF). Once we know the abundance of each of them and compare the reconstructions. Consistent results
taxon, on the horizontal axis, then the ordinate can be found are a clear indication of their robustness.
by projection along the line, and the climate conditions Another problem arises from the fact that the climate
thereby deduced (see Example A whose climate is TA). The variables to be reconstructed are often inter-correlated. If
three cases are represented by three different letters: assemblages are available from only either wet and cold
climates, or dry, hot climates, it would be impossible to
• The fossil assemblage A falls in an area without modern reconstruct wet and warm, or dry and cold climates. If it is
equivalent assemblage, but the TF allows the climate to possible to collect some assemblages which differ from the
be easily inferred by TA even though this value does not dominant gradient, the analysis of point C in Fig. 10.3
shows that the AM was then more efficient than the TF.
A climate reconstruction is calibrated on current data and
applied to fossil data. There is often a gap between the two
situations. For example, in a continental environment,
human activities act as a disruptive factor in the reference
sample, and direct application of this to past data may cause
biased reconstructions. This problem can only be minimized
by selecting current data with limited anthropogenic
influence.
Another problem inherent in any calibration is the risk of
overestimation. In principle, if the number of parameters to
be estimated (here, the weighting coefficients of each of the
taxa in the TF) is high compared to the number of reference
assemblies, it is possible to adjust a TF so that it passes
through almost all the points (in Fig. 10.3, all the points are
located on the line). Unfortunately, this line cannot provide a
reliable forecast. Following principles of statistics, a good
model is based on the lowest number possible of parameters
Fig. 10.3 Schematic representation of the main uncertainties related to
transfer functions to be estimated. An effective way to limit this problem is to
144 V. Masson-Delmotte and J. Guiot

divide the current database into two parts, to calibrate the TF References
on the first part, to apply it to the second, called the vali-
dation sample, and to deduce the average validation error. If Guiot, J., & de Vernal, A. (2007). Transfer functions: Methods for
this error remains reasonable, being inevitably greater than quantitative paleoceanography based on microfossils. In C.
the average calibration error, then the TF is considered to be Hillaire-Marcel & A. de Vernal (Eds.), Developments in marine
geology (Vol. 1, pp. 548–588). Dordrecht: Elsevier.
applicable. The same procedure is repeated after exchanging
Imbrie, J., & Kipp, N. G. (1971). A new micropaleontological method
the calibration and the validation subsets. If the two resulting for quantitative paleoclimatology: Application to a Late Pleistocene
TFs obtained are acceptable, the TF can then be calibrated Caribbean core. In K. K. Turekian (Ed.), The Late Cenozoic glacial
on the full data set and used for fossil assemblies. ages (pp. 71–181). USA: Yale University Press.
Kucera, M., Weinelt, M., Kiefer, T., Pflaumann, U., Hayes, A.,
Other statistical techniques exist (bootstrap, jackknife,
Weinelt, M., et al. (2005). Reconstruction of sea-surface temper-
permutations, etc.), but the basic paradigm is always to atures from assemblages of planktonic foraminifera:
verify the quality of the estimates on data independent from Multi-technique approach based on geographically constrained
those used for the calibration. Another important point is to calibration data sets and its application to glacial atlantic and Pacific
Oceans. Quaternary Science Reviews, 24, 951–998.
systematically provide confidence intervals associated with
Pfister, C. (1980). The climate of Switzerland in the last 450 years.
climate reconstructions. This can be achieved for both the Geographica Helvetica (numéro spécial), 15–20.
TF and the AM. These confidence intervals mean that the
tolerance of the biological assemblies for a more or less wide
range of climatic conditions can be assessed and the
imperfection of the model to adjust all the current reference
points can be taken into account.
 Air-Ice Interface: Polar Ice
11
Valérie Masson-Delmotte and Jean Jouzel

The deposition and preservation of layers of snow, year after

Melt Index and Borehole Temperatures
year, enable many climate and environmental parameters to
be recorded in the structure and composition of the ice, as
Certain glaciers and ice caps from the Arctic coastal areas
well as in its gas inclusions and impurities. This section is
form a first collection of records where the structure of the
devoted specifically to information on climate variables; the
ice allows the changes in summer temperatures to be esti-
composition of the atmosphere and biogeochemical cycles
mated. Many of these locations are characterized by a
are discussed in Chap. 1 of this volume and Chap. 1 of
summer temperature that can exceed 0 °C. In this case, the
Volume 2.
surface snow melts, percolates into the deeper colder layers
Polar ice contains records of past changes in numerous
and then refreezes. Identifying the freeze-back layers in the
climate variables, some specific to that particular site, such
physical structure of the ice allows the indices of summer
as temperature and accumulation with others relevant to a
melting intensity to be reconstructed. This method was used
larger geographical scale, such as atmospheric circulation
to estimate changes in summer temperatures over time scales
and the monsoon regimes. In a unique way, reconstructions
of a few centuries, such as in Spitsbergen (Svalbard) or in
of local variables, temperature and accumulation, are drawn
the Russian Arctic, and during the Holocene, in the Cana-
from physical processes. Estimates of the accumulation of
dian Arctic (Koerner and Fisher 2002). In Greenland and
snow can be derived from the dating of ice (see volume 1,
Western Antarctica, some low-altitude sites (Siple Dome,
Chap. 9), for example, from the identification of seasonal
Antarctica peninsula) are also characterized by the regular or
cycles or reference horizons. Variations in accumulation
occasional occurrence of summer melting. However, the vast
throughout the ages are also estimated from changes in
majority of deep core samples taken from the Antarctica and
temperature in the past, through the dependent relationship
Greenland ice are from sites where the temperature remains
between the saturation vapor pressure in the air and tem-
below 0 °C throughout the year and where there are no
perature. This chapter focuses on the various methods used
freeze-back layers in the ice. In these cases, several methods
to quantify temperature variations. We will discuss the
have been used to estimate past changes in the local
exploitation of freeze-back layers caused by summer melt in
temperature.
some polar regions, the inversion of temperature profiles
The diffusion of heat through the structure of the ice causes
measured in the boreholes, the analysis and modeling of the
fluctuations in the vertical temperature profile, which, once
stable isotope composition of the ice, and finally, the anal-
drilling operations have been completed, can be measured
ysis of the isotopic composition of nitrogen and argon in the
with an accuracy of one thousandth of a degree in the liquid in
air trapped in the ice.
the drill holes. The numerical inversion of the temperature
profile allows, in principle, the large variations of past surface
temperatures to be estimated. However, this problem is poorly
constrained and requires general assumptions to be made
about the shape of the function sought. There is a broad
uncertainty associated with the temperature estimates both in
terms of amplitude and chronology. Through this method,
V. Masson-Delmotte
J. Jouzel (&)
Laboratoire des Sciences du Climat et de l’Environnement, estimates of temperature variations over the last century, and
LSCE/IPSL, CEA-CNRS-UVSQ, Université Paris-Saclay, even over the last millennia, have been made in some sites with
91190 Gif-sur-Yvette, France high accumulation (Dahl-Jensen et al. 1999) as well as
e-mail: [email protected]

© Springer Nature Switzerland AG 2021 145

G. Ramstein et al. (eds.), Paleoclimatology, Frontiers in Earth Sciences,
https://doi.org/10.1007/978-3-030-24982-3_11
146 V. Masson-Delmotte and J. Jouzel

estimates of the glacial-interglacial amplitude in the center of allow the distillation process behind this spatial relationship
Greenland (Dahl-Jensen et al. 1998) and Antarctica (Salamatin to be understood (Fig. 11.2).
et al. 1998). The reconstruction of past temperatures is based on
measuring the isotopic ratio of a thin strip of ice taken along
the length of the cores. The isotope-temperature relationship
Stable Isotopes of Water and Temperature is then applied to this isotopic measurement. The estimate of
changes in past temperatures relies on the assumption that
The most commonly used method to reconstruct variations in the current spatial relationship is applicable to an estimate of
past temperatures at the center of Antarctica and Greenland is the difference in temperature between any two given periods
based on analyzing the isotopic composition of the ice. The at the drilling site; it assumes that this ‘temporal’ slope is
study of the abundance of the isotopic forms of water equal to the spatial slope. For changes at the
molecules in precipitations, initiated in the 1950s (Dansgaard glacial-interglacial scale, a correction linked to variations in
1953), helped to highlight a spatial relationship between the isotopic composition of the ocean (Jouzel et al. 2003)
depletion in heavy isotopes and site temperature, a relation- needs to be taken into account. In the best of cases, the
ship on which the concept of the ‘isotopic thermometer’ is accuracy of measurements by mass spectrometry is ± 0.5‰
based. Natural waters, formed mainly of H16 2 O molecules for dD and ± 0.05‰ for d18O. Temporal resolution is very
(99.7%), also present some rarer stable isotopic forms, variable. In sites where the accumulation rate is high (more
including 0.2% of H18 16
2 O and 0.03% of HD O (D represents than 10 cm per year), it is possible to find a sub-annual
2
deuterium H). The isotopic concentrations are expressed as (seasonal) resolution. However, the diffusion of water vapor
the deviation in permil in d notation (dD and d18O) against an in the upper layers of the firn quickly brings about a
international standard, the V-SMOW. At temperate and polar ‘smoothing’ of the isotopic composition and a loss of
latitudes, a linear relationship is observed between the iso- information with each snowfall. In low-accumulation sites,
topic ratios in precipitations today, dD or d18O, and the the redistribution of surface snow by the winds makes cli-
temperature of the site. Figure 11.1 illustrates the ‘isotopic mate reconstruction on a time scale of less than twenty years
thermometer’ in Antarctica, where more than 900 sites were impossible.
sampled. The spatial gradients observed are of the order of The uncertainty in the estimation of changes in past
6‰/ °C for dD and 0.8‰/ °C for d18O. temperatures is not dependent on the accuracy of the mea-
Modeling the isotopic composition of precipitations has surements but rather on the different parameters that can
been developed using conceptual distillation models (Ciais influence the isotope-temperature relationship. Through the
and Jouzel 1994) and atmospheric general circulation mod-
els incorporating the representation of the cycles of the
different isotopic forms of the water molecule (Joussaume
et al. 1984). These digital tools take into account the effect of
different fractionations related to the differences between the
saturation vapor pressure (equilibrium effect) and the diffu-
sivity in the air (kinetic effect) of the relevant molecules, and

Fig. 11.2 Modeling of the isotopic composition of precipitations

(d18O) in the Vostok region (central plateau of Eastern Antarctica)
using the ECHAM atmospheric general circulation model which
includes explicit modeling of the cycle of the stable isotopes of water.
This model was forced by boundary conditions (sunlight, ice caps,
surface sea temperatures, sea ice, and atmospheric composition)
estimated for different time intervals (current, pre-industrial, 6, 11,
14, 16, 21, 175 ky BP). The squares represent the results obtained for
Fig. 11.1 Spatial relationship found between dD of surface snow and each simulation for the Vostok region: the temporal slope between d18O
average annual temperature for 900 sites in Antarctica where both and temperature can then be estimated, and it can be seen that it is very
variables were measured (see Masson-Delmotte et al, 2008) close to the modern spatial slope
11 Air-Ice Interface: Polar Ice 147

ages, changes in the seasonality of precipitation, in the suggest that in central Greenland, the two slopes differ sig-
altitude of condensation, in the trajectory and origin of nificantly, by up to a factor of 2 (Dahl-Jensen et al. 1998;
precipitation, can all have significant effects on the rela- Masson-Delmotte et al. 2005); the conventional approach
tionship between isotopic composition and local surface underestimates variations in temperature by this amount.
temperature. Isotopic distillation models and atmospheric Currently, the oldest isotopic composition profiles go
general circulation models, equipped with an explicit rep- back to 800,000 years in Antarctica on the Dome C site
resentation of stable water isotopes allow the impact of these (Jouzel et al. 2007a) (Fig. 11.3); in Greenland, the deepest
factors to be estimated through sensitivity studies. A set of core, drilled at NorthGRIP, provides about 123,000 years of
simulations conducted using the ECHAM-iso model, archives of the isotopic composition of ice (NorthGRIP-
(Werner et al. 2001; Jouzel et al. 2007a; Werner et al. 2017) community-members 2004).
has helped to highlight the stability of the temperature- Cores from Summit, in Greenland, have revealed even
isotope relationship for the Antarctic central plateau, older but discontinuous ice segments, identified by com-
between the glacial and current climates, and to show that paring the composition of the air to reference series obtained
the temporal slope is very similar to the spatial slope. This in Antarctica. The study of the isotopic composition of ice
supports the ‘isotopic thermometer’ approach for sites on the has thus allowed reconstructions of changes in local tem-
Antarctic plateau with an accuracy of between −10 and peratures in the past to be proposed (see Masson-Delmotte
+30%, according to Jouzel et al. (2003). et al. 2006), with a high level of consistency between cores
However, this approach is debatable for climates warmer from Eastern Antarctica, Vostok, Dome C (Watanabe et al.
than the present. Sime et al. (2008) examined a scenario 2003) and Dome Fuji, where a new core now covers the last
showing an increase in atmospheric concentration of CO2 720,000 years (Dome Fuji Ice Core Project Members 2017).
with the HadAM3-iso model. This simulation suggests a The combined study of the different isotopic forms of water,
reduction in the isotope-temperature relationship in the also gave rise to a parameter of the second order, deuterium
Dome C region of Antarctica, in the context of a warmer excess defined as d = dD−8d18O (Dansgaard 1964). This
global climate. This result remains difficult to apply to parameter is strongly conditioned by the evaporation con-
measurements from ice cores, because of a lack of analogy ditions of atmospheric water vapor masses, and it has been
between climate changes caused by modifications in the used to estimate the changes in the origin of polar precipi-
Earth’s orbit (‘warm’ interglacial periods) and those caused tation over time (Jouzel et al. 2007b; Vimeux et al. 2001).
by an increase in the greenhouse effect. Changes in ‘source temperatures’ calculated in this way, are
Similarly, climate-isotope modeling and the comparison difficult to compare with the reconstructions of surface
of isotopic analysis with other paleothermometry methods temperature of the oceans (Chap. 10), because they

Fig. 11.3 Isotopic recordings of the EPICA Dome C drilling in corresponding temperature change (relative to the current temperature,
Antarctica (Jouzel et al. 2007b). Data are shown as a function of time in °C), calculated using the current spatial gradient (see Fig. 11.1), and
(x-axis, in thousands of years before the present, i.e. before the year adjusted for changes in isotopic composition of seawater, is indicated in
1950). The isotopic composition of the ice samples is indicated by the gray (on the right axis)
black line, on the left axis (dD, in ‰). The estimate of the
148 V. Masson-Delmotte and J. Jouzel

correspond to a source of precipitations which may vary resolution in order to quantify isotopic diffusion, a process
geographically over time. In addition, the isotopic compo- which depends on the temperature; analysis of oxygen-17 in
sition of the ice may be affected by changes in altitude and water to more accurately estimate the evaporation conditions
run-off, bringing to a drill site on the side of a dome, old ice (temperature, relative humidity) at the surface of the ocean;
previously formed in a distant area, effects which need to be continuous analysis of the isotopic composition of argon,
taken into consideration (Masson et al. 2000). nitrogen and noble gases in order to characterize the thermal
and gravitational fractionations of the firn. Much remains to
be learned about the spatial and temporal variability of
Stable Isotopes of Air and Temperature temperatures in Greenland and Antarctica, both over recent
centuries and in the ancient climate cycles recorded in the
New methods to quantify abrupt changes of temperature polar ice.
have been implemented, taking advantage of the thermal
fractionation of nitrogen or argon (whose isotopic compo-
sition is stable in the atmosphere over these time scales) that References
occurs when air is trapped in the firn (Severinghaus et al.
1998). This approach reveals temperature change markers in Caillon, N., Severinghaus, J. P., Jouzel, J., Barnola, J. M., Kang, J. and
the gas phase, and also allows a more detailed characteri- Lipenkov, V. Y. (2003). Timing of atmospheric CO2 and Antarctic
zation of the phase shifts between changes in atmospheric temperature changes across termination III. Science, 299, 1 728–1
731.
composition (concentrations of greenhouse gases measured Capron, E., et al. (2010). Millennial and sub-millennial scale climatic
in the gas phase) and changes in polar temperature. variations recorded in polar ice cores over the last glacial period.
This method does not allow an accurate estimate of the Climate of the Past, 6, 345–365.
temperature changes in the Antarctic, where gravitational Ciais, P., Jouzel, J. (1994). Deuterium and Oxygen 18 in Precipitation:
An isotopic model including mixed cloud processes. Journal of
fractionation effects dominate, because changes in tempera- Geophysics Research, 99, 16 793–16 803.
ture are slower and less intense. However, it was used to Dahl-Jensen, D., Morgan, V. I., & Elcheikh, A. (1999). Monte carlo
determine the phase shift between Antarctic warming and inverse modeling of the law dome (Antarctica) temperature profile.
increasing atmospheric concentrations of carbon dioxide Annals of Glaciology, 29, 145–150.
Dahl-Jensen, D., Mosegaard, K., Gundestrup, N., Clow, G. D.,
during deglaciations (Caillon et al. 2003). Johnsen, S. J., Hansen, A. W., et al. (1998). Past temperatures
In Greenland, study of the thermal fractionation of gases directly from the greenland ice sheet. Science, 282, 268–271.
was conducted on a series of rapid events, recorded in the ice Dansgaard, W. (1953). The abundance of 18O in atmospheric water and
cores from GISP2, GRIP and NorthGRIP (Capron et al. water vapour. Tellus, 5, 461–469.
Dansgaard, W. (1964). Stable isotopes in precipitation. Tellus, 16, 436–
2010). Estimates of temperature changes through thermal 468.
fractionation of the air have led to an upward revision of the Dome Fuji Ice Core Project Members. (2017). State dependence of
intensity of temperature changes during the warm climatic instability over the past 720,000 years from Antarctic ice
Dansgaard-Oeschger events, reaching an amplitude of 8 to cores and and climate modeling. Science Advances, 3, e1600446.
Joussaume, S., Jouzel, J., & Sadourny, R. (1984). A general circulation
16 ± 3 °C, and have called into question the presumed model of water isotope cycles in the atmosphere. Nature, 311,
stability of the cold phases as they appear in the continuous 24–29.
recordings of the stable isotopes of water. The apparent Jouzel, J., Vimeux, F., Caillon, N., Delaygue, G., Hoffmann, G.,
discrepancy between the quantifications from the current Masson-Delmotte, V., & Parrenin, F. (2003). Magnitude of the
isotope-temperature scaling for interpretation of central antarctic ice
spatial relationship between water isotopes and temperature cores. Journal of Geophysical Research, 108, 1029–1046.
and this alternative paleothermometry method can be Jouzel, J., Masson-Delmotte, V., Cattani, O., Dreyfus, G., Falourd, S.,
explained by significant changes in the seasonality of the Hoffmann, G., et al. (2007a). Orbital and millennial Antarctic
deposition of snow in Greenland over the ages, a process climate variability over the past 800,000 years. Science, 317, 793–
797. https://doi.org/10.1126/science.1141038.
highlighted by the climate-isotope models between the Last
Jouzel, J., Stiévenard, M., Johnsen, S. J., Landais, A.,
Glacial Maximum and present times (Krinner 1997; Werner Masson-Delmotte, V., Sveinbjornsdottir, A., et al. (2007b).
et al. 2001). The GRIP deuterium-excess record. Quaternary Science Reviews,
26, 1–17.
Koerner, R. M., & Fisher, D. A. (2002). Ice-Core evidence for
widespread arctic glacier retreat in the last interglacial and the early
Conclusions Holocene. Annals of Glaciology, 35, 19–24.
Krinner, G., C. Genthon, and J. Jouzel (1997). GCM analysis of local
It is important to note that a number of methods may be influences on ice core d signals. Geophysical Research Letters, 24,
2 825–2 828.
applied to polar ice cores to quantify temperature changes.
Masson, V., Vimeux, F., Jouzel, J., Morgan, V., Delmotte, M.,
New avenues are being explored to improve quantifications Hammer, C., et al. (2000). Holocene Variability in Antarctica based
of temperature change: isotopic measurements at very high on 11 ice core isotopic records. Quaternary Research, 54, 348–358.
11 Air-Ice Interface: Polar Ice 149

Masson-Delmotte, V., Jouzel, J., Landais, A., Stievenard, M., Johnsen, deep boreholes at Vostok station (East Antarctica). Journal of
S. J., White, A., et al. (2005). Deuterium excess reveals millennial Geophysical Research, 103, 8 963–8 977.
and orbital scale fluctuations of greenland moisture origin. Science, Severinghaus, J., Sowers, T., Brook, E. J., Alley, R. B., & Bender, M.
309, 118–121. (1998). Timing of abrupt climate change at the end of the younger
Masson-Delmotte, V., Braconnot, P., Dreyfus, G., Johnsen, S., Jouzel, dryas interval from thermally fractionated gases in polar ice. Nature,
J., Kageyama, M., et al. (2006). Past temperature reconstructions 391, 141–146.
from deep ice cores: relevance for future climate change. Climate of Sime, L. C., Tindall, J. C., Wolff, E. W., Connolly, W. M., & Valdes,
the Past, 2(2), 145–165. P. J. (2008). Antarctic isotopic thermometer during a CO2 forced
Masson-Delmotte, V., Hou, S., Ekaykin, A., Jouzel, J., Aristarain, A., warming event. Journal Geophysical Research, 113, D24119.
Bernardo, R. T., Bromwhich, D., Cattani, O., Delmotte, M., Vimeux, F., Masson, V., Delaygue, G., Jouzel, J., Petit, J.-R.,
Falourd, S., Frezzotti, M., Gallée, H., Genoni, L., Isaksson, E., &Stievenard, M. (2001). A 420,000 year deuterium excess record
Landais, A., Helsen, M., Hoffmann, G., Lopez, J., Morgan, V., from East Antarctica: Information on Past Changes in the Origin of
Motoyama, H., Noone, D., Oerter, H., Petit, J. R., Royer, A., Precipitation at Vostok. Journal of Geophysical Research, 106, 31
Uemura, R., Schmidt, G. A., Schlosser, E., Simões, J. C., Steig, E., 863–31 873.
Stenni, B., Stievenard, M., van den Broeke, M., van de Wal, R., van Watanabe, O., Jouzel, J., Johnsen, S., Parrenin, F., Shoji, H., &
den Berg, W. J., Vimeux and F., White, J. W. C. (2008). A review Yoshida, N. (2003). Homogeneous climate variability across East
of antarctic surface snow isotopic composition: observations, Antarctica over the past three glacial cycles. Nature, 422, 509–512.
atmospheric circulation and isotopic modelling. Journal of Climate, Werner, M., Heimann, M., & Hoffmann, G. (2001). Isotopic compo-
21, 13, 3 359–3 387. sition and origin of polar precipitation in present and glacial climate
NorthGRIP-community-members. (2004). High resolution climate simulations. Tellus, 53B, 53–71.
record of the northern hemisphere reaching into last interglacial Werner, M., Jouzel, J., Masson-Delmotte, V., & Lohmann, G. (2018).
period. Nature, 431, 147–151. Reconciling glacial Antarctic water stable isotopes with ice sheet
Salamatin, A. N., Lipenkov, V. Y., Barkov, N. I., Jouzel, J., Petit, J. R. topography and the isotopic paleothermometer. Nature Communi-
& Raynaud, D. (1998). Ice-core age dating and paleothermometer cations, 9, 3537. https://doi.org/10.1038/s41467-018-05430-y
calibration on the basis of isotope and temperature profiles from
 Air-Vegetation Interface: Pollen
12
Joël Guiot

morphologies depending on the plant type. The palynologist

From the Production of Pollen to Sediment
counts each pollen type to work out the relative abundance
of the species of trees or grasses that produced the grains. It
In order to persist in difficult conditions, most plants and
is not always possible to determine each species, and many
terrestrial animals contain hard parts that are preserved in the
plants can only be recognized at the level of the genus or
sediments after death. For higher-level terrestrial plants,
even family. Because of this heterogeneity in the classifi-
these parts consist mainly of pollen grains and spores which
cation, the term ‘pollen taxon’ is used to characterize the
provide a widely-used tool in paleoclimatology, thanks to
type of plant that produced it. The total number of grains
their abundance in wet sediments. These grains, which are
counted varies depending on the diversity of the vegetation:
dispersed over variable distances depending on their shape
tropical vegetation is more diverse than temperate vegetation
and size (from 5 to 100 microns), are an essential factor in
and therefore requires a much higher total number, some-
the reproduction of higher-level plants. Their outer envelope
times more than a thousand grains compared with a few
(exine) is composed of sporopollenin, a highly resistant
hundred in a temperate vegetation, and this ensures good
substance once it is protected from oxidation. Lakes and
statistical significance of the various fluctuations detected.
bogs are particularly good environments for the conservation
The relative abundance of each taxon makes up a pollen
of these plant remains. These grains are scattered by wind,
assembly or spectrum. It provides information on the relative
insects, birds, water. In temperate regions, wind is the pre-
composition of the surrounding vegetation, but this signal is
dominant vector, and because of its relative inefficiency, lots
influenced by the abundance of pollen productivity, its mode
of pollen grains and spores are found not far from their
and capacity of dispersion. Statistical methods are needed to
source, in continental sediments and marine sediments near
reliably decode the pollen spectrum (Moore et al. 1991).
the coast. In rainforests, animals play a much more important
role, so that many species are underrepresented in sediments.
After choosing a site representative of the surrounding
vegetation, with good conservation of pollen and sufficient
The Pollen Diagram
accumulation rates to permit studies over the desired time
The set of pollen spectra along the core is presented
scale, cores are extracted, usually at the center of the lake or
graphically to provide a pollen diagram whose complexity
bog. These cores are studied stratigraphically and samples
depends on the number of taxa counted. It can be simplified
are dated in order to establish an absolute chronology of
either by grouping together similar taxa (i.e. taxa
climate events. Samples are taken at regular intervals along
co-evolving in similar environments), or by representing
the core. They are subjected to physical and chemical
only the most important ones. Figure 12.1 shows an example
treatments to make the pollen grains clearly visible for
of a simplified diagram. The interpretation of the diagram is
examination under a microscope. The grains are then iden-
complex because of the large number of processes occurring
tified on the basis of their exine which have different
between the pollen production by the vegetation and its
record in the sediment. As in most paleontological disci-
J. Guiot (&) plines, interpretation is done by comparison with modern
European Centre for Research and Teaching in Environmental
Geosciences CEREGE, Aix-Marseille University, CNRS, IRD,
data. This supposes that the principle of uniformity, that is,
INRAE, Collège de France, BP 80, 13545, Aix-en-Provence, where the present is the key to the past, applies. This is
Cedex 04, France generally the case for data from the Quaternary, and in
e-mail: [email protected]

© Springer Nature Switzerland AG 2021 151

G. Ramstein et al. (eds.), Paleoclimatology, Frontiers in Earth Sciences,
https://doi.org/10.1007/978-3-030-24982-3_12
152 J. Guiot

particular for continuous sequences that have been correlated the problem can be treated quantitatively and objective keys
over 500 000 years (Tzedakis et al. 1997). To verify this to interpretation can be established. We develop this per-
assumption, pollen samples are taken from modern or recent spective below.
moss and their pollen spectra are compared with current Figure 12.1 shows that the vegetation around Lake Rot-
vegetation. From a qualitative point of view, we can already see, Switzerland (Lotter and Zbinden 1989) was dominated
sketch the relationship between the abundances of pollen by herbaceous vegetation before 11,000 years BP: sagebrush
and the composition of the vegetation. For example, the (Artemisia) and grasses (Poaceae), indicating an open
study of an altitudinal transect provides an impression of environment with few trees. Around 11,500 years BP, birch
what happened during a cooling climate. Many published (Betula) and, a century later, pine (Pinus) flourished until
articles are based on this approach. But if current pollen they occupied nearly 80% of the assemblage. This pioneer
spectra are collected from vegetation as diverse as possible, arboreal vegetation thrived due to a warmer climate. Around

Fig. 12.1 Simplified pollen

diagram of Lake Rotsee (8.33° E,
47.08° N, 419 m) in the Swiss
Alps (Lotter and Zbinden 1989).
The time period covered is from
12 000 years before now (BP) in
calendar age to the present day.
AP is the sum of all the arboreal
pollen abundances
12 Air-Vegetation Interface: Pollen 153

11,100 years BP, strong warming encouraged the prolifera- continents (Jolly 1998; Tarasov 1998). The types of plants
tion of the hazel (Corylus), the oak (Quercus) three centuries defined in this way, known as plant functional types (PFTs),
later, and the lime tree (Tilia) five centuries later again, can be directly compared to simulations by vegetation models
replacing these pioneers. This succession over less than a based on the same typology. These PFTs are used to define the
millennium is not necessarily exclusively due to climate: vegetation of a site in the form of a biome: a bio-geographic
species expand geographically from refuges more or less area characterized by the species of plants (and animals) that
distant from the studied area at a speed that is specific to live there. In Fig. 12.2 four such PFTs have been reproduced.
them. Another, less dramatic change in climate (probably a The biome is determined by comparing them. Before
wetter and cooler climate) around 8200 years BP enabled the 11,500 years BP, herbaceous plants dominated, as is currently
fir (Abies) and alder (Alnus) to become established. Around the case in the arctic tundra and alpine grasslands. This period
6300 years BP, it was the beech (Fagus) which progressed is called the Younger Dryas. Then, the presence of boreal
in the region. Around 2000 years BP, during the Roman deciduous trees, followed by conifers, indicated a warming
period, the proportion of grasses increased and trees sufficient for the taïga, as the forest in Northern Europe is
decreased (AP), a sign of anthropogenic deforestation. called today, to develop. From 11,100 years BP, at the
Comparison of the different graphs in Fig. 12.1 shows that beginning of the Holocene, the temperate forest became
some taxa evolve together while others are diachronous. Taxa established. The arrival of conifers around 8200 years BP,
that are present at the same time on the same site will likely probably due to a well-known abrupt cooling (Tinner and
thrive in the same climatic conditions: they will be temperate Lotter 2006) transformed the landscape from a deciduous
or boreal, they will be resistant to drought or only survive in forest to a mixed forest which lasted until about 5000 years
wet conditions. If, to this, we add characteristics related to the BP. The consequent growth of deciduous trees was disturbed
size of the plant (tree, shrub, grass), phenology (evergreen around 4000 years BP and even more so around 2000 years
plant or deciduous), type of leaf (needles or broad leaves), a BP, as a result of widespread deforestation by man. The
reasonable classification can be made. Prentice et al. (1996) reconstructed biome then became the steppe, although it was
have proposed one for Europe. This was then applied to other not exactly a proper one, being a mix of grasslands and forests.

Fig. 12.2 Evolution of four Vegetation and climate at Rotsee

functional plant types at Lake
25

Rotsee. Continuous line gray:

0

herbaceous steppe and tundra,

20

temperate deciduous
dashed dotted line: boreal
−10

deciduous trees; dashed line:

scores

15

steppes
Tjan

conifers; dotted line: temperate

deciduous trees. On the same
conifers
−20

figure, the biomes that can be

10

inferred from this are represented:

steppe, temperate forest, mixed
5

−30

forest, taiga, tundra. The other boreal deciduous

three panels represent the three
climate parameters reconstructed 0 2000 6000 10000 0 2000 6000 10000
with their error bars: Pann (annual
precipitation in mm/year), Tjan Time (years BP) Time (years BP)
(January temperatures in °C),
15

Tann (annual average temperature

1200

in °C). The error bar is given by

10

the total variability between the

analogs
Pann

5
Tann
800

0
400

−10

0 2000 6000 10000 0 2000 6000 10000

Time (years BP) Time (years BP)
154 J. Guiot

Reconstruction of Climate respond to climate change in the same way when atmo-
spheric CO2 levels change. This problem was solved by the
These interpretations are essentially qualitative. Transfer use of mechanistic models of vegetation (Guiot et al. 2000).
functions and the analog method (see Chap. 10, Sect. 10.1.3) Another avenue takes pollen dispersal and associated biases
provide quantitative information on climate. Here, we use into account when reconstructing the landscape (Sugita
the analog method, not on the percentages of taxa, but on the 2007; Trondman et al. 2015). The relative abundance of
PFT scores (Fig. 12.2.), which has many advantages: it pollen of a taxon depends on the height and distance of the
reduces the number of variables and groups together the taxa productive plant, its pollen productivity, the weight and
with similar behavior, making the approach more robust. shape of the pollen grain, the dimensions of the lake into
Twelve analogs for each fossil spectrum were retained. which it falls.
Figure 12.2 shows these reconstructions with the shaded To improve climate reconstructions without resorting to
area representing the range of variability between the ana- overly complex models, the ‘multiproxy’ approach is rec-
logs. The quality of the reconstructions is controlled by ommended. This involves taking several simultaneous cli-
applying the same method to modern data: for each modern mate indicators into account: pollen, macro remains of plants
spectrum, the best analogs are determined (obviously or animals, carbon isotopes, lake levels, sedimentological
excluding the spectrum itself) and the present climate is parameters etc. as was done by Cheddadi et al. (1997). This
reconstructed so that it can be compared to direct observa- multiproxy approach is also supported by mechanistic
tions. In this case, we find a coefficient of determination (r2) models (Rousseau et al. 2006. Guiot et al. 2009). These
of 0.64 for the annual precipitation (Pann), 0.89 for tem- avenues of research are being explored both to reconstruct
peratures in January (Tjan) and 0.93 for annual average the climate of the past and to understand how it has influ-
temperatures (Tann). The estimates obtained for precipita- enced the vegetation.
tion will therefore contain larger errors than the other vari-
ables. This is confirmed by Fig. 12.2. In terms of climate, we
see that the Younger Dryas was very cold (14 ± 7 °C colder References
than now) and dry (400 ± 400 mm/year less rainfall than
now over the annual average), but that the uncertainties are Birks, H. J. B. (2011). Strengths and weaknesses of quantitative climate
large (in this method, this is the variability between analogs reconstructions based on late-quaternary biological proxies. The
Open Ecology Journal, 3, 68–110. https://doi.org/10.2174/
rather than an actual error bar). The temperature maximum
1874213001003020068.
(4 ± 2 °C more than now) occurred at around 10,000 years Brewer, S., Guiot, J., Barboni, D. (2007). Pollen Data as Climate
BP, when oak dominated, and the rainfall maximum Proxies. In Encyclopedia of Quaternary Sciences, Elsevier.
(100 ± 250 mm more) occurred around 8200 years BP, Cheddadi, R., Yu, G., Guiot, J., Harrison, S. P., & Prentice, I. C.
(1997). The Climate of Europe 6000 years ago. Climate Dynamics,
with the arrival of the mixed forest. However, the 8200 BP
13, 1–9.
event does not appear to have been short-lived since this Guiot, J., Torre, F., Jolly, D., Peyron, O., Boreux, J. J., & Cheddadi, R.
forest continued for several millennia longer. (2000). Inverse vegetation modeling by monte carlo sampling to
Several other methods have been proposed to reconstruct reconstruct palaeoclimates under changed precipitation seasonality
climate from pollen data (see Brewer et al. 2007; Birks 2011; and CO2 conditions: Application to glacial climate in mediterranean
region. Ecological Modelling, 127, 119–140.
Juggins 2013). They all have their strengths and weaknesses. Guiot, J., Wu, H., Garreta, V., Hatti, L., & Magny, M. (2009). A few
It is not always easy to find the optimal method. It is rec- prospective ideas on climate reconstruction from a statistical single
ommended to try several and compare the results. The proxy approach towards a multi-proxy and dynamical approach.
convergence of estimates is an indication of the robustness Climate of the Past, 5, 571–583.
Jolly, D. et al. (1998). Biome reconstruction from pollen and Plant
of the reconstruction, and their divergence is often a sign that Macrofossil data for Africa and the Arabian Peninsula at 0 and 6 ka.
the initial assumptions used for the reconstruction were not Journal of Biogeography, 25, 1 007–1 028.
entirely valid. In particular, when the climate changes Juggins, S. (2013). Quantitative reconstructions in palaeolimnology:
rapidly, vegetation adapts with a certain delay, which makes New paradigm or sick science? Quaternary Science Review, 64.
https://doi.org/10.1016/j.quascirev.2012.12.014.
it difficult to find current analogs. Another problem is esti- Lotter, A. F., & Zbinden, H. (1989). ‘Late-glacial pollen analysis,
mating the impact of non-climate constraints. For example, Oxygen-Isotope and Radiocarbon stratigraphy from Rotsee
atmospheric CO2 concentration during the Quaternary (Lucerne), Central Swiss Plateau. Ecologae Gologicae Helvetiae,
glaciations was lower than the current concentration (about 82, 191–202.
Moore, P. D., Webb, J. A., & Collinson, M. E. (1991). Pollen Analysis.
200 ppm instead of more than 280 ppm). We know that this Oxford: Blackwell Scientific Publications.
level is an activator of photosynthesis and so it is unlikely Prentice, I. C., Guiot, J., Huntley, B., Jolly, D., & Cheddadi, R. (1996).
that the principle of uniformity applies: vegetation will not Reconstructing biomes from Palaeoecological data: A general
12 Air-Vegetation Interface: Pollen 155

method and its application to European pollen data at 0 and 6 ka. Tinner, W., & Lotter, A. (2006). Holocene expansions of Fagus
Climate Dynamics, 12, 185–194. silvatica and Abies alba in Central Europe: where are we after eight
Rousseau, D., Hatté, C., Guiot, J., Duzer, D., Schevin, P. and Kukla, G. decades of debate? Quaternary Science Reviews, 25(5–6), 526–549.
(2006). Reconstruction of the Grande Pile Eemian Using Inverse https://doi.org/10.1016/j.quascirev.2005.03.017.
Modeling of Biomes and d13C. Quaternary Science Reviews, 25, 2 Trondman, A. K., Gaillard, M. J., Mazier, F., Sugita, S., Fyfe, R.,
806–2 819. Nielsen, A. B., et al. (2015). Pollen-based quantitative reconstruc-
Sugita, S. (2007). Theory of quantitative reconstruction of vegetation I: tions of Holocene regional vegetation cover (plant-functional types
Pollen from Large Sites REVEALS regional vegetation composi- and land-cover types) in Europe suitable for climate modelling.
tion. The Holocene, 17, 229–241. Glob. Chang. Biol. 21. https://doi.org/10.1111/gcb.12737.
Tarasov, P. E. et al. (1998). Present-Day and Mid-Holocene Biomes Tzedakis, P. C., Andrieu, V., de Beaulieu, J.-L., Crowhurst, S., Follieri,
reconstructed from pollen and plant macrofossil data from the M., Hooghiemstra, H., et al. (1997). Comparison of terrestrial and
Former Soviet Union and Mongolia. Journal of Biogeography, marine record of changing climate of the last 500,000 years. Earth
25, 1 029–1 054. and Planetary Science Letters, 50, 171–176.
 Ground-Air Interface: The Loess Sequences,
Markers of Atmospheric Circulation 13
Denis-Didier Rousseau and Christine Hatté

Abstract moraines of the gigantic ice sheets that developed on these

continents during the ice ages. In Asia, however, the situa-
Atmospheric circulation is responsible for the rapid
tion is different, since the Loess Plateau is located south of
distribution of heat and moisture across the Earth and
the deserts of northern China and southern Mongolia. The
hence determines our weather and regional climate, today
loess sediments thus cover vast regions of the northern
and in the past. During past climate cycles, the atmo-
hemisphere, at latitudes where very few other records of the
sphere has been much more dustier, except during the
glacial paleoclimates are available, while in more southern
interglacials, inducing uncertainties about the impact of
latitudes, lakes, peat bogs, speleothems and marine sedi-
mineral aerosols on the past climate dynamics. There are
ments collected close to the coast provide continental
abundant traces of the combination of past atmospheric
records of a high quality. In Southern Hemisphere, loess is
dynamics and paleodust cycle such as eolian mineral
mostly associated with fluvial and piedmont deposits in the
material transported and deposited in terrestrial archives
Pampean plains in South America, and fluvial abrasion and
as loess. Records from loess deposits consistently suggest
glacial grinding in New Zealand.
that atmospheric dynamics was highly variable, during
Loess is a fine sediment transported mainly by wind to
the past climate cycles, much more than presently where
various altitudes depending on the particle grain size, but
the only sources of dust are the major deserts. In this
also on the state of the substratum and on the environmental
chapter we explained the four main categories of
conditions in the region of origin of the material (Fig. 13.2).
parameters allowing to identify loess deposits as reliable
Dust emission is favored by a low rate of vegetation
markers for past air circulation.
development and extension of river-emerged banks, source
of fine mineral dust. Loess deposits are thus specific to
glacial times with low sea-levels thus high river incision and
weak vegetation development.
Overview of Loess
To simplify, coarse material is carried over distances of
less than hundred kilometers and transport is mainly by
Loess is an aeolian sediment which is relatively common in
saltation, that is to say, by a series of jumps. However, fine
continental areas (Fig. 13.1). In Europe and North America,
material (clay), which may be of local or remote origins,
it is dispersed near former polar deserts or near the frontal
may equally have been transported over hundreds of kilo-
meters to altitudes of several kilometers in the form of
mineral dust and may have been redeposited through wet or
D.-D. Rousseau (&) dry deposition (depending on whether or not they are
École Normale Supérieure de Paris, Laboratoire de Météorologie associated with precipitations). Loess is therefore usually a
Dynamique, UMR CNRS 8539, Université Paris Sciences et marker for air circulation in the past.
Lettres, 24 Rue Lhomond, 75231 Paris, France
e-mail: [email protected] To illustrate our words and the research potential of loess
sequences, most of our examples will be drawn from the
Lamont-Doherty Earth Observatory of Columbia University,
Palisades, NY 10964, USA extensively studied Nussloch loess sequence (Rhine Valley,
Germany).
C. Hatté
Laboratoire des Sciences du Climat et de l’Environnement,
LSCE/IPSL, Université Paris-Saclay, 91190 Gif-Sur-Yvette,
France

© Springer Nature Switzerland AG 2021 157

G. Ramstein et al. (eds.), Paleoclimatology, Frontiers in Earth Sciences,
https://doi.org/10.1007/978-3-030-24982-3_13
158 D.-D. Rousseau and C. Hatté

Fig. 13.1 Global distribution of loess sediments and equivalents. 1: loess sediments; 2: loess derivatives (according to Pécsi (1990))

or a tundra gley (hydromorphic soil) are indicative of more

Paleoclimate Indicators temperate interstadial intervals (Antoine (2009)).
Since they present detailed records of past climate,
Different paleoenvironmental indicators can be identified for reproducing the same pedostratigraphic and sedimentologi-
this type of material, deposited at medium and high latitudes cal units with the same succession of these units, the
with climate conditions unfavorable for biological develop- pedosedimentary record of loess sequences can be used as a
ment (glacial periods). These parameters allow their inter- powerful tool for chronostratigraphical correlation (Antoine,
pretation in term of atmospheric circulation. Four major et al, (1999); Rousseau et al. (2007a, b)).
categories are presented below.
The grain size index

Sedimentological Indicators The study of the grain size, calculated through determination
of the dominant classes, their mode distribution and rela-
The conceptual view tionships between them, can characterize the relative
strengths of the winds that created the deposits (Rousseau
Even the observation of a loess sequence is instructive. et al. (2007a, b)). These wind dynamics are linked to
Indeed, when a record is of one or more climate cycles, this changes in the general atmospheric circulation (Fig. 13.3).
is never made up of a single stratigraphic unit. On a finer scale, Nussloch shows a progressive coarsening
While loess sediment is characteristic of cold and quite of the loess deposits between ca 30 and 22 ka (Rousseau
arid periods, other warmer periods are marked by paleosols. et al. (2007b)). This coarsening trend ends with a short but
Thus, if at low resolution, the loess-paleosol sequence cor- major decrease in grain size, followed by an increase to a
responds to one or more climatic cycles; at higher resolution, new maximum at 20 ± 2 ka (‘‘W’’ shape). Correlation
the identification of more precise events is achieved by between the loess grain-size index and the Greenland
observing the sequence of soils or a particular hierarchy of ice-core dust records suggests a global connection between
pedo-sedimentary units. (Kukla and An (1989)). In Europe, a North Atlantic and Western European global atmospheric
leached brown paleosol (Bt), at the base of a brown soil, circulation and wind regimes (Antoine (2009); Rousseau
indicates an interglacial level, while a humus-rich forest soil et al. (2007b)).
13 Ground-Air Interface: The Loess Sequences, Markers … 159

Fig. 13.2 Representation of the different types of dust transportation during the Quaternary (from Pye and Zhou (1989) modified).
contributing to the formation of loess deposits. A—Two modes of B—Diagram showing the various modes of transport of wind sediment
transport and deposition of wind-blown dust from the northwestern particularly to the lower levels of the atmosphere (from Pye (1995)
deserts of China to the Loess Plateau and the North Pacific Ocean modified)

The loess deposit thickness cold climate states (Greenland stadial and Heinrich event)
generally become more than twice as high as those for the
The thickness of the units is another characteristic which relatively warmer Greenland interstadial, in agreement with
also allows the link with wind dynamics to be established. the observed loess data (Sima et al. (2009)) (Fig. 13.4).
Indeed, monitoring a particular unit in a given territory
allows the characterization of gradients which will be ori- The mineralogy
ented according to the prevailing winds, the thickest part
being upwind (Rousseau et al. (2007a)) (Figure 13.4). The mineralogy, in particular its composition of heavy
Furthermore, the cyclic variation of the sedimentation minerals, also helps to trace the origin of certain deposits and
rate has been shown to be a potential response to the to deduce the prevailing winds responsible for their transport
North-Atlantic rapid climate changes, i.e. the Greenland (Lautridou (1985)).
stadial/interstadial cycles and the Heinrich events. This
hypothesis has been tested by modeling the impact of
North-Atlantic climate variations on dust emissions. This Biological Indicators
study clearly highlighted that, besides wind, precipitation,
soil moisture and snow cover showing some differences in These are relatively diverse. Although the remains of
the dust emission intensity, vegetation cover is the main micro-mammals: bones, teeth or skulls, or of larger mam-
impacting parameter (Sima et al. (2009)). Dust fluxes for the mals are identified quite sporadically, other fossils are more
160 D.-D. Rousseau and C. Hatté

Fig. 13.3 Correlation of variations in particle size index (IGR) as Nussloch, corresponding to a brown boreal soil called Lohne Boden
defined at Nussloch with variations in calcium concentration (repre- (LB) or to well-developed tundra gleys (G1, G2a, G2b, G3, G4 and
senting transported dust) from the GRIP survey. The T1 threshold G7). The T2 threshold defines intervals with very high dust concen-
applied to Greenland data highlights the main intervals with low dust tration in Greenland, corresponding to the DO stadials, as well as some
concentration, which correspond to Dansgaard-Oeschger (DOI) inter- less significant peaks in dust (according to Rousseau et al. (2007b))
stadials 8 to 2. These are correlated to intervals where the IGR is low in

frequent. Among these, terrestrial mollusks form populations developed (Rousseau (1987)). Two examples follow: the
typical of diverse environments (Wu (2001); Rousseau first, in China, illustrates the long-term variations that can be
(1987)). correlated to terrestrial orbital frequencies and the second
shows the short-term variation relating to the internal vari-
Mollusk assemblages ability of the climate.

Usually identified at the species level, these organisms have The long-term variability
the advantage of persisting to the present day. In accordance
with the principle of actualism, it is possible to apply modern Using the ecological requirements of taxa identified in the
requirements and ecological characteristics to individual Chinese loess, it was possible to define environmental groups
fossils. Multivariate statistical analyses are used to recon- that have proven to be reliable indicators of the summer and
struct the environment in which a fossil community winter monsoons through the ages (Fig. 13.5). Transfer
13 Ground-Air Interface: The Loess Sequences, Markers … 161

Fig. 13.4 Map of European loess deposits indicating the maximum oceanic influences: GS, HE-GS and GI-GS with GS = Greenland
expansion of the Fennoscandian and British ice sheets during the last stadial, HE = Heinrich stadial, GI—Greenland interstadial (according
glacial maximum (map by Antoine in [10] modified), with wind speed to Sima et al. (2009) modified)
reconstructed for three climate conditions corresponding to northern

functions from terrestrial mollusks and using the principle of also been studied from their signature left in amino acids.
modern analogs, were developed to reconstruct the seasonal This approach helps to distinguish important differences
temperatures from the European loess sequences over three between one climate cycle and another (Oches and McCoy
climate cycles (Rousseau (1991); Moine et al. (2002)). (1995)).
Another paleoclimate index also studied in loess sedi-
The short-term variability ments are phytoliths, siliceous concretions present in
superficial plant tissue. In some cases, they are well pre-
The high-resolution study (1 sample every 10 cm) of served and identified, particularly in the Chinese loess
malacological assemblages from the Nussloch loess sequences, where they helped in the reconstruction of the
sequence allows vegetation change to be described along the temperature and precipitation of the last climatic cycle (Lu
70 to 34 kyr cal BP period, recorded in 6 meters of sediment et al. (2007)). Pollen, in contrast, are poorly preserved in
(Moine et al. (2005)). The mollusk changes reflect three loess sequences and are the subject of very few studies
short phases of vegetation development and climatic (Gerasimenko and Rousseau (2008); Rousseau et al.
improvement related to soils of the interstadials. A steppe to (2001)).
herb/shrub tundra shift characterizes the Lower-Middle Recently, earthworm granules have been established as a
Pleniglacial transition and is followed by a decline in veg- new biological proxy (Prud’homme et al. (2015)). This
etation and humidity increase ending with a new increase in proxy is not based on species recognition but is based on
temperature and vegetation cover (Moine et al. (2005)). abundance. Counts of earthworm granules reveal a link
Many other methods have been developed using between their abundance and the nature of the stratigraphic
micro-mammals or beetles, with the current distribution of units and their associated climate conditions. They are in
species and the average ranges of associated climate acting very high abundances in tundra gley and boreal brown soil
as a reference (Liu et al. (1985)). Terrestrial mollusks have horizons, i.e. during Greenland interstadial intervals and are
162 D.-D. Rousseau and C. Hatté

Fig. 13.5 Variations in the abundance of mollusks, characteristic of insolation at 30° N. A—Variations in the abundance of Vallonia tenera
ancient summer and winter monsoons over the last three climate cycles and Pupilla aeoli, indicators of ancient winter monsoons. B—
in two sequences of the Chinese Loess Plateau (Changwu and Variations in the abundance of Punctum orphana, indicator of ancient
Luochuan) and their relationship to astronomical parameters and summer monsoons (according to Wu (2001))

almost absent in typical calcareous loess, associated with paleosols caused the magnetic grains to become concen-
Greenland and Heinrich stadials (Prud’homme et al. (2015)). trated. This led Kukla to propose a chronological model,
independent of any astronomical calibration, while assuming
continuous and complete sequences. However, the discovery
Geophysical Indicators in the paleosols of bacteria secreting magnetic grains called
this debatable assumption into question (Zhou et al. (1990)).
This is mainly the magnetic sensitivity in low fields which On the other hand, research on modern Chinese soils along
allows, firstly, the different lithological units present in the gradients, reflecting the impact of the summer monsoon, has
same sequence to be characterized, and secondly, to identify enabled the calibration of the sensitivity signal and the
the source of the matter through the magnetic particle size establishment of a transfer function for the region to
(Lagroix and Banerjee (2002)). In general, typical loess units reconstruct annual paleoprecipitation associated with the
have relatively low field magnetic susceptibility, unlike variations in the East Asian monsoon (Maher and Thompson
paleosols, which are characterized by significantly higher (1995)). A new method of characterizing the origin of loess
values. Widely used for the Chinese sequences, is to work directly taking a quartz grain and to study both the
low-frequency magnetic sensitivity was interpreted by Kukla crystallinity index and the intensity of the spin resonance
et al. (1990) as corresponding to a relatively constant supply signal. This new technique makes it possible to differentiate
of fine matter through the ages. According to this theory, the between the origins of the grains and thus to follow the
formation of soil in the various Chinese interglacial variations in the source of the transported material. Applied
13 Ground-Air Interface: The Loess Sequences, Markers … 163

Fig. 13.6 Variation versus depth of rock magnetic parameters allowing the waterlogging in different tundra gleys from the Nussloch loess
sequence to be characterized. From Taylor et al. (2014)

for the first time in the Chinese loess sequences, this method Geochemical Indicators
has helped to clarify the origin of the quartz grains trans-
ported to the Loess Plateau (the Gobi desert during stadial The fourth category of indicators concerns geochemical
periods, deserts of northern China during the warmer peri- indices and, in particular, isotopic parameters which con-
ods) (Sun (2008)). More recently, the study of the mineral stitute a powerful tool for interesting research perspectives.
magnetic composition completed the classical interpretation
of the magnetic susceptibility record by allowing waterlog- Tracing paleodust sources
ging processes in tundra gley supporting the correlation of
these paleosols with Greenland interstadials to be charac- As representatives of the age of the geological formation, the
terized (Taylor et al. (2014)) (Fig. 13.6). dust particles derived from Sr and Pb isotopes are commonly
164 D.-D. Rousseau and C. Hatté

Fig. 13.7 Reconstructing the paleoprecipitation at Nussloch over the Greenland ice (GISP2). During periods of high sea level, the warm
last 80 ka. A—d13C over time. The range of values obtained phases of DO events result in a net increase in rainfall (+30 to 40%),
corresponds to plants of C3 photosynthetic type. B—Reconstruction while during periods of low sea level, the distance from the coastline
of paleoprecipitation by inverse modeling of the isotopic signal insulates Nussloch from any climatic improvement that might result
(according to Hatté and Guiot (2005)). C—Comparison with d18O in from a warm episode

regarded as source tracers. A recent study has been per- of carbonated concretions developed around herbaceous
formed on loess samples, dated from the Last Glacial rootlets (for example, Wang and Follmer (1998); Hatté et al.
Maximum, located along a 50° N transect (from English (2001a)). The values of this isotopic ratio also allow the
Channel to Ukraine), chosen to represent the geographic and characterization of the type of vegetation which trapped the
petrographic variability of the European loess belt. Geo- dust during its deposition, and therefore the associated
chemical results combined with dust emission simulations environmental and climatic conditions. Distinctions may be
revealed that the dust was transported only over a few made at the level of the photosynthetic cycle (plants in C3
hundred kilometers. Moreover, the results highlighted that compared to those in C4) or within a similar photosynthetic
the sources were regional and that distinct sources prevailed type through the definition of seasonal variations in tem-
within the European continent (Rousseau et al. (2014)). perature or precipitation.
The inversion of vegetation models incorporating an
Tracing past precipitation and seasonality isotopic fractionation module allows the reconstruction of
fluctuations in the annual rainfall patterns occurring during
The analysis of d13C from organic matter preserved in very the deposition of the sequence (Hatté and Guiot (2005))
small quantities in the loess sediment indicates the presence (Fig. 13.7). Analysis of other chemical elements, such as
of vegetation; this can also be observed through the presence rare earth elements, or other isotopes, contributes to the
13 Ground-Air Interface: The Loess Sequences, Markers … 165

Fig. 13.8 14C dates obtained from earthworm granules collected in the different paleosols from the Nussloch sequence. Correlation with the
NGRIP climate changes described by the dust and the d18O variations from Moine et al. (2017)

search for the origin of the transported matter, using samples interstadial has been estimated to have been 10–12 ± 4 °C
taken from potential source areas, and thus, to estimate the (Prud’homme et al. (2016)). In line with loess organic d13C
transport-related mechanisms (Rousseau et al. (2014); Gallet interpretation, d13C of earthworm granules has been inter-
et al. (1996); Guo et al. (2002)). preted as a proxy of paleoprecipitation. Thanks to transfer
functions, the past precipitation in Nussloch during the
Tracing past precipitation and past temperature Greenland interstadial has been estimated at about 159–
574 mm/yr (Prud’homme et al. (2018)).
At the same time as the abundance count of earthworm
calcitic granules emerged as a paleoclimate proxy, study of
their isotopic properties has also developed (Prud’homme Loess Chronology
et al. (2016), (2018)). The d18O granules and interlinked
transfer functions between water cycle, air and soil temper- Loess chronology remains a major challenge since OSL
atures allow the estimation of air temperatures. In Nussloch, returns a wide range of value depending on the choice of the
the mean summer air temperature during Greenland mineral (type and grain size), the choice of light ray (IRSL,
166 D.-D. Rousseau and C. Hatté

OSL,), and the analytical choice to reconstruct the evolution Kukla, G. J., & An, Z. S. (1989). Loess stratigraphy in central China.
through time of the ambient dose rate. Nevertheless, it Palaeogeography, Palaeoclimatology, Palaeoecology, 72, 203–
225.
provides a reliable frame allowing comparisons with spe- Lagroix, F., & Banerjee, S. K. (2002). Paleowind directions from the
leothem or ice-core chronologies (Lang et al. (2003); Fuchs magnetic fabric of loess profiles in Central Alsaka. Earth and
et al. (2013); Rousseau et al. (2013)). Planetary Science Letters, 195, 99–112.
14
C is limited by its 50–0 kyr range and by the fact that Lang, A., et al. (2003). High-resolution chronologies for loess:
Comparing AMS 14C and optical dating results. Quaternary
reliable supports such as charcoal or wood are rare in loess Science Reviews, 22, 953–959.
sequences. Trials on ubiquitous supports (loess OM bulk Lautridou, J. P. (1985). Le Cycle Périglaciaire Pléistocène en Europe
(Hatté et al. (2001b)) and alkane (Häggi (2014)) were not that du Nord-Ouest et plus particulièrement en Normandie (pp. 908).
conclusive, highly dependent on the status of the loess sec- Centre Géomorphologie Caen: Thèse Etat, Université Caen, Caen.
Liu, T. S., et al. (1985). Loess and the environment (pp. 251). Beijing:
tion studied and on the cleaning method of the outcrop. Even China Ocean Press.
in the best conditions, they show poor reliability around Lu, H. Y., et al. (2007). Phytoliths as quantitative indicators for the
(paleo) soil. However, the 14C dates obtained from the cal- reconstruction of past environmental conditions in China II:
citic earthworm granules allow perfect dating of the various Palaeoenvironmental reconstruction in the loess plateau. Quater-
nary Science Reviews, 26, 759–772.
paleosols (embryonic soils, tundra gleys and arctic brown Maher, B. A., & Thompson, R. (1995). Paleorainfall reconstructions
paleosols) from which they are collected, as was recently from pedogenic magnetic susceptibility variations in the Chinese
demonstrated in the Nussloch sequence (Fig. 13.8) (Moine Loess and Paleosols. Quaternary Research, 44, 383–391.
et al. (2017)). This new approach opens the door to precise Moine, O., et al. (2002). Paleoclimatic reconstruction using mutual
climatic range on terrestrial mollusks. Quaternary Research, 57,
and independent chronologies of loess sequences. 162–172.
Moine, O., et al. (2005). Terrestrial molluscan records of Weichselian
Lower to Middle Pleniglacial climatic changes from the Nussloch
References loess series (Rhine Valley, Germany): the impact of local factors.
Boreas, 34, 363–380.
Moine, O., et al. (2017). The impact of Last Glacial climate variability
Antoine, P., et al. (1999). Last interglacial-glacial climatic cycle in in west-European loess revealed by radiocarbon dating of fossil
loess-palaeosol successions of north-western France. Boreas, 28, earthworm granules. Proceedings of the National Academy of
551–563. Sciences, 114, 6209–6214.
Antoine, P., et al. (2009). Rapid and Cyclic Aeolian Deposition During Oches, E. A., & McCoy, W. (1995). Amino Acid geochronology
the Last Glacial in European Loess: A High-Resolution Records applied to the correlation and dating of central European loess
from Nussloch, Germany. Quaternary Science Reviews, 28, 2 955– deposits. Quaternary Science Review, 14, 767–782.
2 973. Pécsi, M. (1990). Loess is not just the accumulation of dust.
Fuchs, M., et al. (2013). The loess sequence of Dolni Vestonice, Czech Quaternary International, 7(8), 1–21.
Republic: A new OSL based chronology of the Last Climatic Cycle. Prud’homme, C., et al. (2015). Earthworm calcite granules: a new
Boreas, 42, 664–677. https://doi.org/10.1111/j.1502-3885.2012. tracker of millennial-timescale environmental changes in Last
00299.x. Glacial loess deposits. Journal of Quaternary Science, 30, 529–536.
Gallet, S., et al. (1996). Geochemical characterization of the Luochuan Prud’homme, C., et al. (2016). Palaeotemperature reconstruction during
Loess-Paleosol sequence, China, and Paleoclimatic Implications. the Last Glacial from d18O of earthworm calcite granules from
Chemical Geology, 133, 67–88. Nussloch loess sequence, Germany. Earth and Planetary Science
Gerasimenko, N., & Rousseau, D.-D. (2008). Stratigraphy and Letters, 442, 13–20.
Paleoenvironments of the Last Pleniglacial in the Kyiv Loess Prud’homme, C., et al. (2018). d13C of earthworm calcite granules: a
Region (Ukraine). Quaternaire, 19(4), 293–307. new proxy for palaeoprecipitation reconstructions during the Last
Guo, Z. T., et al. (2002). Onset of Asian desertification by 22 Myr ago Glacial in Western Europe. Quaternary Science Reviews, 179, 158–
inferred from loess deposits in China. Nature, 416, 159–163. 166.
Häggi, C., et al. (2014). On the stratigraphic integrity of leaf-wax Pye, K. (1995). The nature, origin and accumulation of loess.
biomarkers in loess paleosols. Biogeosciences, 11, 2455–2463. Quaternary Science Reviews, 14, 653–667.
Pécsi, M. (1990). Loess Is not just the accumulation of dust. Pye, K., & Zhou, L. P. (1989). Late Pleistocene and Holocene Aeolian
Quaternary International, 7/8, 1–21. Dust Deposition in North China and the Northwest Pacific-Ocean.
Hatté, C., et al. (2001a). d13C Variation of loess organic matter as a Palaeogeography, Palaeoclimatology, Palaeoecology, 73, 11–23.
potential proxy for paleoprecipitation. Quaternary Research, 55, Rousseau, D.-D. (1987). Paleoclimatology of the Achenheim Series
33–38. (Middle and Upper Pleistocene, Alsace, France). A Malacological
Hatté, C., et al. (2001b). Development of an accurate and reliable 14C Analysis. Palaeogeography, Palaeoclimatology, Palaeoecology, 59,
chronology for loess sequences. Application to the loess sequence 293–314.
of Nußloch (Rhine valley, Germany). Radiocarbon, 43, 611–618. Rousseau, D.-D. (1991). Climatic transfer function from quaternary
Hatté, C., & Guiot, J. (2005). Palaeoprecipitation reconstruction by mollusks in european loess deposits. Quaternary Research, 36, 195–
inverse modelling using the isotopic signal of loess organic matter: 209.
application to the Nussloch Loess Sequence (Rhine Valley, Rousseau, D.-D., et al. (2001). Late Pleistocene environments of the
Germany). Climate Dynamics, 25, 315–327. central Ukraine. Quaternary Research, 56, 349–356.
Kukla, G., et al. (1990). Magnetic susceptibility record of Chinese Rousseau, D.-D. et al. (2007a). Evidence of cyclic dust deposition in
Loess. Transactions of the Royal Society Edinburgh: Earth the US great plains during the last deglaciation from the
Sciences, 81, 263–288. high-resolution analysis of the peoria loess in the eustis sequence
13 Ground-Air Interface: The Loess Sequences, Markers … 167

(Nebraska, USA). Earth and Planetary Science Letters, 262, 159– Sun, Y. et al. (2008). Tracing the provenance of fine-grained dust
174. deposited on the central Chinese loess plateau. Geophysical
Rousseau, D.-D. et al. (2007b). Link Between European and North Research Letters, 35. https://doi.org/10.1029/2007GL031672.
Atlantic Abrupt Climate Changes over the Last Glaciation. Taylor, S. N., et al. (2014). Mineral magnetic characterization of the
Geophysical Research Letters, 34, https://doi.org/10.1029/ Upper Pleniglacial Nussloch loess sequence (Germany): an insight
2007gl031716. into local environmental processes. Geophysical Journal Interna-
Rousseau, D. D., et al. (2013). Major dust events in Europe during tional, 199, 1463–1480.
marine isotope stage 5 (130–74 ka): A climatic interpretation of the Wang, H., & Follmer, L. R. (1998). Proxy of monsoon seasonality in
‘markers. Climate of the Past, 9, 2213–2230. carbon isotopes from paleosols of the Southern Chinese loess
Rousseau, D.-D., et al. (2014). European Glacial Dust Deposits: plateau. Geology, 26, 987–990.
Geochemical Constraints on Atmospheric Dust Cycle Modeling. Wu, N. Q. et al., (2001). Orbital forcing of terrestrial mollusks and
Geophysical Research Lettters, 41, 7666–7674. https://doi.org/10. climatic changes from the loess plateau of China during the past
1002/2014GL061382. 350 ka. Journal of Geophysical Research—Atmospheres, 106, 20
Sima, A., et al. (2009). Imprint of North-Atlantic millennial-timescale 045–20 054.
variability on Western European loess deposits as viewed in a dust Zhou, L. P., et al. (1990). Partly Pedogenic Origin of Magnetic
emission model. Quaternary Science Reviews, 28, 2851–2866. Variations in Chinese Loess. Nature, 346, 737–739.
 Air-Ground Interface: Reconstruction
of Paleoclimates Using Speleothems 14
Dominique Genty and Ana Moreno

system); (3) once it arrives in the underground gallery, there

Speleothems: Description, Distribution,
is degassing of CO2 and precipitation of CaCO3.
Formation and Preservation
Drops of water, emerging from micro-fissures, form sta-
lactites at the ceiling of the gallery, and stalagmites on the
The term speleothem refers to carbonate deposits in caves:
ground. It is the latter which is most frequently studied
mainly stalactites, stalagmites and stalagmite floors. It is
because of its simple structure (resembling inverted inter-
taken from the English (speleothem) which has its roots in
locking calcite cups). The physical and chemical parameters
Greek signifying ‘subject’ (thema) and ‘cave’ (spelaion).
that allow us to reconstruct paleoclimates from speleothems
Composed of calcium carbonate, speleothems are most
(stable isotopes, trace elements, growth rate) depend partly
commonly made of calcite (ones in aragonite are rarer and
on the precipitation conditions of the calcite (the temperature
less studied). Carbonate massifs which contain caves, within
in the cave, flow rate, moisture level) and, partly on the
which speleothems are found, are widely scattered around
geochemical characteristics of the water supplying the spe-
the globe, and are found at all latitudes and in all continents,
leothems (McDermott 2004; Fairchild and Baker 2012).
although less frequently in the southern hemisphere
Speleothems, once formed, seldom undergo subsequent
(Fig. 14.1).
changes (no erosion, no internal recirculation, some rare
Speleothems acquire their geochemical and structural
examples of diagenesis, such as, for example, the transfor-
characteristics as a result of the infiltration of rainwater
mation of aragonite into calcite). The calcite in speleothems
through a limestone or dolomitic environment. The forma-
is a material that can be dated as far back as 500 ka using
tion process is basically comprised of three steps: (1) at
radiometric methods (U-Th series dating) or even as far back
ground level, CO2 (produced by the roots of plants and by
as several million years depending on their geochemical
microorganisms in the soil) is dissolved in rainwater; (2) the
characteristics (U-Pb dating).
dissolution of the surrounding rock (limestone, dolomitic
limestone, calcareous dolomites) either at ground level
(known as dissolution in an open system – at the meeting of
Growth and Chronology of Speleothems
the three elements: air, water and rock), or at the level of the
many tiny fissures in the surrounding carbonate (closed
The growth of speleothems is determined by the presence of
infiltrating water. As a result, extremely dry or cold climates
(water freezing above ground), causes a halt in growth
D. Genty (&) (hiatus) except in certain exceptional circumstances found in
Laboratoire des Sciences du Climat et de l’Environnement, the mountains (Luetscher et al. 2015). The speed of growth
LSCE/IPSL, CEA-CNRS-UVSQ, Université Paris-Saclay, of a stalagmite depends not only on the amount of water
91190 Gif-sur-Yvette, France
e-mail: [email protected];
infiltrating into the karst, but also on a number of environ-
[email protected] mental factors such as the Ca2+ concentration of the infil-
EPOC (Environnements et Paléoenvironnements Océaniques et
trating water, the thickness of the film of water on the
Continentaux), Université de Bordeaux, bat. B18N, Allée Geoffroy surface of the speleothem, the temperature and the partial
Saint-Hilaire, 33615 Pessac Cedex, France pressure of CO2 in the cave’s atmosphere. Geochemical
A. Moreno modeling of growth shows that the dominant factors are the
Instituto Pirenaico de Ecología—CSIC, Avda. Montañana 1005, flow rate and the Ca2+ content of the water (Dreybrodt
50059 Saragossa, Spain

© Springer Nature Switzerland AG 2021 169

G. Ramstein et al. (eds.), Paleoclimatology, Frontiers in Earth Sciences,
https://doi.org/10.1007/978-3-030-24982-3_14
170 D. Genty and A. Moreno

Fig. 14.1 Map showing the distribution of carbonate massifs (in black) where speleothems can be found (according to Ford and Williams (2007)
and the University of Auckland http://web.env.auckland.ac.nz/our_research/Karst/)

1988). This model was verified by comparing the theoretical time can be several years (Genty and Deflandre 1998; Genty
growth rates with the actual growth rates measured on et al. 2014). These have resulted in a seasonal variation in
modern calcite deposits at various latitudes. In a cold and the growth rate and are probably responsible for the for-
humid climate (Scotland), the average vertical growth of a mation of visible or luminescent annual growth laminae
stalagmite is only 20 microns/year, while it can reach (Genty and Quinif 1996). When such laminae are present,
1 mm/year in the caves of southern France (Baker et al. their identification (a clear and dark lamina is deposited each
1998). The sensitivity of the growth rate to environmental year) can provide a precise timeline and allows the study of
conditions provides an indicator of paleoclimatic conditions: the evolution of an isotopic or geochemical parameter to the
more speleothems with a faster growth rate are produced in nearest year. But annual lamination of stalagmites is not
warm and humid periods than in cold, dry periods. It appears systematic and is often broken; examples rarely go back
that below a certain rainfall threshold, the growth of spe- further than 1000 years (Baker et al. 2015).
leothems may slow down due to under-saturation of the The most common way to date speleothems is to measure
infiltration water. the isotope series of uranium: 234U, 238U, 230Th (See
The temperature of a cave is generally stable throughout Chap. 6). Uranium, which is soluble, enters the karst system,
the year and is close to the average annual exterior tem- while thorium, insoluble, remains above ground. The 230Th
perature. Depending on the depth of the cave, the exterior measured in the stalagmites is therefore, theoretically, the
thermal wave which determines the cave temperature may result of the disintegration of 234U. Sometimes, thorium is
take between several months and several years to travel also brought by infiltrated water, often along with some
through the thickness of the rock. However, there are detrital clay; a correction must then be made to take account
marked seasonal variations in the water infiltration rate and of 232Th (linked to detrital contamination) in the light of the
in the concentrations of the different chemical elements initial 230Th/232Th ratio (Hellstrom 2006). Finally, uncer-
(Ca, Sr, Mg, U, MO, etc.) even if the average infiltration tainty on the age of a calcite fragment is linked to the size of
14 Air-Ground Interface: Reconstruction of Paleoclimates … 171

the correction for detritus, and to the uranium concentration an understanding of the dynamics of infiltration, especially
in the speleothem (varying from *30 ppb to *1 ppm). when the elements measured are linked to the residence time
This is, on average, between less than 1% and 5% for the last of the infiltrating water. A new model called I-Stal was
three climate cycles, as far back as *500 ka. The most recently developed to investigate the factors involved in the
common current technique to measure these isotopes is by interpretation of trace elements in speleothems (Stoll et al.
MC-ICP-MS; this requires between a few milligrams to tens 2012). To measure these elements, precise techniques are
of milligrams of calcite to make a measurement. Moreover, used to analyze points of a few micrometers in diameter on a
thanks to the technical progress in ICP-MS equipment and polished section of stalagmite (LA-ICP-MS, XRF, X
an improved understanding of half-life constants of fluorescence by synchrotron, ionic microprobe). During the
radioactive isotopes in recent years, the error in U-Th dates transfer of these trace elements, organic colloids play a
can be even lower (Cheng et al. 2013). Other methods are major role. The sensitivity of each one to environmental
used incidentally, such as 14C, whose concentration at the conditions is different; for example, Ba, Na and Sr may be
moment of deposition of the calcite depends on the pro- sensitive to the speleothem growth rate while Mg and U may
portions of atmospheric CO2 and of CO2 without 14C, reflect paleohydrology. Some, such as P, Zn and Cu, are
coming from the dissolution of limestone (Genty et al. related to the vegetation above ground. Trace elements are
1999), 226Ra for recent millennia (Ghaleb et al. 2004), and also chronological markers: when the annual laminae are not
the U-Pb method which can trace back more than a million visible in the calcite structure, then the analysis of trace
years (Woodhead et al. 2006). elements can reveal seasonal variations, thereby providing a
relative or absolute chronology accurate to the nearest year.
This type of analysis is used to determine the duration of a
Paleoclimate Reconstruction: A Qualitative rapid climate event or of a transition (Bourdin et al. 2011).
Approach The stable isotopes of calcite are the most commonly
used to reconstruct climate variations, even though inter-
The proxies used in speleothems to reconstruct past climates preting them into temperature and precipitation terms is not
are isotopes of calcite (d18Oc, d13Cc), isotopes of fluid easy. The d18Oc of the calcite, when it precipitates at isotopic
inclusions trapped in the calcite (dD, d18O), trace elements equilibrium (see below), depends on the temperature of the
(Sr, Ba, Mg, U), organic matter (lipids, amino acids), pet- precipitation of calcite (temperature in the cave and therefore
rography and the growth rate. the average annual exterior temperature) and on the d18Ow of
The most common crystalline fabrics (layout and shape the infiltrating water. The latter is linked to the exterior
of the calcite crystals) in stalagmites is the palisading temperature above the site, to the amount of water extracted
columnar one: large elongated crystals perpendicular to the from the cloud masses between the source of evaporation
growth laminae. However, the addition of detrital compo- and the site and to the isotopic composition of the source,
nents, variations in humidity, in temperature or in pCO2 over usually the ocean (isotope distillation process). During the
the course of climate variations can produce a variety of precipitation of calcite, there is an inverse relationship
crystalline fabrics (dendritic, fibrous, microcrystalline, etc.). between d18Oc and temperature (*−0.24‰/°C), whereas
Even if we could establish links between crystal structures the relationship is direct between the rainfall d18O and out-
and environmental conditions through the study of modern side temperature of (e.g. 0.3–0.7‰/°C). In summary:
calcite deposits (Frisia et al. 2000), this type of relationship

is complex and should be examined with caution. Recently, d18 Oc ¼ f d18 Ow ; Tcave ; isotopic equilibrium
a new approach to stalagmite characterization consists of
with:
precisely describing the petrography and categorizing the
stalagmites and their bounding surfaces into six different

d18 Ow ¼ f d18 Orain ; evapotranspiration in some cases
classes (from individual crystallites to major nonconformi-
ties) (Martín-Chivelet et al. 2017). and
Speleothem calcite contains, either in its crystalline

defects, or by substitution of Ca in the crystal lattice, minor d18 Orain ¼ f Text :; quantity of rain, trajectory of cloud masses,d18 Osource :
or trace elements, which provide information on the pale-
oenvironment (Fairchild et al. 2000). Interpretation of this is Consequently, depending on the location of the site rel-
complex because it involves several factors: composition of ative to the main source of evaporation and on the prevailing
the soil and of the surrounding limestone, the intensity of conditions when the water masses were transported, the
dissolution and the precipitation conditions of the calcite relationship between d18Oc and climate will be more or less
(temperature, supersaturation, growth rate, etc.). At the marked. Thus, in the stalagmites of Southeast Asia the
annual level, analysis of these trace elements contributes to variations in intensity of the monsoon over the last two
172 D. Genty and A. Moreno

climate cycles can be seen remarkably well: the d18Oc is evolution of the d18Oc of the speleothem calcite is not uni-
systematically depleted by *4‰ during periods of strong form from one region to another: in general the d18O
summer monsoon, mainly due to the effect of mass (the d18O decreases when the climate becomes warm and humid (in
of the rain is inversely proportional to the volume of rainfall) temperate regions for example), while this trend may be
but also due to changes in the source (Wang et al. 2008). reversed (d18O increases as the temperature increases) at
Thanks to the many U-Th datings carried out on several certain altitudes as has been observed in the Alps (Boch et al.
stalagmites in caves in Sanbao, Hulu and Dongge (25° N to 2011; Moseley et al. 2014) and in eastern Europe on the
32° N, China), it has been shown that the d18Oc of the calcite coast of the Black Sea (Fleitmann et al. 2009).
had a periodicity of 23 ka and was directly correlated with The d13Cc of calcite may also react to climate changes,
changes in insolation at 65° N, demonstrating that variations and sometimes in a more obvious way than d18Oc. The
in the monsoons were caused by orbital changes (Fig. 14.2) carbon atoms of the calcite molecule, CaCO3, which makes
(Wang et al. 2008). Superimposed on these large climate up the speleothems come from two main sources: CO2 from
variations, millennial climate events were also recorded and the soil and CaCO3 from the surrounding rock. Soil CO2,
linked to Dansgaard-Oeschger events (D/O) detected in the produced by plant roots and microbial activity, has a d13Cc
Greenland ice, which shows the strong connection between close to −24‰ (for C3 type plants, most frequently found in
the climate systems of South Asia and the North Atlantic. temperate areas), CaCO3 from marine limestone has a d13Cc
This interconnection between the monsoon regime and the of between *−2 and +2‰. It has been shown that the main
North Atlantic region was confirmed by one of the longest source of carbon in speleothems is the CO2 from the soil
and most precise paleoclimate recordings produced by the which can represent up to 90% of C contained in the CaCO3
study of Chinese speleothems (Cheng et al. 2016). The of speleothems (Genty et al. 1998). In several sites in the
variations in monsoon intensity were reconstructed for the South of France, the carbon from the dissolution of lime-
past 640,000 years with an unmatched precision. The mul- stone (also called dead carbon because it contains no 14C)
tiplicity of samples, precise U-Th dating and the very high represents, in this particular case, only 15 ± 5% of the C in
resolution of isotopic analyses suggest new theories on the the speleothems. Consequently, any change in the vegetation
causes of the major glacial-interglacial climate cycles and above a cave brought about by climate change, such as
millennial variability. One of these theories posits that the proportion of type C3 to C4 or vegetation density, will have
period between terminations (glacial-interglacial transitions) an impact on the d13C of the CO2 in the soil and thus on the
is equal to a multiple of the length of the precession cycles; d13Cc of the speleothems. In summary:
moreover, a very strong teleconnection between the d13Cc = f (type of vegetation, density of vegetation,
dynamics of the ice caps of the northern hemisphere, the hydrology, isotopic equilibrium).
circulation of the Atlantic Ocean and the Asian monsoon is The d13Cc in the stalagmites in the Villars cave (South-West
implied. Similarly, the variation in monsoon intensity in France) shows abrupt changes from −2‰ to −5‰ over the last
South America was recorded in stalagmites from southern 80 ka. These have been linked to Dansgaard-Oeschger events
Brazil (24° S to 27° S), at the orbital and the millennial scale recorded in Greenland ice cores and to temperature recon-
(Cruz et al. 2005). In this case, the latitudinal changes in the structions using analyzes of pollen from lakes and marine cores
position of the ITCZ (intertropical convergence zone) seem (Genty et al. 2003).
to be the root cause of these variations. The value of this comparison is, firstly, to test the
At higher latitudes, in Europe for example, the d18Oc of chronology from other archives against the absolute
calcite has a less pronounced response to changes in climate, chronology provided by speleothems, and if necessary, to
probably because of the opposite effects controlling the adjust it. Refining the age of an abrupt climatic transition,
d18Oc of calcite (e.g. external temperature impacts on the such as the one which occurred at the beginning of D/O 12
d18O of the rain while cave temperature impacts on d18Oc of (Fig. 14.3), is important in order to find out its cause by
calcite). However, the study of several European samples comparing a sequence of climate events with the external
has revealed a certain logic within the paleoclimate chroni- forcings and with archives from other latitudes. Comparison
cles of varying appearance: there is a variable gradient in the with pollen reconstructions shows the close link between
d18Oc speleothems during the Holocene along a west-east changes in vegetation brought on by variations in tempera-
transect (McDermott et al. 2011). This is related to differ- ture and humidity, and the d13Cc recorded in speleothems.
ences in warming during the Holocene depending on lon- Other examples show that the d13Cc recorded the last
gitude and also on different atmospheric circulations. This deglaciation, in New Zealand as well as in Europe (Genty
logic shows that even if speleothems do not precipitate at et al. 2006; Moreno et al. 2010). Thus, a stalagmite from the
isotopic equilibrium, they contain valuable information on Chauvet cave (Ardèche, France) records the climate events
atmospheric paleocirculation. As a result of the influence of (e.g. Bølling-Allerød, Intra Allerød Cold Period,
these multiple climatic, hydrological and kinetic factors, the Younger-Dryas) that punctuated the last deglaciation with a
14 Air-Ground Interface: Reconstruction of Paleoclimates … 173

Fig. 14.2 After Wang et al, 2008—Example of a recording of the the insolation, dominated here by the precession (cycles of 23 ka). The
variations in intensity of the Asian monsoon using the d18Oc of Chinese d18Oatm from Vostok reflects the impact of the precession on low
stalagmites from Sanbao and Hulu caves (bottom graph). Comparison latitude water cycle and productivity of the biosphere (See Chap. 11 on
with July insolation at 65°N and the d18Oatm of Vostok ice core, polar ice)
Antarctica (dashed line). There is a good correlation between d18Oc and

resolution comparable to that obtained in ice cores from calculate the temperature of calcite formation in the cave,
Greenland, with, however, some differences in the trends which is equivalent to the average annual exterior tempera-
caused by different climate gradients. ture. For this second case, it is necessary for the calcite
precipitation to have occurred at thermodynamic equilibrium
(isotopic, by extension), in other words, that the exchanges
Paleoclimate Reconstruction: A Quantitative between the different carbon species (e.g.HCO 2
3 , CO3 ; CO2
Approach gas) have been completed.
Examples using this new technique are still rare, but are
During its growth, the speleothem calcite traps water in the among the only ones, on land, to express the evolution of
form of microscopic fluid inclusions (1–10 microns large) temperature (from direct measurements) over a precise
or, more rarely, macroscopic ones (several mm). This water absolute time scale. A stalagmite from Peru has thus shown
comes from the rainwater contemporaneous to the cal- that the isotopic composition of the rainwater followed the
cite deposition and can therefore be dated indirectly by local winter insolation (6° S) due to changes in the intensity
dating the surrounding calcite. With an average composition of convective rainfall during the Holocene, themselves
of only a few nL of water per gram of calcite, the technical linked to latitudinal variations in the ITCZ (Van Breukelen
difficulties of extracting and analyzing this water have only et al. 2008). The temperature, calculated using the above
recently been resolved (Verheyden et al. 2008; Vonhof et al. method, varied little (±2 °C) over the last 13.5 ka, unlike at
2006; Affolter et al. 2014; Arienzo et al. 2013). There are higher latitudes, as is shown by another example from
many reasons for measuring the isotopic composition (dD, Vancouver Island (Canada 49° N), where the temperature
d18O) of the fluid inclusions of speleothems: (1) the value of has varied by more than 10°C between 6 ka and 10 ka
the d18O of the inclusion water is close to that of the rain- (Zhang et al. 2008).
water and so is a tracer of atmospheric circulation; (2) in However, this method cannot be applied to all spe-
conjunction with the d18Oc of the calcite, it is possible to leothems. Indeed, conditions of low humidity or low pCO2
174 D. Genty and A. Moreno

Fig. 14.3 Comparison between

the d13C from Villars
Cave stalagmites, the d18O from
NGRIP (Greenland), the
temperature reconstructions from
Bouchet Lake (Massif Central)
(Guiot et al. 1989) and marine
core ODP976 from the south of
Spain (Combourieu Nebout et al.
2002). The dots at the bottom of
the graph show the U-Th datings
with 2r error bars

inside the cave can cause rapid degassing of CO2, and thus CO2 (13C18O16O) produced by acidc attack of carbonate
lead to a kinetic effect causing a thermodynamic disequilib- minerals. When applied to speleothems which were depos-
rium. There are several equations linking the isotopic com- ited at thermodynamic equilibrium, precipitation tempera-
positions of calcite and of water with temperature, allowing tures can be reconstructed independently of the previous
this equilibrium to be checked (Kim and O’Neil 1997; Tre- method which uses the d18O of the calcite and of the fluid
maine et al. 2011). However, the study of deposits of modern inclusions. However, tests on modern deposits from caves in
calcite shows that many of them are not deposited at isotopic the South of France and in vitro experiments show the
equilibrium, but rather reflect a kinetic effect (Genty 2008; existence of a kinetic effect during precipitation (Daeron
Mickler et al. 2006). In this case, calculation of the tempera- et al. 2011, 2008). In this case, the temperature can only be
ture is not valid (the d18Oc measured from calcite is generally calculated by using both the Delta 47 measurement and the
too high, causing the calculated temperature to be too low). measurement of the d18O in calcite and fluid inclusions. The
There is a recent, promising method to reconstruct pale- value of this is twofold: in addition to calculating the pre-
otemperatures using only isotopes from the mineral phase of cipitation temperature, this method quantifies the state of
the calcite, thus overcoming the difficulty of extracting water thermodynamic disequilibrium, often difficult to detect, and
from the fluid inclusions (Daeron et al. 2008, 2011). It which is also a reflection of the conditions of the
involves the clumped isotopes of the CO2 3 molecules and is
paleoenvironment.
based on the thermo-dependence of the isotopic exchanges Another way of quantifying climate parameters is by
between these different molecules: calibrating proxies from the growth laminae of fast-growing
modern stalagmites (Baker et al. 2007; Domínguez-Villar
13
C16 O2 12 18 16 2 13 18 16 2 12 16
3 þ C O O2 \ ¼ [ C O O2 þ C þ C O3
12 16 2
et al. 2018). The method consists of finding the best corre-
lation between the signal measured on the stalagmite
Called the Delta 47 method, this method reflects the (e.g. the thickness of the annual growth laminae) and the
statistical overabundance of mass 47 clumped isotopes in climate signal from instrumental measurements outside
14 Air-Ground Interface: Reconstruction of Paleoclimates … 175

(e.g., rainfall). However, the water arriving at the end of a References

stalactite is, in fact, a mixture of water from a variety of
networks of large and small micro-fissures which are more Affolter, S., Fleitmann, D., & Leuenberger, M. (2014). New online
or less complex, with a variable transfer time from the method for water isotope analysis of speleothem fluid inclusions
exterior to the inner gallery (ranging from a few hours to using laser absorption spectroscopy (WS-CRDS). Climate of the
Past, 10, 1291–1304. https://doi.org/10.5194/cp-10-1291-2014.
few weeks, months or years). This mixture was modeled
Arienzo, M. M., Swart, P. K., & Vonhof, H. B. (2013). Measurement of
by A. Baker in a cave in Ethiopia, with components of d18O and d2H values of fluid inclusion water in speleothems using
different ages: a small proportion (10–30%) of water cavity ring-down spectroscopy compared with isotope ratio mass
flowing quickly through the rock (from a few days to a few spectrometry. Rapid Communications in Mass Spectrometry, 27,
2616–2624. https://doi.org/10.1002/rcm.6723.
weeks) and a greater proportion (70–90%), of the water
Baker, A., Asrat, A., Fairchild, I. J., Leng, M. J., Wynn, P. M., Bryant,
passing through slowly (months to a few years). The C., & Umer, M. (2007). Analysis of the climate signal contained
variation in the amount of water arriving at the particular within delta O-18 and growth rate parameters in two ethiopian
stalagmite being studied (which largely determines the stalagmites. Geochimica et Cosmochimica Acta, 71, 2975–2988.
Baker, A., Genty, D., Dreybrodt, W., Barnes, W. L., Mockler, N. J., &
thickness of the annual laminae) is calculated by applying
Grapes, J. (1998). Testing theoretically predicted stalagmite growth
this mixture model to the rainfall provided by meteoro- rate with recent annually laminated samples: Implications for past
logical data. The proxy (the thicknesses, in this case) is stalagmite deposition. Geochimica et Cosmochimica Acta, 62,
then compared with the amount of water supplying the 393–404.
Baker, A., Hellstrom, J. C., Kelly, B. F. J., Mariethoz, G., & Trouet, V.
stalagmite (modeled using this simple mixture model). The
(2015). A composite annual-resolution stalagmite record of North
timeframe is given by the annual growth laminae. This Atlantic climate over the last three millennia. Scientific Reports, 5,
new method, applied to stalagmites in Ethiopia, has helped 10307. https://doi.org/10.1038/srep10307.
to reconstruct summer rainfall over the last 80 years Boch, R., Cheng, H., Spötl, C., Edwards, R. L., Wang, X., &
Häuselmann, P. (2011). NALPS: A precisely dated European
(Baker et al. 2007). Other work is being developed around
climate record 120–60 ka. Climate of the Past Discussions, 7,
the modeling of the d18O of infiltration water involving 1049–1072. https://doi.org/10.5194/cpd-7-1049-2011.
reservoirs in the unsaturated zone (Fairchild and Baker Bourdin, C., Douville, E., & Genty, D. (2011). Alkaline-earth metal
2012) or by empirical adjustment with modern climate data and rare-earth element incorporation control by ionic radius and
growth rate on a stalagmite from the Chauvet Cave, Southeastern
(Domínguez-Villar et al. 2018).
France. Chemical Geology, 290, 1–11.
Cheng, H., Edwards, R. L., Sinha, A., Spötl, C., Yi, L., Chen, S., et al.
(2016). The Asian monsoon over the past 640,000 years and ice age
Conclusion terminations. Nature, 534, 640–646. https://doi.org/10.1038/
nature18591.
Cheng, H., Lawrence Edwards, R., Shen, C.-C., Polyak, V. J.,
Speleothems constitute continental paleoclimate records Asmerom, Y., Woodhead, J., et al. (2013). Improvements in
that allow the reconstruction of climate changes on a precise 230Th dating, 230Th and 234U half-life values, and U-Th isotopic
and absolute chronological timescale over the last measurements by multi-collector inductively coupled plasma mass
half-million years or more. They help to improve the accu- spectrometry. Earth and Planetary Science Letters, 371–372, 82–
91. https://doi.org/10.1016/j.epsl.2013.04.006.
racy of chronologies derived from other climate archives Comas Bru, L., Burstyn, Y., & Scroxton, N. (2017). From caves to
and, depending on the site, can be used to reconstruct vari- climate: Creating the SISAL global speleothem database. Past
ations in rain intensity, rain source and vegetation den- Global Change Magazine, 25, 156–156. https://doi.org/10.22498/
sity on the continent. However, to correctly interpret the pages.25.3.156.
Combourieu Nebout, N., Turon, J. -L., Zahn, R., Capotondi, L.,
proxies measured in the speleothems, it is essential to know Londeix, L., & Pahnke, K. (2002). Enhanced aridity and atmo-
the present day environment of the studied site: climate, spheric high-pressure stability over the western Mediterranean
seasonality, hydrology, origin of the precipitations. Recent during the North Atlantic cold events of the past 50 ky. Geology, 30
methods of analysis of fluid inclusions and of Delta 47 create (10), 863–866.
Cruz Jr, F. W., Burns, S. J., Karmann, I., Sharp, W. D., Vuille, M.,
opportunities to establish long series of continental temper- Cardoso, et al. (2005). Insolation-driven changes in atmospheric
ature changes and of the isotopic composition of rainwater. circulation over the past 116,000 years in subtropical Brazil.
The recent proliferation of studies of speleothems and the Nature, 434, 63–66.
increasingly structured organization of the scientific com- Daeron, M., Guo, W., Eiler, J., Genty, D., Blamart, D., Boch, R., et al.
(2011). (13)C(18)O clumping in speleothems: Observations from
munity working on this archive (Karst Record symposiums natural caves and precipitation experiments. Geochimica et Cos-
for example) have led to the discovery of important results mochimica Acta, 75, 3303–3317.
on climate change and their causes. Interactions between Daeron, M., et al. (2008). Absolute speleo-thermometry, using clumped
increasingly large datasets (INTIMATE-Moreno et al. 2014, isotope measurements to correct for kinetic isotope fractionations
induced by CO2 degassing. Geochimica et Cosmochimica Acta, 72,
SISAL database being constructed, Comas Bru et al. 2017) A193–a193.
and climate models are multiplying so as to better under- Domínguez-Villar, D., Lojen, S., Krklec, K., Kozdon, R., Edwards, R.
stand the processes and causes of climate variations. L., & Cheng, H. (2018). Ion microprobe d18O analyses to calibrate
176 D. Genty and A. Moreno

slow growth rate speleothem records with regional d18O records of Martín-Chivelet, J., Muñoz-García, M. B., Cruz, J. A., Ortega, A. I., &
precipitation. Earth and Planetary Science Letters, 482, 367–376. Turrero, M. J. (2017). Speleothem Architectural Analysis: Inte-
https://doi.org/10.1016/j.epsl.2017.11.012. grated approach for stalagmite-based paleoclimate research. Sedi-
Dreybrodt, W. (1988). Processes in karst systems. Springer Verlag, mentary Geology, 353, 28–45. https://doi.org/10.1016/j.sedgeo.
288p. 2017.03.003.
Fairchild, I. J., & Baker, A. (2012). Speleothem science: From process McDermott, F. (2004). Paleo-climate reconstruction from stable isotope
to past environments. Wiley. variations in speleothems: A review. Quaternary Science Reviews,
Fairchild, I. J., Borsato, A., Tooth, A. F., Frisia, S., Hawkesworth, C. J., 23, 901–918.
Huang, Y., et al. (2000). Controls on Trace Element (Sr-Mg) McDermott, F., Atkinson, T. C., Fairchild, I. J., Baldini, L. M., &
compositions of carbonate cave waters: Implications for speleothem Mattey, D. P. (2011). A first evaluation of the spatial gradients in
climatic records. Chemical Geology, 166, 255–269. d18O recorded by European Holocene speleothems. Global and
Fleitmann, D., Cheng, H., Badertscher, S., Edwards, R. L., Mudelsee, Planetary Change, 79, 275–287. https://doi.org/10.1016/j.
M., Gokturk, O. M., et al. (2009). Timing and climatic impact of gloplacha.2011.01.005.
Greenland interstadials recorded in stalagmites from northern Mickler, P. J., Stern, L. A., & Banner, J. L. (2006). Large kinetic
Turkey. Geophysical Research Letters, 36. isotope effects in modern speleothems. Geological Society of
Ford, D. C. & Williams, P. (2007). Karst Geomorphology and America Bulletin, 118, 65–81.
Hydrology, Wiley, 562p. Moreno, A., Stoll, H. M., Jiménez-Sánchez, M., Cacho, I.,
Frisia, S., Borsato, A., Fairchild, I. J. & McDermott, F. (2000). Calcite Valero-Garcés, B., Ito, E., & Edwards, L. R. (2010).
Fabrics, Growth Mechanisms, and Environments of formation in A speleothem record of rapid climatic shifts during last glacial
Speleothems from the Italian Alps and Southwestern Ireland. period from Northern Iberian Peninsula. Global and Planetary
Journal of Sedimentary Research, 70, 1183–1196. Change, 71, 218–231. https://doi.org/10.1016/j.gloplacha.2009.10.
Genty, D. (2008). Paleoclimate research in Villars Cave (Dordogne, 002.
SW-France). International Journal of Speleology, 37, 3. Moreno, A., Svensson, A., Brooks, S. J., Connor, S., Engels, S.,
Genty, D., Blamart, D., Ghaleb, B., Plagnes, V., Causse, C., Fletcher, W., et al.(2014). A compilation of Western European
Bakalowicz, M., et al. (2006). Timing and dynamics of the last terrestrial records 60–8 ka BP: towards an understanding of latitu-
deglaciation from European and North African d13C stalagmite dinal climatic gradients. Quaternary Science Reviews, Dating,
profiles—Comparison with Chinese and South Hemisphere stalag- Synthesis, and Interpretation of Paleoclimatic Records and
mites. Quaternary Science Reviews, 25, 2118–2142. https://doi.org/ Model-data Integration: Advances of the INTIMATE project
10.1016/j.quascirev.2006.01.030. (INTegration of Ice core, Marine and Terrestrial records, COST
Genty, D., Blamart, D., Ouahdi, R., Gilmour, M., Baker, A., Jouzel, J., Action ES0907) 106, 167–185. https://doi.org/10.1016/j.quascirev.
et al. (2003). Precise dating of Dansgaard-Oeschger climate oscilla- 2014.06.030.
tions in western Europe from stalagmite data. Nature, 421, 833–837. Moseley, G. E., Spötl, C., Svensson, A., Cheng, H., Brandstätter, S., &
Genty, D., & Deflandre, G. (1999). Drip flow variations under a Edwards, R. L. (2014). Multi-speleothem record reveals tightly
stalactite of the Père Noël cave (Belgium). Evidence of seasonal coupled climate between central Europe and Greenland during
variations and air pressure constraints, Journal of Hydrology, 211, Marine Isotope Stage 3. Geology, 42, 1043–1046. https://doi.org/
208–232. 10.1130/G36063.1.
Genty, D., Labuhn, I., Hoffmann, G., Danis, P. A., Mestre, O., Bourges, Stoll, H. M., Müller, W., & Prieto, M. (2012). I-STAL, a model for
F., et al. (2014). Rainfall and cave water isotopic relationships in interpretation of Mg/Ca, Sr/Ca and Ba/Ca variations in speleothems
two South-France sites. Geochimica et Cosmochimica Acta, 131, and its forward and inverse application on seasonal to millennial
323–343. https://doi.org/10.1016/j.gca.2014.01.043. scales: I-STAL SPELEOTHEM TRACE ELEMENT MODEL.
Genty, D., Massault, M., Gilmour, M., Baker, A., Verheyden, S., & Geochem. Geophys Geosystems 13, n/a-n/a. https://doi.org/10.
Kepens, E. (1998). Calculation of past dead carbon proportion and 1029/2012GC004183.
variability by the comparison of AMS 14C and TIMS U/Th ages on Tremaine, D. M., Froelich, P. N., & Wang, Y. (2011). Speleothem
two Holocene stalagmites. Radiocarbon, 41, 251–270. calcite farmed in situ: Modern calibration of d18O and d13C
Genty, D., & Quinif, Y. (1996). Annually laminated sequences in the paleoclimate proxies in a continuously-monitored natural cave
internal structure of some Belgian stalagmites—Importance for system. Geochimica et Cosmochimica Acta, 75, 4929–4950. https://
paleoclimatology. Journal of Sedimentary Research, 66, 275–288. doi.org/10.1016/j.gca.2011.06.005.
Ghaleb, B., Pons-Branchu, E., & Deschamps, P. (2004). Improved Van Breukelen, M. R., Vonhof, H. B., Hellstrom, J. C., Wester, W.
method for radium extraction from environmental samples and its C. G., & Kroon, D. (2008). Fossil dripwater in stalagmites reveals
analysis by thermal ionization mass spectrometry. Journal of Holocene temperature and rainfall variation in Amazonia. Earth and
Analytical Atomic Spectrometry, 19, 906–910. Planetary Science Letters, 275(1–2), 54–60.
Guiot, J., Pons, A., de Beaulieu, J. L., & Reille, M. (1989). Verheyden, S., Genty, D., Cattani, O., & van Breukelen, M. R. (2008).
A 140,000-year continental climate reconstruction from two Water release patterns of heated speleothem calcite and hydrogen
European pollen records. Nature, 338, 309–313. isotope composition of fluid inclusions. Chemical Geology, 247,
Hellstrom, J. (2006). U-Th dating of speleothems with high initial 266–281.
Th-230 using stratigraphical constraint. Quaternary Geochronol- Vonhof, H. B., van Breukelen, M. R., Postma, O., Rowe, P. J.,
ogy, 1, 289–295. Atkinson, T. C., & Kroon, D. (2006). A continuous-flow crushing
Kim, S.-T., & O’Neil, J. R. (1997). Equilibrium and nonequilibrium device for on-line delta H-2 analysis of fluid inclusion water in
oxygen isotope effects in synthetic carbonates. Geochimica et speleothems. Rapid Communications in Mass Spectrometry, 20,
Cosmochimica Acta, 61, 3461–3475. 2553–2558.
Luetscher, M., Boch, R., Sodemann, H., Spötl, C., Cheng, H., Edwards, Wang, Y., Cheng, H., Edwards, R. L., Kong, X., Shao, X., Chen, S., &
R. L., et al. (2015). North Atlantic storm track changes during the An, Z. (2008). Millennial- and orbital-scale changes in the East
Last Glacial Maximum recorded by Alpine speleothems. Nature Asian monsoon over the past 224,000 years. Nature, 451, 1090–
Communications, 6, 7344. https://doi.org/10.1038/ncomms7344. 1093.
14 Air-Ground Interface: Reconstruction of Paleoclimates … 177

Woodhead, J., Hellstrom, J., Maas, R., Drysdale, R., Zanchetta, G., inferred from fluid inclusion D/H ratios of a stalagmite from
Devine, P., & Taylor, E. (2006). U-Pb geochronology of speleothems Vancouver Island. British Columbia, Canada, Geochimica Cos-
by MC-ICPMS. Quaternary Geochronology, 1, 208–221. mochimica Acta, 72, 1014–1026.
Zhang, R., Schwarcz, H. P., Ford, D. C., Schroeder, F. S., & Beddows,
P. A. (2008). An absolute paleotemperature record from 10 to 6 Ka
 Air-Interface: d18O Records of Past
Meteoric Water Using Benthic Ostracods 15
from Deep Lakes

Ulrich von Grafenstein and Inga Labuhn

of the primary signal (d18OP) on its way to through the water

Introduction
cycle (hydrological effects) and during the formation of
material in which it will be preserved (isotope fractionation
Oxygen-isotope records offer an alternative way to quanti-
effects).
tatively reconstruct paleoclimate, which, at first view, is
Lake sediments very often cover reasonably long periods
independent of an ecosystem’s reaction to climate change.
with relatively high accumulation rates, allowing sampling
The ice core records from high latitude inland ice, where the
resolutions from individual years to decades. Ostracod
former precipitation and its isotopic composition (d18OP) are
valves preserved in those sediments can, with respect to their
preserved in an almost original state, are widely accepted as
dominance and habitat, be considered as the lacustrine
valuable sources of paleoclimate information. Worldwide
equivalent to benthic foraminifera in marine sediments (De
systematic observation of d18OP (Rozanski et al. 1992),
Deckker 2002). As such, they are ideally suited for the
starting in the late sixties, together with the incorporation of
geochemical characterization of former lake water, including
the water isotopes in several generations of general circu-
stable isotope studies (Holmes and Chivas 2002). A major
lation models (Hoffmann et al. 1998; Jouzel et al. 1987) and
advantage they have compared to other materials, such as
models of intermediate complexity have, not only increased
bulk carbonate or organic matter, is that ostracod valves can
the confidence in d18OP as a powerful paleotemperature
be relatively easily separated, cleaned (Danielopol et al.
indicator in a number of key regions, but have also
2002), and assigned to different species for which biological
demonstrated its importance as a primary hydrometeoro-
and behavioural information is available from the study of
logical parameter in regions where temperature dependence
modern materials (Horne et al. 2002). Moreover, the specific
is less evident or absent.
fractionation can be calibrated using the valves of individ-
In non-polar regions, investigations have been conducted
uals that grew and calcified under known conditions (Chivas
on lake sediment (carbonate, bulk organic matter, cellulose,
et al. 2002; von Grafenstein et al. 1999a, b; Xia et al. 1997).
diatoms), speleothems, tree rings and soil carbonate. While
The same applies to benthic molluscs belonging to the
all these terrestrial records respond to changes in the isotopic
family of sphaeriides, which frequently can be found toge-
composition of past precipitation (d18OP), a number of sec-
ther with ostracods.
ondary effects can alter the original atmospheric signal and
The focus of this chapter is on the special situation of
thus limit or exclude a quantitative interpretation. Detecting
ostracods and molluscs in the profundal sediments of deep
past changes of d18OP therefore not only provides a valuable
lakes, where seasonal and long-term temperature changes are
paleoclimate proxy, but also largely facilitates the quantifi-
almost absent and where the d18O of the local water regu-
cation of secondary effects in complex isotopic records. The
larly approaches the average of the entire water column due
problem is to find material, like polar ice, which contains
to the cold season overturn. The fossil valves of these ani-
quantitative information about the isotopic composition of
mals can be used to reconstruct precisely the isotopic com-
past precipitation and from which secondary effects can
position of the lake water (d18OL) and, depending on the
either be excluded or quantified. Those include the alteration
lake’s hydrologic setting, the isotopic composition of former
precipitation (d18OP). The first part of this chapter provides
U. von Grafenstein (&)
I. Labuhn
Laboratoire des Sciences du Climat et de l’Environnement,
an overview of published deep-lake oxygen-isotope records.
LSCE/IPSL, CEA-CNRS-UVSQ, Université Paris-Saclay, 91190 The following two parts more specifically address the pro-
Gif-sur-Yvette, France cesses implicated in the signal transfer from the atmospheric
e-mail: [email protected]

© Springer Nature Switzerland AG 2021 179

G. Ramstein et al. (eds.), Paleoclimatology, Frontiers in Earth Sciences,
https://doi.org/10.1007/978-3-030-24982-3_15
180 U. von Grafenstein and I. Labuhn

precipitation to the ostracod valve, i.e. (1) the major BP to 3000 14C-years BP, which the authors attribute to the
hydrological processes controlling the present and past links slow accumulative evaporative concentration following a
between d18OL and d18OP, and (2) the effects related to the major reduction in humidity in the region. This hypothesis is
valve formation (temperature-dependent fractionation and strengthened by the fact that the reconstructed
vital effects). Most of the data used to illustrate those pro- oxygen-isotope values of the lake water, prevailing for the
cesses are from lakes situated within a small region of last 3000 years, are close to the endpoint of isotopic
southern Germany and are the author’s published or enrichment under modern conditions. Short term excursions
unpublished material. The data was collected to support and from this trend and a substantial variability before 8000
refine the quantitative interpretation of the d18O records from 14
C-years BP are interpreted as episodic changes in the
sub-recent and late Glacial to Holocene mainly from one of lake’s water balance and water level variations. However, it
those lakes (Ammersee), which is presented in the fourth cannot be excluded that shifts of 18OP and/or d18OA (which
part of this chapter. The reason for this regional concentra- are documented in the d18O record from the Guliya ice cap
tion is the lack of case studies as comprehensive from other (Thompson et al. 1997) or changes of the bottom water
regions, combining investigation of the modern isotope temperature (the lake today is only 23 m deep) might have
hydrology, the physical limnology, and the ostracod geo- contributed to both the long-term shifts and to the
chemistry within the settings of a fossil record. The last part small-scale fluctuations of d18OL.
of the chapter aims to show how d18OP reconstruction from In Europe, early monospecific deep-lake records exist
deep lake ostracods could be further developed and exploi- from Lake Zürich (Lister 1988) and Lake Lugano (Niessen
ted to address secondary effects in more complex archives. and Kelts 1989), both have a strong glacial to interglacial
This chapter is an update of a previously published book shift but lack resolution of abrupt events during the transi-
section (von Grafenstein 2002). It incorporates major tions. Two moderately better-resolved d18OL histories of two
advances concerning our understanding of the oxygen iso- neighbouring lakes in the northern alpine foreland
tope signal preserved in ostracod valves and gives an over- (Ammersee and Starnberger See, southern Germany) (von
view of new and upcoming records in Europe. Grafenstein et al. 1992) provided evidence of climate-
induced changes of d18OP consistent with the pollen-inferred
local climate history, including a strong negative excursion
Existing Deep-Lake Oxygen-Isotope Records during the Younger Dryas. They also show systematic and
constant offsets between the d18O values of different taxa
The first published ostracod stable isotope record was from from the same sediment layer, indicative of physiologically
Lake Erie (Fritz et al. 1975). This showed a 4‰ shift at the controlled fractionation in addition to temperature-dependent
late Wisconsin to Holocene transition, which was interpreted fractionation. d18O records of monospecific ostracod sam-
as a temperature controlled change in d18OP. Later, ples from a deep-lake core from Lac Neuchâtel (Switzer-
deep-lake oxygen-isotope ostracod records from the Great land), depicted similar millennial-scale, late-glacial
Lakes demonstrated that the Late Wisconsin and Holocene variations, in addition to sudden changes in the lake’s water
re-arrangements of the hydrological pathways and episodic balance and shifts of the mean isotopic composition of the
drainage from pro-glacial lakes largely controlled the iso- input, due to the episodic connection to the Aare river sys-
topic composition of the lakes’ water (Colman et al. 1994; tem (Schwalb et al. 1994). Shallow-water, late-glacial
Dettman et al. 1995; Forester et al. 1994; Lewis and records from the Ammersee (von Grafenstein et al. 1994),
Anderson 1992; Rea et al. 1994). Although it is almost from southern Sweden (Hammarlund et al. 1999), and
impossible to extract information on d18OP from those Switzerland (von Grafenstein et al. 2000, 2013) give evi-
ostracod records, they are still excellent examples of suc- dence that the large shifts of d18O, bracketing the Younger
cessful and very useful reconstructions of relative d18OL Dryas cold period, were accompanied by relative changes in
changes. Data for the isotopic composition of modern summer water temperatures, consistent with air-temperature
ostracod valves have been reported for Lake Huron (Dett- controlled shifts of d18OP. Quantitative reconstruction of
man et al. 1995). However, whilst these data suggested that d18OP from those records remains biased by the temperature
the most important taxa showed vital offsets, they were effects, by seasonal variation of shallow water d18O, and
insufficient for these offsets to be quantified. often by unknown changes in the lakes’ water balances. The
Lister et al. (1991) presented an oxygen-isotope record best-resolved European oxygen-isotope record from
from Lake Qinghai (Qinghai-Tibetan Plateau) based on deep-lake ostracod valves is from Ammersee (von Grafen-
measurements of Limnocythere inopinata and Eucypris stein et al. 1996, 1998, 1999a, b), and the accompanying
inflata, which alternately dominate the lake’s benthic ostra- hydrological and isotope-geochemical calibration (von
cod assemblage throughout the past 15,000 years. The Grafenstein et al. 1999a, b) will be used below as an
record shows an overall increase of 6‰ from 8000 14C-years example to discuss the possible effects in more detail.
15 Air-Interface: d18O Records of Past Meteoric Water Using … 181

A lower-resolved record from Lake Constance (Schwalb times of water within the catchment and in the lake are long
2003) is in good agreement with the Ammersee record. enough, so that the effective input into the lake carries an
A low-resolution record also exists from Lake Geneva average isotope signal over several years. This average is
covering the Holocene. More recently, the first results from a often close to the average of d18OP, but might be offset by
long-term project aiming to produce a new high-resolution preferential losses of a part of the precipitation by evapo-
record using a new sediment core from Mondsee (Austria) ration and transpiration (see below). In the special case of
show a remarkable synchrony for the negative excursion short catchment retention times combined with a short res-
during the 8.2 ky-event (Andersen et al. 2017) and over the idence time in the lake, the isotopic composition of the lake
entire glacial-interglacial transition (Lauterbach et al. 2011) water (d18OL) can respond with oscillations either in or out
Currently, a multidisciplinary team continues to work on the of phase with the seasonal d18OP cycle.
complete high-resolution d18OP records from Ammersee
(Germany), Mondsee (Austria) and Lac d’Annecy (France)
for the last 15,000 years. Catchment Effects

The term ‘catchment effect’ has been introduced and dis-

Hydro-Meteorological Effects cussed in detail by Gat et al. (1995). It can be defined as the
sum of effects from all hydrological processes occurring in
The Isotopic Composition of Atmospheric the drainage basin that might alter the relation between long
Precipitation term d18OP and d18OL. Besides the ‘catchment effect’ sensu
stricto, which is the measurable difference between the iso-
The isotopic composition of atmospheric precipitation at a topic compositions of the drainage basin runoff d18OI and
given place and time is controlled by various fractionation d18OP (DP−I), we also have to consider those processes that
effects, including (1) the moisture formation in the (mainly control the amount of runoff (I) and the spatial and temporal
oceanic) source regions, (2) loss through precipitation along variability of both I and d18OI.
the moisture transport pathways, (3) re-evaporation and Potentially responsible for a significant DP−I are losses via
(4) the condensation and precipitation process at the sam- evaporation or sublimation, which can lead to an isotopic
pling site. Thirty years of monthly observation at meteoro- enrichment of the remaining water. The importance of these
logical stations around the world show that, in high to mid losses can be estimated by comparing the deuterium excess
latitudes, losses of the original moisture content along hor- of the runoff with that of the mean precipitation. Usually
izontal temperature gradients seem to prevail, leading to the they are small in humid climates and they depend on the
existence of an overall positive correlation between d18OP proportion of open-water surfaces within a drainage basin.
and local air temperatures on seasonal and inter-annual time Alternatively, non-fractionating losses such as the uptake
scales as well as geographically (Dansgaard 1964; Rozanski and consequent transpiration and evaporation of water by
et al. 1992). In low latitudes, a negative correlation between plants, or the recharge of deep groundwater, can be selective
d18OP and the amount of precipitation is observed and is against a portion of the annual precipitation and thus can
explained by the vertical temperature gradients. These lead to a relative shift of d18OP. This effect theoretically
climate-d18OP relations are, however, overlain by noise due could be greater than 1‰, if the evapotranspiration is
to high short-term variability, reflecting the effects of the restricted to the summer period and only summer precipi-
different condensation processes and the admixture of vapor tation is taken up. In practice, the largest part of water for
from different marine or continental sources during a single evapotranspiration comes from the soils, which retain sig-
precipitation event, and through spatial variability, due to nificant amounts of water over the year and where the sea-
relief effects. Evidently, the quantitative relation between sonal variation of d18OP and, in consequence, the effect on
d18OP and climatic parameters, like temperature and pre- DP−I is considerably reduced.
cipitation, has to be checked for each basin before conclu- Stronger effects are to be expected in high-altitude drai-
sions in terms of paleoclimate can be drawn from a record. nage basins, including those in high alpine areas, where
However, even if such a relation is weak or absent, the build-up of glaciers can retain significant amounts of
reconstructed d18OP itself remains a valuable information isotopically-depleted winter precipitation and thus shift
source for paleoclimate analyses and for the quantitative d18OI to more positive values. Glacier growth is a relatively
interpretation of isotopic records with strong secondary long-term process, which takes place over tens to hundreds
effects. of years during relatively cold periods, whereas ablation can
It is important to be aware of the seasonal variations of be almost spontaneous with a consequent discharge of
d18OP, which are large even with respect to the greater shifts isotope-light water easily exceeding the amount of yearly
recognised in paleo-records. In many cases, the retention retention during glacier growth. The isotopic composition of
182 U. von Grafenstein and I. Labuhn

runoff from partially glaciated areas may therefore be biased groundwater, DP−I is less than +0.3‰ for the present-day
to more positive values compared to d18OP during cold conditions in the Ammersee drainage basin. A small
phases with glacier growth, and marked by strong negative decrease in the deuterium excess of the mean runoff (8‰)
spikes during subsequent warm periods. compared to that of the precipitation over the last decade
Environmental changes including climate shifts can (10‰) might indicate that this probable catchment effect in
increase or decrease the runoff without significantly affecting the case of the Ammer river is due to surface evaporation
DP−I. The most evident is a change in precipitation, but (from some smaller lakes in the basin).
evapotranspiration and groundwater recharge and discharge In summary, catchment effects, even if they are seemingly
can also change either in concert with precipitation, or small as in our example, have to be at least considered as a
independently, as a response to climate change, vegetation possible reason for changes of d18OL in the past. Quantifi-
change and human activity. Such variations of the runoff will cation is however rather complicated and has to be based on
alter the lake water balance and thus the link between d18OP a proper description of the modern geomorphologic condi-
and the isotopic composition of lake water d18OL by an tions and the vegetation cover of the drainage basin.
amount dependent on the hydrological sensitivity of the lake
(see below). Changes in the seasonal runoff characteristics
can affect the link between d18OP and d18OL. These may Lake Water Balance Effects
occur without any measurable changes in the annual runoff
and mean annual d18OI, by moderation of the short-term The isotopic composition of lake water is controlled by the
reservoirs (soil, snow cover). Such effects are important in amounts and the isotopic compositions of the inflowing
lakes with water residence times of close to or less than one water (I, I) and loss via evaporation (E, E). In steady state, i.e.
year, but may also be visible in lakes with longer response if the environmental conditions are considered as stable for
times. some time, dL approaches a constant value representing a
Probably the most efficient way to characterize the mixture between dI (weighted by I) and d* (weighted by the
catchment effects in a given basin is through a survey of the evaporation and the atmospheric water vapor deficit):
isotopic composition of both the precipitation and the river
runoff. Figure 15.1 gives an example for the Ammer River dL ¼ ðdP I þ d Eh=ðl  hÞÞ=ðI þ Eh=ðl  hÞÞ ð1Þ
(southern Germany), the main contributor to Lake Ammer-
d* is the isotopic composition of water being in isotopic
see. River water was sampled just upstream of the lake in
equilibrium with the atmospheric water vapor and thus the
two-weekly intervals for more than two years and on a daily
maximum that can be reached by evaporation:
base for the second of these two years (von Grafenstein et al.
1996). Atmospheric precipitation is collected routinely at d ¼ ðh dA þ eÞ=ðh  eÞ ð2Þ
one meteorological station (Hohenpeissenberg) within the
drainage basin and at two additional stations (Neuherberg where h is the relative humidity, dA the isotopic composition
and Garmisch), about 50 km to the north and south, of the atmospheric water vapour and e the sum of the equi-
respectively, providing monthly values for d18OP since 1970 librium and kinetic fractionation between water and vapour
AD. The oxygen isotope composition of the total two-years (Gat et al. 1994; Gibson et al. 1993) Several important facts
of runoff is −10.3‰, with a tendency to increase from with respect to the reconstruction of dP can be derived from
−10.5‰ to −10.1‰ during this period. Interestingly, the these relations: (1) the isotopic enrichment due to evaporation
two-year precipitation mean for the same period is −10.1‰, is independent of the residence time of water (I/V), but
indicating that at least a part of the runoff is from longer term depends, in addition to the atmospheric moisture conditions,
(groundwater) reservoirs. Indeed, the daily record of the on I (which is the product of P-ET, Precipitation-
second year allows the spontaneous, soil water and EvapoTranspiration, and the surface of the catchment basin)
groundwater components to be separated out with a mean and E (which is the product of evaporation per surface unit
retention time of 9 months for the soil water component and and the lake’s surface). Lakes with differing ratios between
of at least several years for the groundwater component. In the lake surface AL and drainage basin area AC will therefore
the long term, the total contribution of soil water and have significantly differing dL, even if the atmospheric con-
spontaneous runoff is about 30%. It is therefore impossible ditions including dP, dA and the evaporative flux are equal.
to define an exact estimate of DP−I without considering the (2) Any change of P-ET or E will be weighted by this factor
history of d18OP for the past few decades. The best expla- AL/AC, i.e. will induce a stronger reaction of dL for lakes with
nation for the tendency of d18OI over more than 4 years a larger AL/AC. The smaller the lake surface compared to the
(including lower frequency measurements from 1989 on) is a drainage basin, the smaller will be the influence of changing
mean age of three years for the groundwater component. hydrology on the quantitative link between dI and dL (and
However, even if we consider a much longer retention in the consequently the link between dP and dL). (3) Comparison of
15 Air-Interface: d18O Records of Past Meteoric Water Using … 183

Fig. 15.1 a Daily runoff of the Ammer River (solid line) and the d18O line). d18O of days with runoff not exceeding 12 m3/s are connected
of monthly precipitation at the meteorological station Hohenpeis- with the thick line and supposed to represent d18O of the base-flow;
senberg (stippled line); b d18O of the Ammer River sampled daily c d18O of surface and deep water in Ammersee (redrawn from von
(continuous line) and in two-week intervals (diamonds and stippled Grafenstein 2002)
184 U. von Grafenstein and I. Labuhn

dL from three neighbouring lakes with differing AL/AC subject to significant inter-annual variability as shown in
should allow quantification of the evaporative enrichment Fig. 15.1c. Hypolimnetic water, in consequence, is the better
and, in consequence, of DP−I, P − ET, E, and h, if one stable representation of the lake’s reaction to the long-term
water isotope is used (d18O or dD), and from two lakes if both development, being ‘updated’ once a year during the lake
d18OL and dDL are available. overturn in autumn, winter, and/or spring.
Figure 15.2 gives an example of such a set of lakes With respect to dynamic effects, the selection of the
(which in part are also used below for dP reconstruction). ‘ideal’ lake for dP reconstruction is a trade-off between a
These four lakes are fed today by surface water of almost short residence time (I/V), which ensures an optimal
identical composition and are sufficiently close to assume response of dL to a change in dP, on the one hand, and a
that P − ET, h, dA, and evaporation from a surface unit are residence time that is long enough to efficiently suppress the
equal. Individual water samples, taken within the period seasonal variability of dI, on the other hand. A theoretical
from 1989 to 1994 in a dD–d18O diagram scatter around residence time of 2.7 years (as for the Ammersee) seems to
mean values from a lake, are clearly separated from the local be sufficiently long to suppress the present-day seasonal
meteoric precipitation and from lake to lake. Averages of the variability of dI. The reactivity of a lake to a change in P can
samples taken in the winter, when the lakes are well mixed, be described by
are plotted on a line with a slope of 4.9, indicating that the  
distance from the long-term average of the precipitation is dLðtÞ ¼ dLð0Þ etI=V þ dLð1Þ 1  et=V ð3Þ
due to the relative importance of evaporation from the
individual lake surface. Using Eqs. 1 and 2, d18OL mea- where dL(0) is the isotopic composition before a change, dL
surements of the four lakes can be reproduced using mean (∞) is the new steady state after the change, and t is the
climatic conditions for the region and assuming that the elapsed time since the change. For example, after a change
isotopic composition of atmospheric water vapor (dA) is in of 1‰ in d18OP, the isotopic composition of the lake will
equilibrium with the long term mean of the isotopic com- reach a value indistinguishable from the new equilibrium
position of the precipitation (d18OP). Evaporation from a (±0.05‰) after t = −ln (0.05/dL(0) − dL(∞)) * V/I, which
surface unit, as calculated from the isotope balances, is would be 8 years for the Ammersee against 63 years for the
535 mm, close to independent estimates from the water Starnberger See, the lake with the longest residence time in
balances (ca. 600 mm) and from energy balance modelling our selection. All variability of dP with higher frequency will
(580 mm). have a response in dL with reduced amplitude.
Figure 15.3 gives an example of a sensitivity test In order to minimize the overall error of a d18OP recon-
exploiting how the dL of the lakes in Fig. 15.2 will change if struction from lake isotopic records, the ‘ideal’ lake should
hydrologic conditions change (without a change of dP), such have a simple, well-defined drainage basin, which is large
as a doubling of the input (I) or a doubling of the evapora- compared to the surface area of the lake to ensure a small
tion (E). Evidently, the lake with the lowest modern evap- evaporative isotopic enrichment of its water. The lake should
orative offset is also the least sensitive to changing be holomictic with a cold hypolimnion, i.e. a water depth
hydrology, and therefore should give the most accurate exceeding 40 m, and should have a short residence time, but
estimate of dP from reconstructed dL. not much below 2 years. Most of these criteria can often be
verified based on basic field observation and literature, but
should be documented with isotopic determinations of river
Transient Changes and Dynamic Effects and lake water samples if a lake is considered to be a
potential candidate for paleoclimatological or paleohydro-
The scatter of d18OL of individual lake water samples in logical investigations. Such hydrological pre-site studies not
Fig. 15.2 is the result of temperature stratification, which only increase the significance of the interpretation of the
leads to short-term deviations from the ‘steady state.’ In isotopic records in terms of changing d18OP, but may also
summer, the warm epilimnic water body, which is efficiently help to design studies on one or more distinct hydrological
separated from the much larger hypolimnion, is both evap- effects.
oratively enriched and fed by isotopically-enriched summer
precipitation and river runoff. The first effect leads to sys-
tematic shifts parallel to the local evaporation line, whereas Isotope Geochemistry of Benthic Freshwater
the second leads to less systematic scatter along lines con- Ostracods
nection the prevailing L with the respective isotopic com-
position of the precipitation and river water. Under present Theoretically, all benthic organisms living in the hypo-
day conditions, these seasonal deviations of epilimnetic limnion of lakes and producing identifiable fossil remains
water in our examples can add up to at least 1.5‰ and are preserved in the sediments could be considered for the
15 Air-Interface: d18O Records of Past Meteoric Water Using … 185

Fig. 15.2 Lower graph: water isotope ratios (dD versus d18O) of lake data from one lake is due to temporal variability of mean runoff (4 years
water (Ammersee, blue; Pilsensee, magenta; Starnberger See, orange; of observation for Ammersee and Starnberger See) and to vertical
Woerthsee, red), precipitation (IAEA-Station Hohenpeissenberg, large differences in the water column in summer. All lakes have permanent
black crosses), and river water (Ammer, green) from the ‘Fünf-Seen-- outflow. The upper graph shows the calculated isotopic compositions of
Land’ (Five Lakes District) close to Munich, Germany. Large symbols isotopically depleted water vapor produced from the lakes (dE, labelled
mark the average of the lake samples taken during the winter period, as E), of lake water (dL, labelled as L), and of the limiting maximal
when the lakes are well mixed and effects from the thermal stratification enriched water (d*, labelled as *) governed by the mean local climatic
are absent. Note the increasing distance of lake samples from the conditions, by the mean isotopic compositions of the atmospheric water
Meteoric Water Line (MWL) due to evaporative enrichment, propor- vapor (dA, labelled as A) and precipitation (dP, labelled as P). Colours
tional to the respective evaporation to input ratios (E/I). Scattering of the for the different lakes are the same as for the lower graph

reconstruction of the isotopic composition of former lake 1992). Contamination by littoral fauna are also detectable
water. However, at present, studies have been restricted to through the occurrence of species typically restricted to
valves of ostracods and small bivalve molluscs belonging to shallow water. The existence of fossil fauna representing the
the genus Pisidium. One reason for this choice is that these former life assemblage (De Deckker 2002) is itself an
organisms build protective shells or valves from CaCO3, excellent paleolimnological indicator of sufficient oxygen in
which are relatively easy to separate from the rest of the the hypolimnion and thus for a regular winter overturn of the
sediment. In addition, we now have the capacity to measure water column, which is needed for the transfer of the d18OP
oxygen and carbon isotope ratios in quantities of carbonate signal into the hypolimnion.
as small as 10 lg (Chivas et al. 1993; von Grafenstein et al.
186 U. von Grafenstein and I. Labuhn

Fig. 15.3 Reaction of d18OL to

changes in the water balance
(same colours as for Fig. 15.2).
The modern isotopic
compositions are labelled as
‘act.’, those resulting from to
doubled input as ‘I*2’ and those
from doubled evaporation as
‘E*2’. Note that the d18OL of
Ammersee remains within a range
of 1‰ for such hydrologic
extremes, in contrast to the much
stronger reaction of the d18OL of
Wörthsee (almost 3‰)

Vital Effects reflects the oxygen isotopic composition of the sum of

[HCO3−] and [CO32−]. For carbonate-dominated freshwater
However, the formation of the protective calcitic ostracod lakes with a pH of around 8.3 the contribution of [CO32−] to
valves is a physiological process, which may affect the this sum is negligible. The isotopic composition of ostracods
oxygen isotopic composition of the valves through system- valves in these lakes therefore reflects almost exclusively the
atic differences compared to an inorganic calcite formed in isotopic composition of [HCO3-] which in turn depends
isotopic equilibrium. Indeed, such ‘vital effects’ have been solely on d18OL and the water temperatures, whereas, in
postulated based on evidence from constant differences in waters with high and variable salinity, the variable contri-
valve d18O between different taxa in deep lake cores from bution of [CO32−] leads to significant negative excursions in
Starnberger See and Ammersee (von Grafenstein et al. the d18O of the ostracod calcite, thus possibly influencing the
1992). These effects have been quantified for some common reconstruction of the host water d18O. This new calcification
European freshwater species by systematic field observation model nicely explains the apparent and up-to-now still
and collection (von Grafenstein et al. 1999a, b), and by enigmatic, positive offset of valve d18O with respect to an
laboratory cultivation for the North American species Can- ‘equilibrium calcite’ in almost all freshwater settings. It
dona rawsoni (Xia et al. 1997) and the Australian euryhaline could also be the key to understanding the significant dif-
nectic ostracod Australocypris robusta (Chivas et al. 2002). ferences of ‘vital offsets’ within taxonomic groups. For the
The result of the field observations in Ammersee and deep freshwater lake situation, the correction for the
Starnberger See have been confirmed by a similar one-year empirical ‘vital offset’ and for the temperature-dependent
observation in Lake Geneva (Decrouy et al. 2011). However, fractionation between the d18O of ambient water and a cal-
field and laboratory studies on ostracods from sites with cite in isotopic equilibrium should still give an excellent
variable salinity (Li and Liu 2010) and host water pH measure of the d18OL, if the water temperature during the
reporting (Chivas et al. 2002; Marco-Barba et al. 2012) formation of the valve is known, or can be established.
indicate that the ‘vital effect’ might be influenced by host
water conditions. Devriendt et al. (2017) therefore compiled
all data from published ‘calibration’ studies (Chivas et al. Water Temperature Effects
2002; Xia et al. 1997; von Grafenstein et al. 1999a, b;
Decrouy et al. 2011; Li and Liu 2010; Didié and Bauch Holomixis (the full overturn of the water column) of deep
2002; Keatings et al. 2002; Van der Meeren et al. 2011; freshwater lakes occurs when the temperature of the entire
Bornemann et al. 2012) and suggested that ostracod calcite water column approaches the density maximum of
15 Air-Interface: d18O Records of Past Meteoric Water Using … 187

freshwater at 4 °C. The exact temperature for the overturn of summer and winter produced valves (after correction for
might, however, deviate slightly from 4 °C, if the wind vital offsets) are common (von Grafenstein et al. 1994;
forcing during the holomixis is strong enough to overcome 1999a, b, 2013; Dettman et al. 1995). However, a problem
the density gradients. Nevertheless, the possible temperature with such deep-lake studies is the low abundance of ostra-
range of 3–5 °C is relatively small. In regions where air cods compared to littoral sites; this is most probably related
temperatures of the coldest month are equal to, or below, 4 ° to the combination of low population densities of the dif-
C, positive deviations from 4 °C are restricted to a short ferent species with relatively elevated sediment accumula-
period during the onset of the holomixis, and negative tion rates (in average *1 mm per year in core AS96-1 from
deviations are limited to the winter and to cases of a con- Ammersee). High-resolution d18O records approaching the
tinuing wind-forced overturn. Most probably, the deep water hydrologic resolution (ca. 8 years/sample) are therefore only
is at 4 °C at the moment of stabilisation due to spring possible if 10 to 20 valves of juveniles (instars A-5 to A-2)
warming. During summer, the hypolimnion is efficiently of the most abundant species Fabaeformirscandona levan-
insulated from irradiative warming. Small temperature deri, F. tricicatricosa or Candona candida are grouped
increases up to 6 °C are, however, observed in lakes in together to produce samples of >10 lg calcite. This mixing
regions with high geothermal gradients. Warmer of species and instars is not problematic, however, as the
bottom-water temperatures occur in meromictic lakes, in studied European Candonidae share identical vital offsets for
which the winter overturn only reaches down to a limited d18O.
depth and in which a deeper and denser water body (mon- The error of d18OL reconstruction from Ammersee deep
imolimnion) exists. However, as mentioned above, no ben- lake ostracod d18O, is about ±0.2‰. The error of the d18OP
thic in situ ostracod and mollusc fauna will be found there calculation, introduced by assuming that evaporative
due to the absence of oxygen in the monimolimnion. In enrichment and catchment effects were as today, could, in
order to further minimize any error in the reconstruction of times of extreme hydrological conditions, range from
d18OL from benthic carbonate fossils, bottom-water tem- −0.3‰ to +0.6‰.
peratures of the respective lakes should be followed over
several seasons and compared with those simulated by
energy- and water-balance models driven by observed cli- The Fossil Ostracod Record
mate conditions for the same period. The models, validated
in this way, can then be used to estimate bottom water Calibration Against the Instrumental Air
temperatures and thus provide error estimates for d18OL for a Temperature Record
large range of climate conditions. For Ammersee and Lac
d’Annecy, two of the few lakes which should be excellently Despite the efforts to understand and quantify the transfer of
suited for quantitative reconstruction of d18OP, this mod- an atmospheric d18OP signal into a sedimentary archive, two
elling approach was used, confirming deep water tempera- points of significance with respect to paleoclimatic studies
tures astonishingly constant at 4 °C for a very wide range of could not be addressed due to the short period of the relevant
climate conditions, colder than today for Ammersee, but field observation. The first concerns the relationship between
with the risk of meromixis in warmer climates, especially mean annual d18OP and mean annual air temperature based
when winter mean daily temperatures remain significantly on regional and European-wide inter-annual variability of
above 4 °C. In contrast, Lac d’Annecy continues to be the of the last three decades (Rozanski et al. 1992). This
episodically meromictic during modelled warmer periods, comparison had to be extended to at least the duration of
because of its higher transparency during the warmest existing instrumental records, to maximize the range of
summer month, allowing deep water to reach the warmer observed temperature changes and to meet the standards for
winter minimal air temperatures within a couple of years the calibration of other paleotemperature proxies. The sec-
(Danis et al. 2003, 2004) ond open point was the overall reactivity of the coupled
In our example from the Ammersee, the average tem- drainage basin-lake system to changes in d18OP, where,
perature at 80 m over the last 20 years (1980–2000, data especially, the average retention time of the slowest runoff
provided by Dr. B. Lenhardt, WWA Weilheim) was 4.15 ° (groundwater) could only be roughly estimated to *3 years
C, with a standard deviation of 0.45 °C, and extremes of from a tentative match between a four-year shift of d18OL
3.2 °C and 5.0 °C. With such a narrow range of water and long-term averages of d18OP.
temperatures and the relatively constant d18OL of the Both concerns could be addressed by establishing a
hypolimnion (see Fig. 15.1c), the preferred moulting and 300-year long ostracod-derived record of d18OL from the
calcification period of the ostracods and their instars is uppermost 120 cm of sediment in 80 m water depth in
almost irrelevant. This is in strong contrast to sites within the Ammersee (von Grafenstein et al. 1996). The age-depth
epilimnion, where differences of up to 3‰ between the d18O model of the core was based on clearly identifiable annual
188 U. von Grafenstein and I. Labuhn

lamination (Alefs et al. 1996; Czymzik et al. 2013) and the 8200 years BP in Europe and Greenland and its probable
time represented by a sediment sample 1 cm thick was on forcing by the collapse of the Hudson Bay ice dome. The
average 2.5 years. A first, direct comparison of the second (von Grafenstein et al. 1999a, b) makes a comparison
youngest 200 years of the d18OL with the adequately between d18OP in Europe and in Greenland over the period
averaged mean annual air temperature (MAAT) record from between 15,000 and 5500 years B.P. The most striking
Hohenpeissenberg (within the catchment of the lake) gave a feature is probably the great similarity in the records, even at
reasonable correlation (r2 = 0.91), but an apparent d18OL/ high frequencies, providing evidence that the climate of both
MAAT sensitivity, which, with 0.38‰/°C, was significantly regions (and most likely of the entire North Atlantic
below the d18OP/MAAT sensitivity derived from three perimeter) experienced the same decadal variations, gov-
decades of observation in the region and for the rest of erned by the variability of the heat flux from the North
Europe (0.58‰/°C). In addition, all d18OL values younger Atlantic Ocean. Independent confirmation of the details of
than 1920, while still showing the same temperature- the d18O from ice in Central Greenland in a European record
dependence, were systematically displaced by −0.15‰ with significantly increases confidence in the quantitative inter-
respect to the older part of the record. The smaller- pretation of changes of the regional air temperatures and
than-expected amplitude of d18OL can clearly be attributed helps to exclude alternative explanations such as significant
to oversampling of the record compared to the reactivity of changes in the prevailing water vapor sources for precipi-
the entire retention system and, in a next step, can be used to tation in both regions.
better quantify the catchment residence time. The abrupt Like the 8.2-ka-event, the short, abrupt, cold oscillations
relative shift of d18OL is an indication of a change of the during the relatively warm periods of the late Glacial and
d18OP-d18OL difference or of the catchment effect in its early Holocene were probably forced by cataclysmic
wider sense, related to the regulation of the Ammer River, fresh-water discharges into the Atlantic Ocean as a conse-
which started in 1920 and ended in 1922. This regulation quence of the disintegration of the continental ice sheets.
was designed to accelerate the runoff of storm-related flood Despite the high correlation at high frequencies, the quan-
events (which occur primarily in summer) and to allow titative comparison of both d18OP records reveals periods of
cultivation of the extended river plains. Thus, after the reg- systematic change in the differences between Europe and
ulation, a larger portion of isotopically-enriched summer Central Greenland around the Younger Dryas cold period.
water was transferred directly into the lake, was mixed into These might indicate slow systematic changes in the surface
the epilimnion, and became partially lost via the outlet, conditions and circulation in the Greenland-Norwegian Sea,
whereas before regulation, those flood waters could infiltrate governed by the persisting meltwater flux from the Scandi-
into the river plain aquifers and more efficiently change the navian ice sheet during the relatively warm periods. The
longer term d18OI and, in consequence, the d18OL. proposed mechanism (von Grafenstein et al. 1998) is also an
After correction for this ‘summer bypass effect’, the entire attractive hypothesis for ‘Dansgaard-Oeschger-events’,
d18OL record was compared to the d18OL calculated using a occurring frequently during the period from *80,000 to
very basic lake model, by assuming a linear relation between *25,000 yr B.P.
d18OP and the measured air temperature and by taking into Figure 15.5 shows an overview of the state of d18OP
account changes in the seasonal distribution of precipitation reconstruction from Ammersee deep-lake ostracods. The
and the most obvious features of lake water mixing. Fig- lack of resolution for the last 5500 years is evident, even if
ure 15.4 shows the results of this modelling approach, which the available data already show some similarity with the
gives the best fit between calculated and observed d18OL, if corresponding part of the d18OP records from Greenland.
the mean drainage basin retention time is set to 3 years and The highest resolved parts of the record provide probably the
if the temperature dependence of d18OP is assumed to be best constrained and resolved d18OP record existing in
0.58‰/°C, thus confirming the assumption that the corre- Europe. However, there is still uncertainty ranging between
lation based on direct observation during the last decades +0.6‰ and −0.3‰ from potential hydrological changes that
was valid for the last two centuries. have not been considered. In addition, the record might not
be representative of the entire continent, at least during the
late Glacial, when stronger and highly dynamic longitudinal
The Record of d18OP in Central Europe Over climate gradients were very likely to have prevailed. Both
the Past 15,000 Years the errors from hydrological change and those from regional
European climatic gradients may partially explain the
The longer term history of d18OP reconstructed from apparent cross North Atlantic differences. Another problem
Ammersee deep-lake cores has been discussed in two pub- of the existing record from Ammersee is the quality of the
lications, the first of which (von Grafenstein et al. 1998) age model which is based on the few radiocarbon mea-
concentrates on the synchrony of a cold event around surements of macro plant and insect remains found in the
15 Air-Interface: d18O Records of Past Meteoric Water Using … 189

Fig. 15.4 Comparison of the d18O record of Lake Ammersee for the time. The lower box compares the slope of the input relationship
last two centuries derived from the d18O of juvenil Candona (diamonds) with the slopes of the modelled (open circles) and the
(d18OL(from C.sp), crosses) compared to the d18OL as predicted by a ostracod-derived d18OL-MAAT relation. Air-temperature variations
two-box isotope hydrology model (d18OL (model), open circles), if the dominate both the modelled and observed lake responses. The slopes of
lake was supplied with an inflow of river water with an isotopic the relationship of both the modelled and observed lake response to air
composition (d18OI (model), diamonds), controlled by the mean annual temperatures are decreased compared to the d18OP—temperature
air temperature (MAAT) (using d18OP = 0.58‰* MAAT—14.48‰ as relation, due to the incomplete reaction of d18OL to the fast climate
d18OP–MAAT relation). The upper graph shows the development over changes (von Grafenstein et al. 1996)
190 U. von Grafenstein and I. Labuhn

sediment core. We therefore decided to take a new sediment of Pisidium from three cores taken in Ammersee at 6, 7 and
core at a position close to that of core AS96-1, using the 11 m modern water depth, compared with the deep-lake
latest generation of the UWITEC coring system available in d18O measured on juvenile Candona, all corrected for the
2010 which now provides excellent material to improve the respective vital offsets. The difference between littoral and
age model by combining radiocarbon dating and varve profundal d18O values should give the mean water temper-
counting for large parts of the Holocene (Czymzik et al. ature for the shell growth period of Pisidium, i.e. summer (ts)
2013) and through radiocarbon dating and tephrochronology at the littoral sites according to ts = 4 °C − (d18OCs −
18
for the Late Glacial period. Sample treatment and the OPi)/(0.23‰/°C) (von Grafenstein et al. 1999a, b). For the
selection of monospecific shell material from the last sites at 6 and 7 m, summer water temperatures are rather
5500 years are ongoing. More highly-resolved and similar and range between 6 and 12 °C for the period from
well-dated records will come from Mondsee (Austria) and 14.8 to 12.7 ky B.P. (Bølling/Allerød), i.e. on average 4 °C
Lac d’Annecy (France). For both records, all necessary colder than today’s 13 °C at this water depth in Ammersee.
monospecific carbonate samples have been prepared and are During the Younger Dryas they are 6 °C, i.e. −7 °C with
currently being analysed. A new sediment core from Lago respect to modern values. Summer water temperatures reach
d’Iseo (Italy) will provide a d18OP record for the southern present-day values immediately after the end of the Younger
central Alps. However, this record will be discontinuous due Dryas. For the core from 11 m water depth, the early
to the lack of preserved carbonate material in parts of the Holocene summer water temperatures are also close to the
sequence. respective observed modern equivalent (9 ± 3 °C), in con-
trast to the 6 ± 2 °C during the Younger Dryas. Together,
these results indicate a strong reduction in the stability of the
Late Glacial and Early Holocene Shallow-Water thermal summer stratification of the water column due to
Temperatures considerably colder air temperatures during the Younger
Dryas.
As shown above, the d18O of ostracods and molluscs living The summer water temperatures calculated from the dif-
in shallow water carries combined information on the d18OL ferences between valves from littoral and profundal benthic
and the water temperature at the moment of calcification. organisms are an independent confirmation of the climatic
Some of the species and their instars have a seasonal pref- significance of the changes in d18OP. They provide one of
erence for calcification. Therefore, it should be possible to the rare temperature reconstructions for the late Glacial in
calculate the water temperature of this time interval from Central Europe, which is derived from a physicochemical
d18O of fossil valves, if d18OL for the given deposition time process, such as the temperature-dependent oxygen isotope
is known. Figure 15.6 shows, for example, the d18O records fractionation between water and calcite. Theoretically,

Fig. 15.5 Summary diagram showing the state of the d18OP recon- P. to *1700 AD) where further condensation is indicated. The high
struction from Ammersee deep-lake ostracods from AS Tmax (von correlation with the d18OP record from Central Greenland (grey line)
Grafenstein 2002), AS 92-5 (von Grafenstein et al. 1998), and AS96-1 that includes decadal events, provides strong evidence for a common
(von Grafenstein et al. 1998) cores. For sections sampled and measured control mechanism of the climate variability in both regions, probably
with the optimal resolution (*10 years/sample), d18OP is traced by the via the North Atlantic thermohaline circulation
thicker black line. The thin black line indicates the interval (5500 B.
15 Air-Interface: d18O Records of Past Meteoric Water Using … 191

Fig. 15.6 Estimation of

epilimnetic summer water
temperatures in Ammersee by
comparing the d18O of deep-lake
ostracods (represented by d18OL,
black line, data from (von
Grafenstein et al. 1999a, b) with
d18O of littoral Pisidium sp. in
cores taken at water depths of
11 m (crosses), 7 m (filled
circles), and 6 m (open circles)
(von Grafenstein et al. 1994) The
vital offsets and a systematic
offset between epilimnetic
summer d18OL and hypolimnetic
d18OL (0.75‰) are taken into
account

similar calculations could be made for all species and their temperatures were very close to 4 °C. In addition, the spe-
instars present in shallow water cores and so provide water cies selected for isotope measurement are identical for both
temperature information over the entire seasonal cycle and lakes, allowing direct comparison of their d18O values as
over a wide range of water depths. Because these tempera- representative of the d18OL history of the lakes. The only
ture calculations are based on the differences alone, they are restrictions for a direct comparison are the differences in the
essentially independent of the respective d18OL and therefore temporal resolution of the records and the much smaller
an attractive alternative among lake-based measurements, reactivity of the Starnberger See to changes in d18OP. One
where the d18OL itself is dominated by hydrological effects sample from the Starnberger See record represents the
or in regions where the temperature-dependence of d18OP is average d18OL over a period of several hundreds of years,
weak or absent (as for example in the north-American Great compared to about ten years for an Ammersee sample.
Lakes or in closed basins like Issyk-Kul). Similarly, the modern isotopic composition of Starnberger
See water is integrated into the d18OP history of at least the
last 63 years, in contrast to about 8 years for the Ammersee.
Quantification of Hydrological Effects In order to get a meaningful ‘modern’ lake differential based
on the same d18OP reference, we calculated a longer term
Above, we have shown that the sensitivity of d18OL to d18OL average for the Ammersee since *1930 from the
hydrological changes is dependent to a large extent on the ostracod record of the short gravity core TMAX and from
ratio between the lake’s surface and the drainage basin area the modelling of d18OL, which increase the reference lake
(AL/AC), as this controls the relative importance of the offset to 2.5‰ compared to the apparent 2.0‰ offset in
evaporation (E/I). We also mentioned that the quantitative 1994. This is to a large extent due to the fact that Starnberger
comparison of d18OL records from neighbouring lakes with See water could not follow the rapid 0.5‰ increase of d18OP
significantly differing AL/AC would allow detection of between 1990 and 1994.
changes of the runoff (P-ET), the evaporation (E), and/or the The ‘modern’ lake offset seems to be maintained or has
relative humidity (h). These hydrological changes would slightly increased by up to 0.3‰ for the last 7000 years and
lead, in contrast to a changing d18OP and to deviations from for the Late Glacial (except for the Younger Dryas, which is
the present day’s d18OL separation of the lakes. not represented by ostracod valves in the core from Starn-
Figure 15.7 shows an example of a comparison between berger See). At the beginning of the Holocene, the difference
the deep-lake records from Ammersee with a weak isotope- was about 0.8‰ higher and approached the ‘modern’ offset
hydrological sensitivity (E/I = 0.04; D18OL−P = 0.75‰) and between 9000 and 7000 years B.P. Lacking further evidence
Starnberger See (E/I = 0.17; D18OL−P = 2.75‰). Similar to from another neighbouring lake or from dDL records from
Ammersee, the core from Starnberger See comes from a both lakes, we can attribute the maximal increase of the
water depth (88 m) sufficient to assume that calcification difference to a reduction in the runoff (I = P − ET) of 50%,
192 U. von Grafenstein and I. Labuhn

Fig. 15.7 Comparison of the deep-lake ostracod-derived d18OL difference for the last 7000 years and for the first warm period of the
records of the Ammersee (black line) and the Starnberger See (open Late Glacial. The offset is increased by up to 0.8‰ during the Early
circles and line, corrected for the modern d18OL-difference between Holocene (11,500 to 7000 years BP) indicating ‘dryer than modern’
both lakes). The record from Starnberger See follows the low-frequency conditions
changes of d18OL in Ammersee, with an offset very close to the modern

an increase of the evaporation from the lakes’ surfaces (E) of

100% or a reduction of the relative humidity from 0.74 to Conclusions
0.5. Most probably, we need to distribute the overall effect to
more moderate changes of all three parameters, which, in a Oxygen isotope records from deep-lake ostracods can provide
natural world, are closely connected and often correlated. high-resolution quantitative records of d18OP, if those lakes
A reduction in I of 12%, together with an increase in E of that are selected have (1) a small relative contribution from
12%, and a relative humidity of 0.64, would, as would an evaporation to the lake water balance, (2) a well-defined
infinitive number of other combinations, explain the increase drainage basin without evidence of major re-organisation of its
in the lake offset of 0.8‰. However, all changes of these associated fluvial system, (3) a theoretical residence time
parameters had an effect not only on the difference of d18OL between 2 and 10 years, (4) a water depth in excess of 50 m,
jn the lakes but also on the evaporative enrichment and (5) an annual complete overturn, and (6) proven preservation
d18OL–d18OP of the Ammersee, which we considered as of in situ ostracod fauna among its profundal sediments. The
constant for the d18OP reconstruction. The resulting correc- temporal resolution of d18OP from benthic deep-lake ostracods
tion of the d18OP estimate for the Early Holocene would be is limited to roughly ten years, the e-folding time of lakes with
−0.6‰ in the case of highly improbable isolated changes of a buffering volume large enough to suppress eventual bias
either E or I, against only −0.25‰ for our example of from the seasonal variability of d18O of precipitation and river
combined h, I, and E changes. Besides important paleohy- water. It was shown, using the example from Ammersee that a
drological information, the lake-to-lake comparison also maximal theoretical resolution can be practically obtained.
allows us to exclude errors in the d18OP reconstruction Detailed hydrological, limnological, and isotopic investiga-
arising from water balance changes for last 7000 years. For tions can provide the basis for a realistic quality control of the
the early Holocene, the d18OP-record from Ammersee is d18OP reconstruction. For the d18OP record of the last
probably up to 0.25‰ too high. A correction can be pro- 15,500 years, the maximal error from water temperature
posed using all available data points from the Starnberger effects is ±0.2‰, to which must be added uncertainty from
See, which would then largely reduce the hydrological- possible hydrological changes of –0.3‰ to +0.6‰. Correction
induced uncertainties of the Ammersee d18OP-record. In of the record using the quantification of the hydrological
future, a condensation of the Starnberger See record should effects through comparison with the more strongly affected
be considered to increase confidence in the quantitative Starnberger See further reduces the overall error to a maximum
reconstruction of d18OP from its neighbour. of ±0.4‰ for the driest period in the Holocene.
15 Air-Interface: d18O Records of Past Meteoric Water Using … 193

d18OP records provide quantitative hydro-meteorological moment, is considered as a simple, one-step function from
information, which is largely independent of the biological cold, isotopically-depleted to present-day values. They will
response to climate change and thus can be used as a refer- also facilitate quantification and climatic interpretation of
ence in multidisciplinary studies. For Central and Northwest records from less direct continental isotopic archives such as
Europe, those d18OP records can be considered as a reason- tree rings, shallow-water lake records, and stalagmites.
able proxy for the mean air temperature. A calibration of the Although lakes with the potential to provide a quantitative
d18O record from Ammersee for the 200-year long period of d18OP record are characterized by a narrow range of hydro-
instrumental climate observation confirmed the d18OP/tem- logical and geomorphological settings, there are at least ten
perature relationship derived from precipitation measure- suitable candidates identified in Europe, for which a limited
ments (0.58‰/°C) for the last three decades in Europe. dataset of water isotopes exists. Those lakes could in future
Independent qualitative evidence for colder air temperatures be exploited to provide a network of reasonably well dis-
during the Younger Dryas could be extracted from water- and tributed and highly resolved d18OP records. It is likely that at
isotope-based shallow water temperature reconstruction, least a similar number of such lakes can be found in the
which further confirms the temperature-dependence of d18OP temperate regions of North and South America and within
in Europe. However, d18OP records provide excellent quan- other formerly glaciated mountain ranges around the globe.
titative paleoclimatic information even for regions where
such a temperature-dependence is absent or weak. High
quality d18OP records are also very useful to increase the References
significance of d18O records from archives, which are more
affected or dominated by secondary isotopic effects. Deep- Alefs, J., Müller, J., & Wunsam, S. (1996). Die Rekonstruktion der
lake ostracod valves from carefully selected lakes are, at the epilimnischen Phosphorkonzentrationen im Ammersee seit 1958.
moment, the best material to obtain quantitative d18OP Gwf Wasser Abwasser, 137(8), 443–448.
Andersen, N., Lauterbach, S., Erlenkeuser, H., Danielopol, D.L.,
records for significant time periods from temperate, non- Namiotko, T., Hüls, M., et al., (2017). Evidence for
glaciated areas. As such, they merit more consideration as a higher-than-average air temperatures after the 8.2 ka event provided
climate proxy in future research strategies. by a Central European d18O record. Quaternary Science Reviews,
Beside d18OP-reconstruction, deep-lake ostracod-based 172, OP- 96.
Bornemann, A., Pirkenseer, C. M., De Deckker, P., & Speijer, R.
oxygen isotope records have been successfully used to P. (2012). Oxygen and carbon isotope fractionation of marine
unscramble the complex hydrological changes in the Great ostracod calcite from the eastern Mediterranean Sea. Chemical
Lakes Basin, and, in combination with geomorphological Geology, 310–311, 114–125.
lake level reconstruction, to provide information on the Chivas, A. R., de Deckker, P., Cali, J. A., Chapman, A., Kiss, E., &
Shelley, J. M. G. (1993). Coupled stable-isotope and trace-element
water balance development and humidity in hydrologically measurements of lacustrine carbonates as paleoclimatic indicators.
closed basins. In all cases, deep-lake ostracods provide very In P. K. Swart, K. C. Lohmann, J. Mckenzie & S. Savin (Eds.),
precise estimates for the isotopic composition of former lake Climate change in continental isotopic records. School of Marine
water, which can be used as a reference to quantify the and Atmospheric Science.
Chivas, A., De Deckker, P., Wang, S. X., & Cali, J.A.
secondary effects on d18O of all other authigenic material (2002) Oxygen-isotope systematics of the nektic ostracod Australo-
produced in the same lake, such as water temperature- cypris robusta. In J. A. Holmes, A. R. Chivas (Eds.), Ostracoda,
dependent fractionation in the case of littoral and sublittoral applications in quaternary research. Geophysical Monograph
ostracods. Series (Vol. 131, pp. 301–313). American Geophysical Union
(AGU).
Colman, S. M., Keigwin, L. D., & Forester, R. M. (1994). Two
episodes of meltwater influx from Lake Agassiz into the Lake
Perspectives Michigan basin and their climatic contrasts. Geology, 22, 547–550.
Czymzik, M., Brauer, A., Dulski, P., Plessen, B., Naumann, R., von
Grafenstein, U., et al. (2013). Orbital and solar forcing of shifts in
In future, archived and new core material from Ammersee, Mid- to Late Holocene flood intensity from varved sediments of
Mondsee, Lac d’Annecy, and Lago d’Iseo will be used to pre-alpine Lake Ammersee (southern Germany). Quaternary
close the existing gaps and to provide optimally resolved Science Reviews, 61, 96–110.
d18OP records for the entire period since the last glacial Danielopol, D. L., Ito, E., Wansard, G., Kamiya, T., Cronin, T.M., &
Baltanas, A. (2002) Techniques for collection and study of
maximum. Besides their direct application as proxies for
ostracoda. In J. A. Holmes & A. R. Chivas (Eds.), Ostracoda,
mean annual air temperatures, records of decadal d18OP Applications in quaternary research, Geophysical Monograph
variations in Europe offer many possibilities for further Series (Vol. 131, pp. 65–97). American Geophysical Union (AGU).
hydrogeological and paleoclimatological studies. They could Danis, P. A., von Grafenstein, U., & Masson-Delmotte, V. (2003) Sen-
sitivity of deep lake temperature to past and future climatic changes:
serve as ‘marker input function’ into groundwater and thus
A modelling study for Lac d’Annecy, France, and Ammersee,
allow a better exploitation of stable isotope measurements of Germany. Journal of Geophysical Research-Atmospheres, 10(D19).
fossil groundwater for the last 15,500 years, which, for the
194 U. von Grafenstein and I. Labuhn

Danis, P. A., von Grafenstein, U., Masson-Delmotte, V., Planton, S., model: The seasonal cycle for present-day conditions. Journal of
Gerdeaux, D., & Moisselin, J. M. (2004) Vulnerability of two Geophysical Research, 92, 14739–14760.
European lakes in response to future climatic changes. Geophysical Keatings, K. W., Heaton, T. H. E., & Holmes, J. A. (2002). Carbon and
Research Letters, 31(21). oxygen isotope fractionation in non-marine ostracods: Results from
Dansgaard, W. (1964). Stable isotopes in precipitation. Tellus, 16, 436– a ‘natural culture’ environment. Geochimica et Cosmochimica Acta,
468. 66(10), 1701–1711.
De Deckker, P. (2002). Ostracod palaeoecology. In J. A. Holmes & A. Lauterbach, S., Brauer, A., Andersen, N., Danielopol, D. L., Dulski, P.,
R. Chivas (Eds.), Ostracoda, applications in quaternary research. Hüls, M., et al. (2011). Environmental responses to Lateglacial
Geophysical Monograph Series (Vol. 131, pp. 121–134). American climatic fluctuations recorded in the sediments of pre-Alpine Lake
Geophysical Union (AGU). Mondsee (northeastern Alps). Journal of Quaternary Science, 26
Decrouy, L., Vennemann, T. W., & Ariztegui, D. (2011) Controls on (3), 253–267.
ostracod valve geochemistry: Part 2. Carbon and oxygen isotope Lewis, C. F. M., & Anderson, T. W. (1992). Stable isotope (O and C)
compositions. Geochimica et Cosmochimica Acta, 75(22), 7380–7399. and pollen trends in eastern Lake Erie, evidence for a locally
Dettman, D. L., Smith, A. J., Rea, D. K., Moore, T. C., & Lohmann, K. induced climatic reversal of Younger Dryas age in the Great Lakes
C. (1995). Glacial meltwater in Lake Huron during Early basin. Climate Dynamics, 6, 241–250.
Postglacial time as inferred from single-valve analysis of oxygen Li, X., & Liu, W. (2010). Oxygen isotope fractionation in the ostracod
isotopes in ostracodes. Quaternary Research, 43(3), 297–310. Eucypris mareotica: Results from a culture experiment and
Devriendt, L. S., McGregor, H. V., & Chivas, A. R. (2017). Ostracod implications for paleoclimate reconstruction. Journal of Paleolim-
calcite records the18O/16O ratio of the bicarbonate and carbonate nology, 43(1), 111–120.
ions in water. Geochimica et Cosmochimica Acta, 214, 30–50. Lister, G. S. (1988). A 15,000-year isotopic record from Lake Zürich of
Didié, C., & Bauch, H. A. (2002). Implications of upper Quaternary deglaciation and climatic change in Switzerland. Quaternary
stable isotope records of marine ostracods and benthic foraminifers Research, 29(2), 129–141.
for paleoecological and paleoceanographical Investigations Lister, G. S., Kelts, K., Zao, C. K., Yu, J. Q., & Niessen, F. (1991).
(pp. 279–299). American Geophysical Union (AGU). Lake Qinghai, China; closed-basin lake levels and the oxygen
Forester, R. M., Colman, S. M., Reynolds, R. L., & Keigwin, L. D. isotope record for Ostracoda since the latest Pleistocene. In A.
(1994). Lake Michigan’s late Quaternary limnological and climate R. Chivas & P. De Deckker (Eds.), SLEADS conference, Arkaroola,
history from ostracod, oxygen isotope, and magnetic susceptibility. South Aust. (pp. 141–162), Aug. 8–16. 62 Refs.
In: D. W. Folger, S. M. Colman, & P. W. Barnes (Eds.), South. Marco-Barba, J., Ito, E., Carbonell, E., & Mesquita-Joanes, F. (2012).
Lake Michigan Coast. Eros. Study (pp. 93–107). Woods Hole, MA, Empirical calibration of shell chemistry of Cyprideis torosa (Jones,
USA: U. S. Geological Surve. 1850) (Crustacea: Ostracoda). Geochimica et Cosmochimica Acta,
Fritz, P., Anderson, T. W., & Lewis, C. F. M. (1975). Late Quaternary 93, 143–163.
climatic trends and history of Lake Erie from stable isotope studies. Niessen, F., & Kelts, K. (1989). The deglaciation and Holocene
Science, 160, 267–269. sedimentary evolution of southern perialpine Lake Lugano—
Gat, J. R., Lister, G. S., & Frenzel, B. (1995). The ‘catchment effect’ on Implications for Alpine paleoclimate. Eclogae Geologicae Helveti-
the isotopic composition of lake waters; its importance in cae, 82(1), 235–263.
paleolimnological interpretations. In B. Frenzel, B. Stauffer, & M. Rea, D. K., Moore, T. C., Jr., Anderson, T. W., Lewis, C. F. M.,
M. Weiss (Eds.), Problems of stable isotopes in tree-rings, lake Dobson, D. M., Dettman, D. L., et al. (1994). Great Lakes
sediments and peatbogs as climatic evidence for the Holocene. paleohydrology: Complex interplay of glacial meltwater, lake
Issue ESF Proj. ‘European Palaeoclim. Man’ 10 (pp. 1–16). levels, and sill depths. Geology, 22(12), 1059.
Stuttgart/New York: Gustav Fischer Verlag. Rozanski, K., Araguás-Araguás, L., & Gonfiantini, R. (1992). Relation
Gibson, J. J., Edwards, T. W. D., Bursey, G. G., & Prowse, T. D. between long-term trends of oxygen-18 isotope composition of
(1993). Estimating evaporation using stable isotopes: Quantitative precipitation and climate. Science, 258, 981–985.
results and sensitivity analysis for two catchments in Northern Schwalb, A. (2003). Lacustrine ostracods as stable isotope recorders of
Canada. Nordic Hydrology, 24, 79–94. late-glacial and Holocene environmental dynamics and climate.
Hammarlund, D., Edwards, T. W. D., Bjorck, S., Buchardt, B., & Journal of Paleolimnology, 29(3), 267–351.
Wohlfarth, B. (1999). Climate and environment during the Younger Schwalb, A., Lister, G. S., & Kelts, K. (1994). Ostracode carbonate
Dryas (GS-1) as reflected by composite stable isotope records of d18O- and d13C-signatures of hydrological and climatic changes
lacustrine carbonates at Torreberga, Southern Sweden. Journal of affecting Lake Neuchâtel, Switzerland, since the latest Pleistocene.
Quaternary Science, 14(1), 17–28. Journal of Paleolimnology, 11, 3–17.
Hoffmann, G., Werner, M., & Heimann, M. (1998). The water Isotope Thompson, L. G., Yao, T., Davis, M. E., Henderson, K. A.,
module of the ECHAM atmospheric general circulation model—A Mosley-Thompson, E., Lin, P. N., et al. (1997). Tropical climate
study on time scales from days to several years. Journal of instability: The Last Glacial Cycle from a Qinghai-Tibetan Ice Core.
Geophysical Research, 103(16), 816–871, 896. Science, 276, 1821–1825.
Holmes, J. A., & Chivas, A. R. (2002). Ostracod shell chemistry— Van der Meeren, T., Ito, E., Verschuren, D., Almendinger, J. E., &
overview. in: J.A. Holmes, A.R. Chivas (Eds.), Ostracoda, Martens, K. (2011). Valve chemistry of Limnocythere inopinata
applications in Quaternary research, Geophysical Monograph (Ostracoda) in a cold arid environment—Implications for paleolim-
Series (Vol. 131, pp. 1–4). American Geophysical Union (AGU). nological interpretation. Palaeogeography, Palaeoclimatology,
Horne, D. J., Cohen, A., & Martens, K. (2002) Taxonomy, morphology Palaeoecology, 306(3–4), 116–126.
and biology of Quaternary and living Ostracoda. In: J. A. Holmes & von Grafenstein, U. (2002). Oxygen-isotope studies of ostracods from
A. R. Chivas (Eds.), Ostracoda, applications in Quaternary deep lakes. In J. A. Holmes & A. R. Chivas (Eds.), Ostracoda,
research. Geophysical monograph series (Vol. 131, pp. 5–36). applications in quaternary research. Geophysical Monograph
American Geophysical Union (AGU). Series (Vol. 131, pp. 249–265). American Geophysical Union
Jouzel, J., Russell, G. L., Suozzo, R. J., Koster, R. D., White, J. W. C., (AGU).
& Broecker, W. S. (1987). Simulations of the HDO and H18 2 O von Grafenstein, U., Belmecheri, S., Eicher, U., van Raden, U.J.,
atmospheric cycles using the NASA GISS general circulation Erlenkeuser, H., Andersen, N., et al. (2013). The oxygen and carbon
15 Air-Interface: d18O Records of Past Meteoric Water Using … 195

isotopic signatures of biogenic carbonates in Gerzensee, Switzer- von Grafenstein, U., Erlenkeuser, H., Müller, J., & Kleinmann-
land, during the rapid warming around 14,685 years BP and the Eisenmann, A. (1992). Oxygen isotope records of benthic ostracods
following interstadial. Palaeogeography, Palaeoclimatology, in bavarian lake sediments—Reconstruction of late and postglacial
Palaeoecology, 391(Part B OP-In Early Rapid Warning, Palaeo- air temperatures. Naturwissenschaften, 79(4), 145–152.
geography, Palaeoclimatology, Palaeoecology 1 December 2013 von Grafenstein, U., Erlenkeuser, H., Müller, J., Trimborn, P., & Alefs,
391 Part B: 25–32), 25. J. (1996). A 200 year mid-European air temperature record
von Grafenstein, U., Eicher, U., Erlenkeuser, H., Ruch, P., Schwander, preserved in lake sediments: An extension of the d18Op—Air
J., & Ammann, B. (2000). Isotope signature of the Younger Dryas temperature relation into the past. Geochimica et Cosmochimica
and two minor oscillations at Gerzensee (Switzerland): Palaeocli- Acta, 60(21), 4025–4036.
matic and palaeolimnologic interpretation based on bulk and von Grafenstein, U., Erlernkeuser, H., & Trimborn, P. (1999a). Oxygen
biogenic carbonates. Palaeogeography, Palaeoclimatology, and carbon isotopes in modem freshwater ostracod valves: Assessing
Palaeoecology, 159(3–4), 215–229. vital offsets and autecological effects of interest for palaeoclimate
von Grafenstein, U., Erlenkeuser, H., Kleinmann, A., Müller, J., & studies. Palaeogeography, Palaeoclimatology, Palaeoecology, 148
Trimborn, P. (1994). High-frequency climatic oscillations as (1–3), 133–152.
revealed by oxygen-isotope records of benthic organisms (Ammer- von Grafenstein, U., Erlenkeuser, H., Brauer, A., Jouzel, J., & Johnsen,
see, Southern Germany). Journal of Paleolimnology, 11, 349–357. S. (1999b). A mid-European decadal isotope-climate record from
von Grafenstein, U., Erlenkeuser, H., Müller, J., Jouzel, J., & Johnsen, 15,500 to 5,000 years B.P. Science, 284, 1654–1657.
S. (1998). The cold event 8200 years ago documented in oxygen Xia, J., Ito, E., & Engstrom, D. R. (1997). Geochemistry of ostracod
isotope records of precipitation in Europe and Greenland. Climate calcite: 1. An experimental determination of oxygen isotope
Dynamics, 14(2), 73–81. fractionation. Geochimica et Cosmochimica Acta, 61, 377–382.
 Vegetation-Atmosphere Interface:
Tree Rings 16
Joël Guiot and Valérie Daux

Douglas (1920) was the first to recognize the potential of

A Dendrochonological Approach
series of annual growth rings to provide information about
past climates, and he established the fundamentals of what is
In many parts of the world, there is a strong seasonality in
now known as dendroclimatology. The main difficulty is to
the annual distribution of temperatures and rainfall. This
distinguish the impact of climate from other factors. The
seasonality is reflected in the growth of trees, which is the
greater the climate stress the tree is subjected to, the easier
result of the interaction of the tree with its environment, via
this is to decode. Generally, there are two types of climate
its leaves (for carbon and water exchanges) and its roots (for
stress: heat stress and water stress. In arid or semi-arid
nutrients and water). In temperate latitudes, during the
regions, tree growth is limited by the availability of water.
winter, the tree is dormant and woody cells are not produced.
Therefore, this will be the main parameter recorded by the
In spring, when the thermal conditions are met, the tree
tree. Trees in the far north or at high altitudes are constrained
becomes active and starts to produce large, dispersed woody
by temperature. This makes them very good thermometers.
cells (early wood). Towards late spring and early summer,
The climate conditions in the months prior to the growing
the cells produced are denser and smaller (late wood), then,
season also affect the ring. These may be the replenishment
at the end of the summer, cells are no longer produced along
of groundwater reserves, or carbon reserves ready to be
the trunk and the tree starts to store reserves for the fol-
activated for the following year. This further complicates the
lowing year. If a trunk is cut through, alternating light and
decoding of climate information.
dark bands can be observed, which, combined, constitute an
The dendroclimatic approach (Cook and Kairiukstis
annual growth ring. Comparison between rings shows a high
1990; Trouet 2020) involves taking a number of cores from a
level of variability. This variability is the direct consequence
given forest, at a rate of between two and four cores per tree
of the climate conditions (temperature, rainfall, sunlight)
from ten to twenty trees per stand. A core is a small tube
which prevailed during or before the formation of the cells
taken from the bark to the middle of the tree, from which the
(Fritts 1976).
sequence of rings can be read. The trees most likely to
These seasonal growths are produced by tissues in the
provide the best information on the climate parameter to be
cambium, a layer of cells between the wood and the bark
reconstructed are chosen. For example, when reconstructing
which causes expansion in the diameter of the roots, trunk
a water parameter, trees growing in shallow soils which are
and branches of the tree. Besides climate, the thickness of
unable to store much water will be selected. Each core is
the ring also depends on many other parameters including
then dated by counting the rings from the bark to the heart of
the species, the age of the ring, the availability of nutrients in
the tree. Each ring is supposed to be annual, but sometimes
the soil, the tree’s ability to retain water, its exposure etc.
growth stops during the season because of a temporary
J. Guiot (&) drought and resumes if it starts to rain. This resulting growth
European Centre for Research and Teaching in Environmental arrest produces what is called a false ring. For this year, there
Geosciences CEREGE, Aix-Marseille University, CNRS, IRD, are two rings. In other years, conditions are so unfavorable
INRAE, Collège de France, BP 80, 13545 Aix-en-Provence Cedex
04, France
that the ring appears to be missing. These two types of
e-mail: [email protected] phenomena will produce errors in the dating of rings. Cross
V. Daux
dating, that is, the comparison of series between cores, helps
Laboratoire des Sciences du Climat et de l’Environnement, eliminate these errors.
LSCE/IPSL, CEA-CNRS-UVSQ, Université Paris-Saclay, 91190
Gif-sur-Yvette, France

© Springer Nature Switzerland AG 2021 197

G. Ramstein et al. (eds.), Paleoclimatology, Frontiers in Earth Sciences,
https://doi.org/10.1007/978-3-030-24982-3_16
198 J. Guiot and V. Daux

reconstruction because the interactions between climate and

growth are complex.
A network of master timelines for a given region needs to
be established, including different species of trees, in order
to better isolate the relevant climate variable. This is then
followed by a statistical approach called transfer function
(Cook and Kairiukstis 1990). On one side, a matrix of
dendrochronological series, necessarily of variable length, is
established, and on the other side, weather series for the
same region are assembled. Over the time period common to
both the meteorological and tree-ring series, a statistical
relationship by regression methods can be calibrated.
The following example illustrates how it is possible to
reconstruct summer temperatures in Europe from tree-ring
series, supplemented by other proxies (Guiot et al. 2010).
Among these proxies, we used series of harvest dates from
several wine regions in France and Switzerland and the
isotopic series for oxygen-18 in Greenland, which is con-
sidered to be related to the global climate. All these series are
of variable length and resolution. They also represent dif-
ferent characteristics of the climate, but the climate param-
eter that best explains tree growth and the precociousness of
the grape harvest is the temperature between April and
Fig. 16.1 Typical profile of a conifer ring (above) with the corre-
sponding density curve (below) and the main parameters that can be
September. Therefore, it is this summer temperature that was
deduced estimated using a statistical technique of similarities, called
the analog method (Guiot et al. 2010).
The graph (Fig. 16.3) shows that the Little Ice Age (the
Each ring is measured and a set of between twenty and defined span of which varies, but, here, is fixed at 1400–
forty time series is produced for each stand. As well as the 1900, and is marked by advancing glaciers in the Alps)
thickness, the density may also be measured using a was on average 0.35 °C colder than the 1961–1990 refer-
micro-densitometer (see typical profile in Fig. 16.1). One ence period, with a maximum cooling around 1600 (around
can thus obtain the thicknesses of the early wood and the −0.5 °C). The 1940–2007 period was 0.2 °C warmer than
summer wood as well as the average density of the early the reference period, and the last decade of the twentieth
wood and the maximum density. These are the most fre- century topped this with a difference of 0.5 °C. Even tak-
quently used parameters (Schweingruber et al. 1978). ing into account that these averages are calculated over
The next step is to try to reconstruct the climate signal. varying periods, the warming of the late twentieth century
Tree-ring series are often affected by trends associated with is significant. This is confirmed by the polynomial curve
the age of the tree. Statistical methods are used to eliminate indicating a warming since the early nineteenth century of
these. Various methods have been developed: polynomial 0.6 °C for the century.
curves, digital filters, exponential decay are removed from
the raw data by division. This process is called indexing. The
main problem with this is that often a part of the trend linked Dendro-isotopic Analysis
to climate is also removed. Figure 16.2 shows how one can
get an indexed series from an untreated series of rings. The Tree rings are made up of organic material containing
various indexed series from the same stand are averaged to mainly carbon, oxygen and hydrogen. Each one of these
produce a representative series for the stand, called the elements has several isotopes. The isotopes of an element
master timeline, in order to remove intra-site variability. This have chemical properties that are qualitatively identical.
is done in an attempt to maximize the climate signal. Other However, physical, chemical and biological processes can
indexing methods are increasingly used to try to best pre- bring about a fractionation between light isotopes and heavy
serve climate variations over the long-term (Cook and isotopes of the same element during physical or chemical
Kairiukstis 1990). reactions in which this element is involved. As a result, the
A single average series (also called master series) of a isotopic ratios of oxygen (d18O), carbon (d13C) and hydro-
stand is rarely able to provide a reliable climate gen (dD) are a source of environmental information.
16 Vegetation-Atmosphere Interface: Tree Rings 199

Fig. 16.2 Example of indexing

of a tree-ring series: the top graph
shows the raw data for the ring
thicknesses with the trend line
superimposed; the lower curve
shows the indexed series, that is
to say the raw series divided by
the trend

Fig. 16.3 Evolution of the

summer temperatures
reconstructed for Europe [10° O
60° E-and-30° N 85° N] taken
from multiple proxies and
smoothed. Shaded: the associated
uncertainty (at 95%); dashed line:
the corresponding observations
(1850–2007); and, bold line: the
trend line of the reconstruction
(Guiot et al. 2010)

The isotopic composition of oxygen in precipitations is a composition of the oxygen in the water of the xylem,
function of temperature at high and middle latitudes. Water through which water and mineral nutrients are transported, is
from precipitation infiltrates the soil. In the superficial levels, the same as that of the soil water absorbed by the root
evaporation can result in oxygen enrichment of the infiltrated system. Significant fractionation occurs before the water
water compared to the original precipitation. Trees take most isotopes are fixed in the wood structure (for a summary of
of their water from the superficial soil levels. Therefore, the the isotopic composition of wood constituents see Gessler
water they draw can be enriched to some extent in heavy et al. 2014). The critical phase occurs in the leaves where
isotopes compared to the precipitation. Nevertheless, there is evapotranspiration results in a loss of light isotopes. The
little or no enrichment in wooded areas where the soils are extent to which the leaf is enriched with heavy isotopes
covered by thick litter that limits evaporation. The isotopic depends on the difference between the isotopic compositions
200 J. Guiot and V. Daux

of the water in the xylem and in the ambient vapor, as well carboxylation (27‰), pi and pa are the partial pressures of
as on the difference between the vapor pressure inside and CO2 in the substomatal cavity and the atmosphere respec-
outside the leaf. As a result of the evaporative fractionations, tively. The partial pressure of CO2 within the substomatal
the isotopic composition of the glucose produced in the leaf cavity is conditioned by the opening of the stomata (result-
is enriched by 27‰ compared to the water in the leaf. ing from a compromise between water loss and uptake of
During synthesis of cellulose from glucose, about 40% of the CO2 from the ambient air). Additional isotopic fractiona-
oxygen atoms are exchanged with the xylem water. Conse- tions, accentuating the depletion in 13C, occur during the
quently, the isotopic composition of the cellulose in tree synthesis of cellulose and lignin. Plant-air exchanges are
rings reflects that of the source (soil water is more or less determined by the tree’s environment (climate, water status
equivalent isotopically to precipitations) and the degree of of the soil). The outcome is that the d13C of the cellulose in
enrichment by evaporation in the leaf. The link between the tree rings is dependent on atmospheric temperature
isotopic compositions of cellulose and rainfall is complex. (Fig. 16.5).
Nevertheless, a growing number of studies have reported The isotopic composition of hydrogen (dD) is linked to
statistically significant correlations between the d18O of the the atmospheric temperature, as is that of oxygen; it is not
cellulose in tree rings (oak, pine, larch, cedar) and that of the modified in the transfer from soil to tree, but it is affected by
rainfall during the growing season, as well as certain other evapotranspiration, which causes isotopic enrichment of the
climate parameters (atmospheric temperature, relative water in the leaf and is subjected to fractionation during the
humidity, water stress). Figure 16.4 shows examples of process of photosynthesis. During photosynthesis, enzyme
these correlations. activity causes kinetic fractionations which may differ
The d13C of atmospheric carbon is close to −8‰ (relative depending on the position of hydrogen in the glucose
to the standard Pee Dee Belemnite). That of the leaves and molecule (Augusti et al. 2006). During the transformation of
the wood in the trees is in the region of −20‰ to −30‰. The glucose to cellulose in the trunk, the catalytic action of the
isotopic fractionations which create the differences between enzymes generates isotopic exchanges between the sugars
the d13C of CO2 in the air and the d13C of CO2 in the plant, and water in the xylem which involve about 40% of the
occur primarily in the leaf. Farquhar et al. (1982) has hydrogen atoms in the sugars (Waterhouse et al. 2002). As it
expressed this in the following equation: is difficult to distinguish between the climate and physio-
logical influences on the abundance of deuterium, the use of
D13 Cð&Þ ¼ a þ ðb  aÞpi =pa ; isotopic ratios of hydrogen in cellulose is complicated
(Pendall 2000; Augusti et al. 2008).
where D13C represents the discrimination of carbon between
As for the series of ring widths, the d13C cellulose of tree
the d13C of glucose synthesized in the leaf and the CO2 in
rings can show trends linked to age. This so-called ‘juvenile’
the air, a is the fractionation due to diffusion through the
effect is characterized by low values of d13C in wood cel-
stomata (4.4‰), b is the fractionation caused by
lulose produced in the first decades of the tree (Francey and

Fig. 16.4 Linear relationships between the isotopic composition of et al. 2016), b the average maximum temperatures from April to
oxygen in the cellulose of oaks and a the Standardized Precipitation September at Fontainebleau, France
Evapotranspiration Index (SPEI) at Angoulême, France (SPEI; Labuhn
16 Vegetation-Atmosphere Interface: Tree Rings 201

Isotopic Reconstruction of the Variations

in Climate Parameters Over Time

The variations in the isotopic compositions of the cellulose

in rings are interpreted in different ways. Indeed, the
reconstructed climate parameters are, depending on the case:
temperature, relative humidity, sunshine, the amount of
summer or winter rainfall, the average isotopic composition
of this rainfall. The isotopic fractionations during the man-
ufacture of cellulose are determined, as we have seen, by
several factors. The conditions under which the growth takes
place will determine which factor is dominant. For example,
for trees in areas experiencing drought, stomatal conduc-
tance tends to dominate the carbon isotope fractionation
processes. In this case, the environmental control factors are
the relative humidity of the air and soil moisture; when, on
the contrary, trees are growing in conditions where water is
not limited, the fractionations are instead conditioned by the
rate of photosynthesis which itself depends on the incident
Fig. 16.5 Correlation between the isotopic composition of the carbon radiation and, to a lesser extent, by the temperature. The
in the cellulose of larches in the Névache region (Alps, France) and the
average maximum temperature in July–August (Daux et al. 2011) disparity in the reconstructions mentioned earlier, is a
reflection of environmental disparities. The possibility of
changes in the dominant factor over time must be contem-
Farquhar 1982). In most cases, the isotopic compositions of plated when reconstructing temporal variations in the cli-
the oxygen and hydrogen in the cellulose are not affected by mate parameters.
the juvenile effect, which makes these parameters particular Two reconstructions of atmospheric temperature are
interesting for the reconstruction of low-frequency climate presented below, one from the d13C of the cellulose in
variations. Fitzroya cupressoides (Patagonia cypress, Argentina) cov-

Fig. 16.6 Evolution of summer temperatures (December–February) over the last two centuries in northern Patagonia (Argentina), calculated from
the isotopic composition of oxygen in the cellulose of Patagonian cypresses
202 J. Guiot and V. Daux

Fig. 16.7 Values for the d13C of

cellulose taken from the summer
wood of forest oaks (Quercus
robur) and from buildings in
Rennes (France) over the period
1610–1996 AD. The values of the
reconstructed temperatures are
indicated on the scale on the left.
Instrumental temperatures,
measured by Météo-France, from
1890 to 2003, are shown for
comparison (Masson-Delmotte
et al. 2005)

ering the period 1800–2011 (Lavergne et al. 2017) Daux, V., Edouard, J. L., Stievenard, M., Mestre, O., Guibal, F.,
(Fig. 16.6) and the other from the d18O of the cellulose of Masson-Delmotte, V., & Thomas, A. (2011). Ring width, and
carbon and oxygen Isotopic Composition of the cellulose in Larix
Quercus (forest oaks and beams from old buildings in the decidua as climatic proxies: A case study in the French Alps. Earth
Rennes region covering the last 400 years (Raffali-Delerce and Planetary Science Letters.
et al. 2004; Masson-Delmotte et al. 2005) (Fig. 16.7). The Douglass, A. (1920). Evidence of climatic effects in the annual rings of
relationship between isotope data and temperature was Trees. Ecology, 1(1), 24–32.
Farquhar, G. D., O’Leavy, M. H., & Berry, J. A. (1982). On the
established through calibrations carried out over the 1931– relationship between carbon isotope discrimination and intercellular
2011 period for the cypress from Patagonia, and over the carbon dioxide concentration in leaves. Australian Journal of Plant
1951–1996 period for the oaks in Brittany. For Lavergne Physiology, 9, 121–137.
et al. (2018), the evolution of d13C in the cellulose of the Francey, R. J., & Farquhar, G. D. (1982). An explanation of 13C/12C
variations in Tree rings. Nature, 297, 28–31.
cypresses implies that the summers (December to February) Fritts, H. C. (1976). Tree-rings and climate (p. 567). New York:
of the nineteenth century, particularly in the second half, Academic Press.
were cool (an average temperature of 13.4 °C) and the Gessler, A., Ferrio, J. P., Hommel, R., Treydte, K., Werner, R. A., &
summers of the twentieth and twenty-first centuries have Monson, R. K. (2014). Stable isotopes in tree rings: towards a
mechanistic understanding of isotope fractionation and mixing
higher temperatures (average temperature of 14.2 °C). processes from the leaves to the wood. Tree Physiology, 34, 796–
The reconstruction produced from the d18O of the cellu- 818.
lose of the oaks in Brittany also shows an increase in sum- Guiot, J., Corona, C., & ESCARSEL Members. (2010). Growing
mer temperatures from the end of the nineteenth century season temperatures in Europe and climate forcings over the last
1400 years. PLoS-one, 5(4), e9972. https://doi.org/10.1371/journal.
onwards, in line with thermometer data. The similarity in the pone.0009972.
general shape of the graphs of the oak d18O and the Labuhn, I., Daux, V., Girardclos, O., Stievenard, M., Pierre, M., &
instrumental temperatures shows that the dendro-isotopic Masson-Delmotte, V. (2016). French summer droughts since 1326
parameter can be used with a high degree of confidence to CE: a reconstruction based on tree ring cellulose d18O. Climate of
the Past, 12, 1101–1117.
reconstruct climate trends on a multi-year scale. Lavergne, A., Daux, V., Villalba, R., Pierre, M., Stievenard, M., &
Srur, A. M. (2017). Improvement of isotope-based climate recon-
structions in Patagonia through a better understanding of climate
References influences on isotopic fractionation in tree rings. Earth and
Planetary Science Letters, 459, 372–380.
Lavergne, A., Daux, V., Pierre, M., Stievenard, M., Srur, A. M., &
Augusti, A., Betson, T. R., & Schleucher, J. (2006). Hydrogen Villalba, R. (2018). Past summer temperatures inferred from
exchange during cellulose Synthesis Distinguishes climatic and dendrochronological records of on the eastern slope of the northern
biochemical Isotope fractionation in tree rings. New Phytologist, patagonian andes. Journal of Geophysical Research: Biogeo-
172, 490–499. sciences, 123(1), 32–45.
Augusti, A., Betson, T. R., & Schleucher, J. (2008). Deriving correlated Masson-Delmotte, V., Raffali-Delerce, G., Danis, P., Yiou, P.,
climate and physiological signals from deuterium isotopomers in Stievenard, M., Guibal, F., et al. (2005). Changes in European
tree rings. Chemical Geology, 252, 1–8. precipitation seasonality and in drought frequencies revealed by a
Cook, E. R., & Kairiukstis, L. A. (Eds.). (1990). Methods of four-century-long tree-ring isotopic record from Brittany, Western
dendrochronology: Applications in the environmental sciences. France. Climate Dynamics, 24, 57–69.
Boston, MA: International Institute for Applied Systems Analysis, Pendall, E. (2000). Influence of precipitation seasonality on Piñon pine
Kluwer Academic Publishers. cellulose dD values. Global Change Biology, 6, 287–301.
16 Vegetation-Atmosphere Interface: Tree Rings 203

Raffali-Delerce, G., Masson-Delmotte, V., Dupouey, J. L., Stievenard, Trouet, V. (2020). Tree Story, the History of the World Written in Rings
M., Breda, N., & Moisselin, J. M. (2004). Reconstruction of (p. 256). Johns Hopkins University Press.
SUMMER droughts using tree-ring cellulose isotopes: A calibration Waterhouse, J. S., Switsur, V. R., Barker, A. C., Carter, A. H. C., &
study with living Oaks from Brittany (Western France). Tellus, 56, Robertson, I. (2002). Oxygen and hydrogen isotope ratios in tree
160–174. rings: How well do models predict observed values? Earth and
Schweingruber, F. H., Fritts, H. C., Braeker, O. U., Drew, L. G., & Planetary Science Letters, 201, 421–430.
Schaer, E. (1978). The X-ray technique as applied to dendroclima-
tology. Tree-Ring Bulletin, 38, 61–91.
 Air-Vegetation Interface: An Example
of the Use of Historical Data on Grape 17
Harvests

Valérie Daux

Historical data provide clues, often precise, of climate phe- commence the harvest at will. However, the use of an official
nomena experienced by societies. The example below date was maintained in most parishes, for reasons of security
illustrates the scientific approach applied to records of grape and public order. After 1791, it was the mayor, advised by
harvest dates in order to quantitatively estimate the evolution the vineyard owners, who fixed the date of the beginning of
of temperature in a region. the harvest. The actual date of harvesting could take place
The vine is a hardy perennial Mediterranean plant, well some days after this date. The official and real dates of
adapted to hot dry weather conditions. Temperature is a key harvesting have been recorded in different registers
environmental factor which triggers its various phenological depending on the place and time: these might be in monastic
stages. Buds appear when daytime temperatures reach at records, chapters of canons, hospital records (such as those
least 10 °C for an extended period, usually in April. Flow- from the Hospices de Beaune, Burgundy, France), municipal
ering dates occur about two months later (usually June), registers or private records (by the vineyards). During the
when the daily temperatures cumulatively reach a certain ‘Medieval Climate Optimum’ (around 10th to thirteenth
threshold. The date of the onset of ripening, veraison (in centuries), vineyards were planted at latitudes as far north as
August), is dependent on the date of flowering and the the shores of the Baltic or the south of England. Conversely,
subsequent temperatures. Finally, the grape reaches maturity due to the cooling of the ‘Little Ice Age’ (from the 14th to
about thirty days after veraison, most often in September. the nineteenth centuries), most northern vineyards collapsed,
Harvesting takes place when the grapes are perfectly ripe. and the vine growing season was shortened so much that the
The time between veraison and harvesting is fairly constant; harvests were difficult even in some southern vineyards.
it depends on the temperature, on the phytosanitary status of The series of published harvest dates are mostly com-
the vineyard and the amount of rainfall. The time periods posite series constructed from multiple sets of dates from
given for the different phenological stages are approximate different sites of the same vineyard. Ideally, the composite
and vary according to grape varieties and regions. series should be established from local series of harvesting
Harvest dates, since they depend mainly on the temper- dates of the same grape variety, to avoid introducing bias
atures during the growth and maturation of the vine, can be related to a phenology which may vary by grape variety. The
regarded as thermal indicators. Variations in harvest dates longest and most comprehensive series published to date is
over the past centuries reflect historical variations in spring that of the vineyards of Burgundy (Fig. 17.1). This is a
and summer temperatures. composite series constructed from data from multiple sites in
the Côtes de Beaune and perhaps also from the Côtes de
Dijon. The grapes harvested in that area since the fourteenth
Historical Series of Harvest Dates century are mostly pinot noir and chardonnay.

In the Middle Ages, the official opening dates of harvest

were set by informed feudal decision. After the Revolution, Reconstruction of Spring-Summer
this tradition was theoretically abandoned; each owner could Temperatures Based on Grape Harvesting
Dates
V. Daux (&)
Laboratoire des Sciences du Climat et de l’Environnement, Grape harvesting data from the twentieth century, from
LSCE/IPSL, CEA-CNRS-UVSQ, Université Paris-Saclay, 91190
Gif-sur-Yvette, France different vineyards and varieties, were compared with
e-mail: [email protected] regional meteorological data (Daux et al. 2007).

© Springer Nature Switzerland AG 2021 205

G. Ramstein et al. (eds.), Paleoclimatology, Frontiers in Earth Sciences,
https://doi.org/10.1007/978-3-030-24982-3_17
206 V. Daux

Fig. 17.1 Bottom: Harvest dates in Burgundy since 1370 (Daux et al. 1570 to 1630 (with a slight warming around 1600–1620), in 1675,
2012). Dates are arbitrarily expressed in days after August 31st. Top: during the 1690s and in the following century, from 1709 to 1715;
Anomalies in the corresponding temperatures, calculated using a heatwaves in summers of the 1630s, 1660s and 1680s; warming during
phenological model. The anomalies are calculated compared to the the eighteenth century, particularly obvious in the years 1704–1707,
reference period 1960–1989; according to Garcia de Cortazar et al. 1718–1719, in the 1720s and 1730s, the years 1757–1765 and
(2010). The bold lines are 29-year moving averages. The major climate especially from 1778–1781 and during the 1780s; nevertheless, years
events of the past seven centuries may be described as follows: of cool-cold-wet (1725, 1740, 1770); the Little Ice Age, which never
warming in the 1380s, and from 1415 to 1435, a period that includes quite ceased, became vigorous again from 1812 to 1860; added to this
the wheat ‘blast’ that led to the famine of 1420; cooling during the were the very cold snaps of 1812–1817; heatwave of 1846, affecting
second half of the ‘Quattrocento’ particularly evident in the famine of grain harvests; the definitive end of the Little Ice Age in 1860; warming
1481, arising from rain and cold conditions; hot spells during the during the twentieth century from around 1900 onwards with an
1520s, 1530s and 1550s; the strong cold surge of the Little Ice Age acceleration since 1976 and the 1990s. After Le Roy Ladurie et al.
(LIA) from 1560. A cold seventeenth century’? This was the case from (2006)

The harvest date series and temperatures present many temperatures for the growing season (April in August; see
similarities, at the annual, decadal and multi-decadal scale also for example Cook and Wolkovich 2016).
(see example in Fig. 17.2). The best correlations (statisti- The phenology of woody species can be simulated by
cally significant with a maximum R) between harvest dates models that express the relationship between the matura-
and climate data are those with the average maximal tem- tion process and daytime temperatures. A model is adap-
peratures from April to August. For these correlations, the ted to describe the phenology of a species, or a variety in
slopes are around −6 to −10 days/°C for all the vineyards the case of the vine. Chuine et al. (2004) have thus used
(Fig. 17.3). In other words, regardless of the earliness or recent observations of phenological development of the
lateness of the vines and regardless of the geographic loca- pinot noir variety, made between 1964 and 2001 in Col-
tion (including soil, orientation and meteorology), a varia- mar (France), by INRA, to model the phenology of this
tion of six to ten days on the date of the harvest reflects a variety and to calculate the veraison and harvest dates
difference of about 1 °C from the average maximum from temperatures. The inverted model has been used here
17 Air-Vegetation Interface: An Example … 207

Fig. 17.2 Series of harvest dates (bold lines with symbols, scale on right; in days after August 31) and maximum daytime temperatures for the
corresponding region (dotted lines, scales on left) versus time

Fig. 17.3 Correlations among

harvest dates (in days after
August 31) and average monthly
daytime temperatures from April
to August (in °C) for the
vineyards of the Champagne,
Burgundy, Alsace, Bordeaux and
Loire valley regions

to reconstruct variations in spring-summer temperatures in the start of a long cool period that continued until 1970,
Burgundy since the fourteenth century from the series of was by a little more than one degree. The exceptionally
dates shown in Fig. 17.1. This reconstruction highlights hot summer of 2003 also shows up very clearly (Chuine
that the cooling of the climate in the period 1690–1800, et al. 2004).
208 V. Daux

Limitations of These Reconstructions: References

The Anthropic Effects
Chuine, I., Yiou, P., Viovy, N., Seguin, B., Daux, V., & Le Roy
The decision to harvest is based on a simple principle: the Ladurie, E. (2004). Grape ripening as a past climate indicator.
grapes need to be harvested when they are ripe and not too Nature, 432, 289–290.
Cook, B. I., & Wolkovich, E. M. (2016). Climate change decouples
long afterwards so as to avoid exposure to unfavorable drought from early wine grape harvests in France. Nature Climate
weather conditions, such as hail or heavy rain. Judgment of Change, 6, 715–719.
the ideal state of ripeness may have evolved over time Daux, V., Garcia de Cortazar-Atauri, I., Yiou, P., Chuine, I., Garnier,
depending on the political and economic context, cultivation E., Le Roy Ladurie, E., et al. (2012). An open-access database of
grape harvest dates for climate research: data description and quality
techniques, oenological practices and consumer tastes. assessment. Clim. Past, 8, 1403–1418.
Variations over time in these parameters can create trends Daux, V., Yiou, P., Le Roy Ladurie, E., Mestre, O., Chevet, J.-M., &
and cause breaks in the data series. These variations caused the OPHELIE team. (2007). Température et dates de vendanges en
by factors not related to the climate may not necessarily be France’, Articles du colloque ‘Changement climatique, quels
impacts probables sur le vignoble?’, Dijon, 28–29 mars 2007, 10p.
detectable in a single series. To be as rigorous as possible, a Garcia de Cortazar-Atauri, I., Daux, V., Garnier, E., Yiou, P., Viovy,
climate reconstruction must be based on a collection of N., Seguin, B., et al. (2010). An assessment of error sources when
series (ideally of the same variety, from the same wine using grape harvest date for past climate reconstruction. Holocene,
region), which is tested for disruptions and 20, 599–608.
Le Roy Ladurie, E., Daux, V., & Luterbacher, J. (2006). Le climat de
inter-correlations. Historical information of this type can Bourgogne et d’ailleurs. Histoire, Economie et Sociétés, 3, 421–
provide insights and help to identify the proportion of 436.
variability attributable to anthropic effects.
 Air-Ground Interface: Sediment Tracers
in Tropical Lakes 18
David Williamson

The climate archives found at low latitudes are mostly sedi- physical and chemical methods of measuring the variability
mentary or geomorphological. Their formation and preser- of depositional environments, the measurement of magnetic
vation depend fundamentally on favorable geomorphological sedimentary properties (low-magnetic field susceptibility,
and climatic conditions. Therefore, they differ according to induced and remanent magnetizations) is a particularly
the environment, the region and the climate and include appropriate one, because multiple measurements can be
moraines left behind by mountain glaciers, ancient shorelines performed. The magnetic signature of sediments—and other
of old lakes, dunes, flood levels, archeological sites occupied surface materials—depends on changes in the concentration,
by animals or humans, continental carbonates (speleothems, grain size and mineralogy of iron minerals (mainly oxides,
travertine, stromatolites) and sedimentary sequences col- sulfides, carbonates). Iron mineral assemblages closely
lected in cores from lakes and bogs. Additional archives depend on the iron biogeochemical cycle, in particular on the
consist of coastal deposits (e.g. corals) and marine hemi- weathering of primary minerals, pedogenetic processes,
pelagic sediments with a high deposition rate, especially near biomass fires, gravity deposition, redox changes, microbial
river deltas or coastal upwellings off Mauritania, Benguela activity, authigenesis and early diagenesis. Magnetic meth-
and Somalia. The dating and study of aquifers, as well as of ods thus allow for the reconstruction of the dynamics of
tree rings complete the range of climate records currently surface processes recorded along sedimentary sequences,
available and studied, particularly in semi-arid areas. and the identification of the main modes of environmental
The physical, chemical and biological controls of the and climate variability (Thompson and Oldfield 1986).
sedimentary records vary greatly according to the local or
regional specificity of the areas of deposition. For example,
the mixing or overturning of the water column of a lake is Intertropical Hydrological Variability in Africa
generally recorded by biological (diatoms) or organic
proxies. It may be activated by an earthquake, outgassing Africa is characterized by a wide variety of climates, mainly
from anoxic hypolimnion or by sediments, but also by the defined by the amount and distribution of rainfall at the
cooling of surface waters, the melting of ice, by wind seasonal, annual, and year-to-year level (from less than
intensity or by a drop in lakewater level. Hence, the climate 100 mm per year to more than 3000 mm per year) and
or hydrological proxies (e.g. physicochemical properties or temperature (especially at mid-latitudes and at altitude).
isotopic signatures of water and sediments, terrestrial or From the humid equatorial climate of the Congo basin to the
aquatic vegetation markers, erosional or atmospheric inputs) hyperarid or subtropical climates of the Sahel/Sahara and of
do not have the same fundamental meaning from one site to the Kalahari/Namibia, regional climates respond to the
another. annual and seasonal characteristics of global and tropical
To reduce the uncertainties in timing and interpretation, it atmospheric circulation (in particular the activity of the
is essential to undertake a detailed reconstruction of the Intertropical Convergence Zone, ITCZ). The temperatures
deposition processes in several sites characteristic of the and surface currents of the neighboring oceans and the
region and for a given climate. Among the non-destructive regional orographies (e.g. East-African rift, great escarpment
of Namibia) determine the regional rainfall patterns. Toge-
D. Williamson (&) ther with the local geomorphology, these regional climates
World Agroforestry Centre, Research Institute for Development, have a determining influence on the presence, nature and
UMR LOCEAN, Université Pierre et Marie Curie, P.O.
Box 30677-00100 Nairobi, Kenya historical continuity of most of the land-based climate
e-mail: [email protected] archives.

© Springer Nature Switzerland AG 2021 209

G. Ramstein et al. (eds.), Paleoclimatology, Frontiers in Earth Sciences,
https://doi.org/10.1007/978-3-030-24982-3_18
210 D. Williamson

Sahara, Kalahari and Arid Zones:

Discontinuous Evidence of Hydrological
Inversions

Since the 1950s and 1960s, the first robust evidence of major
hydrological changes during the late Pleistocene and Holo-
cene, have been found in the two African subtropical deserts
in the form of proof of the existence of gigantic interdune
lakes; flora and fauna assemblages; sites demonstrating a
sedentary human presence and a positive water balance over
centuries or millennia. A key outcome of this research is
indications that, during the first half of the Holocene, there
was an intensification of the African monsoon in the Sahara
and a shift in the tropical summer rains (at the northern limit
of the ITCZ) north of 20°N as well as a significant reduction
in the Sahara as a biogeographical barrier during this period
(Lézine et al. 1990; Petit-Maire and Riser 1981). Very
similar evidence of major hydrological changes has been Fig. 18.1 Chronostratigraphy comparing lacustrine sequences from
the Kalahari (Thomas et al. 2003) and Lake Chad (Servant and
obtained for the Kalahari (Thomas et al. 2003). As shown in Servant-Vildary 1980). Black represents high water level; gray, the
Fig. 18.1, these records show the asynchronism of the intermediate level; and white the low water level or drying out. Note the
humidity ‘optima’ during the Holocene, north (from 11 to phase difference between the water balances of the southern tropical
6 ka BP) and south (after 6 ka BP) of the equator, illus- region (Kalahari) and the edge (Chad)
trating the forcing of the changes in regional insolation on
the monsoon circulation and summer tropical rainfall (from Holocene from Lake Chad (Servant and Servant-Vildary
June to September in the Sahara and from December to 1980). The coupling of the millennial climate variability in
March in the Kalahari). Furthermore, at the millennial scale, the high latitudes during the glacial period with the humidity
numerous sedimentary discontinuities suggest the occur- of the Sahara is clearly demonstrated from marine sediments,
rence of hydroclimatic thresholds, apparently linked with in particular through erosion markers (Adegbie et al. 2003)
rapid changes in activity of the ITCZ from one latitude to and ecosystems (Schefuß et al. 2005). However, the infor-
another, corresponding to rapid climate fluctuations in the mation obtained often lacks spatial accuracy, and the critical
high latitudes (Mayewski et al. 2004). However, the diffi- contribution of transportation processes with regard to the
culty with these archives remains in the discontinuity of the interpretation of markers often remains hypothetical
records and the lack of detailed chronological precision on (Pichevin et al. 2005).
dry periods. The preserved deposits contain the initial stages
of wet periods, but the preservation of the signal, in partic-
ular for the phases of aridification, is hypothetical. (Sub)Equatorial Zone: Changes in the Activity
The depositional gaps and the differential removal of and Position of the ITCZ
remaining deposits do not allow a continuous hydrological
and climate dynamic to be reconstructed, particularly for the The continental archives providing the most continuous
glacial period. recording of climate variability over the last 25 ka are
The dynamics of the African deserts must then be located in sites that have remained relatively humid, located
reconstructed from the proximate and hemipelagic marine in equatorial regions or at altitude, and where the hydric
field, especially at the mouths of the major rivers (Senegal, balance [Precipitation—Evaporation, P-E) remained posi-
Niger, Orange River, Zambezi) and in upwelling zones tive. These are cores taken either from large terminal lake
(Mauritania, Benguela), where sedimentation rates permit a reservoirs found in central Africa (Lake Victoria) or the East
resolution at the millennial scale, coupled with isotopic African Rift (e.g. Lakes Tanganyika and Malawi), or from
stratigraphy for the last ice age(s). The data obtained confirm small lake basins suitable for the reconstruction of regional
the important role of insolation in the low latitudes in conditions. As shown from the complete drying up of Lake
monsoon circulation and the hydrology of the subtropical Victoria at 18 and 16 ka BP (Stager et al. 2002), the drop by
areas. They also confirm the instability of the edges of the several hundred meters of the levels of lakes Tanganyika and
Sahara and the Kalahari at the millennial scale (DeMenocal Malawi during the last glacial period, or the rapid rise 18
et al. 2000; Little et al. 1997), already highlighted for the (over a few decades) of great lakes that are today very
18 Air-Ground Interface: Sediment Tracers in Tropical Lakes 211

shallow or drying out (Rukwa, Victoria, Chad) (Williamson taro, rice, and, more recently, plants from the new world
et al. 1993), the whole intertropical zone has recorded major such as maize, tobacco, tomatoes. However, it is only since
and rapid changes in its water balance, linked mostly to the late nineteenth century and especially after the Second
changes in regional temperature. The changes affected the World War, that the archives show radical changes in the
continental surfaces, the soils, the distribution and the bio- environment caused by man (Marchant et al. 2018): the
diversity of terrestrial and aquatic ecosystems. For example, organic components (on land and in water) and inorganic
the iron mineralogy of soils was modified, and variations of components, dissolved or suspended (notably magnetic),
several orders of magnitude of the concentration in magnetic present in the environment are affected (Garcin et al. 2007).
minerals are evidenced (Williamson et al. 1993). In this context, just as changes in atmospheric composi-
Owing to unrivaled chronological and spatial resolution, tion have been reconstructed from ice cores, the land-based
the data obtained from peat bogs or crater lakes, most ‘pre-industrial’ archives recording changes in the climate
especially the molecular markers of microbial origin (Cof- and the tropical environment are now the local and regional
finet et al. 2014), show that these rapid changes in atmo- references for environments relatively untouched by man,
spheric circulation and in the composition of the atmosphere from the decadal to the millennial scale. Such references are
(water, CO2) were responsible for a fundamental reorgani- required to establish a frame of reference so as to identify the
zation of ecosystems, particularly with regard to the distri- various controls of climate and society on the structural
bution of mountain ecosystems and most wetland areas. changes in the landscape and in biodiversity, in vegetal
These records confirm the dominant forcing of insolation production, in erosional processes, or in stored material
and monsoon circulation over tropical hydrology. At the (including pollutants), and thus to assess the extent and
decennial to millennial scale, however, the depositional sustainability of the changes imposed by humans on the
record clearly indicates the coupling of the tropical climate Earth’s surface.
with the climate in the high latitudes, and with the activity of
the thermohaline circulation, particularly during the glacial
interval where the Dansgaard-Oeschger variability is iden- References
tified in numerous hydrological records. New high resolution
chronological records are required to identify the mecha- Adegbie, A. T., Schneider, R. R., Röhl, U., & Wefer, G. (2003). Glacial
nisms and the potential phase-lags linked with such changes millennial-scale fluctuations in central African precipitation
(Zhang et al. 2018). recorded in terrigenous sediment supply and freshwater signals
offshore came.
Coffinet, S., Huguet, A., Williamson, D., Fosse, C., & Derenne, S.
(2014). Potential of GDGTs as a temperature proxy along an
Recent Anthropological Influence on Climate altitudinal transect at Mount Rungwe (Tanzania). Organic Geo-
Archives: Both a Proof and a Tool to Assess chemistry, 68, 82–89.
DeMenocal, P., Ortiz, J., Guilderson, T., Adkins, J., Sarnthein, M.,
the Impacts of Local and Regional Baker, L., et al. (2000). Abrupt onset and termination of the African
Development humid period: Rapid climate responses to gradual insolation
forcing. Quaternary Science Reviews, 19, 347–361.
Superimposed on natural climate stress, the human impact Garcin, Y., Williamson, D., Bergonzini, L., Radakovitch, O., Vincens,
A., Buchet, G., et al. (2007). Solar and anthropogenic imprints on
on the hydrology and land-based environments of Africa lake masoko (Southern Tanzania) during the last 500 years. Journal
remains poorly documented, and presents a challenge to a of Paleolimnology, 37, 475–490.
better understanding of the interactions between climate, Lézine, A. M., Casanova, J., & Hillaire-Marcel, C. (1990). Across an
environment and society in developing countries. The first early holocene humid phase in Western Sahara: Pollen and isotope
stratigraphy. Geology, 18, 264–267.
large-scale human impacts on ecosystems (livestock Little, M. G., Schneider, R. R., Kroon, D., Price, B., Summerhayes, C.
domestication and pastoralism) are recorded as early as ca. P., & Segl, M. (1997). Trade wind forcing of upwelling, seasonality,
4000 yr B.P., especially during the last 2500 yr with and heinrich events as a response to sub-milankovitch climate
deforestation and biomass burning (Thevenon et al. 2003) variability. Paleoceanography, 12, 568–576.
Marchant, R., Richer, S., Boles, O., Capitani, C., Courtney-Mustaphi,
being associated with iron metallurgy, the expansion of C. J., Lane, P., et al. (2018). Drivers and trajectories of land cover
farming activities and demography: rapid increases in the change in East Africa: Human and environmental interactions from
fluxes of microcharcoal, detrital inputs (especially magnetic) 6000 years ago to present. Earth-Science Reviews, 178, 322–378.
and pollen from ruderal vegetation (near habitations) and Mayewski, P. A., Rohling, E. E., Stager, J. C., Karlén, W., Maasch, K.
A., Meeker, L. D., et al. (2004). Holocene climate variability.
pioneer taxa (Zhang et al. 2018). In East Africa, the Quaternary Research, 62, 243–255.
expansion of trade routes between the coast and the interior Petit-Maire, N., & Riser, J. (1981). Holocene lake deposits and
at least 1300 years ago increased the abundance of exotic palaeoenvironments in central Sahara, Northeastern Mali. Palaeo-
plants, especially cultivated plants from Asia such as banana, geography Palaeoclimatology Palaeoecology, 35, 45–61.
212 D. Williamson

Pichevin, L., Cremer, M., Giraudeau, J., & Bertrand, P. (2005). A 190 NW Kalahari: an integrated study from the Tsodilo Hills, Botswana.
Ky record of lithogenic grain-size on the namibian slope: Forging a Quaternary International, 104, 53–67.
tight link between past wind-strength and coastal upwelling Thompson, R., & Oldfield, F. (1986). Environmental magnetism.
dynamics. Marine Geology, 218, 81–96. London: Allen and Unwin.
Servant, M., & Servant-Vildary, S. (1980). L’Environnement quater- Williamson, D., Taieb, M., Damnati, B., Icole, M., & Thouveny, N.
naire du bassin du Tchad. The Sahara and the Nile. In M. (1993). Equatorial extension of the younger dryas event:
A. J. Williams, & H. Faure (Ed.), (pp. 133–162). Leiden: Balkema. Rock-magnetic evidence from lake Magadi. Global and Planetary
Schefuß, E., Schouten, S., & et Schneider, R. R. (2005). Climatic Change, 7, 235–242.
controls on central African hydrology during the Past 20,000 Years. Williamson, D., Jelinowska, A., Kissel, C., Tucholka, P., Gibert, E.,
Nature, 437, 1003–1006. Gasse, F., et al. (1998). Mineral-magnetic proxies of
Stager, J. C., Mayewski, P. A., & Meeker, L. D. (2002). Cooling erosion/oxidation cycles in tropical maar-lake sediments (Lake
cycles, heinrich event 1, and the desiccation of lake victoria. Tritrivakely, Madagascar): Paleoenvironmental implications. Earth
Paleogeography, Paleoclimatology, Paleoecology, 183, 169–178. and Planetary Science Letters, 155, 205–219.
Thevenon, F., Williamson, D., Vincens, A., Taieb, M., Merdaci, O., Zhang, Y. C., Chiessi, C. M., Mulitza, S., Sawakuchi, A. O., Haggi, C.,
Decobert, M., et al. (2003). A late holocene charcoal record from Zabel, M., et al. (2018). Different precipitation patterns across
Lake Masoko, SW Tanzania: Climatic and anthropologic implica- tropical South America during Heinrich and Dansgaard-Oeschger
tions. The Holocene, 15, 785–792. stadials. Quaternary Science Reviews, 177, 1–9.
Thomas, D. S. G., Brook, G., Shaw, P., Bateman, M., Appleton, C.,
Nash, D., et al. (2003). Late pleistocene wetting and drying in the
 Air-water Interface: Tropical Lake Diatoms
and Isotope Hydrology Modeling 19
Florence Sylvestre, Françoise Gasse, Françoise Vimeux,
and Benjamin Quesada

Regardless of the timescale, a distinctive characteristic of the climate with often dramatic fluctuations, accompanied by
tropics is a high variability in the water cycle, in particular in profound changes in chemistry, biology, water-mass
the P-E water balance [precipitation minus (evapora- dynamics and water-atmosphere heat transfers. Indeed, one
tion + evapotranspiration)]. This variability has more impact way which is often forgotten that lakes can have a significant
on life in the tropics than the variability in temperature interaction on the continental hydrological cycle is through
which only has a moderate influence compared to the high the source of water vapor they represent (Vallet-Coulomb
and middle latitudes. The time frame, speed and magnitude et al. 2008). In this chapter, we will examine specifically the
of hydrological events need to be established so that their example of a hydro-isotopic reconstruction established from
causes and mechanisms can be understood. These events are sedimentary records from a paleolake located in the tropical
recorded in terrestrial sediments (lacustrine deposits, Andes (Fig. 19.1) and its impact on the local hydrological
deposits in caves), in fossil waters in deep aquifers, and cycle during the last glacial-interglacial transition (18–
imprinted in the landscape by geomorphological features 12 ka) (Quesada et al. 2015). Over this period, this paleo-
(dried-up river channels, ancient lake shorelines). lake, called ‘Tauca’ (from the name of a village where the
Lacustrine archives are particularly interesting because eponymous outcrops were discovered, Servant and Fontes
they contain the imprint of hydrological variations recorded 1978) underwent a transgression phase due to an increase in
in lakes. They preserve a wide range of complementary rainfall in the tropical Andes (Sylvestre et al. 1999; Blard
indicators (or proxies), providing access to different envi- et al. 2011), before disappearing abruptly around 14.3 ka.
ronmental and climate parameters if they can be calibrated. Here, we propose a reconstruction of the isotopic composi-
Lakes are today numerous in the humid tropics, and tion of the lake water and we explore how its disappearance
evidence of ancient lakes (i.e. paleolakes) are plentiful in might have disturbed the local atmospheric water cycle. In
certain regions that are now arid. They react to changes in particular, we explore the possible link between a sudden
evaporation of the Tauca paleolake and an abrupt event
found in the isotopic composition (oxygen 18) of the glacier
covering Mount Sajama overlooking the former lake
(Thompson et al. 1998).

F. Gasse—Deceased.
Site Selection and Collection of Samples
F. Sylvestre (&)  F. Gasse  B. Quesada
Aix-Marseille Université, CNRS, IRD, Collège de France,
INRAE, CEREGE, Europôle Méditerranéen de l’Arbois, 13545 A fundamental criterion for the selection of study sites is the
Aix-en-Provence cedex 4, France sensitivity of the lake system (the lake and its watershed) to
e-mail: [email protected] variations in P-E. The change in volume of a lake during the
F. Vimeux time interval Dt is represented by the equation:
Institut de Recherche pour le Développement (IRD), Laboratoire
HydroSciences Montpellier (HSM), UMR 5569 (CNRS, IRD, DVL =Dt ¼ SL ðPL  EL Þ þ A  D: ð19:1Þ
UM), 34090 Montpellier, France
F. Vimeux  B. Quesada where VL is the volume of the lake (m3) and SL is the surface
Institut Pierre Simon Laplace (IPSL), Laboratoire des Sciences du area (m2). VL and SL are functions of the level of the lake
Climat et de l’Environnement, LSCE/IPSL, CEA-CNRS-UVSQ, (HL). PL and EL are, respectively, the rates of precipitation
Université Paris-Saclay, 91190 Gif-sur-Yvette, France

© Springer Nature Switzerland AG 2021 213

G. Ramstein et al. (eds.), Paleoclimatology, Frontiers in Earth Sciences,
https://doi.org/10.1007/978-3-030-24982-3_19
214 F. Sylvestre et al.

Fig. 19.1 a The Bolivian

Altiplano (MODIS image
modified from http://www.nssl.
noaa.gov/projects/pacs/web/
ALTIPLANO/) with 1. Lake
Titicaca, 2. Lake Poopó, 3. The
Salar de Coipasa, 4. The Salar of
Uyuni. The red triangles indicate
the position of the ice cores taken
from the glaciers of Illimani,
Mount Sajama and Tunupa.
b Diagram of the Bolivian
Altiplano indicating the
contemporary lakes, the outline of
the watershed (black line) and the
extent of the Tauca paleolake
shown in blue. The study sites are
indicated by orange dots and
yellow squares

and evaporation at the surface of the lake (m), A (m3) is the In a paleohydrological study of lacustrine archives, the
sum of the surface and subterranean inflows. A is dependent maximum morphological, geological, meteorological,
on the P-E balance of the watershed (PB − EB) and its hydrological and hydrochemical information on the lake
surface SB. D (m3) is the sum of the losses from the surface watershed is collected by satellite imagery, aerial pho-
and from infiltration at the bottom of the lake. Equa- tographs, field observations, and instrumental data. The
tion (19.1) shows that closed lakes (D = 0), common in spatial distribution of exposed lake deposits and former
semi-arid or arid areas, are the most sensitive because they perched shorelines (beach deposits, notches eroded in the
cumulate variations in the P-E of the lake and of its water- cliffs by waves, etc.) is measured by GPS and localized
shed. Thus, they amplify the responses to climate change. using a digital terrain model (DTM). Contemporary samples
The Bolivian Altiplano is located in the central tropical of water and silts are collected as references for the inter-
Andes between 15°S and 22°S latitude and 65°W and 69°W pretation of past tracers. Sediments are taken at the outcrop
longitude at an average altitude of 3800 m. Today, it is and by coring. The sites are chosen according to the
occupied by a chain of four lakes of decreasing altitude and bathymetry and the continuity of the identified sediments.
increasing salinity from north to south: Lake Titicaca, a deep On the Bolivian Altiplano, the samples were selected from
lake of fresh water at 3810 m; Lake Poopó, a very shallow, outcrops according to their altitude around the lake, making
hypersaline lake at 3686 m; and the Coipasa and Uyuni it possible to quantify the depth of the lake during its evo-
lakes at 3657 m and 3653 m which are now covered by two lution through time (Fig. 19.1b) (Sylvestre et al. 1999).
large salt crusts of 2,500 km2 and 11,000 km2 respectively,
the remaining legacy following evaporation of the Tauca
paleolake (Fig. 19.1a). This hydrological system is endor- Reconstruction of Paleohydrological
heic, and Lake Titicaca can flow into Lake Poopó, which Conditions
itself, in very rainy seasons, can feed the Salars of Coipasa
and Uyuni (Fig. 19.1a). The Bolivian Altiplano receives The study commences with a lithological description of the
most of its precipitation during the rainy season which samples, a prerequisite to understanding the organization of
occurs from November to March. Today, precipitation the sedimentary deposits, before selecting the levels to
comes mainly from the Atlantic, transported to the Altiplano establish their absolute chronology using radiometric meth-
by the easterly winds (Garreaud et al. 2003). ods such as radiocarbon. In stratigraphy, we find the strata
19 Air-water Interface: Tropical Lake Diatoms and Isotope … 215

with remarkable sedimentary facies that allow the different the water during formation have been experimentally
profiles to be correlated. Indicators of a small sliver of water established for lacustrine diatoms under contemporary
or of a drought located in the profile (mollusk shells, coarse environmental conditions, taking the seasonal to
sand beds, or paleosols) as well as former perched shorelines multi-annual variations into account (Fig. 19.2).
provide direct data on lake level fluctuations. The miner- These relationships all show a low dependence on tem-
alogical composition of the sediments, their grain size, the perature, of around −0.20‰/°C. On the other hand, the
geochemical properties of the various fractions (mineral or fractionation factors are very different, which may be due
organic, detrital or authigenic), and the assemblages of fos- either to sampling problems or to different methodological
silized organisms are indirect indicators of past environ- approaches when carrying out these calibrations (Crespin
mental conditions. et al. 2010; Alexandre et al. 2012). Nevertheless, applied in
In the outcrops of the Bolivian Altiplano, we focused our different contexts, this method proved relevant to the
investigations on the diatoms. These are microscopic uni- reconstruction of the isotopic composition of lakes in the
cellular algae (3–100 lm), which are very sensitive to past (Leng and Barker 2006). In addition, the application of
variations in environmental parameters, such as the depth or these different calibrations enables sensitivity tests to be
salinity of water, and each cell produces an easily fossilized conducted depending on the potential temperature range
bivalve siliceous outer skeleton. This external skeleton, applicable during the period considered.
called a frustule, consists of an inner layer of silica tetrahedra So that the thermo-dependence of the isotopic fractiona-
(Si2O, nH2O) bonded to each other, and of a hydrated outer tion between the diatoms and the formation water is taken
layer made of organic material, mainly proteins and into account, the temperature of the lake water in which they
polysaccharides. Diatom valves are extracted from sedi- develop must also be known. Diatoms are photosynthetic
ments by physicochemical treatment. In optical microscopy, organisms which grow in the epilimnion and reflect the
the species are identified by the structure of their valves: surface temperatures of the lake. For current periods, field
their absolute (number of valves/g) and relative (percentage) observations are essential; for past periods, assumptions
abundance can then be estimated. Knowing the self-ecology based on simulated atmospheric temperature data from cli-
of each species identified makes it possible to characterize mate models and data obtained from other proxies in the
the aquatic environment in which they developed. same region are used.
The presence of oxygen in the inner layer of the frustule
allows the isotopic composition in oxygen 18 to be mea-
sured. Preliminary dehydration is necessary because of the Hydro-Isotopic Modeling and Paleoclimatic
hydrated nature of the silica so that only the isotopes in the Interpretation
oxygen of the internal structure are measured. After extrac-
tion of oxygen and purification in a preparation line, the O2 A quantified estimate of the hydrological balance of lakes or
molecule is analyzed by mass spectrometry (Crespin et al. of the P-E is obtained by hydrological modeling. The more
2008, Crespin et al. 2010). or less sophisticated models used all rely on the balance
equation for lake water (Eq. 1). The first step is to derive the
contemporary water balance from the available instrumental
Quantification of the Oxygen Isotopic data and the relationships between HL, VL and SL con-
Composition of Lakes structed from a digital terrain model (DTM). Estimating
evaporation, a function of temperature, solar radiation,
In lacustrine environments, the isotopic oxygen composition humidity, wind and vegetation cover, is always complex. It
of diatoms (d18Odiatoms) varies with the temperature and involves the use of different hydrological and climatic con-
isotopic composition of the lake water (d18Olake water). cepts and methods. It is better constrained by connecting the
Diatomaceous silica precipitates in isotopic equilibrium with water balance to the salt balance (Vallet-Coulomb et al.
the lake water, and calibration studies have established the 2001) with a watershed runoff model (Legesse et al. 2004) or
thermo-dependent relationship [([d18Odiatoms − d18Olake with an energy balance model (Kutzbach 1980). Application
water] (‰ vs. VSMOW) = a*T lake water (°C) + b)] to of the model to the past is usually done for a time interval
express the variation in the isotopic composition of diatoms where the lake is considered to have been in equilibrium
as a function of temperature (Brandriss et al. 1998; Moschen (DVL = 0). We know SB, HL (from which we calculate VL
et al. 2005; Crespin et al. 2010; Dodd and Sharp 2010; Dodd and SL), the solar radiation provided by astronomical cal-
et al. 2012; Alexandre et al. 2012). These relationships culations, paleotemperatures and paleo- vegetation on the
between the oxygen isotopic composition of the diatoms and watershed can be deduced from data e.g. from palynology.
216 F. Sylvestre et al.

Fig. 19.2 Calibration of the

thermo-dependent relationship
between d18Odiatoms and d18Olake
water based on different
calibrations of freshwater diatoms
Modified from Crespin et al.
(2010), Alexandre et al. (2012)

On the Bolivian Altiplano, paleohydrological data the lake during its stability phases. This isotopic enrichment
(Fig. 19.3) show that the Bolivian Altiplano experienced a of lake waters is caused by interplay between lake evapo-
first lacustrine transgression at 18,500 BP, interrupted ration and precipitation processes, both of which alter the
between 18,100 BP and 15,800 BP by a phase of stagnation isotopic composition of the lake differently. This confirms
of the lake level, followed by a subsequent phase of maxi- the establishment of the Tauca paleolake, probably caused
mum extension up to 15,000 BP years. After this date, this by a massive influx of precipitation (significant decrease in
lake dried up. A second lacustrine transgression of lesser d18O of the water of the paleolake) from the tropical Atlantic
magnitude, named ‘Coipasa’, took place between 12,500 BP in the Altiplano watershed. The abrupt disappearance of the
and 11,900 BP. paleolake is contemporaneous with an isotopic excursion of
The reconstruction of the isotopic composition of the lake +7‰ recorded in the ice of Mount Sajama (Thompson et al.
produces a very interesting result showing a spectacular 1998). Calculation of the isotopic composition of the vapor
decrease in its isotopic composition during the transgression flux produced by evaporation of the paleolake, with its
phases (increase of the lake level) in response to rainfall volume and surface taken into account, shows that this
inputs, followed by phases of isotopic enrichment during the isotopic excursion could have been caused by an evaporation
phases when its level stabilizes (Fig. 19.3). of 5–60% of the total volume of the lake. If this hypothesis is
Simple hydro-isotopic modeling has shown that the proven correct, it shows that the Tauca paleolake has cer-
increase in precipitation during the establishment of the lake tainly influenced the local, or even regional, hydrological
is quantitatively consistent with the decrease recorded in the cycle.
reconstruction of d18Olake water (Quesada et al. 2015). It also In conclusion, this example shows the potential impact of
partly explains the increase in the isotopic composition of a lake on the local hydrological conditions, especially during
19 Air-water Interface: Tropical Lake Diatoms and Isotope … 217

Fig. 19.3 Reconstruction of lake levels (m) and the isotopic oxygen for T = 5 °C, T = 7.5 °C and T = 10 °C, and (2) the isotopic oxygen
composition d18Olake water (‰ vs. V-SMOW) from (1) the isotopic composition of ostracods (green squares) and carbonates (red squares;
oxygen composition of diatoms (blue lines) using the Blard et al. 2011) for T = 7.5 °C
thermo-dependent relationship established by Dodd and Sharp (2010)

strong fluctuations in its level. Sites where several climate Younger Dryas: New insights from 14C, U-Th dating and d18O
archives co-exist, allowing the large-scale climate signal to of carbonates. Quaternary Sciences Review, 30, 3973–3989.
Brandriss, M. E., O’Neil, J. R., Edlund, M. B., & Stoermer, E. F.
be differentiated from local conditions are relatively rare. (1998). Oxygen isotope fractionation between diatomaceous silica
The example discussed here shows the benefits to be gained and water. Geochimica Cosmochimca Acta, 62, 1119–1125.
from the co-location of a glacier archive and a lacustrine Crespin, J., Alexandre, A., Sylvestre, F., Sonzogni, C., Paillès, C.,
archive. Garreta, V. (2008). IR-laser-extraction technique applied to oxygen
isotopes analysis of small biogenic silica samples. Analytical
Chemistry, 80, 2372–2378.
Crespin, J., Sylvestre, F., Alexandre, A., Sonzogni, C., Paillès, C., &
References Perga, M.-E. (2010). Re-examination of the temperature-dependent
relationship between d18Odiatoms and d18Olake water and implications
for paleoclimate inferences. Journal of Paleolimnology, 44, 547–557.
Alexandre, A., Crespin, J., Sylvestre, F., Sonzogni, C., & Hilbert, D. Dodd, J. P., & Sharp, Z. D. (2010). A laser fluorination method for
W. (2012). The oxygen isotopic composition of phytolith assem- oxygen isotope analysis of biogenic silica and a new oxygen isotope
blages from tropical rainforest soil tops (Queensland, Australia): calibration of modern diatoms in freshwater environments.
Validation of a new paleoenvironmental tool. Climate of the Past, 8, Geochimica Cosmochimica Acta, 74, 1381–1390.
307–324. Dodd, J. P., Sharp, Z. D., Fawcett, P. J., Brearley A. J., & McCubbin F.
Blard, P.-H., Sylvestre, F., Tripati, A. K., Claude, C., Causse, C., M. (2012). Rapid post-mortem maturation of diatom silica oxygen
Coudrain, A., et al. (2011). Lake highstands on the altiplano isotope values. Geochemistry, Geophysics, Geosystems, 13. https://
(Tropical Andes) contemporaneous with Heinrich 1 and the doi.org/10.1029/2011gc004019.
218 F. Sylvestre et al.

Garreaud, R. D., Vuille, M., & Clement, A. C. (2003). The climate of d18O of the Andean glaciers during the last deglaciation (18.5–
the Altiplano: Observed current conditions and mechanisms of past 11.7 ka): Diatom-inferred d18O values and hydro-isotopic model-
changes. Palaeogeography, Palaeoclimatology, Palaeoecology, ing. Quaternary Science Review, 120, 93–106.
194, 5–22. Servant, M., & Fontes, J. C. (1978). Les lacs quaternaires des hauts
Kutzbach, J. E. (1980). Estimate of past climate at paleolake Chad, plateaux des Andes boliviennes: Premières interprétations paléocli-
North Africa, based on a hydrological and energy balance model. matiques. Cahiers ORSTOM. Série Géologie, 10, 9–23.
Quaternary Research, 14, 210–223. Sylvestre, F., Servant, M., Servant-Vildary, S., Causse, C., Fournier,
Legesse, D., Vallet-Coulomb, C., & Gasse, F. (2004). Analysis of the M., & Ybert, J.-P. (1999). Lake-level chronology on the Southern
hydrological response of a tropical terminal lake, lake Abiyata Bolivian Altiplano (18–23°S) during late-glacial time and the early
(Main Ethiopian Rift Valley) to changes in climate and human holocene. Quaternary Research, 51, 54–66.
activities. Journal of Hydrological Processes, 18, 487–504. Thompson, L. G., Davis, M. E., Mosley-Thompson, E., & Sowers, T.
Leng, M. J., & Barker, P. A. (2006). A review of the oxygen isotope A. (1998). A 25,000-Year tropical climate history from bolivian ice
composition of lacustrine diatom silica for palaeoclimate recon- cores. Science, 282, 1858–1864.
struction. Earth-Science Review, 75, 5–27. Vallet-Coulomb, C., Legesse, D., Gasse, F., Travi, Y., & Chernet, T.
Moschen, R., Lücke, A., & Schleser, G. (2005). Sensitivity of biogenic (2001). Lake evaporation estimates in tropical Africa (Lake Ziway,
silica oxygen isotopes to changes in surface water temperature and Ethiopia). Journal of Hydrology, 245, 1–18.
palaeoclimatology. Geophysical Research Letters, 32, L07708. Vallet-Coulomb, C., Gasse, F., & Sonzogni, C. (2008). Seasonal
https://doi.org/10.1029/2004GL022167. evolution of the isotopic composition of atmospheric water vapour
Quesada, B., Sylvestre, F., Vimeux, F., Black, J., Paillès, C., Sonzogni, above a tropical lake: Deuterium excess and implication for water
C. et al. (2015). Impact of Bolivian paleolake evaporation on the recycling. Geochimica Cosmochimica Acta, 72, 4661–4674.
 Air-Ice Interface: Tropical Glaciers
20
Françoise Vimeux

Just as the polar caps are excellent archives of information propose to review these and to present the main results in
on past climates, so too are the high-altitude tropical gla- terms of climate variability. Although complementary, we
ciers. Indeed, the prevailing temperature and humidity con- will not discuss here the qualitative results from the chemical
ditions there usually ensure very good preservation of analysis of ice, which essentially informs us about changes
chemical and isotopic tracers. These archives have therefore in the environmental, atmospheric transport processes and
been used for the past 25 years to study the variability of the air pollution.
tropical climate over past centuries and millennia. They are Over recent centuries, the seasonal cycles of the chemical
mostly found in the Andes in South America between 0° and elements and stable isotopes in precipitation allow us to date
20° S (Vimeux 2009; Vimeux et al. 2009), although drilling the annual layers with a relatively good degree of accuracy
has been conducted on Kilimanjaro (Thompson et al. 2002) (±5 to 10 years around 1900). It is thus possible to calculate
and the southern Himalayas (Thompson et al. 2000, 2006). the annual net accumulation. To correct for the effects of
The focus of this chapter is on climate information taken snow compaction in the depths, an ice flow model is applied,
from the Andean glaciers. or alternatively, the thinning of the layers is corrected by
The rapid dynamic of these glaciers, the high level of snow directly observing the relationship between the annual
accumulation per year (0.5–1 m) and reduced ice thickness thickness and the depth, along the core. The latter method
(100–150 m) mean that it is not possible to access climate cannot however be used to discuss climate trends which are,
periods as old as those possible in polar cores. The oldest cores in principle, corrected using the same method. The question
date back to the last glacial maximum, about 20 000 years ago, that then arises is: what does the net accumulation represent
with the Sajama core (Bolivia, 6542 m, Western Cordillera, given that it is actually the combination of the total accu-
18° 06′ S, 68° 53′ W) dating back to * 25,000 years mulation (controlled by the precipitation and wind) minus
(Thompson et al. 1998). On the other hand, the tropical ice removal (sublimation, erosion by wind), and that the latter
allows our climate to be studied with a very good temporal processes can have different seasonalities? If one considers
resolution, at the seasonal level for recent centuries. sites where accumulation is low (0.31 m of water/year) with
a high level of sublimation throughout a long dry season,
such as at Cerro Tapado (Chile, arid diagonal, 5550 m, 30°
Paleoclimate Markers 08′ S, 69° 55′ W), it is difficult to use this parameter as a
marker of the amount of deposited precipitation. This is less
Several types of quantifiable climate variables can be the case for sites such as the one at Illimani (Bolivia, eastern
extracted from tropical glaciers: the net accumulation, the cordillera, 6350 m, 16° 37′ S, 67° 46′ W) where the rainfall
temperature in the borehole and regional precipitation. We season is longer, the annual snow accumulation heavier
(0.58 m of water/year) and the sublimation is concentrated
over a shorter time of the year which is different to the
F. Vimeux (&) accumulation season (Ginot et al. 2006). Only the end of the
Laboratoire HydroSciences Montpellier (HSM), Institut de rainy season which represents about 10% of the annual
Recherche pour le Développement (IRD), accumulation may be truncated in the records. In that case,
Montpellier, 34095, France
e-mail: [email protected] we can we assume that large variations in the net accumu-
lation are representative of the amount of precipitation. The
Laboratoire des Sciences du Climat et de l’Environnement (LSCE),
Institut Pierre Simon Laplace (IPSL), accumulation estimates made on the Quelccaya core (Peru,
91191 Gif-sur-Yvette, France 5670 m, 13° 56′ S, 70° 50′ W, the only core with dating by

© Springer Nature Switzerland AG 2021 219

G. Ramstein et al. (eds.), Paleoclimatology, Frontiers in Earth Sciences,
https://doi.org/10.1007/978-3-030-24982-3_20
220 F. Vimeux

Fig. 20.1 Example of relationship between the isotopic composition isotopic composition is strongly depleted of heavy isotopes, while
of deuterium (‰) in rainfall collected in the Zongo Valley at multiple during the dry season, it is enriched. The correlation with local
sites (Bolivia, 16° S, a valley linking the Andean peaks to the Amazon) precipitation explains only 50% of the isotopic signal. The remainder of
(connected points) and the quantity of precipitation (mm/month) (bars), the variance can be explained by precipitation at the regional level. This
on a monthly scale over several years. During the rainy season, the figure is adapted from Vimeux et al. 2005

seasons over the past 1500 years) show sequences of high observations, through rainfall collection network systems,
(the seventeenth and eighteenth centuries and in the twen- and modeling of atmospheric cycling of stable isotopes of
tieth century) and low (the nineteenth century) accumula- water with a hierarchy of models (general circulation
tions (Thompson et al. 2006). atmospheric model, mesoscale model correctly representing
As in polar ice, it is possible to measure the temperature the topography and one dimension convection model) has
in the borehole. This method does not allow past temperature shown that at the seasonal and interannual scale, the isotopic
changes to be reconstructed with a high temporal resolution, composition of Andean snow is strongly related to precipi-
but it does offer the possibility of measuring slow fluctua- tation upstream from the drilling sites along the trajectories,
tions. In high altitude glaciers, this profile depends on the in the Amazon and over the tropical Atlantic regions where
energy balance at the surface and on the flow of geothermal the most intense convection phenomena are located (Vimeux
heat in the depths. The proximity of the ice to the bedrock et al. 2005; Vuille and Werner 2005; Vimeux et al. 2011)
does not allow these profiles to be applied to the second half (Fig. 20.1). It was shown that the relationship between water
of the cores although the surface profiles can be interpreted. isotopes in the Andes and precipitation is strongly dependent
In the cores where these profiles are available, an increase in on convection conditions (re-evaporation of water droplets
temperature is observed for recent decades (Vimeux et al. and recycling of the resulting vapor in the convective col-
2009), reaching 1.1 °C in the twentieth century. umn) (Risi et al. 2008).
The measurement of the isotopic composition (d18O and
dD) of tropical ice provides information on rainfall patterns
caused by the air mass along its trajectory. The linear rela- Some Important Results
tionship between the isotopic composition of snow and from the Interpretation of Andean Isotopic
surface air temperature, well established for the polar Records
regions, does not hold true in the tropics. This relationship is
mainly due to the fact that, in the middle and high latitudes, Recent studies have sought to link the changes in precipi-
the amount of precipitation formed and air temperature are tation in tropical South America to larger scale processes
closely linked as per the Clausius-Clapeyron law. This is not over the last century. Most of the interannual variability in
the case in the tropics, where the majority of precipitation is rainfall for this region is linked to variations in intensity and
convective and where the water cycle is complex (recycling geographical extension of the ascending and convective
of water vapor from the surface). The coupling of branch of the Hadley-Walker cell, affecting the South
20 Air-Ice Interface: Tropical Glaciers 221

Fig. 20.2 The Andean Isotopic Index (AII) is built from the isotopic components of global precipitation (EOF1) which reflects the first mode
composition of four Andean ice cores having similar interannual of interannual climate variation that is ENSO; b) the isotopic
variations and a sufficiently precise dating over the last century (in composition of the oxygen-18 in Amazonian water vapor, simulated
Bolivia: Illimani (16° S, 6300 m) and Sajama (18° S, 6542 m) in Peru: by the atmospheric model ECHAM-4. The lines in bold represent a
Huascaran (9° S, 6048 m) and Quelccaya (14° S, 5670 m)). It is moving average over 5 years. This figure is adapted from Hoffman
compared here to: a) the first component of an analysis of the principal et al. (2003)

American monsoon pattern. This cell is strongly disrupted were wetter (depletion of isotopes and higher accumulation
by anomalies in the temperature of surface waters in the on the glaciers to increase the mass at their base) and colder
tropical Pacific. (maximum spread of the glaciers).
We therefore think that ENSO events are likely to mark
the isotopic composition of Andean ice (Bradley et al. 2003; At the glacial-interglacial scale, isotope profiles obtained
Hoffmann et al. 2003). The isotopic composition recorded in on three Bolivian and Peruvian cores (Illimani; Huascaran,
several Andean cores over the last century effectively shows Peru, 6050 m, 9° 06′ S, 77° 30′ W and Sajama) show a
a common decadal signal, called Isotopic Andean Index common isotope signal in terms of variability and amplitude
strongly linked to rainfall variations in the Amazon, caused (Fig. 20.3). This signal, very similar to that recorded by the
by the ENSO phenomenon (Fig. 20.2). However, variations isotopic composition in the polar cores, highlights highly
in temperature of the surface waters of the tropical Atlantic depleted values during the last glacial maximum and a
also have a strong impact on the South American monsoon gradual enrichment during deglaciation, with a return to
system (and so on regional precipitation), and it becomes depleted conditions before reaching an optimum around
difficult to untangle the different causes in the isotopic signal 11,000 years. This structure is similar to that described in
(Hoffmann 2003). polar ice, where known climate periods are recorded (glacial
The Illimani and Quelccaya ice cores with their precise maximum, Younger Dryas and the Holocene optimum),
dating back to about 1700 show a significant loss of about although the interpretation of the isotopes in Andean ice
1.5‰ of d18O between the late seventeenth and early nine- (moisture) is different from that of polar ice (temperature).
teenth century. This period also corresponds to the maxi- The translation of this glacial-interglacial variation in terms
mum spread of glacial moraines in Bolivia, Peru and of humidity shows that 20,000 years ago, the air masses
Ecuador, dated by lichenometry (Jomelli et al. 2009). This sustained a more major drain along their trajectories, corre-
comparison between the spread of glaciers and the d18O of sponding to a precipitation increase of about 10% (Vimeux
ice suggests that during the Little Ice Age, the tropical Andes et al. 2005).
222 F. Vimeux

Fig. 20.3 Isotopic composition

in deuterium (‰) over the last
meters of the Illimani core and
isotopic composition of oxygen
(‰) of air trapped in the bubbles

A study using coupled ocean-atmosphere climate models over the last millennium and palaeoclimatic implications: A review.
for this glacial period shows that rainfall was indeed more Palaeogeography, Palaeoclimatology, Palaeoecology, 281, 269–
282. https://doi.org/10.1016/j.palaeo.2008.10.033.
intense 20,000 years ago in Northeast Brazil and the tropical Khodri, M. (2009). Sensitivity of South American tropical climate to
South Atlantic, the cradle of Andean precipitation (Khodri last glacial maximum boundary conditions: Focus on teleconnec-
2009). tions with tropics and extratropics. In F. Vimeux, F. Sylvestre, & M.
Khodri (Eds.), Developments in paleoenvironmental research series
(DPER) (Vol. 14, XVII, 418 p. 106 illus., 61 in color). Springer.
ISBN 978-90-481-2671-2.
References Risi, C., Bony, S., & Vimeux, F. (2008). Influence of convective
processes on the isotopic composition (d18O and dD) of precipi-
tation and water vapor in the tropics, part 2: Physical interpretation
Bradley, R. S., Vuille, M. Hardy, D., & Thompson, L. G. (2003). Low of the amount effect. Journal of Geophysical Research, 113,
latitude ice cores record Pacific sea surface temperatures. Geophys- D19306. https://doi.org/10.1029/2008JD009943.
ical Research Letters, 30. https://doi.org/10.1029/2002gl016546. Thompson, L. G., Davis, M. E., Mosley-Thompson, E., Sowers, T.A.,
Gilbert, A., Wagnon, P., Vincent, C., Ginot, P., & Funk, M. (2010). Henderson, K. A, Zagorodnov, V. S., Lin, P.-N., Mikhalenko, V.
Atmospheric warming at a high-elevation tropical site revealed by N., Campen, R. K., Bolzan, J. F., Cole-Dai, J., & Francou, B.
englacial temperatures at Illimani, Bolivia (6340 m above sea level, (1998). A 25,000-Year tropical climate history from bolivian ice
16° S, 67° W). Journal Geophysical Research, 115, D10109. cores. Science, 282, 1858–1864.
https://doi.org/10.1029/2009JD012961. Thompson, L. G., Yao, T., Mosley-Thompson, E., Davis, M. E,
Ginot, P., Kull, C., Schotterer, U., Schwikowski, M., & Gäggeler, H. Henderson, K. A., & Lin, P. -N. (2000). A high-resolution
W. (2006). Glacier mass balance reconstruction by sublimation millennial record of the South Asian monsoon from himalayan ice
induced enrichment of chemical species on Cerro Tapado (Chilean cores. Science, 289, 1916–1919.
Andes). Climate of the Past, 2, 21–30. Thompson, L. G., Mosley Thompson, E., Davis, M. E., Henderson, K.
Hoffmann, G., Ramirez, E., Taupin, J. D., Francou, B., Ribstein, P., A., Brecher, H. H., Zagorodnov, V. S., et al. (2002). Kilimanjaro ice
Delmas, R., Dürr, H., Gallaire, R., Simoes, J., Schotterer, U., core records: Evidence of Holocene climate change in tropical
Stievenard, M., & Werner, M. (2003). Coherent isotope history of Africa. Science, 298, 589–593.
Andean ice cores over the last century. Geophysical Research Thompson, L. G., Mosley Thompson, E., Brecher, H., Davis,M., Leon,
Letters, 30, 1179–1184. B., Les, D., Lin, P.-N., Mashiotta, T., & Mountain, K. (2006).
Hoffmann, G. (2003). Taking the pulse of the tropical water cycle. Abrupt tropical climate change: Past and present. Proceedings of the
Science, 301, 776–777. National Academy of Sciences, 103, 10536–10543.
Jomelli, V., Favier, V., Rabatel, A., Brunstein, D., Hoffmann, G., & Vimeux, F., Gallaire, R., Bony, S., Hoffmann, G., Chiang, J., &
Francou, B. (2009). Fluctuations of glaciers in the tropical andes Fuertes, R. (2005). What are the climate controls on isotopic
20 Air-Ice Interface: Tropical Glaciers 223

composition (dD) of precipitation in Zongo Valley (Bolivia)? and surrounding regions. In F. Vimeux, F. Sylvestre, M. et Khodri
Implications for the Illimani ice core interpretation. Earth and (Eds.), Developments in paleoenvironmental research series
Planetary Science Letters, 240, 205–220. (DPER) (Vol. 14, XVII, 418 p., 106 illus., 61 in color.). Springer.
Vimeux, F., Ginot, P., Schwikowski, M., Vuille, M., Hoffmann, G., ISBN 978-90-481-2671-2.
Thompson, L. G., et al. (2009). Climate variability during the last Vimeux, F., Tremoy, G., Risi, C., & Gallaire, R. (2011). A strong
1000 years inferred from Andean ice cores: A review of method- control of the South American see-saw on the intra-seasonal
ology and recent results. Palaeogeography, Palaeoclimatology, variability of the isotopic composition of precipitation in the
Palaeoecology, 281, 229–241. https://doi.org/10.1016/j.palaeo. Bolivian Andes. Earth and Planetary Sciences Letters, 307, 47–58.
2008.03.054. Vuille, M., & Werner, M. (2005). Stable isotopes in precipitation
Vimeux, F. (2009). Similarities and discrepancies between polar and recording South American summer monsoon and ENSO variability:
andean ice cores over the last deglaciation in past climate variability Observations and model results. Climate Dynamics, 25. https://doi.
from the last glacial maximum to the holocene in South America org/10.1007/s00382-005-0049-9.
 Climate and the Evolution of the Ocean: The
Paleoceanographic Data 21
Thibaut Caley, Natalia Vázquez Riveiros, Laurent Labeyrie, Elsa Cortijo,
and Jean-Claude Duplessy

expedition of 1947–1948 collected over 300 different cores

Introduction: The Development of Tools from various deep ocean basins that became the basis of the
and Concepts first studies on the geological history of the oceans. In par-
allel, during the 1950s, Maurice Ewing, the founder of the
The idea of reconstructing the history of oceans and climates Lamont-Doherty Geological Observatory (USA), and one of
in the past using marine sediment cores arrived quite late the developers of seismic sediment mapping of the ocean
after the beginnings of oceanography. It was initiated in the floor, initiated the first systematic collection of marine sed-
twentieth century, well after the first attempts to measure iment cores. The first descriptions of the main sedimentary
variations in seawater temperature down the water column, systems, changes in fossil faunas and the timeframe for the
which date back to the eighteenth century with the great first biostratigraphic age scales were proposed based on
circumnavigation expeditions. Land geologists were the first these cores. One of the main results of these studies was the
to propose paleoceanographic reconstructions from exposed continuous reconstruction of the alternating warm and cold
marine series, limiting the collected information to former phases that took place during the Pleistocene.
coastal waters. The first reconstructions of past seawater Unquestionably, the honor for the initiation of quantitative
temperatures were made possible by the piston corer paleoceanography belongs to Cesare Emiliani. After the dis-
developed by the Swedish oceanographer Kullenberg, cap- covery of isotopic fractionation and the development of an
able of collecting continuous sedimentary deposits without accurate method to measure isotopic ratios, Harold Urey and
layer disruption. Geologists were therefore able to collect his group in Chicago refined the use of the isotopic ratio
18 16
uninterrupted sedimentary series, sometimes over 20 m in O/ O in fossil carbonates as a paleothermometer. They
length, for laboratory analysis, and thus study long records realized that the 18O/16O ratio of foraminiferal shells and other
of the environmental conditions from the time the sediments carbonates depended on two variables: the temperature and the
18 16
were deposited. With this type of corer, the Swedish O/ O ratio of the water where the carbonate was formed.
Changes in water temperature are reflected in variations in the
isotopic fractionation between the carbonate and the water
Thibaut Caley, Natalia Vázquez Riveiros and Laurent Labeyrie— during the formation of the shell: for water with a given isotopic
These authors have contributed equally to this chapter. composition, the higher the temperature, the lower the 18O/16O
T. Caley (&)
ratio (Epstein et al. 1951, 1953). Emiliani (1955) applied this
EPOC, UMR 5805, CNRS, University of Bordeaux, Pessac, tool to foraminifera shells sampled along a sediment core from
France the Caribbean Sea to propose the first reconstruction of the
e-mail: [email protected] variations in sea surface temperature (SST) over the past
N. V. Riveiros
L. Labeyrie
E. Cortijo
J.-C. Duplessy 400 ka (Fig. 21.1). He also established the major method-
Laboratoire des Sciences du Climat et de l’Environnement, ological guidelines for this type of study: use of continuous
LSCE/IPSL, CEA-CNRS-UVSQ, Université Paris-Saclay,
91190 Gif-sur-Yvette, France
records, precise dating, and interpretation of parameters
quantitatively linked to key variables of the climate system.
N. V. Riveiros
Institut Français de Recherche pour l’Exploitation de la Mer
To extract a temperature signal from the 18O/16O ratio of
(Ifremer), Unité de Geosciences Marines, Pointe du Diable, planktonic foraminifera, Emiliani had to constrain the
29280 Plouzané, France changes in the isotopic composition of the ocean water in
L. Labeyrie which the foraminifera developed. The 18O/16O ratio of
LGO Université Bretagne Sud, 56000 Vannes, France

© Springer Nature Switzerland AG 2021 225

G. Ramstein et al. (eds.), Paleoclimatology, Frontiers in Earth Sciences,
https://doi.org/10.1007/978-3-030-24982-3_21
226 T. Caley et al.

Fig. 21.1 a Cesare Emiliani,

founder of isotopic marine
paleoclimatology, in the early
1950s at the University of
Chicago. (Photo from the
archives of the Rosenstiel School
of Marine and Atmospheric
Science, University of Miami).
b First attempt to evaluate surface
water temperature changes in the
Caribbean Sea, Emiliani (1955).
Subsequent studies have shown
that the timescale was
underestimated by about 25% (the
last interglacial, called Sangamon
in American literature, is dated at
about 125,000 years and not
100,000 years), and that the
amplitude of temperature
variations, calculated from a
simple model (see below), was
overestimated

seawater is affected by evaporation and precipitation: the The growth and melting of ice caps, which involves
vapor phase is depleted in the heavy 18O isotopes relative to considerable volumes of water (several million cubic kilo-
the liquid phase. Conversely, when the water vapor in clouds meters), directly affect the salinity and the average 18O/16O
condenses, the precipitation is richer in 18O than the vapor. ratio of the ocean, and therefore that of the foraminifera that
Thus, the transport of air masses from low to high latitudes is develop there. Regional climate changes are also accompa-
accompanied by a large-scale isotopic distillation process in nied by local variations in evaporation and precipitation,
the water vapor that results in the gradual decline of the which induce further regional variations in the salinity of
18 16
O/ O ratio in precipitation. For this reason, the 18O/16O surface seawater and its 18O/16O ratio.
ratio of snowfall feeding the high-latitude ice caps is From the data available at the time, Emiliani (1955)
depleted by more than 30‰ compared to that of the tropical estimated that the development of large ice sheets covering
ocean. Canada (the Laurentide ice sheet) and northern Europe (the
21 Climate and the Evolution of the Ocean: The Paleoceanographic … 227

Fennoscandian ice sheet) at the maximum of the glaciation associations and isotopic composition have been widely
caused an enrichment in 18O of the global ocean of +0.4‰. used as paleoclimate indicators for decades, and the advent
We now know that the enrichment in 18O in the glacial of relatively new indicators implies an undiminished interest.
ocean was in fact close to +1.0‰ (Schrag et al. 2002). Other indicators such as corals will be briefly discussed in
Despite these inaccuracies, the work of Emiliani was the first association with some of the tracers.
to demonstrate from paleoclimate observations that glacial—
interglacial periods indeed oscillated with cyclicities pre-
dicted by the Milankovitch theory several decades earlier Sea Surface Temperature
(see Chap. 28). Emiliani also proposed the ‘Marine Isotopic
Stage’ (MIS) nomenclature, now universally adopted, to The surface temperature of the ocean is an essential climate
characterize the alternation of warm and cold Pleistocene parameter that governs heat exchange with the atmosphere.
phases, with odd numbers for interglacial periods and even SST also modulates the solubility of gases, oxygen and CO2
numbers for glacial ones (1 for the Holocene, 2 for the last in particular, and their exchange rates with the atmosphere.
glacial period, and so on). He also discovered that the last The amplitude of the spatial variability of SST is well
interglacial, or MIS 5 (Fig. 21.1), was interrupted by two known: it ranges between −1.96 °C, freezing point for sea-
colder periods, which led him to divide it into three warm water at 35 psu, and 30–35 °C, maximum temperature
subperiods (designated 5a, 5c and 5e from the most recent to recorded for the open ocean. However, its temporal vari-
the oldest) and two cold ones (5b and 5d). The term ‘5e’ is ability is more difficult to constrain because it not only varies
still frequently used, as it has been incorporated into the on a daily basis, but also seasonally and annually. In situ
European continental reconstructions as the equivalent to the measurement sensors are precise to ±0.001 °C at a given
Eemian warm period. The isotopic stratigraphy formalism location, water depth and time. Satellite data provide global
has since been generalized, with subdivisions either num- coverage and allow long-term monitoring of the evolution of
bered as decimals between alternating warmer (e.g., 5.1 for SST, but their accuracy is, at best, close to 0.1 °C, and
5a, 5.3 for 5c and 5.5 for 5e) and colder (5.2 and 5.4) periods surface values are averaged over tens or hundreds of square
(Pisias et al. 1984) or as letters (Railsback et al. 2015). kilometers. Paleoceanographers cannot aim to reconstruct
With the assumption that past variations in foraminiferal SST variations with this level of precision. Nonetheless,
18 16
O/ O ratios in cores from different ocean basins had to be given the magnitude of changes in the past, relevant infor-
approximately synchronous across global climate changes, mation may be acquired when SST changes are estimated to
Emiliani paved the way for a global marine isotopic the nearest degree. Paleoceanographers also aim to estimate,
stratigraphy. The demonstration that the volume of ice caps whenever possible, the amplitude of the seasonal cycle, and
was indeed the dominant component of the isotopic signal the temperature distribution of the upper water column.
recorded in marine cores reinforced the stratigraphic value of For these studies, two major types of paleotemperature
the marine isotopic stage age scale, which became a major indicators are used: (i) changes in the distribution of fossil
reference tool for past climate change studies. The routine planktonic flora or fauna (foraminifera, diatoms, dinoflag-
use of drilling ships as part of the International Ocean ellates, radiolarians), and (ii) geochemical tracers produced
Drilling Program has allowed the recovery of sediment cores by these organisms or recorded in their fossil skeletons.
that cover the last tens of millions of years, extending the
isotopic sequences not only to the Quaternary (Fig. 21.2),
but as far back as the Paleocene, 60 Ma ago. The Distribution of Marine Fauna and Flora
By 1970–1980, the paleoclimate community had arrived
at the conclusion that variations in the oxygen isotopic The distribution of the various groups that make up the marine
composition provided a remarkable stratigraphic tool to planktonic ecosystem was extensively studied during the
establish long-term correlations. However, new tools still major exploration campaigns that marked the nineteenth and
needed to be developed to precisely reconstruct past SST, as early twentieth centuries. Foraminifera, single-celled proto-
well as variations in other oceanic features such as salinity, zoans that secrete a calcareous shell, were the most generally
or the direction and intensity of deep-water currents. recorded group. They are very diverse and inhabit all the
This chapter will focus on the development of classic and oceans, from the coldest to the warmest. However, each
new paleoceanographic tracers over the last decades. We individual species has a limited tolerance to environmental
will mainly, but not exclusively, focus on tracers that are changes, particularly temperature, which allowed biologists at
based on foraminifera, since these abundant microfossils the end of the nineteenth century to highlight the zonal dis-
have been extensively used because of their ubiquity in the tribution of many species. They also established existing
oceans and their great preservation potential. Their faunal relationships between climate and the abundance of certain
228 T. Caley et al.

Fig. 21.2 Variations in the

isotopic composition of the
oxygen in benthic foraminifera
from 70 million years ago
(Zachos et al. 2008) with a detail
of the last 1.8 Ma showing
periodicities predicted by the
astronomical theory of
paleoclimate (Lisiecki and Raymo
2005). Presence of Northern
Hemisphere and Antarctic ice
sheets are from Zachos et al.
(2008). Geological epochs
(Pl. = Pleistocene,
Plio = Pliocene) are indicated at
the bottom of the top panel

species. These observations, along with the statistical method paleotemperature to be estimated using the ecological equa-
of ‘transfer functions’ developed by Imbrie and Kipp, served tions calculated from the modern core tops. This work led to
as a basis for the quantitative reconstruction of surface tem- the great success of the CLIMAP group (Climate Long-range
peratures from fossil fauna found in marine sediments. Investigation, Mapping And Prediction), which reconstructed
Transfer functions allow the estimation of seawater tempera- the first global map of summer and winter SST distribution
ture during both the cold and warm seasons, when adequate during the Last Glacial Maximum (LGM) period, about 21 ka
data is available. The basic principle for the reconstruction of ago (Fig. 21.3).
SST changes in the past is to assume that the ecological One of the main problems with this paleo-reconstruction
requirements of the species present nowadays have not chan- method is that it is based on the a priori stable statistical
ged since the past period under consideration. The specific correlation between changes in the specific factors defined
abundance of planktonic foraminifera samples from recent by the species distributions and the arbitrarily chosen envi-
core tops may be expressed as a matrix (x species relative ronmental parameter, temperature, in this case. However,
distributions within each of n sampling stations) from which other environmental factors, such as the availability of food
the vectors corresponding to the main factors that describe the supply, may also be involved and change the sensitivity of
specific faunal (or floral) variance can be extracted. The foraminifera to temperature from one region to another.
method proposed by Imbrie and Kipp (1971) provides the best There is also another potential problem with the transfer
correlation between changes in these factors and the associated function approach: it will provide accurate estimates only if
changes within the modern environmental parameters, sys- modern conditions are good analogs of the past hydrological
tematically selected for the n stations (such as summer and conditions. This is not always necessarily the case. For
winter temperatures, or other parameters, provided they are example, the fossil fauna of the eastern Mediterranean Sea
statistically independent). A similar factor analysis, when during the LGM period has no modern analogs. Indeed, this
applied to fossil assemblages down sediment cores, allows the basin experienced hydrological and climate conditions very
21 Climate and the Evolution of the Ocean: The Paleoceanographic … 229

Fig. 21.3 Reconstructions of temperature anomalies (°C) between the figure). While the two reconstructions have many features in common,
Last Glacial Maximum (LGM) and modern surface waters as obtained the cores studied under the MARGO program revealed the presence of
by the CLIMAP (1981) (August temperatures, top figure) and MARGO larger longitudinal gradients in SST in all ocean basins than the
(2009) programs (July–August–September temperatures, bottom estimates of CLIMAP (1981)

different from the modern ones, due to the development of analogs are defined using a mathematical distance calcula-
large ice sheets over Northern Europe and to a very different tion. The method is based on the assumption that the closer a
hydrological cycle. In fact, the absence of direct analogs fossil assemblage is to one or more modern references, the
becomes the rule rather than the exception for the distant more similar their optimal growth conditions are (tempera-
past: most fossil species from periods earlier than the Qua- ture and nutrient supply, in particular). The distance to the
ternary are not found in recently deposited sediments. best analogs and the dispersion of associated environmental
This is why the method used by Imbrie and Kipp (1971) conditions provide an estimate of the reconstruction uncer-
was gradually replaced by the best analogs method. Its tainties (Waelbroeck et al. 1998).
principle is simple: to compare the changes in fossil In parallel with these statistical developments, recent
assemblages to modern references, without a preconceived advances in the field of artificial intelligence and neural
idea of the origin of the observed changes. The closest networks have helped to improve paleoceanographic
230 T. Caley et al.

reconstructions without fundamentally changing the princi- Geochemical Methods

ple. Still based on the comparison between fossil fauna and
modern reference fauna, they do not require the establish- Organic Tracers
ment of specific mathematical relationships, but establish The organic geochemistry of marine sediments provides a
their own learning from available databases through mini- different set of tracers. The most common so-called ‘bio-
mization of the uncertainties (Malmgren et al. 2001). marker’ is based on the changes in the abundance ratio of di-
These various methods, applied here to foraminifera, and tri-unsaturated alkenones (molecules with 37 carbon
have also been used for diatom flora, coccoliths, dinoflag- atoms containing two or three double bonds). This ratio is a
ellate cysts and radiolarians. Diatoms, dominant in cold function of the growth temperature of the synthesizing
waters rich in silica, have been used in particular to recon- organisms, a group of algae called coccolithophorids, and in
struct variations in sea ice cover in polar regions. The latest particular of the species Emiliania huxleyi for the modern
LGM SST global distribution, as reconstructed within the ocean. The number of double bonds is inversely related to
MARGO program, used several types of indicators and is the temperature: the lower the temperature, the higher the
presented in Fig. 21.3. number of double bonds (Prahl and Wakeham 1987). The
Statistical methods have many limitations: (i) they can abundance ratio of di- and tri-unsaturated alkenones is
only be used if the faunal assemblages are close to modern k0
conventionally expressed by the index U37 :
assemblages; (ii) the genetic diversity of species among
0
different ocean regions may induce variations in their faunal k
U37 ¼ ½C37:2 =½C37:2 þ C37:3 
responses to temperature and the corresponding statistical 0
k
links; (iii) temperature reconstructions based on variations in The initial calibration of the U37 index is based on E.
the abundance of fossil fauna or flora assume that other huxleyi cultures in controlled conditions (Prahl and Wake-
factors, such as productivity for example, have no significant ham 1987) (Fig. 21.4), and has been subsequently verified
influence on the relative abundances of the different species; using samples collected from ocean water or from the sed-
(iv) due to the activity of burrowing animals (bioturbation), iment surface (Müller et al. 1998; Conte et al. 2006; Tierney
marine sediments are usually mixed over several centime- and Tingley 2018).
ters, so that the same stratigraphic level of a sediment core
represents a mix of fauna that lived in different centuries (or
even several thousand years apart if the sedimentation rate is
low); (v) transfer function calibration is based on the
assumption that sediment core top assemblages reflect
modern hydrological conditions. This latest assumption
ignores in particular ocean and climate changes which
occurred over the last millennia, and that could have been
significant enough to bias calibrations. These limitations
have encouraged the development of new reconstruction
methods based on either biological or geochemical
mechanisms.
The biological approach is still in its infancy, and it is
derived from ecological studies of the requirements of the
different species in the modern ocean. A first approach
directly calibrates the proxies (foraminifera or others) from
controlled laboratory cultures with varying physiological
and geochemical constraints that duplicate those observed in
the marine environment. A more theoretical approach com-
plements these calibrations by modeling the growth condi-
tions within the natural environment, using the
experimentally calibrated variables. Using such methods, it
0
may be possible to obtain a reliable reconstruction of the Fig. 21.4 Relationship between the unsaturation index U37 k
and SST.
hydrology corresponding to the specific habitat of the dif- The line represents the temperature calibration curve based on cultures
ferent species of planktonic foraminifera (e.g., Lombard of E. huxleyi grown under laboratory conditions (Prahl and Wakeham
1987). Natural particulate samples collected are indicated (Prahl and
et al. 2009). Wakeham 1987)
21 Climate and the Evolution of the Ocean: The Paleoceanographic … 231

0
k
However, like all paleoclimate indicators, the U37 index Chemical Tracers
also presents various biases that limit its use in specific The chemical composition of the carbonate from for-
oceanographic contexts. A few of the most significant ones aminiferal tests and coral skeletons may also provide pale-
are as follows: otemperature or paleoenvironment indicators. For example,
the concentration of magnesium incorporated in the calcium
– The temporal evolution of the alkenone-producing spe- carbonate of foraminifera is an empirical function of the
cies. E. huxleyi, which is currently the main producer of temperature at which that foraminifer crystallized its test. On
alkenones, was not present prior to MIS 8 (Thierstein time scales where the Mg/Ca of the oceans has remained
et al. 1977). Alkenones are also produced by other spe- constant, the sensitivity of Mg/Ca to temperature has been
cies such as Gephyrocapsa oceanica, but with different determined using either a culture-based, sediment trap or
k0 core top calibrations (see for example, Lea et al. 1999;
temperature—U37 index relationships. Care must there-
k0 Elderfield and Ganssen 2000; Anand et al. 2003; Mashiotta
fore be taken when applying the U37 method to ancient
et al. 1999) (Fig. 21.5), and it takes the form:
sediments.
– The tiny coccolithophorids are easily transported by sea Mg/Ca ¼ B expðA  TÞ
currents, so these algae can travel long distances between
their place of production and place of sedimentation. where A and B are the exponential and pre-exponential
A significant portion of the residual input to sediments constants, respectively, and T is the temperature in °C.
may thus originate from remote areas with very different Magnesium replaces calcium more easily at high than low
hydrological conditions. This problem is particularly temperatures, so the Mg/Ca ratio from carbonates increases
significant in areas of low productivity or in frontal zones with temperature at the time of calcite formation. Thermo-
that separate two distinctly different water masses (Sicre dynamic considerations suggested an exponential tempera-
et al. 2005). ture dependence of Mg uptake into calcite (Rosenthal
– Reconstructions could be biased toward a specific season et al. 1997).
(Rosell-Melé and Prahl 2013) and a degree of nonlin- However, the growth temperature is not the only factor to
earity may exist in the relation between alkenones and be considered. Seawater salinity and alkalinity have also
SST at the higher and lower ends of the temperature been shown to significantly alter the Mg/Ca ratio in
range (Conte et al. 2006).

Another organic tracer to reconstruct past SST is based on

the quantification of the average number of cyclopentane
rings found in glycerol dialkyl glycerol tetraethers (GDGTs)
of archaea membrane lipids. An index, called TEX86, was
deduced after analyzing the GDGTs distribution in marine
surface sediments in comparison to annual mean SSTs
(Schouten et al. 2002).
Recently, a number of different TEX86 calibrations have
been developed (Kim et al. 2010; Tierney and Tingley 2014;
Ho and Laepple 2016), in response to possible differences in
membrane adaptation of the resident archaea communities at
different temperatures, and to the differences found between
the TEX86 ratio and other SST reconstruction proxies.
A number of pre- and post-depositional processes can
influence the TEX86 ratio. For some processes, this influence
can be constrained. For example, the BIT index is used to
track the amount of terrestrial GDGT input, using a ratio of
branched versus isoprenoid GDGTs (Weijers et al. 2006;
Schouten et al. 2013). Nonetheless, the scientific under- Fig. 21.5 Mg-temperature calibration results from culturing experi-
standing of TEX86 remains imperfect, particularly since the ments with one species of planktonic foraminifera (Globigerina
effects of environmental factors such as salinity, nutrient bulloides) (black dots) and core top samples (orange dots). Mg/Ca
results are plotted versus calibration temperature (black dots) or World
concentrations, and water column structure may modulate Ocean Atlas mean annual SST (orange dots) (modified from Mashiotta
the TEX86–SST relationship (Tierney and Tingley 2014). et al. 1999)
232 T. Caley et al.

foraminiferal tests (Elderfield et al. 2006; Mathien-Blard and

Bassinot 2009; Hönisch et al. 2013). Another issue is the (Standard Mean Ocean Water) international standard.
observed offsets in both culture and field studies between These standards are distributed by international agen-
Mg/Ca ratios among individual species, that indicate the cies for laboratory calibration. To properly apply the
need for single-species calibrations. In addition, the geo- paleotemperature formula, which presumes that all
graphical extension of genotypes must be assessed when isotopic compositions are expressed relative to the
choosing to develop calibrations (Vázquez Riveiros et al. same standard, it is necessary to compare the isotopic
2016). Recent studies try to simultaneously assess the rela- ratio of the CO2 extracted from the PDB standard by
tionship between foraminiferal Mg/Ca, and temperature, controlled phosphoric acid attack with the isotopic
salinity, and the carbonate system using statistical approa- ratio of the CO2 isotopically equilibrated with the
ches (Khider et al. 2015; Gray et al. 2018). SMOW standard. The latter is lower in 18O content by
Mg/Ca in foraminifera has been cited here as a main 0.27‰ than the CO2 extracted from PDB, so that for
example of a geochemical temperature tracer. Other ratios, every water sample:
such as Sr/Ca or Li/Mg in corals, are also used as tracers of
dw ðvs:PDB  CO2 Þ ¼ dw ðvs:SMOW  CO2 Þ0:27:
temperature (Corrège 2006; Montagna et al. 2014).
ð21:2Þ
Isotopic Tracers If, as in the paleotemperature formula, PDB is used
As presented in this chapter’s introduction, the first isotopic as the standard for carbonates and SMOW as the
approach developed was the relationship between tempera- standard for waters, then Shackleton’s equation
ture, the isotopic composition of seawater and the isotopic becomes:
composition of the biocarbonate that developed within that
water. Traditionally, isotopic compositions are expressed T ¼ 16:94:38  ðdc  dw þ 0:27Þ
using the notation d, which is the relative difference (ex- þ 0:10  ðdc  dw þ 0:27Þ2 :
pressed in parts per thousand) between the isotope ratio R of
the sample and that of a reference standard: One major disadvantage of the paleotemperature formula

is that temperatures can only be determined if the isotopic
d ¼ ½ Rsample =Rst  1  1000
composition of the water is known, which is almost never
The d18O of the water is denoted dw, that of a carbonate, the case for geological samples.
dc, and the relationship between temperature T, dw and dc is A more recent isotopic method, still under development,
known as the ‘paleotemperature equation’. This relationship is expected to overcome this constraint (Ghosh et al. 2006;
was experimentally determined by Urey’s group in the Schauble et al. 2006). The crystal lattice of a carbonate
1950s and improved by Shackleton (1974) in the form consists of CO2 2+
3 groups and of cations (Ca , for example).
below: Among the CO2 3 ions in a sample, the heavy isotopes
13
C
18
and O do not spread out randomly. Their relative abun-
T ¼ 16:9  4:38  ðdc  dw Þ þ 0:10  ðdc  dw Þ2 ð21:1Þ dance will depend on the isotopic equilibrium reaction:
In this empirical formula, dc represents the d18O of the 13 ð2Þ ð2Þ ð2Þ ð2Þ
C16 O3 þ 12 C18 O16 O2 ,13 C18 O16 O2 þ 12 C16 O3
CO2 extracted from the carbonate through dissolution with
phosphoric acid, and dw is the d18O of the CO2 obtained by
equilibration with the seawater to be analyzed. dc and dw are so that the distribution of these four isotopic species depends
measured by mass spectrometry using the same CO2 labo- on their own binding energy, itself a function of temperature.
ratory standard. Other d18O—temperature relationships The abundance of the various isotopic species is assessed
defined in the last decades (Bemis et al. 1998; Mulitza et al. by dissolving the carbonate with phosphoric acid and mea-
2003; Marchitto et al. 2014) use the same terminology. suring the abundance of 13C18O16O molecules (with a mass
of 47) in the extracted CO2, and comparing this to the
abundances of other isotopic species with masses 45 and 46.
Box 1. Practical Application of the Paleotem- The ‘stochastic’ state is taken as a reference and is defined
perature Formula by a random distribution of the isotopes of C and O within
Nowadays, isotopic geochemistry laboratories have the molecules.
adopted the convention of expressing the dc isotopic The thermodynamic variable, denoted as D47, which
compositions against the PDB (Pee-Dee Belemnite) describes the state of the carbon dioxide and from which we
international standard and dw against the SMOW deduce a paleotemperature, is defined by the relationship:
21 Climate and the Evolution of the Ocean: The Paleoceanographic … 233

D47 ¼ ½ðR47/R47  1Þ  ðR46/R46  1Þ Sea Surface Salinity

ðR45/R45  1Þ  1000
While the temperature of seawater varies over a range of
where R45, R46 and R47 are the abundance ratios of masses more than 30 °C in the open ocean, salinity changes much
45/44, 46/44 and 47/44, respectively, in the CO2, and where less (between 33 and 38 g of salt per liter (psu)). Salinity is
R 45*, R 46* and R 47* denote these ratios for a gas with highest in tropical areas, where evaporation exceeds pre-
the same overall composition but in ‘stochastic’ state (Ghosh cipitation (Fig. 21.7), and it decreases where precipitation
et al. 2006). dominates, in the equatorial belt and at high latitudes. As the
The main advantage of this method (called the D47 hydrological cycle is greatly affected by the glaciations,
method) is that the thermodynamic-measured value reflects significant variations in the salinity of the ocean during the
an internal balance of the crystal lattice and requires no Quaternary are to be expected.
knowledge of the composition of the original water. This Temperature and salinity jointly determine the density of
method is based on properties obeying thermodynamical seawater, the driver of deep ocean circulation. Dense waters
principles, so it can be applied to a variety of environments sink at high latitudes and are progressively redistributed by
without changes (Fig. 21.6). deep currents through the various basins of the world’s
However, as it is the case for the paleotemperature for- oceans. The reconstruction of the distribution of surface
mula, the calibration performed by Ghosh et al. (2006) only water salinity in the past would thus contribute to the
applies to carbonates precipitated at thermodynamic equi- understanding of why and how ocean circulation changed
librium. Therefore, testing for the possible existence of when climate conditions were different from today. It will
effects of parasitic isotopic fractionation of kinetic, biologi- also provide modelers with quantitative estimates to use as a
cal or diagenetic origin should be carried out (Eiler et al. forcing of numerical climate models.
2014; Saenger et al. 2012). Estimating the surface water salinity distribution of past
The temperature sensitivity of the D47 proxy is low oceans is difficult, partly due to the close correlation of
(*0.003‰/°C) (Kele et al. 2015). It requires high mea- temperature and salinity. Because of this correlation, chan-
surement precision, which is commonly achieved by ges in plankton distribution and transfer functions do not
increasing counting times and/or the number of replicates differentiate between changes in SST and changes in salin-
analyzed per sample, a challenge for foraminifer-based ity. Moreover, the dominant signal recorded by most indi-
reconstructions that use low carbonate samples. Recent cators is often temperature.
studies have focused on the development of precise stan- Nonetheless, reconstructions of past sea surface salinity
dardized calibrations that are applicable to paleoceano- (SSS) using transfer functions of dinoflagellate or diatom
graphic studies (Peral et al. 2018). assemblages have been proposed in specific marine envi-
ronments (DeSève 1999; De Vernal et al. 2001) with an
accuracy of ±1.8 psu for the present day (De Vernal et al.
2001). However, these methods are difficult to extrapolate
unambiguously to a global scale due to non-analogue situ-
ations in the past.
The most common method presently used to reconstruct
past SSS is the calibration of salinity against stable oxygen
isotope ratios measured on foraminifera (Duplessy et al.
1991; Malaizé and Caley 2009). Geochemical methods
based on the analysis of trace metals have recently been
developed; we will briefly discuss these two approaches in
the sections that follow.

Isotopic Methods

Paleosalinities from Stable Oxygen Isotopes (d18O)

In the open ocean, the isotopic composition of seawater is
Fig. 21.6 D47 of CO2 extracted from calcites grown from aqueous closely correlated to salinity (Figs. 21.7 and 21.8): the vapor
solution and of deep-sea corals (in orange) and surface corals (in black),
plotted against 106/T2, where T is the known growth temperature in
pressure of H18 16
2 O being lower than that of H2 O, isotopic
Kelvin (modified from Ghosh et al. 2006) ratio of vapor in the atmosphere is systematically lower than
234 T. Caley et al.

Fig. 21.7 Surface salinity of the modern oceans (WOA 2013: Zweng et al. 2013) and gridded data set of surface d18O of seawater (dw) (Legrande
and Schmidt 2006)

in the condensed phase (ocean and rain). The more evapo- In the past, dw has been seen to vary globally. When the
ration exceeds precipitation, the higher the surface water ice caps grew on land, they trapped snow poor in 18O, and so
isotopic ratio. At local to regional scale, there is a linear the dw of the ocean increased. Thus, we observe a simulta-
relationship between salinity and dw (Fig. 21.8). This is the neous drop in sea level and an increase in dw. Conversely,
relationship which allows the estimation of past seawater when the ice caps melt, sea level rises and dw decreases. The
salinities. most recent studies (see Chap. 24) estimate that sea level
The major uncertainty in the ‘paleotemperature formula’ dropped by about 120 m during the LGM, and that the
now becomes an advantage. dw can be determined when dc average isotopic composition of the ocean was then at
is measured if the foraminiferal habitat temperature T is +1.0‰ SMOW (while the current value is 0‰ SMOW by
estimated independently (by Mg/Ca ratios analysis of the definition). Various approaches aim to reconstruct changes
same shells for example). in sea level linked with changes in continental ice volume.
21 Climate and the Evolution of the Ocean: The Paleoceanographic … 235

Fig. 21.8 The relationship

between surface dw and salinity
for the global ocean as measured
within the international
GEOSECS program (Ostlund
1987), by Craig and Gordon
(1965), and for the South Indian
Ocean with the GISS database
(Schmidt et al. 1999)

It is then feasible to estimate the change in mean ocean dw

related to variations in continental ice volume and sea level. salinity of 34.7 psu, and since the amount of salt in
Locally, changes in the hydrological cycle (evaporation, the ocean remains constant, a drop in sea level of
precipitation, water mass movements, melting events) can 120 m is also accompanied by an increase in
cause additional variations in dw, of both climatic and salinity. The average ocean salinity becomes:
hydrological origin.
ð34:7  3900Þ=3780 ¼ 35:8 psu:
Reconstructing the evolution of the isotopic composition
of seawater in the past is in itself an interesting task because
dw tracks changes in the hydrologic cycle. However, we can Salinity has thus increased by about 1.1 psu and
try to qualitatively interpret a record of paleo dw in terms of dice
w by +1.0‰. Coupled models of the Northern
paleosalinities (see Box 2). Hemisphere ice sheets allow the ice-sheet contri-
bution to the variability in oxygen isotope com-
position and sea level changes to be determined
Box 2. Practical Calculation of Paleosalinities (Bintanja et al. 2005).
Estimating paleosalinity changes along a sediment – Estimate the variation in local isotopic composition
core requires the following steps: dlocal due to hydrological changes by subtracting
w
ice
dw changes from the dw value reconstructed for
– Measure dc and T at each level so as to derive a each core level. The corresponding change in
recording of paleo dw over the time period covered salinity can be estimated from current observations
by the core, using the paleotemperature equation (Fig. 21.7).
(Eq. 21.1).


– Estimate global dw changes dice w related to con- The statistical error associated with this approach is
tinental ice volume variations over the study period high and rarely permits meaningful quantitative
using known records of changes in sea level. salinity reconstructions because of the associated large
A drop of 120 m in sea level is accompanied by an uncertainties (Rohling and Bigg 1998; Schmidt 1999;
increase in dice
w of +1.0‰. Given an average depth Rohling 2000; Legrande and Schmidt 2011; Caley and
of the modern ocean of *3900 m and an average Roche 2015). The structural/analytical error is in the
236 T. Caley et al.

(foraminifera are zooplankton and coccoliths are phyto-

range of 0.8–1.8 psu (Schmidt 1999). In addition, the plankton) could emerge, together with differences in disso-
spatial and temporal evolution of the slope of the dw- lution and bioturbation in a sediment core.
salinity relationship, tested in isotope-enabled numer-
ical climate models, can lead to very large errors
(Legrande and Schmidt 2011), up to 25 psu in certain Chemical Methods
regions for the LGM (Caley and Roche 2015).
To reduce these very large errors on past SSS Calibrations established using the modern Ba/Ca-salinity
reconstructions, model-derived temporal slopes of the relationship (Carroll et al. 1993; Weldeab et al. 2007)
dw-salinity relationship can be used directly in the (Fig. 21.9) have suggested that the Ba/Ca ratio of for-
calculation. This approach has been tested with suc- aminiferal CaCO3 can be used as a proxy for river runoff.
cess for the LGM on a marine sediment record located This approach is limited to coastal regions affected by river
in Gulf of Guinea and influenced by West African runoff (i.e. prone to relatively large salinity changes) and
monsoon hydrology (Caley and Roche 2015). How- assumes that (1) the Ba/Ca ratio in planktonic foraminifera
ever, allowing model-derived regional dw-salinity shells is dominated by the Ba/Ca concentration of seawater
relationships to vary through time can lead to signifi- (Hönisch et al. 2011) and not by other factors and (2) that the
cant uncertainties related to the shortcomings of the present-day calibration is applicable to the past.
models, so complementary approaches should also be Another recent study has established the potential of the
developed. Na/Ca ratio of foraminiferal calcite as a quantitative proxy
for past salinities. In culture experiments, Wit et al. (2013)
Paleosalinities from Stable Hydrogen Isotopes studied sodium incorporation in the benthic foraminifera
(d2H) Ammonia tepida at a range of salinities and suggested that
Another method uses hydrogen isotope changes to recon- foraminiferal Na/Ca could serve as a robust and independent
struct paleosalinities. Culture experiments have found a proxy for salinity. More recently, the field study of Mezger
constant offset between the hydrogen isotopic composition et al. (2016) on planktonic foraminifera also suggested that
of water and the hydrogen isotopic composition of alkenones salinity controls foraminiferal Na/Ca. Incorporation of Na in
synthesized in that water (Paul 2002; Englebrecht and Sachs foraminiferal calcite could therefore constitute a potential
2005). Schouten et al. (2006) demonstrated that this offset proxy for salinity, although species-specific calibrations are
was dependent on salinity via biological fractionation pro- still required and more research on the effect of temperature
cesses. Reconstructing salinity by using the biological frac- is needed.
tionation factor that is linked to it requires information on the
past hydrogen isotope ratio of seawater (d2Hw).
Isotope-enabled climate model results indicate a rather
stable dependence between d2H and surface dw in the past
(Caley and Roche 2015). As dw can be reconstructed (see
Sect. “Chemical Methods”) this suggest that d2Hw can also
be obtained. An estimation of paleosalinities based on d2H
measurements in alkenones might therefore be possible if the
slope and the intercept of the regression between the bio-
logical fractionation factor and salinity can be sufficiently
constrained. The impact of species composition and growth
phase on the use of alkenone d2H to reconstruct paleosalinity
currently requires further investigations (Wolhowe et al.
2009; Chivall et al. 2014; M’Boule et al. 2014).
Pairing information from water isotopes, d18O and d2H
(isotopologues), could yield better estimates for paleosalinity
(Rohling 2007; Leduc et al. 2013). Numerical modeling
experiments for the Holocene and the LGM periods have
demonstrated that this combination of water isotopologues
may indeed allow for a better estimation of paleosalinity
variability (Legrande and Schmidt 2011; Caley and Roche
Fig. 21.9 Relationship between Ba in seawater and salinity in the Bay
2015). Nonetheless, ecological biases introduced by com- of Bengal (Carroll et al. 1993)
bining proxies based on two different organisms
21 Climate and the Evolution of the Ocean: The Paleoceanographic … 237

Reconstruction of the Hydrology of the Deep dependent on the climate of the high latitudes of the
Ocean Northern Hemisphere (North Atlantic Ocean, Norwegian
and Labrador Seas) and of the Southern Ocean in the
Main Features of Modern Circulation Southern Hemisphere.

On a rotating planet like the Earth, surface ocean circulation

is governed by the winds and the position of the continents Reconstructing the Temperature and Salinity
that define the shape of the ocean basins. Deep circulation, of Deep Waters
on the other hand, is governed by the small variations in the
density of the water masses. Since density depends on The past evolution of the deep ocean has been the subject of
temperature and salinity, the term ‘thermohaline circulation’ extensive research. However, the reconstruction of the basic
is used. Reconstructing past changes in temperature, salinity properties of bottom waters has been hindered by the lack of
and density in the intermediate and deep ocean is thus transfer functions linking the abundance of benthic species
needed to understand the temporal variations of the circu- to the temperature of seawater in the vicinity of the sedi-
lation and distribution of water masses; it also provides a ment. In many environments, these abundances are essen-
benchmark for simulations provided by general circulation tially governed by the availability of food and by the
models. proportion of dissolved oxygen. However, paleoceanogra-
As we saw in Chap. 1, deep waters are currently formed phers have attempted to apply other approaches to the
in winter in highly localized areas of the high latitudes: the reconstruction of bottom water temperature.
Norwegian Sea and the Labrador Sea in the northern
hemisphere and the Weddell Sea and the Ross Sea near Searching for a Reference Zone with Constant
Antarctica. During winter, surface waters here become Temperature
denser as they cool down but also because the formation of As early as 1967, Shackleton (1967) suggested that the d18O
sea ice is accompanied by a release of salt. When surface of benthic foraminifera must closely resemble the d18O of
waters become as dense as deep waters, large-scale con- deep waters, because these waters are formed close to
vection movements are initiated and the waters sink into the freezing conditions and their temperature cannot drop much
abyssal depths. Once at depth, very small changes in density further during a glacial period. Labeyrie et al. (1987) further
of the various deepwater masses govern their circulation elaborated on this concept by analyzing the d18O of benthic
through the different basins. Surface waters sinking in the foraminifera from a core in the Norwegian Sea where tem-
Norwegian Sea cross the sills separating it from the Atlantic perature, well below 0 °C in the deep basins, is constrained
Ocean to form North Atlantic Deep Water (often referred to by exchanges with the ice. By comparing the isotopic record
by the acronym NADW). This water mass then follows the from this core with others from the Pacific and Indian
American coast to join the Southern Ocean and is caught up Oceans, the authors were able to demonstrate, contrary to the
in the Antarctic divergence, a large upwelling zone, where it assumptions of Shackleton, that deep water temperature in
mixes with the surface waters of southern high latitudes. the major ocean basins did change significantly at the
Here, these very cold surface waters increase their density beginning of the last glaciation, with a cooling in all deep
through winter sea ice formation, and sink along the oceans to a temperature close to the freezing point (*−1 °
Antarctic continental shelf forming the densest waters in the C) during the LGM. This result has been confirmed since by
world. This water mass is called Antarctic Bottom Water (or other tracers such as the Mg/Ca ratio of benthic foraminifera
AABW), and it lines the bottom of all ocean basins. At (see Sect. “Estimating the Temperature Independently of the
present, abyssal waters are not formed from the surface Paleotemperature Formula”). Unfortunately, the isotopic
waters of the Indian and Pacific Oceans. The Antarctic benthic record of the Norwegian Sea is far from continuous
waters that rush into these basins create the Pacific and and this method could not be successfully applied to
Indian Deep Waters (PDW and IDW, respectively) by reconstruct the evolution of the abyssal water temperature
mixing with the waters of the main thermocline, and then over the whole of the Quaternary.
return to the Southern Ocean at around 3 km depth. We can It should be emphasized, however, that very low
devise a simplified view of the global ocean circulation, deep-water temperatures during glacial periods are to be
where the Norwegian Sea is the main source of deep waters, expected. Indeed, under current conditions, NADW is
and the Southern Ocean acts as a recirculation pump formed from very cold water (close to the freezing point)
returning to the depths the surface waters surrounding that overflows from the Norwegian Sea through the sills
Antarctica that have received upwelled NADW via the located between Scotland, the Faroe Islands, Iceland and
Antarctic divergence. This circulation pattern is critically Greenland. However, the water that crosses these shallow
238 T. Caley et al.

sills (with a depth of less than one kilometer) mixes with the salinities from one basin to another, with the Southern Ocean
much warmer waters of the North Atlantic, so that the newly being the saltiest, in contrast with the modern situation.
formed NADW is characterized by a temperature of +3 °C
and a salinity of 34.95 psu. Its density remains high, close to Reconstructing Changes in Water Mass
but slightly less than that of AABW (temperature of −1 to Distribution
0 °C, salinity of 34.6 psu). It is therefore AABW that lines Lynch-Stieglitz et al. (1999) and (2014) showed that an
the great ocean depths, and, in the Atlantic Ocean, is topped approximate direct relationship could be established between
by NADW. During the last glaciation, the sinking of very the d18O of benthic foraminifera and the density of seawater,
cold water directly into the North Atlantic (and not into the within the temperature range where the temperature-salinity-
Norwegian Sea anymore) explains why the deep waters of density relationship is roughly linear (T greater than 2 °C).
the world are all found to be at temperatures close to the In this way, the authors were able to study the geostrophic
freezing point. deformations of the deep thermocline in the Straits of Florida
and propose estimates of the changes in the meridian flow
Estimating the Temperature Independently linked to the Atlantic thermohaline circulation between the
of the Paleotemperature Formula LGM and the present.
The simplest way to determine the temperature of the water Reconstructions using benthic foraminiferal d18O have
close to the sediment is to use the concentration of trace also shown marked changes in the distribution of deep and
metals (Mg/Ca) contained in the carbonate shells of benthic intermediate water masses during the LGM compared to the
foraminifera. This independent estimate of bottom water present day. In the Atlantic Ocean, the temperature gradient
temperature allows the calculation of the dw of deep water currently observed at the base of NADW at around 3000 m
using the paleotemperature formula. was to be found at around 2000 m, and was much more
However, the temperature dependence of Mg incorpora- pronounced than the one that currently separates NADW and
tion in benthic foraminiferal tests is species-specific, and AABW (Labeyrie et al. 1992). In the Indian Ocean, a strong
may depend on different hydrological factors such as salinity gradient separated two water masses with distinctly different
or carbonate ion saturation (Elderfield et al. 2006). In characteristics at a depth of around 2000 m (Kallel et al.
addition, the expected bottom water temperature variations 1988). More recent research even suggests the presence of a
during glacial-interglacial cycles are small compared to third deep water mass in the deepest North Atlantic at the
surface temperature changes, implying relatively small LGM, formed by brine rejection and not by heat loss to the
Mg/Ca variations. The recently discovered D47 method is atmosphere (Keigwin and Swift 2017).
only starting to be applied to this problem (Peral et al. 2018),
although more precise measurements in benthic foraminifera
are needed to confirm the utility of this technique. Reconstructing the Circulation of Deep Waters

Searching for the Geochemical Signature Searching for Lines of Current from the d13C
of Ancient Waters in Pore Waters of Benthic Foraminifera
Adkins et al. (2002) found that in long cores extracted by An original approach, independent of temperature and
drilling ships, pore water trapped within the sediments salinity, tries to characterize the main features of deep water
shows measurable differences in salinity and in dw. These circulation without trying to a priori understand the under-
differences increase initially with core depth, then reach a lying physical mechanisms governing it. It is based on the
maximum after which they decrease slowly. They inter- carbon cycle and its tracer, the 13C/12C ratio usually denoted
preted this maximum as the signature of highly saline water as d13C. At the ocean surface, waters easily exchange their
from the LGM that had diffused into the sedimentary col- gas content with the atmosphere; they contain carbon diox-
umn. Using a simple diffusion model, the authors estimated ide and are rich in dissolved oxygen. During photosynthesis,
the values for the salinity and the dw of the bottom waters phytoplankton preferentially absorbs 12CO2 over 13CO2. The
20 ka ago. Using this estimate of the dw of bottom waters, organic material thus produced has a d13C close to −20‰,
and combining it with d18O measurements on the calcite of while the d13C of dissolved CO2 in surface waters varies
benthic foraminifera, the paleotemperature formula con- between +1 and +2‰. This surface organic matter forms the
firmed that the deep waters of the glacial ocean were actually base of the ocean’s food chain, and eventually falls to the
at a temperature near freezing point and hypersaline. The depths carried in fecal pellets of zooplankton and higher
sediment cores that have been measured for dw are too few animals. In the water column, settling organic matter
to give a complete picture of the ocean during the last undergoes a slow remineralization, which consumes any
glaciation. However, they show a significant disparity in the dissolved oxygen that may remain and produces CO2
21 Climate and the Evolution of the Ocean: The Paleoceanographic … 239

depleted in 13C. Consumption of dissolved oxygen and when well ventilated (with a high d13C) waters of the
production of CO2, accompanied by a decrease in d13C, take Atlantic Ocean formed Glacial North Atlantic Intermediate
place in the deep waters (Fig. 21.10). The d13C of dissolved Waters (GNAIW), at a shallower depth than today’s NADW,
CO2 in deep waters is thus lower than that of the surface while deep waters (AABW) were even more poorly venti-
waters. lated than today (Fig. 21.11). The understanding of this
The remineralization of organic matter is a slow process. variability in the thermohaline circulation, which is a major
This is why deep waters are characterized by a high d13C in regulatory mechanism of climate, is the subject of substantial
regions close to their formation area (this is the case for the research, both analytical and in modeling.
North Atlantic Ocean). Gradually, as they move away and
circulate at depth, without the opportunity to exchange with Using Trace Elements Measured in Benthic
the atmosphere, they become increasingly deprived of dis- Foraminifera
solved oxygen, while their d13C decreases through the In the modern ocean, geochemists have showed that cad-
mechanism described above. To give an order of magnitude, mium (Cd) is included in organic matter, so that its cycle
we can consider that the d13C of deep water decreases by follows that of phosphate. The concentration of dissolved Cd
about 1‰ per thousand years. in ocean waters shows therefore very similar variations to
It is therefore the waters of the deep basins of the Pacific that of dissolved phosphate, a nutrient with a well-known
and Indian Oceans, at the end of the circulation scheme cycle. It is assimilated by phytoplankton to ensure growth,
described in Sect. “Main Features of Modern Circulation”, so, as for all organic matter formed by photosynthesis, it falls
which have the lowest d13C. The evolution of d13C in the into the water column with organic debris and is gradually
deep ocean can thus be used as a tracer to characterize the released in the deep waters as bacteria oxidize it. Conse-
lines of current and the exchanges between the various deep quently, in the deep waters of the ocean, the consumption of
water masses. Epibenthic foraminifera, such as the species dissolved oxygen and production of carbon dioxide (de-
Cibicides wuellerstorfi, reflect this evolution of the water in pleted in 13C as we have seen) occur in parallel with
which they grew, and variations in their d13C in cores increases in phosphate and cadmium.
extracted from different ocean basins are used to reconstruct The Cd ion has a charge and an ionic radius similar to that
changes in ocean circulation through time (Duplessy et al. of Ca. It is therefore easily incorporated in trace amounts
1984; Schmittner et al. 2017). For example, this proxy has into the carbonate shells of benthic foraminifera, so that their
been used to reconstruct ocean circulation during the LGM, Cd/Ca ratio reflects the concentration of Cd in the seawater

Fig. 21.10 Variations in the

concentration of dissolved
oxygen, the concentration of total
dissolved CO2, the 13C/12C ratio
of the total dissolved CO2 and
phosphate with respect to depth at
GEOSECS station 322 in the
Pacific Ocean (43.0 °S/129.9 °
W). The oxygen minimum at
depth indicates consumption by
marine bacteria. The resulting
CO2 production is characterized
by a maximum of dissolved
inorganic carbon concentrations
and by a minimum of d13C (since
carbon in organic matter is
depleted by about 20‰ relative to
dissolved inorganic carbon)
240 T. Caley et al.

Fig. 21.11 (Top) Modern distribution of dissolved phosphate estimates of the Cd concentration (nmol/kg) during the LGM based on
(lmol/liter) in the western Atlantic; (middle) reconstructed benthic the ratio of Cd/Ca in the shells of benthic foraminifera (figure modified
d13C in the western and central Atlantic during the LGM; (bottom) from Lynch-Stieglitz et al. 2007)
21 Climate and the Evolution of the Ocean: The Paleoceanographic … 241

in which the foraminifera developed. The Cd/Ca ratio is and it may be difficult to obtain the amount of carbonate
therefore a tracer for nutrient concentration of the water required for analysis from a normal-size sample, even using
masses, present and past (Marchitto and Broecker 2006; the most sensitive technique, accelerator mass spectrometry.
Fig. 21.11). Second, the 14C ages of foraminifera, once in the sediment,
Cd and 13C have similar geochemical behaviors, with the are very sensitive to bioturbation: the abundance of one
essential difference that surface waters can exchange their species shows considerable variation over time, and shells
dissolved carbon dioxide with the atmosphere, while Cd is that are found at one level may have been displaced by the
not involved in ocean-atmosphere exchanges. In general, activity of burrowing animals and therefore come from a
there is an excellent anti-correlation between the variations significantly different age level than the selected one.
in the Cd/Ca ratio and the 13C/12C of benthic foraminifera Finally, the atmospheric 14C concentration has undergone
measured in sediment cores. One notable exception is the large scale changes, so much so that the difference in 14C age
Southern Ocean, where the benthic foraminifera that lived between planktonic and benthic foraminifera does not
during the last ice age have very negative d13C, while their directly reflect the residence time of the waters at depth.
concentrations in Cd are very similar to those of recent Another approach to reconstruct the dynamics of
sediments from the Holocene. Despite inter-specific differ- deep-water masses makes use of the geochemical behavior
ences that may have affected d13C reconstructions based on of the decay chain of uranium in seawater. The isotopic
benthic foraminifera (Gottschalk et al. 2016), or diagenetic composition of dissolved uranium is constant throughout the
and metabolic effects influencing the incorporation of trace ocean. Two of the isotopes of uranium, 235U and 234U, decay
metals into biogenic calcite (Marchitto and Broecker 2006; producing 231Pa and 230Th respectively, with an output ratio
McCorkle et al. 1995), this discrepancy between the two that is constant and equal to 0.093. 231Pa and 230Th are very
indicators remains still to be explained. reactive to particles sinking in the water column: they are
adsorbed on their surface and settle as sediment along with
Reconstructing the Dynamics of Water Masses them. However, 231Pa is less reactive than 230Th, so that the
The tracers we have described so far allow the reconstruction residence time in seawater of dissolved 231Pa is close to
of specific physical or chemical characteristics of water 200 years, while that of dissolved 230Th is only thirty years
masses, but they do not convey information on their (Yu et al. 1996). The residence time of 231Pa is close to that
dynamics. In this section, we do not discuss in detail tracers of NADW in the Atlantic. Because of this, a fraction of the
related to particle transport at the ocean floor (particle size dissolved 231Pa is advected out of the North Atlantic Ocean
distribution, magnetic grain size, sortable silt). However, we by NADW (about 50% in the modern ocean), while 230Th is
will discuss two unstable radioactive tracers in the ocean: the unaffected and settles completely with the particles. The net
concentration in 14C of benthic foraminifera and the excess loss of 231Pa in the water column at depths greater than or
231
Pa/230Th ratio in sediments. equal to the level at which NADW flows, leads to a deficit of
231
When surface waters exchange carbon dioxide with the Pa in the sediments and therefore in the 231Pa/230Th ratios
atmosphere, they absorb 14C. Currently, the 14C concentra- below the production ratio (0.093). If the circulation of
tion in the surface ocean is 95% of that of the atmosphere. NADW becomes slower, less 231Pa is advected out of the
When surface waters sink, they bring with them the dis- basin, and the 231Pa/230Th ratio of the particles settling to the
solved carbon dioxide as well as the 14C they contain. Once sediment increases to values closer to the production ratio.
they reach the bottom of the ocean, these waters are isolated It should be noted that sediments also contain 231Pa and
from the atmosphere, and 14C decreases due to its own 230
Th, present as daughter isotopes of the uranium in clays
radioactive decay, with its period of 5720 years. The oldest that have reached secular equilibrium with their relevant
waters in the northern Pacific and Indian Oceans have an parents. The addition of 231Pa and 230Th from settling par-
apparent age of around 800 years (see Chap. 4). The ticles thus produces an excess of these two radioisotopes in
planktonic foraminifera (that live in surface waters) and the the sediment. Measuring the ratio of excess 231Pa/230Th in
benthic foraminifera (that live on the ocean floor) incorpo- the sediments therefore allows the variations in the circula-
rate the 14C of the waters around them into their shells. By tion of deep waters from the North Atlantic towards the
comparing, at the same level of sediment, the 14C ages of Southern Ocean to be traced, and so to detect the variability
planktonic and benthic foraminifera, we can estimate the associated with major changes in climate (McManus et al.
apparent age of the deep waters over the last *40,000 2004; Gherardi et al. 2009; Lippold et al. 2016). This
years. This apparently simple method presents in fact many technique has been used with success to reconstruct the
difficulties. Firstly, despite recent analytical developments, ‘strength’ of the thermohaline circulation back to *140 ky
benthic foraminifera are not always sufficiently abundant, ago (Guihou et al. 2010, 2011; Böhm et al. 2015).
242 T. Caley et al.

Major Fields of Paleoceanography ago). This decline is characterized by an increase in d18O

over time that reflects the drop in ocean temperatures at
A prerequisite to reconstruct ocean circulation in the past is higher latitudes (where deep waters are formed) in a world
that marine sediments have not been buried in the mantle at without major glaciers, followed by the growth of ice caps.
subduction zones. Due to the renewal of the ocean floor by The d18O of benthic foraminifera therefore dropped from an
plate tectonics, the oldest sediments date back to the Trias- average of *0.2‰ during the climate optimum of the early
sic, about 200 million years ago. These very old sediments Eocene, to an average of *4‰ at the end of the Quaternary.
are rare, and found only in the Pacific Ocean. In practice, we This slow climate decline is generally attributed to two
can hope to obtain global reconstructions for the whole main factors: a gradual reduction in atmospheric CO2 con-
Tertiary era, but they become increasingly scarce as we go centration, and the growing thermal isolation of the Antarctic
back to the Secondary era. Here, we will restrict ourselves to continent due to the widening of the ocean passages that
the analysis of some of the major aspects of the evolution of surround it.
climate and paleoceanography over the last 25 million years, Relatively abrupt incidents (that is, events happening over a
which correspond to the progressive establishment of glacial much shorter time scale than the general trend), of large
conditions in the high latitudes of both hemispheres. amplitude, are superimposed onto this slow drift, indicating
that other climate drivers such as thresholds or rapid feedbacks
are also involved. The first of these changes occurred at the
From ‘The Greenhouse Effect Era’ to very end of the Paleogene, at the boundary of the
the ‘Ice Ages’ Eocene/Oligocene, about 33.5 Ma ago. It resulted in a rapid
increase in benthic foraminiferal d18O from +1.6 to +2.8‰
The Quaternary period, which covers the last 2.6 million over only 100–200 ka (Fig. 21.2). This change in d18O is
years of the history of our planet, is characterized by per- attributed to the development of the Antarctic ice cap. The
sistent major ice sheets in the high latitudes of both hemi- northward drift of the Australian continent allowed the open-
spheres. The volume of these caps and their geographical ing of the Strait of Tasmania and the establishment of the
expansion fluctuate over timescales of 104 to 105 years, in Antarctic Circumpolar Current (Zachos et al. 2001). The iso-
response to changes in insolation controlled by the orbital lation of this large land mass resulted in a drop in temperatures
parameters of the Earth (see Chap. 28). and the establishment of a permanent ice cap on East
The origin of these major glacial phases is discussed in Antarctica, with an ice volume that may have reached about
several chapters that present the point of view of geo- 50% of its current size. After a period of ten million years when
physicists, geochemists and modelers (see Chaps. 22, 26 and the climate changed little, two warming phases, one at the end
27). At time scales exceeding a million years, the major of the Oligocene and the other during the Middle Miocene,
causes of the development of glaciations are feedbacks caused a significant reduction of the Antarctic ice sheet. The
related to plate tectonics. Among these feedbacks, the fol- increase in temperature and the decrease in ice volume were
lowing may be highlighted: reflected in a 1.2‰ decrease in the benthic d18O signal. The
causes of these climatic changes are not clear yet.
– the varying shapes and positions of landmasses and ocean The final phase of the climatic decline leading to the
basins, major glaciations of the Quaternary began from the Middle
– the existence of passages between basins, Miocene, between 14.2 and 12.2 Ma, and was marked by an
– the location and altitude of mountain ranges, both in the increase in benthic foraminiferal d18O of 1.0‰ over two
ocean and on land, affecting oceanic and atmospheric million years. Paleotemperature reconstructions based on
circulation and heat transfer, Mg/Ca ratios in benthic foraminifera suggest that deep water
– volcanism and erosion, and their impact on atmospheric temperature varied little, and that the growth of the Antarctic
chemistry and on the pCO2 of the atmosphere. ice cap was the main contributor to the benthic d18O change
(around *0.8‰). If these estimates were correct, the ice cap
The record describing the evolution of the d18O of ben- located on the eastern part of Antarctica would have reached
thic foraminifera over time shown in Fig. 21.2 is the result 85% of its current volume. As for the West Antarctic ice cap,
of a compilation of analyses made in more than forty it seems to have only developed from 6 Ma on, as evidenced
ocean-drilling sites. This global compilation registers both by the first coarse sediment deposits in the Weddell Sea,
the changes in dw and temperature, in accordance with the coming from melting icebergs emitted at the edge of this
paleotemperature formula (Eq. 21.1). It shows that the cur- cap. These sediments transported by drifting ice result from
rent climate is the result of a long decline that started at the land erosion caused by the friction of glaciers, and are often
end of the climate optimum of the early Eocene (52–50 Ma referred to by their acronym IRD (ice rafted debris).
21 Climate and the Evolution of the Ocean: The Paleoceanographic … 243

Although the development of a permanent ice cap on large portion of North America and Europe. During the
Antarctica started early, during the early Neogene, the Holocene, the interglacial period we currently live in, these
development of perennial continental ice caps in the high caps were largely diminished; the meltwater derived from
latitudes of the northern hemisphere did not occur until the them has caused sea level to rise by 120 m since the LGM.
end of the Neogene. The first deposits of IRD in the Nor- However, an ice cap of 2.8 million km3 continues to exist on
wegian Sea, proof of the early development of an ice cap Greenland that would cause a further sea level rise of about 7
(although perhaps not a permanent one) on Greenland are meters, if it were to completely melt.
not observed before 5.5 Ma (Jansen and Sjoholm 1991). The The amplitude of the glacial-interglacial oscillations
rapid increase in d18O from *3.2 Ma onwards may be increased sharply between 1.2 and 0.6 Ma (Fig. 21.2).
interpreted as the beginning of permanent glaciation in high During this period, called ‘the Middle Pleistocene Transi-
latitudes of the northern hemisphere. This glaciation inten- tion’ (Clark et al. 2006; McClymont et al. 2013), a threshold
sified rapidly around 2.1–2.6 Ma, as evidenced by the response to longer-term atmospheric CO2 decline has been
massive IRD deposits in the Norwegian Sea. proposed (Raymo et al. 1997). However, recent atmospheric
Many studies have focused on the hypothesis of the partial pressure CO2 reconstructions have failed to show this
‘closure of the Panama isthmus’ as a potential trigger for the long-term decrease during the Pleistocene (Hönisch et al.
development of northern hemisphere ice sheets. The mech- 2009). The gradual increase in glacial-interglacial amplitude
anism would involve warm intertropical waters no longer is mainly due to increasingly high values of d18O during
being able to cross from the Atlantic to the Pacific. They glacial periods. The few available reconstructions of deep-
would therefore deviate into the North Atlantic, increasing water temperature during this period indicate near-freezing
oceanic evaporation in this basin and thus snow accumula- temperatures at every glacial maximum instead of a gradual
tion in the high latitudes. However, recent studies suggest cooling (Elderfield et al. 2012), which suggests that an
that the closure of the Panama isthmus could have occurred increase in Antarctic ice volume would be responsible for
during the Miocene, well before the intensification of the rapid and steep increase in seawater d18O at 0.9 Ma.
glaciations (Montes et al. 2015). The recent work of Rohling This change in amplitude of glacial-interglacial oscilla-
et al. (2014) also observes a large temporal offset during the tions is accompanied by a disruption in the frequency con-
onset of the Plio-Pleistocene ice ages, between a marked tent of the global d18O signal. While benthic foraminifera
cooling step at 2.73 My ago and the first major glaciation d18O oscillations show mainly a cycle of *41 ka over most
starting 2.15 My ago. Other theories indicate that a decrease of the Neogene and early Quaternary, the last 600,000 years
in atmospheric CO2 may have been responsible for a cool- are dominated by oscillations with a cyclicity of *100 ka
ing, an increase of deep water formation in the North (Fig. 21.2). Some authors have agreed on the progressive
Atlantic and a change of circulation that together induced the nature of the ‘Middle Pleistocene Transition’, with the
start of the glaciations. amplification of the 100 ka cycles occurring over hundreds
of thousands of years. However, in some ocean regions, the
records fail to demonstrate this progressive nature. This is
The ‘Middle Pleistocene Transition’ the case, for example, in the equatorial Atlantic, where the
and the Establishment of 100-Ka Cycles dynamics of the thermocline, reconstructed from micropa-
leontological tracers, suddenly change its variability
The trend towards the climatic decline (seen as an increase in around *930 ka. The mechanisms responsible for this
benthic d18O) discussed in Sect. “From ‘The Greenhouse transition are still unclear, although it appears that an
Effect Era’ to the ‘Ice Ages’” continued over the last two important role can be attributed to the enormous Laurentide
million years, as is shown in detail in Fig. 21.2. Superim- ice sheet, which may have favored the frequency of 100 ka
posed on this trend are quasi-periodic oscillations. They through its inertia (Clark and Pollard 1998) (see Chap. 28).
reflect the alternating glacial periods—corresponding to a
cooling of deep waters and an increase in ice volume at high
latitudes—and interglacial periods, with warming and rela- The Last Glacial Maximum (LGM)
tive melting of the ice caps. It should be emphasized here
that the use of the terms ‘interglacial’ and ‘glacial’ does not The LGM has long been, and still remains, a major area of
imply a total melting of ice sheets. During interglacial interest in paleoclimatology, in particular because it presents
periods, ice sheets do not disappear, even if they are greatly another extreme on the climate spectrum on which Earth
reduced in the northern hemisphere. For example, during the System models can be validated (Kageyama et al. 2018).
LGM, the ice sheets in the northern hemisphere covered a Early studies defined this period as the time encompassing
244 T. Caley et al.

the last great cold maximum (as recorded by micropaleon- It therefore became essential to reconsider the recon-
tology and pollen) as well as the maximum spread of the ice struction of the surface ocean during the LGM. The latest
sheets (marked by the position of moraines on land masses). and most comprehensive synthesis was carried out in the
Radiocarbon dating has placed the maximum at around MARGO program (MARGO Project Members 2009), which
16-20 ka 14C (equivalent to 18–23 ka in calendar age). This focused on the period 19–23 ka corresponding to the LGM
period was the first to be the subject of a global paleoclimate sensu stricto. This period corresponds to the maximum
study, thanks to the CLIMAP group. The isotopic maximum expansion of ice sheets, as opposed to the coldest conditions
in d18O of planktonic and benthic foraminifera, interpreted of Heinrich Stadials. The LGM MARGO reconstructions
as reflecting the cumulative effects of the cold and ice vol- (Fig. 21.3) agree relatively well with CLIMAP, but they also
ume maxima, was used as a stratigraphic marker, and revealed some important differences:
summer and winter sea surface temperatures were deter-
mined from micropaleontological transfer functions. This – the northern seas were ice-free during the summer;
established the CLIMAP maps (CLIMAP 1981) (Fig. 21.3) – latitudinal and longitudinal thermal gradients were
that served as boundary conditions for the first comprehen- strong; the mid-latitudes of the North Atlantic Ocean
sive paleoclimate modeling experiments. The CLIMAP experienced the strongest cooling (*−10 °C);
results have had a profound impact. For the first time, the – the decrease in temperature was generally larger on the
magnitude of the temperature change between an ice age eastern side than on the western side of the oceans; this
(LGM) and an interglacial period (modern times) could be was particularly marked along the African margin,
quantified: the average global temperature dropped by 6 °C. especially in coastal upwelling zones of Namibia and
However, this cooling was far from uniform: it exceeded South Africa;
10 °C at high northern latitudes, while it was only a few – the cooling of tropical waters was close to 2 °C, although
degrees in the intertropical region. This intense cooling was some localized waters of the Pacific and Indian Oceans
associated with the development of large ice sheets over the experienced moderated warming;
landmasses of the northern hemisphere, which, with about – in the Southern Ocean, a cooling of 2–6 °C marked a
50 million cubic kilometers of ice, was the most glaciated northward displacement of the polar front.
hemisphere. In addition, analyses of Antarctic ice showed
that atmospheric CO2 concentration was about 100 ppmv Other studies have focused on the deep ocean during the
below pre-industrial values (Petit et al. 1999). LGM. We have mentioned several of them in the description
It quickly became necessary to expand these early stud- of the various methodological techniques that have been
ies. Continental tracers in tropical regions, such as pollen developed over the last forty years. Significant differences
series or concentration of noble gases in aquifers (which are between the LGM and the present day include:
dependent on the temperature of the rains feeding these
aquifers), indicated a cooling of 3–6 °C during the LGM. In – the downwelling of surface waters in the North Atlantic
the nearby ocean, sea surface temperature alkenone recon- happened in open ocean, leading to the formation of very
structions indicated a cooling of only about 2 °C, and cold deep water that found its density equilibrium at
micropaleontological transfer functions showed little or no 2000 m depth;
change. Detailed studies were therefore conducted in later – the very cold and dense bottom waters formed in the
decades to explain the observed differences. It appeared that southern hemisphere spread throughout the deep ocean,
many cores taken from the tropical Pacific Ocean and used occupying a much larger volume than today;
for the CLIMAP reconstruction had very low sedimentation – the boundary between deep and bottom waters was
rates, so that bioturbation caused the contrasts in fauna over characterized by a much stronger gradient of physical (T,
time to disappear. In addition, the fauna from warm waters S, density) and geochemical (d18O, d13C) properties than
exhibited variability that did not solely respond to temper- today;
ature changes, with the result that micropaleontological – the ventilation and renewal rates of deep waters are still
transfer functions became insensitive at the temperatures poorly constrained because of conflicting information
above 25 °C common in tropical regions. At the same time, from different tracers with a complex geochemical
high-resolution studies started to indicate strong climatic behavior (14C, 231Pa/230Th); this uncertainty is also
variability between 17 ka and 25 ka. In the North Atlantic, reflected in the simulations from general circulation
for example, the LGM does not correspond to the coldest models of the ocean and coupled ocean-atmosphere
conditions, which are instead associated with two periods models. Most proxies do indicate, however, lower ven-
framing the LGM: Heinrich Stadial (HS) 2 at around 24– tilation of the deep ocean and a resulting large accumu-
22 ka and HS1 at around 19–17 ka. lation of carbon dioxide in the deeper waters.
21 Climate and the Evolution of the Ocean: The Paleoceanographic … 245

The Last Deglaciation taken (the proglacial lake of Bølling and the city of Allerød
in Denmark).
Several decades ago, continental paleoclimatologists The drivers and feedbacks that led to this specific
described the warming by steps that occurred during the last sequence of events are still being actively studied. The start
disappearance of the large northern ice caps, a period of the deglaciation is linked to the evolution of the astro-
extending from 20 to 8 ka. This ‘last deglaciation’ was also nomical parameters, with a strong increase in summer
identified by paleoceanographers in marine sediment cores insolation in the northern hemisphere between 20 and 10 ka
with high sedimentation rates (Fig. 21.12). The terminology (Milankovitch’s theory), and aided by pulses of increases in
for this succession of warming and relative cooling comes atmospheric CO2 likely released from the CO2-rich deep
directly from the first descriptions made in the continental waters of the Southern Ocean (Marcott et al. 2014). How-
records based on pollen assemblages: Older Dryas, Middle ever, the mechanisms that explain the phase differences
Dryas, Bølling-Allerød and Younger Dryas all take their between the two hemispheres (Fig. 21.12) during the
names from plant pollens (in the case of Dryas, it is asso- deglaciation are still unclear. A distinct warming trend
ciated with the reappearance of the cold flower Dryas appeared in Antarctica around 19 ka, while simultaneously,
octopetala), or from locations from which the samples were the northern hemisphere, after a brief warming trend, cooled

Fig. 21.12 Records of the last

deglaciation in the high latitudes
of both hemispheres.
(Top) Variations in d18O in the
ice from the NorthGRIP site
(Greenland), a proxy for the local
variations in atmospheric
temperature (Rasmussen et al.
2014); (middle-top) variations in
the percentage of the cold species
Neogloboquadrina pachyderma
in core NA87-22 from the North
Atlantic (55 °30 ′N, 14 °42 ′W,
2161 m deep) (Waelbroeck et al.
2001; Vázquez Riveiros et al.
2013); (middle-bottom) variations
in the isotopic composition of
hydrogen dD of ice from Dome C
(Antarctica), a proxy for local
variations in atmospheric
temperature (EPICA 2004);
(bottom) variations in the
percentage of the cold species
Neogloboquadrina pachyderma
and SST estimated by the Mg/Ca
method in core MD07-3076Q
from the Southern Ocean (44 °
09 ′S, 14 °13 ′W, 3770 m deep)
(modified from Vázquez Riveiros
et al. 2010)
246 T. Caley et al.

and presented armadas of icebergs linked to the HS1 event Interglacial Periods, the Holocene and the Last
(see Chap. 29). Following this event, around 14 ka, northern Two Millennia
hemisphere warming was strongly amplified with a culmi-
nation during the Bølling-Allerød, while temperatures in the In order to explain the succession of glacial and interglacials
southern hemisphere stabilized and even dropped in periods over the last million years, Milankovitch developed
Antarctica during the so-called Antarctic Cold Reversal the astronomical theory of paleoclimate. Since then, con-
(Fig. 21.12). During this early phase of the deglaciation, the ceptual models have been able to describe the general trends,
North-South antiphase is similar to what is observed during as well as the dominant periodicities centered around 100, 40
abrupt events of the last ice age, with the exception of the and 20 kyr fairly accurately (see Chap. 28). Although
general deglacial warming trend. Recent studies point to changes in ice cap volume during glacial periods and the
CO2 as a key mechanism of global warming during the last time constants of their response to changes in insolation are
deglaciation. An anti-phased hemispheric temperature relatively well understood, the same cannot be said for the
response to ocean circulation changes, superimposed on evolution of climate during interglacial periods. In particular,
globally in-phase warming driven by increased CO2 con- the mechanisms causing the differences in duration, in
centrations, is an explanation for much of the temperature temperature of the atmosphere and ocean, and in ocean
change at the end of the most recent ice age (Barker et al. circulation are not well understood, even though differences
2009; Shakun et al. 2012). in forcing are precisely calculated (Past InterGlacialS
However, at the end of the Bølling-Allerød warm event, Working Group of PAGES 2016). This lack of under-
at about 12.5 ka, the ice caps stopped melting, the sea level standing is derived in part from the small ocean temperature
stabilized and the deglaciation stopped: this was the differences between past interglacial periods and the present
Younger Dryas period (Fig. 21.12), characterized by a return day, with temperature changes that remain close to the error
to very cold conditions for about 1.5 kyr, despite insolation of temperature reconstructions with the usual tracers
reaching maximum values. This return to almost ice age (Sect. “Sea Surface Temperature”). A further complication
conditions still raises many questions. The most commonly arises because the internal mechanisms in the climate system
accepted explanation is a sudden change in the path taken by must be investigated through its various components (at-
meltwater from the Laurentide ice sheet (Leydet et al. 2018). mosphere, ocean, continent), which involves the construc-
Until about 12.5 ka, this huge flow of water was transported tion of time scales common to the various archives used to
by the Mississippi River. Released into the Gulf of Mexico, reconstruct each of them, and makes the study of the inter-
the fresh water was drawn in by the circulation of the surface glacial periods prior to the Holocene particularly difficult.
currents of the Atlantic (Gulf Stream followed by the North In this section, we limit ourselves to the analysis of the
Atlantic Drift) and was very gradually diluted by the salty last two interglacial periods: the Last Interglacial (also called
tropical waters without any major climate impact. During the the Eemian), about 125 ka ago, and the Holocene, period in
Younger Dryas, however, the flow rate of the Mississippi which we now live. Eemian and Holocene, the terms used in
River dropped considerably, which led to the hypothesis that this chapter, are names borrowed from palynologists to
the watershed of the meltwater plume changed and flowed identify these two interglacial periods. A short subsection
instead through the St. Lawrence River to the northwest of will finally be devoted to results recently obtained for the last
the Atlantic Ocean. The salinity in this higher latitude area two millennia, which has the advantage of presenting a wide
was reduced, interrupting deep water formation and thus the range of continental and marine records that can, in some
thermohaline circulation, and causing cooling and the cases, be compared with recorded meteorological data.
growth of some glaciers. This hypothesis has been supported
by simple ocean circulation models, although marine sedi- The Last Interglacial Period
ment cores recovered from the likely North Atlantic zone of Before presenting our understanding of this period of time, it
evacuation of meltwater have, as of yet, failed to yield traces is important to define what an interglacial is. It may in fact
of this event. Recent study suggests multiple causes of the be defined in a number of ways depending on whether one
Younger Dryas cold period: a weakened Atlantic Meridional considers, for example, variability in flora, ocean circulation,
Overturning Circulation, moderate negative radioactive atmospheric temperature or ocean temperature (Past
forcing and an altered atmospheric circulation (Renssen et al. InterGlacialS Working Group of PAGES 2016). If we take
2015). The detailed study of this event could help us to ice volume as a marker, an interglacial period sensu stricto is
better understand the interactions between ocean, ice and the time interval during which the ice volume is at its
atmosphere under conditions of strong insolation. minimum and remains constant for several millennia.
21 Climate and the Evolution of the Ocean: The Paleoceanographic … 247

Strictly speaking, the interglacial comparable to the per- led to the persistence of cold conditions in the northern high
iod we live in, and defined by an ice volume minimum is latitudes and the early warming of southern high latitudes
called the Last Interglacial, and runs from about 129–116 ka during the early phase of the Last Interglacial (Capron et al.
(Govin et al. 2015; Dutton et al. 2015). From 115 ka, the 2014, 2017) (Fig. 21.13).
midpoint of the transition marking the entry into MIS 5d, ice Nevertheless, despite greenhouse gas concentrations that
volume had already increased significantly, so much so that were similar to pre-industrial times, the larger increase in
sea level dropped by as much as −40 m at the height of MIS summer insolation in the northern hemisphere with respect
5d at about 110 ka. to the current situation did have an impact on the climate of
During the Last Interglacial, the insolation forcing was the Last Interglacial optimum. Surface water temperatures
characterized by a relatively high eccentricity, the combi- were 1–2 °C warmer in the North Atlantic, the Nordic Seas
nation of a strong inclination and a perihelion close to the and the Southern Ocean than during the Holocene (Capron
summer solstice. This orbital configuration triggered an et al. 2014, 2017; Hoffman et al. 2017). Such warmer high
increase in summer insolation in the northern hemisphere of latitudes during the Last Interglacial had a double impact:
more than 30 W/m2 compared to the present day. Despite
these differences in forcing, the general evolution of the Last – the warming by about 0.4 °C of the temperatures of the
Interglacial climate is to a first degree quite similar to that of deep Atlantic waters, which was then carried into
the Holocene: high temperatures at higher northern latitudes Antarctic circumpolar deep waters (Duplessy et al. 2007);
until about 123 ka (in line with higher insolation and higher – the partial melting of Greenland and West Antarctica
elevation of the sun on the horizon), followed by a gradual (Dutton et al. 2015).
cooling linked to the decline in boreal summer insolation in
parallel with the progressive growth of glacial conditions These two combined actions brought about a rise in sea
(Cortijo et al. 1999). However, the Last Interglacial tem- level of 6–9 m (Dutton et al. 2009) compared to current
perature peak was reached at about 126 ka in the North levels.
Atlantic against 129 ka in the southern high latitudes. This The Last Interglacial is a good case study to test our
hemispheric asynchrony is related to the disruption of the mechanistic understanding of the effect of warmer-than-
Atlantic overturning circulation due to freshwater discharges present polar climate on sensitive components of the Earth
into the North Atlantic (in response to ice sheet melting) that system (e.g. ice sheets, sea level). It has recently sparked

Fig. 21.13 Temperature anomalies at 127 ka compared to preindus- 127 ka compared to preindustrial, in response to the high boreal
trial times (1870–1899 CE) in the northern and southern high latitudes summer insolation. The few cold anomalies suggest remnants of
(modified after Capron et al. 2017). Negative (positive) temperature freshwater discharge into the North Atlantic, Nordic Seas and Labrador
anomalies are shown in blue (red). The bigger the dot, the stronger the Seas
temperature anomaly. Most records indicate warmer conditions at
248 T. Caley et al.

interest in the community, as shown by multiple paleo-data Sahelian vegetation in turn disappeared about 2.7 ka ago to
compilations and model-data comparison exercises (Otto- make way for the desert conditions present today. This major
Bliesner et al. 2017). change could be related to the gradual decrease in insolation
over the past 10 ka aided by the albedo feedback induced by
The Holocene the gradual disappearance of vegetation. Alternatively, a
The Holocene period started about 11 ka ago. The last major rapid termination of the AHP could have been triggered by
ice sheets had not completely disappeared, but major chan- northern-latitude cooling combined with biogeophysical
ges had occurred since the early deglaciation, both in terms feedbacks (Collins et al. 2017).
of sea level and continental and oceanic temperatures. At During this wet period, the Mediterranean Sea received
first look, the climate over these past 11 ka seems stable, but more fresh water, especially in the eastern basin (Kallel et al.
this apparent stability hides very pronounced regional vari- 1997). The sinking of well-ventilated, shallow water masses in
ations in the hydrological cycle, in the circulation of surface winter became impossible in the Levantine basin, and bottom
waters (especially during the final stage of melting of the waters there became completely anoxic, leading to the disap-
residual ice caps), and in the general circulation of the pearance of benthic fauna below 800 m depth. A layer of black
Mediterranean Sea, a basin surrounded by land and with sediment rich in organic matter, called a sapropel, marks this
limited connections to the open ocean and thus very strongly event (Rossignol-Strick et al. 1982; Rohling et al. 2015).
affected by changes in rainfall intensity over its watershed. Although ventilation of the eastern waters of the Mediter-
The Holocene is a period of major movement and ranean resumed at 6 ka, the deep fauna of this basin, whose
development of populations. However, for the most part of colonization rate is slow, is still very poor.
this period, human activities still had a negligible impact on In addition to these long-term reorganizations, the
the global environment, so the study of climate changes over Holocene also recorded an abrupt event of short duration
recent millennia provide a benchmark against which dis- 8.2 ka ago. Without reaching the amplitude of the rapid and
turbances caused by industrial and agricultural activities can sudden climate changes of the last ice age, this event still left
be detected. The reconstruction of Holocene climate changes a significant imprint on northern hemisphere temperatures.
is facilitated by the precise chronology offered by carbon-14 Like its glacial counterparts, the ‘8.2 ka event’ is associated
analysis. with a freshwater discharge, in this case due to the rupture of
The forcing of summer insolation at 65°N at the begin- a proglacial reservoir, Lake Agassiz, formed by the retreat of
ning of the Holocene reached more than 390 W/m2 and the Laurentide ice sheet (Barber et al. 1999; Wiersma and
caused a global warming that would last until about 6 ka. Renssen 2006; Hoffman et al. 2012). The sudden release of
The temperature optimum affected the high latitudes of the tens of thousands of km3 of water (estimates vary from
North Atlantic basin, including Iceland, the Norwegian Sea 50,000 to 120,000 km3) over just 1–5 yrs had strong con-
and the Scandinavian coast (Koç et al. 1993). In the Barents sequences, such as a reduction in the SST (about 1 °C) and
Sea, the temperature maximum was limited to the period salinity of the North Atlantic, a reduction of 2–6 °C in the
from 7.9 to 6.9 ka due to the dissipation of the heat brought atmospheric temperature above Greenland, a decrease in the
by the North Atlantic Drift by the melting of the surrounding temperature of air and water in the lakes of western Europe,
ice. At lower latitudes, the temperature increase was and a decrease in the intensity of ocean circulation for a
accompanied by a northward shift of the Intertropical Con- period of about 100 yrs after the freshwater discharge.
vergence Zone (ITCZ) and a major change in monsoon The study of the 8.2 event has shown that interglacial
dynamics, and therefore in the atmospheric water cycle. The ocean circulation, such as the one of the early Holocene,
increase in the thermal contrast between ocean and conti- may also be sensitive to an intense, although brief, fresh-
nent, for example, accentuated the African monsoon as far as water discharge. Recent studies have pointed out that this
the center of the continent. may also have been the case during earlier interglacial
The study of sediment cores from both the Mediterranean periods (Galaasen et al. 2014).
Sea and African lakes indicates the existence of major cli- The climate of the last two millennia has also been the
mate reorganizations. For example, before 6 ka, the Sahara subject of much focus, since it provides a relatively
was not the wide-ranging desert that it is today, but grassland long-term perspective for recent observations from the
dotted with lakes conducive to farming settlement. This World Meteorological Organization (WMO) network (re-
period is called the African Humid Period (AHP). Around stricted to the last 150 years) and from satellites dedicated to
6 ka, this wet period ended and conditions degraded at a rate the observation of the Earth (limited to a few decades). The
that is still debated (Collins et al. 2017; Shanahan et al. reconstructions of air temperature in the northern hemi-
2016; Tierney and deMenocal (2013). The tropical vegeta- sphere, used as projections for the whole planet, have pri-
tion of canopy forests along the rivers declined, and the marily been based on continental data (Mann et al. 1998).
21 Climate and the Evolution of the Ocean: The Paleoceanographic … 249

Fig. 21.14 Standardized SST anomalies over the last 2000 years been averaged into 200-year long bins (e.g. 1–200 CE). The thick black
(modified after McGregor et al. 2015). Thin colored lines represent line is the area-weighted median SST value
individual SST reconstructions from different ocean basins, which have

Reconstructing and understanding changes in the ocean over References

the last two millennia is particularly difficult, since oceano-
graphic observations are only available for the last century at Abram, N. J., McGregor, H. V., Tierney, J. E., Evans, M. N., McKay,
most, and for paleoceanographers this period is recorded in N. P., Kaufman, D. S., & The Pages 2 k Consortium. (2016). Early
the uppermost portion of the sediment which is often poorly onset of industrial-era warming across the oceans and continents.
consolidated or lost. Nature, 536, 411. https://doi.org/10.1038/nature19082.
Adkins, J. F., McIntyre, K., & Schrag, D. P. (2002). The salinity,
Recently, much effort has been put on the gathering of the temperature and d18O of the glacial deep ocean. Science, 298,
best time-series of the last two millennia as part of the Past 1769–1773.
Global Changes (PAGES) 2 k network (PAGES 2k Consor- Anand, P., Elderfield, H., & Conte, M. H. (2003). Calibration of Mg/Ca
tium 2013, 2017). Continental-scale temperature reconstruc- thermometry in planktonic foraminifera from a sediment trap time
series. Paleoceanography, 18(2).
tions provide evidence of twentieth century warming over all Barker, S., Diz, P., Vautravers, M. J., Pike, J., Knorr, G., Hall, I. R.,
reconstructed regions except Antarctica (PAGES 2k Consor- et al. (2009). Interhemispheric Atlantic seesaw response during the
tium 2013). A global SST compilation shows a 1800-year last deglaciation. Nature, 457(7233), 1097.
long cooling of the surface ocean over the pre-industrial past Barber, D. C., Dyke, A., Hillaire-Marcel, C., Jennings, A. E., Andrews,
J. T., Kerwin, M. W., et al. (1999). Forcing of the cold event of
2000 years (Fig. 21.14), and that the cooling from 801 to 1800 8200 years ago by catastrophic drainage of Laurentide lakes.
CE was likely caused by volcanic eruptions (McGregor et al. Nature, 400, 344. https://doi.org/10.1038/22504.
2015). A more recent synthesis of paleoclimate records since Bemis, B. E., Spero, H. J., Bijma, J., & Lea, D. W. (1998).
1500 CE has identified that sustained industrial-era warming Reevaluation of the oxygen isotopic composition of planktonic
foraminifera: Experimental results and revised paleotemperature
of the tropical oceans first developed during the mid- equations. Paleoceanography, 13, 150–160.
nineteenth century and was nearly synchronous with North- Bintanja, R., van de Wal, R. S., & Oerlemans, J. (2005). Modelled
ern Hemisphere continental warming (Abram et al. 2016). atmospheric temperatures and global sea levels over the past million
If we look at the deep ocean, the relative strength of the years. Nature, 437, 125–128.
Böhm, E., Lippold, J., Gutjahr, M., Frank, M., Blaser, P., Antz, B.,
meridional overturning circulation has also been recently et al. (2015). Strong and deep Atlantic meridional overturning
assessed for the last 1.6 ka (Thornalley et al. 2018). The circulation during the last glacial cycle. Nature, 517, 73–76.
authors suggest that, while it was relatively stable between Caley, T., & Roche, D. M. (2015). Modeling water isotopologues
400 and 1850 CE, it has declined in strength by *15% at during the last glacial: Implications for quantitative paleosalinity
reconstruction. Paleoceanography and Paleoclimatology, 30(6),
the beginning of the industrial era. In addition, the com- 739–750.
parison of SST patterns in the North Atlantic with model Capron, E., Govin, A., Stone, E. J., Masson-Delmotte, V., Mulitza, S.,
simulations points to an additional weakening over the last Otto-Bliesner, B., et al. (2014). Temporal and spatial structure of
150 yrs (Caesar et al. 2018). multi-millennial temperature changes at high latitudes during the
Last Interglacial. Quaternary Science Reviews, 103, 116–133.
Compilation studies like the ones presented above high- Capron, E., Govin, A., Feng, R., Otto-Bliesner, B., & Wolff, E. W.
light the need for paleoclimate reconstructions, which can be (2017). Critical evaluation of climate syntheses to benchmark
compared to instrumental records that at present are too short CMIP6/PMIP4 127 ka Last Interglacial simulations in the high
to comprehensively assess anthropogenic climate change. latitude regions. Quaternary Science Reviews, 168, 137–150.
250 T. Caley et al.

Carroll, J., Brown, E. T., & Moore, W. S. (1993). The role of the climate change during the penultimate interglacial. Nature Geo-
Ganges-Brahmaputra mixing zone in supplying barium and 226Ra to science, 2, 355–359.
the Bay of Bengal. Geochimica et Cosmochimica Acta, 57, 2981– Dutton, A., Carlson, A. E., Long, A. J., Milne, G. A., Clark, P. U.,
2990. DeConto, R., Horton, B. P., Rahmstorf, S., & Raymo, M. E. (2015).
Caesar, L., Rahmstorf, S., Robinson, A., Feulner, G., & Saba, V. Sea-level rise due to polar ice-sheet mass loss during past warm
(2018). Observed fingerprint of a weakening Atlantic Ocean periods. Science, 349. https://doi.org/10.1126/science.aaa4019.
overturning circulation. Nature, 556, 191–198. Eiler, J. M., Bergquist, B., Bourg, I., Cartigny, P., Farquhar, J.,
Chivall, D., M’Boule, D., Sinke-Schoen, D., Sinninghe Damsté, J. S., Gagnon, A., et al. (2014). Frontiers of stable isotope geoscience.
Schouten, S., & van der Meer, M. T. (2014). The effects of growth Chemical Geology, 372, 119–143. https://doi.org/10.1016/j.
phase and salinity on the hydrogen isotopic composition of chemgeo.2014.02.006.
alkenones produced by coastal haptophyte algae. Geochimica et Elderfield, H., Yu, J., Anand, P., Kiefer, T., & Nyland, B. (2006).
Cosmochimica Acta, 140, 381–390. Calibrations for benthic foraminiferal Mg/Ca paleothermometry and
Clark, P. U., & Pollard, D. (1998). Origin of the middle Pleistocene the carbonate ion hypothesis. Earth and Planetary Science Letters,
transition by ice sheet erosion of regolith. Paleoceanography and 250, 633–649.
Paleoclimatology, 13(1), 1–9. Elderfield, H., & Ganssen, G. (2000). Past temperature and d18O of
Clark, P. U., Archer, D., Pollard, D., Blum, J. D., Rial, J. A., Brovkin, surface ocean waters inferred from foraminiferal Mg/Ca ratios.
V., et al. (2006). The middle Pleistocene transition: Characteristics, Nature, 405, 442–445.
mechanisms, and implications for long-term changes in atmospheric Elderfield, H., Ferretti, P., Greaves, M., Crowhurst, S., McCave, I. N.,
pCO2. Quaternary Science Reviews, 25, 3150–3184. Hodell, D. A., et al. (2012). Evolution of ocean temperature and ice
CLIMAP. (1981). Seasonal reconstructions of the Earth’s surface at the volume through the mid-Pleistocene climate transition. Science, 337
last glacial maximum. Geological Society of America, Map and (6095), 704–709.
Chart Series, 36. Englebrecht, A. C., & Sachs, J. P. (2005). Determination of sediment
Collins, J. A., Prange, M., Caley, T., Gimeno, L., Beckmann, B., provenance at drift sites using hydrogen isotopes and unsaturation
Mulitza, S., et al. (2017). Rapid termination of the African Humid ratios in alkenones. Geochimica et Cosmochimica Acta, 69, 4253–
Period triggered by northern high-latitude cooling. Nature Commu- 4265.
nications, 8, 1372. EPICA. (2004). Eight glacial cycles from an Antarctic ice core. Nature,
Conte, M. H., Sicre, M. A., Rühlemann, C., Weber, J. C., Schulte, S., 429, 623–628.
Schulz-Bull, D., & Blanz, T. (2006). Global temperature calibration Epstein, S., Buchsbaum, R., Lowenstam, H., & Urey, H. C. (1951).
of the alkenone unsaturation index (UK′37) in surface waters and Carbonate-water isotopic temperature scale. Geological Society of
comparison with surface sediments. Geochemistry, Geophysics, America Bulletin, 62(4), 417–426.
Geosystems, 7(2). Epstein, S., Buchsbaum, R., Lowenstam, H. A., & Urey, H. C. (1953).
Corrège, T. (2006). Sea surface temperature and salinity reconstruction Revised carbonate-water isotopic temperature scale. Geological
from coral geochemical tracers. Palaeogeography, Palaeoclimatol- Society of America Bulletin, 64(11), 1315–1326.
ogy, Palaeoecology, 232(2–4), 408–428. Emiliani, C. (1955). Pleistocene temperatures. Journal of Geology, 63,
Cortijo, E., Lehman, S., Keigwin, L., Chapman, M., Paillard, D., & 538–578.
Labeyrie, L. (1999). Changes in meridional temperature and salinity Galaasen, E. V., Ninnemann, U. S., Irvalı, N., Kleiven, H. K. F.,
gradients in the North Atlantic Ocean (30–72 N) during the last Rosenthal, Y., Kissel, C., et al. (2014). Rapid reductions in North
interglacial period. Paleoceanography, 14, 23–33. https://doi.org/ Atlantic deep water during the peak of the last interglacial period.
10.1029/1998PA900004. Science, 343(6175), 1129–1132.
Craig, H., & Gordon, L. I. (1965). Deuterium and oxygen-18 variations Gherardi, J. M., Labeyrie, L., Nave, S., Francois, R., McManus, J. F., &
in the ocean and marine atmosphere. In E. Tongiorgi (Ed.), Stable Cortijo, E. (2009). Glacial-interglacial circulation changes inferred
isotopes in oceanographic studies and paleotemperatures (pp. 9– from 231 Pa/230Th sedimentary record in the North Atlantic
130). Pisa, Italy: Laboratory of Geology and Nuclear Science. Region. Paleoceanography, 24, PA2204. https://doi.org/10.1029/
DeSève, A. M. (1999). Transfer function between surface sediment 2008pa001696.
diatom assemblages and sea-surface temperature and salinity of the Ghosh, P., Adkins, J., Affek, H., Balta, B., Guo, W., Schauble, E. A.,
Labrador Sea. Marine Micropaleontology, 36, 249–267. Schrag, D. P., & Eiler, J. M. (2006). 13C-18O bonds in carbonate
De Vernal, A., et al. (2001). Dinoflagellate cyst assemblages as tracers minerals: A new kind of paleothermometer. Geochimica et
of sea-surface conditions in the Northern Atlantic, Arctic and Cosmochimica Acta, 70, 1 439–1 456.
sub-Arctic seas: The new n = 677 data base and its application for Gottschalk, J., Vázquez Riveiros, N., Waelbroeck, C., Skinner, L. C.,
quantitative paleoceanographic reconstruction. Journal of Quater- Michel, E., Duplessy, J. C., et al. (2016). Carbon isotope offsets
nary Science, 16, 681–698. between benthic foraminifer species of the genus Cibicides
Duplessy, J. C., Shackleton, N. J., Matthews, R. K., Prell, W. L., (Cibicidoides) in the glacial sub-Antarctic Atlantic. Paleoceanog-
Ruddiman, W. F., Caralp, M., et al. (1984). 13C record of benthic raphy, 31, 1–20.
foraminifera in the Last Interglacial Ocean: Implications for the Govin, A., Capron, E., Tzedakis, P. C., Verheyden, S., Ghaleb, B.,
carbon cycle and the global deep water circulation. Quaternary Hillaire-Marcel, C., et al. (2015). Sequence of events from the onset
Research, 21, 225–243. to the demise of the Last Interglacial: Evaluating strengths and
Duplessy, J. C., Labeyrie, L., Juillet-Leclerc, A., Maitre, F., Duprat, J., limitations of chronologies used in climatic archives. Quaternary
& Sarnthein, M. (1991). Surface salinity reconstruction of the North Science Reviews, 129, 1–36. https://doi.org/10.1016/j.quascirev.
Atlantic Ocean during the last glacial maximum. Oceanologica 2015.09.018.
Acta, 14, 311–324. Gray, W. R., Weldeab, S., Lea, D. W., Rosenthal, Y., Gruber, N.,
Duplessy, J. C., Roche, D. M., & Kageyama, M. (2007). The deep Donner, B., et al. (2018). The effects of temperature, salinity, and
ocean during the last interglacial period. Science, 316, 89–91. the carbonate system on Mg/Ca in Globigerinoides ruber (white): A
Dutton, A., Bard, E., Antonioli, F., Esat, T. M., Lambeck, K., & global sediment trap calibration. Earth and Planetary Science
McCulloch, M. T. (2009). Phasing and amplitude of sea-level and Letters, 482, 607–620. https://doi.org/10.1016/j.epsl.2017.11.026.
21 Climate and the Evolution of the Ocean: The Paleoceanographic … 251

Guihou, A., Pichat, S., Nave, S., Govin, A., Labeyrie, L., Michel, E., Kim, J.-H., van der Meer, J., Schouten, S., Helmke, P., Willmott, V.,
et al. (2010). Late slowdown of the Atlantic Meridional overturning Sangiorgi, F., et al. (2010). New indices and calibrations derived
circulation during the last glacial inception: New constraint from from the distribution of crenarchaeal isoprenoid tetraether lipids:
sedimentary (231 Pa/230Th). Earth and Planetary Science Letters, Implications for past sea surface temperature reconstructions.
289, 520–529. Geochimica et Cosmochimica Acta, 74, 4639–4654. https://doi.
Guihou, A., Pichat, S., Govin, A., Nave, S., Michel, E., Duplessy, J.-C., org/10.1016/j.gca.2010.05.027.
et al. (2011). Enhanced Atlantic Meridional overturning circulation Koç, N., Jansen, E., & Haflidason, H. (1993). Paleoceanographic
supports the last glacial inception. Quaternary Science Reviews, 30, Reconstructions of surface Ocean conditions in the Greenland,
1576–1582. https://doi.org/10.1016/j.quascirev.2011.03.017. Iceland and Norwegian Seas through the Last 14 Ka based on
Ho, S. L., & Laepple, T. (2016). Flat meridional temperature gradient in diatoms. Quaternary Science Reviews, 12, 115–140.
the early Eocene in the subsurface rather than surface ocean. Nature Labeyrie, L. D., Duplessy, J. C., & Blanc, P. L. (1987). Variations in
Geoscience, 9(8), 606. mode of formation and temperature of oceanic deep waters over the
Hoffman, J. S., Carlson, A. E., Winsor, K., Klinkhammer, G. P., past 125,000 years. Nature, 327.
LeGrande, A. N., Andrews, J. T., & Strasser, J. C. (2012). Linking Labeyrie, L., Duplessy, J. C., Duprat, J., Juillet-Leclerc, A., Moyes, J.,
the 8.2 ka event and its freshwater forcing in the Labrador Sea. Michel, E., et al. (1992). Changes in vertical structure of the North
Geophysical Research Letters, 39(18). Atlantic Ocean between glacial and modern times. Quaternary
Hoffman, J. S., Clark, P. U., Parnell, A. C., & He, F. (2017). Regional Science Reviews, 11, 401–413.
and global sea-surface temperatures during the last interglaciation. Lea, D. W., Mashiotta, T. A., & Spero, H. J. (1999). Controls on
Science, 355(6322), 276–279. magnesium and strontium uptake in planktonic foraminifera
Hönisch, B., Hemming, N. G., Archer, D., Siddall, M., & McManus, determined by live culturing. Geochimica et Cosmochimica Acta,
J. F. (2009). Atmospheric carbon dioxide concentration across the 63(16), 2369–2379.
Mid-Pleistocene transition. Science, 324, 1551–1556. Leduc, G., Sachs, J. P., Kawka, O. E., & Schneider, R. R. (2013).
Hönisch, B., Allen, K. A., Russell, A. D., Eggins, S. M., Bijma, J., Holocene changes in eastern equatorial Atlantic salinity as
Spero, H. J., et al. (2011). Planktic foraminifers as recorders of estimated by water isotopologues, Earth Planet. Sci. Lett., 362,
seawater Ba/Ca. Marine Micropaleontology, 79, 52–57. https://doi. 151–162.
org/10.1016/j.marmicro.2011.01.003. LeGrande, A. N., & Schmidt, G. A. (2006). Global gridded data set of
Hönisch, B., Allen, K. A., Lea, D. W., Spero, H. J., Eggins, S. M., the oxygen isotopic composition in seawater. Geophysical research
Arbuszewski, J., et al. (2013). The influence of salinity on Mg/Ca in letters, 33(12).
planktic foraminifers—Evidence from cultures, core-top sediments LeGrande, A. N., & Schmidt, G. A. (2011). Water isotopologues as a
and complementary d18O. Geochimica et Cosmochimica Acta, 121, quantitative paleosalinity proxy. Paleoceanography, 26, PA3225.
196–213. https://doi.org/10.1016/j.gca.2013.07.028. https://doi.org/10.1029/2010pa002043.
Imbrie, J., & Kipp, N. G. (1971). A new micropaleontological method Leydet, D. J., Carlson, A. E., Teller, J. T., Breckenridge, A., J. E.,
for quantitative paleoclimatology: Application to a late Pleistocene Barth, A. M., Ullman, D. J., Sinclair, G., Milne, G. A., Cuzzone,
Caribbean Core. In K. K. Turekian (Ed.), The late Cenozoic glacial J. K., Caffee, M. W. (2018). Opening of glacial Lake Agassiz’s
ages (pp. 71–181). Yale University Press. eastern outlets by the start of the Younger Dryas cold period.
Jansen, E., & Sjoholm, J. (1991). Reconstruction of glaciation over the Geology, 46, 155–158. https://doi.org/10.1130/G39501.1.
6 Myr from ice-borne deposits in the Norwegian Sea. Nature, 349, Lippold, J., Gutjahr, M., Blaser, P., Christner, E., de Carvalho Ferreira,
600–603. M. L., Mulitza, S., et al. (2016). Deep water provenance and
Kageyama, M., Braconnot, P., Harrison, S. P., Haywood, A. M., dynamics of the (de) glacial Atlantic meridional overturning
Jungclaus, J. H., Otto-Bliesner, B. L., et al. (2018). The PMIP4 circulation. Earth and Planetary Science Letters, 445, 68–78.
contribution to CMIP6—Part 1: Overview and over-arching anal- Lisiecki, L. E. & Raymo, M. E. (2005). A Pleiocene-Pleistocene stack
ysis plan. Geoscientific Model Development, 11, 1033–1057. of 57 globally distributed d18O records. Paleoceanography, 20.
https://doi.org/10.5194/gmd-11-1033-2018. https://doi.org/10.1029/2004PA001071.
Kallel, N., Labeyrie, L. D., Juillet-Leclerc, A., & Duplessy, J. C. Lombard, F., Labeyrie, L., Michel, E., Spero, H. J., & Lea, D. W.
(1988). A deep hydrological front between intermediate and (2009). Modelling the temperature dependent growth rates of
deep-water masses in the glacial Indian Ocean. Nature, 333, 651– planktic foraminifera. Marine Micropaleontology, 70, 1–7.
655. Lynch-Stieglitz, J., Curry, W. B., & Slowey, N. (1999). A geostrophic
Kallel, N., Paterne, M., Duplessy, J. C., Vergnaud-Grazzini, C., Pujol, transport estimate for the Florida current from the oxygen isotope
C., Labeyrie, L., et al. (1997). Enhanced rainfall in the Mediter- composition of benthic foraminifera. Paleoceanography, 14, 360–
ranean Region during the last sapropel event. Oceanologica Acta, 373.
20, 697–712. Lynch-Stieglitz, J., Adkins, J. F., Curry, W. B., Dokken, T., Hall, I. R.,
Keigwin, L. D., & Swift, S. A. (2017). Carbon isotope evidence for a Herguera, J. C., Hirschi, J.l.J.M., Ivanova, E. V., Kissel, C.,
northern source of deep water in the glacial western North Atlantic. Marchal, O., Marchitto, T. M., McCave, I. N., McManus, J. F.,
Proceedings of the National Academy of Sciences, 114(11), 2831– Mulitza, S., Ninnemann, U., Peeters, F., Yu, E.-F., & Zahn, R.
2835. (2007). Atlantic Meridional overturning circulation during the last
Kele, S., Breitenbach, S. F. M., Capezzuoli, E., Meckler, A. N., Ziegler, glacial maximum. Science, 316, 66–69. https://doi.org/10.1126/
M., Millan, I. M., et al. (2015). Temperature dependence of oxygen- science.1137127.
and clumped isotope fractionation in carbonates: A study of Lynch-Stieglitz, J., Schmidt, M. W., Gene Henry, L., Curry, W. B.,
travertines and tufas in the 6–95 °C temperature range. Geochimica Skinner, L. C., Mulitza, S., et al. (2014). Muted change in Atlantic
et Cosmochimica Acta, 168, 172–192. https://doi.org/10.1016/j.gca. overturning circulation over some glacial-aged Heinrich events.
2015.06.032. Nature Geoscience, 7, 144–150. https://doi.org/10.1038/ngeo2045.
Khider, D., Huerta, G., Jackson, C., Stott, L. D., & Emile-Geay, Malaizé, B., & Caley, T. (2009). Sea surface salinity reconstruction as
J. (2015). A Bayesian, multivariate calibration for Globigerinoides seen with foraminifera shells: Methods and cases studies. The
ruber Mg/Ca. Geochemistry, Geophysics, Geosystems, 16, 2916– European Physical Journal Conferences, 1, 177–188. https://doi.
2932. https://doi.org/10.1002/2015GC005844. org/10.1140/epjconf/e2009-00919-6.
252 T. Caley et al.

Malmgren, B. A., Kucera, M., Nyberg, J., & Waelbroeck, C. (2001). Mulitza, S., Boltovskoy, D., Donner, B., Meggers, H., Paul, A., &
Comparison of statistical and artificial neuronal network techniques Wefer, G. (2003). Temperature: [delta]18O relationships of plank-
for estimating past sea surface temperatures from planktonic tonic foraminifera collected from surface waters. Palaeogeography,
foraminifera census data. Paleoceanography, 16, 520–530. Palaeoclimatology, Palaeoecology, 202, 143–152. https://doi.org/
Mann, M. E., Bradley, R. S., & Hughes, M. K. (1998). Global-scale 10.1016/S0031-0182(03)00633-3.
temperature patterns and climate forcing over the past six centuries. Müller, P. J., Kirst, G., Ruhland, G., Von Storch, I., & Rosell-Melé, A.
Nature, 392(6678), 779. (1998). Calibration of the alkenone paleotemperature index U 37 K′
Marchitto, T. M., & Broecker, W. S. (2006). Deep water mass based on core-tops from the eastern South Atlantic and the global
geometry in the glacial Atlantic Ocean: A review of constraints ocean (60 N-60 S). Geochimica et Cosmochimica Acta, 62(10),
from the paleonutrient proxy Cd/Ca. Geochemistry, Geophysics, 1757–1772.
Geosystems, 7, Q12003. Otto-Bliesner, B. L., Braconnot, P., Harrison, S. P., Lunt, D. J.,
Marchitto, T. M., Curry, W. B., Lynch-Stieglitz, J., Bryan, S. P., Cobb, Abe-Ouchi, A., Albani, S., et al. (2017). The PMIP4 contribution to
K. M., & Lund, D. C. (2014). Improved oxygen isotope temperature CMIP6—Part 2: Two interglacials, scientific objective and exper-
calibrations for cosmopolitan benthic foraminifera. Geochimica et imental design for Holocene and Last Interglacial simulations.
Cosmochimica Acta, 130, 1–11. Geoscientific Model Development, 10, 3979–4003. https://doi.org/
Marcott, S. A., Bauska, T. K., Buizert, C., Steig, E. J., Rosen, J. L., 10.5194/gmd-10-3979-2017.
Cuffey, K. M., et al. (2014). Centennial-scale changes in the global Ostlund, H. G. (1987). GEOSECS Atlantic, Pacific, and Indian Ocean
carbon cycle during the last deglaciation. Nature, 514, 616–619. Expeditions (Vol. 7).
MARGO Project Members. (2009). Constraints on the magnitude and PAGES 2 k Consortium. (2013). Continental-scale temperature vari-
patterns of ocean cooling at the last glacial maximum. Nature ability during the past two millennia. Nature Geoscience, 6, 339.
Geoscience. https://doi.org/10.1038/NGEO411. https://doi.org/10.1038/ngeo1797.
Mashiotta, T. A., Lea, D. W., & Spero, H. J. (1999). Glacial– PAGES 2 k Consortium. (2017). A global multiproxy database for
interglacial changes in Subantarctic sea surface temperature and temperature reconstructions of the common era. Scientific Data, 4,
d18O-water using foraminiferal Mg. Earth and Planetary Science 170088. https://doi.org/10.1038/sdata.2017.88.
Letters, 170(4), 417–432. Past InterGlacialS Working Group of PAGES. (2016). Interglacials of
Mathien-Blard, E., & Bassinot, F. (2009). Salinity bias on the the last 800,000 years. Reviews of Geophysics, 54.
foraminifera Mg/Ca thermometry: Correction procedure and impli- Paul, H. (2002). Application of novel stable isotope methods to
cations for past ocean hydrographic reconstructions. Geochemistry, reconstruct paleoenvironments: Compound specific hydrogen iso-
Geophysics, Geosystems, 10, Q12011. https://doi.org/10.1029/ topes and pore-water oxygen isotopes. Ph.D. thesis, 149 pp., Swiss
2008gc002353. Federal Institute of Technology.
M’boule, D., Chivall, D., Sinke-Schoen, D., Sinninghe Damsté, J. S., Peral, M., Daëron, M., Blamart, D., Bassinot, F., Dewilde, F.,
Schouten, S., & van der Meer, M. T. (2014). Salinity dependent Smialkowski, N., et al. (2018). Updated calibration of the clumped
hydrogen isotope fractionation in alkenones produced by coastal isotope thermometer in planktonic and benthic foraminifera.
and open ocean haptophyte algae. Geochimica et Cosmochimica Geochimica et Cosmochimica Acta, 239, 1–16. https://doi.org/10.
Acta, 130, 126–135. 1016/j.gca.2018.07.016.
McClymont, E. L., Sosdian, S. M., Rosell-Melé, A., & Rosenthal, Y. Petit, J. R., Jouzel, J., Raynaud, D., Barkov, N. I., Barnola, J. M.,
(2013). Pleistocene sea-surface temperature evolution: Early cool- Basile, I., et al. (1999). Climate and atmospheric history of the past
ing, delayed glacial intensification, and implications for the 420,000 years from the Vostok ice core, Antarctica. Nature, 399,
mid-Pleistocene climate transition. Earth-Science Reviews, 123, 429–436.
173–193. Prahl, F. G., & Wakeham, S. G. (1987). Calibration of unsaturation
McCorkle, D. C., Martin, P., Lea, D. W., & Klinkhammer, G. (1995). patterns in long-chain ketone compositions for paleotemperature
Evidence of a dissolution effect on benthic foraminiferal shell assessment. Nature, 330, 367–369.
chemistry: d13C, Cd/Ca, Ba/Ca, and Sr/Ca results from the Ontong Pisias, N. G., Martinson, D. G., Moore, T. C., Shackleton, N. J., Prell,
Java Plateau. Paleoceanography, 10, 699–714. W. L., Hays, J. D., et al. (1984). High resolution stratigraphic
McGregor, H. V., Evans, M. N., Goosse, H., Leduc, G., Martrat, B., correlation of benthic oxygen isotopic records spanning the last
Addison, J. A., et al. (2015). Robust global ocean cooling trend for 300,000 years. Marine Geology, 56, 119–136.
the pre-industrial Common Era. Nature Geoscience, 8, 671–677. Railsback, L. B., Gibbard, P. L., Head, M. J., Voarintsoa, N. R. G., &
https://doi.org/10.1038/ngeo2510. Toucanne, S. (2015). An optimized scheme of lettered marine
McManus, J. F., Francois, R., Gherardi, J.-M., Keigwin, L. D., & isotope substages for the last 1.0 million years, and the cli-
Brown-Leger, S. (2004). Collapse and rapid resumption of Atlantic matostratigraphic nature of isotope stages and substages. Quater-
meridional circulation linked to deglacial climate changes. Nature, nary Science Reviews, 111, 94–106.
428, 834–837. Rasmussen, S. O., Bigler, M., Blockley, S. P., Blunier, T., Buchardt, S.
Mezger, E. M., Nooijer, L. J., Boer, W., Brummer, G. J. A., & Reichart, L., Clausen, H. B., et al. (2014). A stratigraphic framework for
G. J. (2016). Salinity controls on Na incorporation in Red Sea abrupt climatic changes during the last glacial period based on three
planktonic foraminifera. Paleoceanography and Paleoclimatology, synchronized Greenland ice-core records: Refining and extending
31(12), 1562–1582. the INTIMATE event stratigraphy. Quaternary Science Reviews,
Montagna, P., McCulloch, M., Douville, E., López Correa, M., Trotter, 106, 14–28.
J., Rodolfo-Metalpa, R., et al. (2014). Li/Mg systematics in Raymo, M. E., Oppo, D. W., & Curry, W. (1997). The mid-Pleistocene
scleractinian corals: Calibration of the thermometer. Geochimica climate transition: A deep sea carbon isotopic perspective. Paleo-
et Cosmochimica Acta, 132, 288–310. https://doi.org/10.1016/j.gca. ceanography, 12(4), 546–559.
2014.02.005. Renssen, H., Mairesse, A., Goosse, H., Mathiot, P., Heiri, O., Roche,
Montes, C., Cardona, A., Jaramillo, C., Pardo, A., Silva, J. C., D. M., et al. (2015). Multiple causes of the Younger Dryas cold
Valencia, V., et al. (2015). Middle Miocene closure of the Central period. Nature Geoscience, 8(12), 946–949.
American seaway. Science, 348, 226–229.
21 Climate and the Evolution of the Ocean: The Paleoceanographic … 253

Rohling, E., & Bigg, G. (1998). Paleosalinity and d18O: A critical Shackleton, N. J. (1974). Attainment of isotopic equilibrium between
assessment. Journal Geophysical Research, 103, 1307–1318. ocean water and benthonic foraminifera genus Uvigerina: Isotopic
https://doi.org/10.1029/97JC01047. changes in the ocean during the last glacial. dans Les méthodes
Rohling, E. J. (2000). Paleosalinity: Confidence limits and future quantitatives d’étude des variations du climat au cours du
applications. Marine Geology, 163, 1–11. Pleistocène CNRS, Gif-sur-Yvette, pp. 203–209.
Rohling, E. J. (2007). Progress in paleosalinity: Overview and Shakun, J. D., Clark, P. U., He, F., Marcott, S. A., Mix, A. C., Liu, Z.,
presentation of a new approach. Paleoceanography, 22, PA3215. et al. (2012). Global warming preceded by increasing carbon
https://doi.org/10.1029/2007pa001437. dioxide concentrations during the last deglaciation. Nature, 484,
Rohling, E. J., Foster, G. L., Grant, K., Marino, G., Roberts, A. P., 49–54.
Tamisiea, M. E., et al. (2014). Sea-level and deep-sea-temperature Shanahan, T. M., Hughen, K., McKay, N. P., Overpeck, J., Scholz, C.
variability over the past 5.3 million years. Nature, 508, 477–482. A., Gosling, W. D., et al. (2016). CO2 and fire influence tropical
Rohling, E. J., Marino, G., & Grant, K. M. (2015). Mediterranean ecosystem stability in response to climate change. Scientific
climate and oceanography, and the periodic development of anoxic Reports, 6, 29587.
events (sapropels). Earth-Science Reviews, 143, 62–97. Sicre, M. A., et al. (2005). Mid-latitude southern Indian ocean response
Rosell-Melé, A., & Prahl, F. G. (2013). Seasonality of UK′ 37 to northern hemisphere Heinrich events. Earth and Planetary
temperature estimates as inferred from sediment trap data. Quater- Science Letters, 240(3–4), 724–731.
nary Science Reviews, 72, 128–136. Thierstein, H. R., Geitzenauer, K. R., Molfino, B., & Shackleton, N.
Rosenthal, Y., Boyle, E. A., & Slowey, N. (1997). Temperature control J. (1977). Global synchroneity of late Quaternary coccolith datum
on the incorporation of magnesium, strontium, fluorine, and levels validation by oxygen isotopes. Geology, 5(7), 400–404.
cadmium into benthic foraminiferal shells from Little Bahama Thornalley, D. J. R., Oppo, D. W., Ortega, P., Robson, J. I., Brierley,
Bank: Prospects for thermocline paleoceanography. Geochimica et C., Davis, R., et al. (2018). Anomalously weak Labrador Sea
Cosmochimica Acta, 61(17), 3633–3643. convection and Atlantic overturning during the past 150 years.
Rossignol-Strick, M., Nesteroff, W., Olive, P., & Vergnaud-Grazzini, Nature, 556, 227–232.
C. (1982). After the deluge: Mediterranean stagnation and sapropel Tierney, J. E., & deMenocal, P. B. (2013). Abrupt shifts in Horn of
formation. Nature, 295(5845), 105. Africa hydroclimate since the Last Glacial Maximum. Science, 342
Saenger, C., Affek, H. P., Felis, T., Thiagarajan, N., Lough, J. M., & (6160), 843–846.
Holcomb, M. (2012). Carbonate clumped isotope variability in Tierney, J. E., & Tingley, M. P. (2018). BAYSPLINE: A new
shallow water corals: Temperature dependence and growth-related calibration for the alkenone paleothermometer. Paleoceanography
vital effects. Geochimica et Cosmochimica Acta, 99, 224–242. and Paleoclimatology, 33, 281–301. https://doi.org/10.1002/
Schmidt, G. A., Bigg, G. R., & Rohling, E. J. (1999). Global Seawater 2017PA003201.
oxygen-18 database—v1.21. Available at http://data.giss.nasa.gov/ Tierney, J. E., & Tingley, M. P. (2014). A Bayesian, spatially-varying
o18data/. calibration model for the TEX86 proxy. Geochimica et Cosmochim-
Schmidt, G. A. (1999). Error analysis of paleosalinity calculations. ica Acta, 127, 83–106.
Paleoceanography, 14, 422–429. Vázquez Riveiros, N., Govin, A., Waelbroeck, C., Mackensen, A.,
Schmittner, A., Bostock, H. C., Cartapanis, O., Curry, W. B., Filipsson, Michel, E., Moreira, S., et al. (2016). Mg/Ca thermometry in
H. L., Galbraith, E. D., et al. (2017). Calibration of the carbon planktic foraminifera: Improving paleotemperature estimations for
isotope composition (d13C) of benthic foraminifera. Paleoceanog- G. bulloides and N. pachyderma left. Geochemistry, Geophysics,
raphy, 32, 512–530. Geosystems, 17, 1249–1264.
Schauble, E. A., Ghosh, P., & Eiler, J. M. (2006). Preferential Vázquez Riveiros, N., Waelbroeck, C., Skinner, L. C., Duplessy, J. C.,
formation of 13C–18O bonds in carbonate minerals, estimated McManus, J. F., Kandiano, E. S., et al. (2013). The ‘MIS11
using first-principles lattice dynamics. Geochimica et Cosmochim- paradox’ and ocean circulation: Role of millennial scale events.
ica Acta, 70(10), 2510–2529. Earth and Planetary Science Letters, 371–372, 258–268.
Schouten, S., Hopmans, E. C., Schefuß, E., & Damste, J. S. S. (2002). Vázquez Riveiros, N., Waelbroeck, C., Skinner, L. C., Roche, D. M.,
Distributional variations in marine crenarchaeotal membrane lipids: Duplessy, J. C., & Michel, E. (2010). Response of South Atlantic
A new tool for reconstructing ancient sea water temperatures? Earth deep waters to deglacial warming during terminations V and I.
and Planetary Science Letters, 204(1–2), 265–274. Earth and Planetary Science Letters, 298, 323–333.
Schouten, S., Ossebaar, J., Schreiber, K., Kienhuis, M. V. M., Langer, Waelbroeck, C., Labeyrie, L., Duplessy, J. C., Guiot, J., Labracherie,
G., Benthien, A., et al. (2006). The effect of temperature, salinity M., Leclaire, H., et al. (1998). Improving past sea surface
and growth rate on the stable hydrogen isotopic composition of long temperature estimates based on planktonic fossil faunas. Paleo-
chain alkenones produced by Emiliania huxleyi and Gephyrocapsa ceanography, 13, 272–283.
oceanica. Biogeosciences, 3, 113–119. Waelbroeck, C., Duplessy, J. C., Michel, E., Labeyrie, L., Paillard, D.,
Schouten, S., Hopmans, E. C., & Damsté, J. S. S. (2013). The organic & Duprat, J. (2001). The timing of the last deglaciation in North
geochemistry of glycerol dialkyl glycerol tetraether lipids: A Atlantic climate records. Nature, 412, 724–727.
review. Organic Geochemistry, 54, 19–61. Weijers, J. W., Schouten, S., Spaargaren, O. C., & Damsté, J. S. S.
Schrag, D. P., Adkins, J. F., McIntyre, K., Alexander, J. L., Hodell, D. (2006). Occurrence and distribution of tetraether membrane lipids in
A., Charles, C. D., et al. (2002). The oxygen isotopic composition soils: Implications for the use of the TEX86 proxy and the BIT
of seawater during the last glacial maximum. Quaternary Science index. Organic Geochemistry, 37(12), 1680–1693.
Reviews, 21(1–3), 331–342. Weldeab, S., Lea, D. W., Schneider, R. R., & Andersen, N. (2007).
Shackleton, N. J. (1967). Oxygen isotope analysis and Pleistocene 155,000 years of West African monsoon and ocean thermal
temperature reassessed. Nature, 2151, 15–17. evolution. Science, 316, 1303–1307.
254 T. Caley et al.

Wiersma, A. P., & Renssen, H. (2006). Model–data comparison for the Zachos, J. C., Shackleton, N. J., Revenaugh, J. S., Pälike, H., & Flower,
8.2 ka BP event: Confirmation of a forcing mechanism by B. P. (2001). Climate response to orbital forcing across the
catastrophic drainage of Laurentide Lakes. Quaternary Science oligocene-miocene boundary. Science, 292, 274–278.
Reviews, 25(1–2), 63–88. Zachos, J. C., Dickens, G. R., & Zeebe, R. E. (2008). An early
Wit, J. C., De Nooijer, L. J., Wolthers, M., & Reichart, G. J. (2013). Cenozoic perspective on greenhouse warming and carbon-cycle
A novel salinity proxy based on Na incorporation into foraminiferal dynamics. Nature, 451, 279–283. https://doi.org/10.1038/
calcite. Biogeosciences, 10(10), 6375–6387. nature06588.
Wolhowe, M. D., Prahl, F. G., Probert, I., & Maldonado, M. (2009). Zweng, M. M., Reagan, J. R., Antonov, J. J., Locarnini, R. A.,
Growth phase dependent hydrogen isotopic fractionation in Mishonov, A. V., Boyer, T. P., Garcia, H. E., Baranova, O. K.,
alkenone-producing haptophytes. Biogeosciences, 6, 1681–1694. Johnson, D. R., Seidov, D., & Biddle, M. M. (2013). World Ocean
Yu, E.-F., Francois, R., & Bacon, M. (1996). Similar rates of modern Atlas 2013, Volume 2: Salinity. In S. Levitus & A. V. Mishonov
and last-glacial ocean thermohaline circulation inferred from (Eds.), NOAA Atlas NESDIS 74, p. 39.
radiochemical data. Nature, 379, 689–694.
 Climate Evolution on the Geological
Timescale and the Role of Paleogeographic 22
Changes

Frédéric Fluteau and Pierre Sepulchre

Throughout geological time, major climate changes have perform a “contextual” simulation of a large geologic time
marked the history of the Earth (Fig. 22.1). Although pale- interval (for example, the late Miocene). One can then study
oclimate markers provide us with the broad outlines of these the impact of paleogeographic or geochemical changes
changes, their causes could be manifold as feedback mech- within this interval via sensitivity experiments in which, for
anisms occur between the different compartments of the example, a mountain range is uplifted or lowered, an ocean
climate system. In a system where the solid and fluid passage is opened or closed, the chemical composition of the
envelopes are closely linked, understanding the evolution of atmosphere is changed or the orbital parameters of the Earth
Earth’s climates at the scale of geologic time involves are altered in line with information provided by the geo-
knowing its paleogeographic history. Within this context, logical, geophysical and geochemical data.
climate modelling is presented as a well-adapted tool to help In the first part of this chapter, we present the broad
to understand the factors causing climate change over geo- outlines of the climate history of the earth and in the second
logical time. However, modelling a continuous climate part, we provide examples of how the direct and indirect
evolution over million-year timescales is out of reach, as it couplings between the different envelopes, solid, liquid and
would require a detailed and reliable knowledge of the gaseous, can be studied through modelling and show how
model boundary conditions, e.g. the location of the conti- they contribute to the explanation of the climate history of
nents, the topography, the bathymetry as well as the chem- the Earth over long time scales.
ical composition of the atmosphere. Prior to the recent
Quaternary period, uncertainties generally tend to increase
regarding these conditions, and our knowledge becomes The Evolution of Climate Over the Past 4.54
increasingly fragmentary the further into the past we go. Billion Years
Moreover, even with a perfect knowledge of these condi-
tions, several million years simulations are beyond the Although the Precambrian (4.54–0.54 Ga) represents 88% of
computing capabilities of the supercomputers and the codes the Earth’s history we only have a very fragmentary
used today to simulate climate. This methodological knowledge of the climate during this period. There are very
dilemma is routinely overcome by means of steady-state few records of the first 900 million years of the Earth’s
simulations, called “snapshot experiments”, which simulate history (4.5–3.6 Ga). The oldest geological formations dis-
the response of the climate system to particular boundary covered in northwestern Canada (Acasta Gneiss) and in
conditions, and which require only a few thousand years of Greenland (Isua Greenstone Belt) are dated at 4 Ga and
simulation, corresponding to the time needed to achieve 3.8 Ga respectively (Valley 2006) but do not provide any
equilibrium for all compartments of the climate system. The climate constraints. However, the discovery of zircons in
first step is to establish the boundary conditions so as to Australia in the Archean metaconglomerates of Mount
Narryer and Jack Hills, dated at 4.4 Ga, is evidence of the
F. Fluteau (&) existence of the first granitic proto-continents (sensu lato).
Université de Paris, Institut de Physique du Globe de Paris, CNRS, The oxygen isotopic signature (d18O) of these zircons (5–
75005 Paris, France 7‰) confirms the presence of liquid water, and certainly of
e-mail: [email protected]
oceans, 150 million years after the formation of the Earth.
P. Sepulchre Between 4.3 and 2.8 Ga (Archean), the Earth’s atmosphere
Laboratoire des Sciences du Climat et de l’Environnement,
CNRS-CEA-Université de Versailles Saint Quentin en consisted of a mixture of nitrogen and greenhouse gases,
Yvelines-Université Paris-Saclay, Gif-sur-Yvette, France notably carbon dioxide and methane. There was no oxygen.

© Springer Nature Switzerland AG 2021 255

G. Ramstein et al. (eds.), Paleoclimatology, Frontiers in Earth Sciences,
https://doi.org/10.1007/978-3-030-24982-3_22
256 F. Fluteau and P. Sepulchre

Fig. 22.1 a Estimates of water temperature calculated using maxi- low-latitude planktonic foraminifera younger than 118 Ma and
mum d18O chert value per 50 Ma age intervals, adapted from Tartèse brachiopods older than 117 Ma. The temperature curves were calcu-
et al. (2017). The secular trend shows an overall cooling of Earth lated using a baseline value for Phanerozoic seawater which mimics the
through geological time. Blue bars indicate episods of widespread present day 0‰ SMOW. Grey shading indicate 2-sigma intervals. Blue
glaciation. Dashed lines, and question marks show the apparent bars indicate the presence of ice caps. Adapted from Veizer and
contradiction between poorly dated warm temperature estimates and Prokoph (2015)
the relatively short glacial events. b Evolution of d18O data for
22 Climate Evolution on the Geological Timescale and the Role … 257

There are big uncertainties regarding Earth’s surface tem- atmosphere is the subject of intense debate because a high
perature during the Archean, as deduced from isotopic partial pressure of CO2, necessary to counter the “young
measurements of oxygen in carbonate rocks, with values sun” effect, would have strongly acidified precipitations,
ranging from 26 to 85 °C (Knauth and Lowe 2003; Robert causing the pH to fall to about 3.7, and cause a particularly
and Chaussidon 2006; De Wit and Furnes 2016). However, intense chemical weathering of the rocks, which does not
these values may be inaccurate due to our lack of under- seem to be the case for the 3.5–3 Ga period (Kasting and
standing of the average d18O of the seawater at that time. Howard 2006). Another gas, methane, may therefore have
Indeed, the average 18O concentration in the oceans is played an important role. In the modern atmosphere, the
determined by 18O exchanges with oceanic crustal basalts residence time of methane is around eight years because it is
during the circulation of seawater in the hydrothermal sys- quickly oxidized. However, in an atmosphere devoid of
tems of the ocean ridges. There is no evidence to indicate oxygen or at very low partial pressures (<10−5 bar) as in the
that it was within the range of values observed much more Archean, this residence time is considerably longer. Methane
recently. Measurements of the isotopic ratio of oxygen and can thus accumulate in the atmosphere and significantly
silicon in cherts (siliceous rocks) suggest a paleotemperature increase the greenhouse effect. We know of two sources of
of 60–80 °C (Fig. 22.1a). However, these values may be methane at that time: methanogenic organisms and the ser-
distorted by the contributions of the hydrothermal systems pentinization of ultramafic rocks on the ocean floor. Taking
lining the oceanic crust at that time. To overcome these the results of the work of Haqq-Mishra et al. (2008), with
difficulties, new isotopic tools, called the clumped isotopes, 1% methane in the atmosphere, the CO2 partial pressure
have been developed. A promising paleothermometer, based required to maintain the Earth at 30 °C drops to about 10−3
not on the relative abundance of one single isotope to bar. The contribution of methane to the greenhouse effect is
another, but on the abundance ratio of isotopologues with therefore extremely effective. Even if atmospheric concen-
multiple rare isotopes relative to the expected quantity for a trations remain poorly constrained, a CH4 to CO2 ratio
stochastic distribution of isotopes within a (13C18O16O2)2− greater than 0.2 is impossible (Zerkle et al. 2012).
group (Eiler 2007; Bonifacie et al. 2017). The thermody- Traces of several glacial periods, dated between 3.5 and
namic variable measured and noted D47 is therefore based on 2.2 Ga, have been discovered in South Africa, Europe,
the substitution of carbon and oxygen isotopes within the North America and Australia. The oldest glaciation was
carbonate ions, which is a function of the temperature at discovered in South Africa within the Barberton Greenstone
which these carbonates formed. Although uncertainties Belt. It was dated at about 3.4–3.5 Ga (De Wit and Furnes
remain (Daëron et al. 2019), this method does not require 2016) and in a latitude band between 20° and 40° (Biggin
knowledge of the average isotopic composition of the ocean et al. 2011). It was followed by another glacial episode at
and can therefore be applied to the oceans of the Archean. 2.9 Ga discovered in units of the Mozaan geological group
Although the evolution of temperature during the Archean is in South Africa (Young et al. 1998). The end of the Archean
up for debate, it remained within a range of values able to and the beginning of the Proterozoic produced the Huron
sustain liquid water on the Earth’s surface. This is an glaciation, originally discovered in the province of Ontario
essential condition for the development of life which could in Canada, but identified in South Africa and Australia. It is
have started as early as 3.77 Ga (Dodd et al. 2017). actually a succession of three or four glacial events, dated
Maintaining a temperature conducive to the development between 2.45 and 2.2 Ga (Caquineau et al. 2018). One of
of life on Earth was by no means guaranteed within the these glacial episodes is demonstrated by the presence of
context of the “young Sun”. In fact, the radiation emitted by glacial sediment at low latitudes and at low altitude sug-
a young star is weak, gradually increasing throughout its life. gesting that the Earth could have been completely frozen at
In the case of the Sun, the solar radiation is estimated to have this time. These glaciations are associated with a major event
been 30% lower 4 Ga ago than it is today, and 20% lower in the history of the Earth: the oxygenation of the atmo-
3 Ga ago. In these radiation conditions, maintaining a tem- sphere and of the shallow oceans.
perature that would support the presence of liquid water on The next billion years (between 1.85 and 0.85 Ga) is
the surface of the globe during the Archean would have been often referred to as the ‘boring billion’ due to the apparent
impossible without a powerful greenhouse effect. climatic and environmental stability. Indeed, isotopic mea-
Let’s examine the data to determine the constraints on the surements of carbon (d13C) show no major disturbance of
chemical composition of the atmosphere. Precipitation of the carbon cycle and the absence of any known glacial traces
sodium bicarbonate in the Archean oceans around 3.3 Ga during this period suggest (but do not prove) that the Earth’s
argues in favour of a partial CO2 pressure of between 1.4 and climate had stabilized into a hot configuration. This climate
15% of the atmospheric pressure of the time (Lowe and Tice stability ended with the Neoproterozoic. During this time,
2004), and so lower than the estimates by the models. the Earth experienced three periods of glaciation: the Stur-
However, the chemical composition of the Precambrian tian glaciation from 717 to 659 Ma, the Marinoan glaciation
258 F. Fluteau and P. Sepulchre

from 645 to 636 Ma and finally the Gaskiers glaciation glaciation, which lasted 70 Ma, was punctuated by several
around 582 Ma. Marine glacial sedimentary formations at phases of advance and withdrawal of the ice sheets. This is
low latitudes indicate that during the Sturtian and Marinoan the longest and most important glacial episode of the entire
glaciations, the Earth was completely glaciated: these are the Phanerozoic. Sedimentary formations, striated floors and
famous episodes of Snowball Earth (Hoffman et al. 2017). ‘dropstones’ (pieces of rock deposited onto unconsolidated
Directly on top of these glacial sedimentary formations we marine sediments by icebergs when they melted) are proof of
find carbonate formations (currently, carbonate production is the presence of ice in South America, southern and eastern
mainly located in warm tropical seas). The rapidity of this Africa, on the Arabian peninsula, the Indian subcontinent
transition between glacial and carbonate formations (on the and Australia, that is, the whole southern part of the
scale of geological time) is a peculiarity in the climate his- Gondwana continent then located in the mid and high lati-
tory of the Earth. Numerous studies have been undertaken in tudes of the southern hemisphere.
recent years to understand the entry and exit modalities of At the end of the Carboniferous, while a cold climate
these glacial phases (Hoffman et al. 2017; and Chap. 5). developed at the high and mid latitudes of the southern
The Precambrian/Cambrian boundary (542 Ma) marked a hemisphere, paleoclimatic indicators indicate that there was
new turning point in the Earth’s climate history. According a tropical and humid climate over a part of Europe and North
to oxygen isotopic ratio measurements, the mean global America, then located close to the equator. At this time,
temperatures of the Phanerozoic climate became stable there was a strong contrast between the climates of low and
within a range comparable to that of the modern day high latitudes.
(Fig. 22.1b). This period was punctuated by three major These conditions disappeared in favor of a warmer and dryer
glaciations: one at the end of the Ordovician (around climate during the Late Permian. Wet climates were limited to
443 Ma), during the Permo-Carboniferous (between 335 and narrow bands around the equator and in the mid-latitudes. The
260 Ma) and at the end of the Cenozoic (the last 40 Myr). Paleozoic era ended with two mass extinctions, the first at the
Traces of the Ordovician glaciation are found in Africa, end of the “Guadalupian” (*258 Ma) and the second at the
particularly in the Sahara and South Africa but also on the Permo-Trias boundary (*251 Ma) (Bond et al. 2010; Bond and
Arabian peninsula. This glaciation is estimated to have las- Wignall 2014). This last crisis is the most important mass
ted a little more than one million years, and is marked by at extinction of the Phanerozoic with the disappearance of 90% of
least three glacial cycles alternating with interglacial periods the fauna and flora (Erwin 1994). During this crisis, climate
(Ghienne et al. 2013). The glaciation is associated with a indicators show significant warming, a disruption of the carbon
major disruption in the carbon cycle but also with one of the cycle as well as oceanic anoxia. An exceptionally high level of
five mass extinctions of the Phanerozoic when close to 86% volcanic activity leading to the contemporaneous formation of
of the marine benthic and planktonic species died out. the Siberian large igneous province is considered to be the main
During the Silurian and Devonian periods, the Earth expe- cause of the extinction. Although the relationship between
rienced a warmer global climate. Carbonate platforms, volcanic activity and this crisis is not fully understood, it should
which develop in warm seas, stretched from 45° S to 60° N. be noted that every ecological crisis of the Phanerozoic, what-
An expansion of this scale would never be seen again ever its magnitude, occurred simultaneously with the estab-
(Copper and Scotese 2003). The strong latitudinal expansion lishment of a large basaltic province through particularly intense
of these carbonate platforms suggests weak latitudinal ther- volcanic activity. After the Permo-Trias mass extinction, a warm
mal gradients. The widespread evaporite facies suggest a global climate became established during the Triassic. During
semi-arid to arid climate at subtropical latitudes. the Late Triassic, sedimentary facies in North America suggest a
The Devonian period is also marked by the colonization strong seasonal precipitation caused by “mega-monsoon” pat-
of land surfaces by life. The first forests, made up of terns (Dubiel et al. 1991; Bahr et al. 2020). Subsequently, the
Archeopteris, appeared at the end of the Devonian at around Earth experienced a colder global climate during the Jurassic.
370 Ma (Meyer-Berthaud et al. 1999). During the Silurian, The presence of ice caps at high latitudes has been suggested, but
some plants, including bryophites, were the pioneers of this this hypothesis is based on assumptions that have still to be
land colonization that, until then, had been deserted and confirmed (Dromart et al. 2003).
barren. The emergence of the continental biosphere affected Until the mid-1990s, the Cretaceous (145–65 Ma) was
the carbon cycle and probably influenced the climate of the described as a period with a uniformly warm global climate,
Earth (Le Hir et al. 2011). The end of the Devonian and the but the accumulation of data from different climate indicators
beginning of the Carboniferous mark the return of glacial has completely changed our notions of the climate of this
periods and a more contrasted climate latitudinal gradient. period. The beginning of the Cretaceous period was cold,
The Earth then underwent a glacial period which started in probably with ice sheets, but experienced a particularly warm
the Carboniferous (*335 Ma) and which ended with the period towards the middle of the Cretaceous, before cooling
Permian around 260 Ma (Montañez and Poulsen 2013). This towards the end of the period. On top of this long-term trend,
22 Climate Evolution on the Geological Timescale and the Role … 259

rapid climate variability was superimposed. The isotopic transition of the Middle Miocene is suggested by the sedi-
measurements of the oxygen in carbonate tests of planktonic mentary facies observed in the Arctic Ocean. This cold cli-
foraminifera and fish teeth provide an estimate of the pale- mate impacts on the high latitudes of both hemispheres and
otemperature of the surface waters of the oceans. These data, provides the necessary conditions for the rapid glacial/
as well as studies carried out in continental areas, on paly- interglacial fluctuations during the Pleistocene (but does not
nomorphs, for example, reflect the climate variability of this trigger them).
period (Ladant and Donnadieu 2016). Thus, d18O measure- Paleoclimate indicators can be used to trace the evolution
ments show that the Turonian had a particularly hot climate, of the Earth’s climate, so as to gradually refine its contours
with sea surface temperatures of between 34 and 37 °C. and to observe rapid fluctuations superimposed on
However, the d18O measurements also show the existence of longer-term trends. However, there are areas of uncertainty
a glaciation event lasting less than 200,000 years (Borne- (as will probably always be the case), especially for the
mann et al. 2008). The Cretaceous ended with a rapid cooling oldest periods. The causes and mechanisms of these climate
just before the major mass extinction of the Cretaceous- changes at the scale of geological time are manifold. Data
Tertiary boundary at 66 Ma, during which nearly 60% of the collected in the field allow us to document this evolution
species on Earth died out. As for the other crises of the with increasing accuracy, but it is impossible to isolate with
Phanerozoic, this crisis is synchronous with the establishment certainty the specific cause or causes of these climate dis-
of a large basaltic province, the Deccan traps in India, but ruptions. Since the 1970s, numerical modeling of climates
also, with another event, the fall of an extraterrestrial body has been used, in addition to data, to test the sensitivity of
into the Gulf of Mexico. the climate to different forcings and to try to reproduce
After this crisis, the Earth entered a warm period during numerically the climate changes observed in the field. Over
which signs of glaciation disappeared. This period was punc- long time scales, paleogeographic changes brought about by
tuated by the thermal maximum at the Palaeocene-Eocene plate tectonics have shaped the face of the Earth (Volume 1,
boundary (55 Ma) (Zachos et al. 2001). This was a rapid tran- Chap. 2) and are a major forcing of the Earth’s climate
sient event (at the geological time scale), lasting about 300 ka, history through their direct effects on atmospheric and
marked by an abrupt increase in temperatures (considered as the oceanic circulation. We will also see the indirect effects of
best analogue to current global warming). Tropical flora, but paleogeographic changes induced by feedback mechanisms,
also turtles, crocodiles, and many mammals were discovered on in particular on the regulation of the partial pressure of CO2,
the Ellesmere and Axel Heiberg islands in northern Canada. another major contributor to the climate system.
As early as the Late Eocene (*50 Ma), the Earth expe-
rienced gradual cooling with the onset of a new glacial
period around 40 Ma marked by a first stage of development Some Consequences of Paleogeographic
of the Antarctic ice cap, corresponding probably to the Changes on the Earth’s Climate
appearance of glaciers in the Trans-Antarctic chain. This ice
cap grew quickly and reached the coast, as evidenced by Continental Drift
glacial sediments found off the Antarctic continent, dating
from the Eocene-Oligocene boundary (*34 Ma). Numerous Continental drift leads to changes in the latitudinal and
measurements of the isotopic composition of oxygen in the longitudinal distribution of emerged lands with various
carbonate tests of benthic foraminifera in sediments col- consequences for climate. The main ones are: changes in the
lected at different points around the globe confirm a con- distribution of solar radiation received by the continents;
sistent cooling during the Eocene. At the Eocene-Oligocene changes in the atmospheric dynamics by the uplift or col-
boundary, a significant and rapid cooling of the ocean bot- lapse of mountain ranges or during the formation or
tom waters occurred, estimated at about 6 °C (Hren et al. break-up of supercontinents; changes in ocean dynamics
2013). Henceforth, the isotopic ratio of strontium increased during the opening or closing of basins or ocean passages
significantly in response to intensified continental erosion. (also called seaways), and also indirect effects such as
During the Lower Miocene (23–15 Ma), the global trend changes in weathering fluxes that impact on the carbon cycle
was a slight warming interspersed with brief cooling epi- (Donnadieu et al. 2004). In the Upper Permian (*260 Ma),
sodes. Around 14 Ma, rapid cooling led to a new phase of paleoclimate indicators suggest that a warm, dry climate
development of the Antarctic ice sheet. This climate trend developed over a large part of Gondwana (southern hemi-
accelerated during the Upper Miocene and the Pliocene. The sphere) located between the narrow rainy equatorial band
development of an ice cap on Greenland probably dates from and the narrow temperate mid-latitude band, below the
the Upper Miocene or the Pliocene, but a high probability of theoretical subsidence zone of the Hadley cell. Several
sea ice on the Arctic Ocean starting from the climate numerical climate simulations made it possible to develop
260 F. Fluteau and P. Sepulchre

hypotheses to explain the various mechanisms involved. these paleogeographic changes into account in numerical
Fluteau et al. (2001) suggested that high seasonality in simulations suggests that the establishment of a sea passage
Gondwana was typical at mid and high latitudes due to the between the South Atlantic and Central Atlantic played an
low heat capacity of the continents, and that monsoon-type important role in the formation and oxygenation of deep
atmospheric circulation marked the eastern side of the waters on a global scale (Poulsen et al. 2003).
supercontinent. Other modeling studies (Kiehl and Shields The Cenozoic is also characterized by large-scale climate
2005; Shields and Kiehl 2018) have confirmed these results, change, including global cooling and reorganization of ocean
even suggesting that these characteristics continued despite circulation. Apart from India and Australia, the drift in latitude
a high atmospheric concentration of carbon dioxide of most of the continents is small over the last 60 million years,
(3550 ppm). Climate modeling has also looked at the con- and cannot by itself explain the temperature changes implicit
sequences of supercontinent break-up. By simulating the in the data. However, several openings and closings of ocean
climate response to paleogeographic changes between the passages have altered the exchanges between water bodies and
Triassic and Cretaceous, it was suggested, for example, that the associated heat flows. At the beginning of the Cenozoic, the
the break-up of Gondwana and Laurasia annihilated the distribution of the continents meant that the Pacific, Atlantic
continental effect, preventing the development of large and Indian basins were connected in the tropical band via three
desert bands in the subtropics (Fig. 22.2), and favoring the open ocean passages: The Central American seaway (CAS),
establishment of wet conditions contemporaneous to the the east-Tethys seaway, and the Indonesian passage. Numer-
diversification of flowering plants (Angiosperms) (Fluteau ical simulations suggest that this interoceanic connection
et al. 2007; Chaboureau et al. 2014). operated from east to west at the surface, with the formation of
the circum-equatorial current (CEC). Conversely, in the
southern hemisphere, the Drake and Tasmanian passages
Paleogeographic Changes and Ocean Circulation separating South America and Australia from Antarctica,
respectively, were closed, preventing the formation of a strong
The oceans are a major component of the climate system. Antarctic Circumpolar Current (ACC). At the Eocene-
They ensure, in part, the transport of heat from the low to the Oligocene transition, the circumpolar Antarctic maritime
high latitudes, particularly in modern times, via the Atlantic passages gradually opened, widened and deepened. This epi-
Meridional Overturning Circulation (AMOC). At the geo- sode would be followed by the gradual closure of tropical
logical time scale, large changes in ocean basin geometry passages in the Middle Miocene (15 Ma).
have controlled the dynamics of water bodies and the Numerical simulations show that both events contributed
associated heat and salt fluxes by opening or closing inte- to the establishment of deep water formation in the northern
roceanic connections. This is, for example, the case of the hemisphere and to the cooling of the southern hemisphere
opening of the South Atlantic Ocean which began 135 (see Sijp et al. 2014 for an overview). Since the late 1970s,
million years ago. Exchanges between the South Atlantic the establishment of the ACC in response to the opening of
and Central Atlantic oceans have only happened since the marine passages of the southern hemisphere has been
100 Ma (Murphy and Thomas 2013; Granot and Dyment advanced as a cause of the freeze-up of the Antarctic (Kennett
2015) and led to changes in the global ocean circulation as 1977). Indeed, the opening of the Tasman Sea and the
shown by the isotopic data of neodymium and oxygen (e.g. oceanic exchanges through the Drake Passage in southern
Donnadieu et al. 2016). Although there is no consensus on South America fostered the thermal isolation of Antarctica.
the exact evolution of Cretaceous ocean circulation, taking Recent studies show that the paleogeographic configuration,

Fig. 22.2 Simulated Jurassic

and Cretaceous bioclimatic zones.
The numbered points indicate the
localisation of angiosperm fossils.
Adapted from Chaboureau et al.
(2014)
22 Climate Evolution on the Geological Timescale and the Role … 261

in particular the bathymetry, is decisive to correctly quantify Other maritime passages have seen their configuration
the cooling related to the change of configuration of the change during the Cenozoic. The drift of Australia towards
seaways, but also that the freeze-up of the Antarctic requires, Indochina has progressively restricted the maritime
in addition to paleogeographic changes, a significant exchanges between the Pacific Ocean and Indian Ocean
decrease of the atmospheric partial pressure of carbon diox- through the Indonesian Passage. Numerical simulations
ide from 1160 to 560 ppm (Lefebvre et al. 2012). show that before closure around 4 Ma, a warm ocean current
Numerous modelling studies have also focused on the flowed between the tropical Pacific Ocean and the Indian
impact of the central American seaway (CAS) on climate. Ocean (Cane and Molnar 2001; Brierley and Fedorov 2016).
Early works by Haug and Tiedemann (1998) had suggested After closure, this warm ocean current was blocked and was
that the closure of the isthmus during the early Pliocene led to a replaced by a colder current from the north Pacific. Simu-
reorganization of the Atlantic thermohaline circulation. lated consequences involve a cooling of the surface waters of
Specifically, the authors inferred that the shallowing and clo- the Indian Ocean and a marked drying in East Africa,
sure of the CAS intensified the Gulf stream and strengthened causing tree cover to decrease in favor of savanna vegeta-
deep water formation in the northern Atlantic. More impor- tion. Although this drying has been confirmed by paleocli-
tantly these authors suggested that warmer SST induced by mate indicators (Bonnefille 2010), more proximal tectonics
this strengthening of the Gulf stream increased atmospheric changes, namely the uplift of the east African dome might
moisture content, and ultimately favored the Greenland ice have played a role in this aridification (Sepulchre et al.
sheet growth. Later, numerous model experiments with open 2006).
CAS with various depth and width have been carried out (see
Zhang et al. 2012; Sepulchre et al. 2014 for reviews). Most of
the model indeed showed intensified AMOC with the seaway The Influence of Shelf and Epicontinental Seas
closure, but models run with explicit ice sheet modelling failed
to demonstrate that the closure had a significant impact on the Variations in sea level have marked the history of the Earth.
Greenland icecap onset (Lunt et al. 2008; Tan et al. 2017). Our Reconstructions by Haq et al. (1987) of eustatic variations
knowledge of the timing of CAS closure also evolved during show that a high underlying sea level during the Upper
the last 20 years. Although still very debated (O’Dea et al. Cretaceous (*95 Ma) is responsible for the formation of
2016), authors have suggested that CAS constriction hap- numerous shelf seas. The functioning and role of these shelf
pened much earlier than previously thought (Montes et al. seas is still poorly understood as there is no modern equiv-
2015; Bacon et al. 2015; Jaramillo 2018), with a very restricted alent of these water bodies. From the climate perspective, the
seaway by the late Miocene (ca. 10 Ma, Fig. 23.3). This dif- answer seems simple: the higher the sea level, the less land
ferent chronology has many consequences on our under- surface is exposed, the smoother the seasonal cycle and the
standing of CAS influence on climate, as it involves that its more homogeneous the climate becomes. Simulations con-
constriction has occurred during other major tectonics events, ducted using a general atmospheric circulation model indi-
such as the uplift of the south American cordilleras (Andes). cate that the climate response to the formation of shelf seas is

Fig. 22.3 Sketches of putative ocean currents in different paleogeo- ocean gateway is open between the American continents at 25 Ma,
graphic configurations for Central America between 25 Ma and whereas only a very narrow passage allows water exchange during the
present. Reconstructions of continental areas are from Jaramillo late Miocene. With this latter configuration, climate models suggest that
(2018). Blues arrows show surface currents, and doted-red arrows surface currents flow westward, while undercurrents can bring fresher
show subsurface to mid-depth currents. In these reconstructions, a wide Pacific waters into the tropical Atlantic ocean
262 F. Fluteau and P. Sepulchre

more complex than it might appear (Fluteau et al. 2006). shifted northward by up to 1000 km along the WIS fol-
A massive epicontinental sea in North America, the Western lowing its establishment. Although, these results cannot be
Interior Seaway (WIS), stretched across the North American generalized since they depend on the location of the shelf
continent from the Arctic ocean to the Gulf of Mexico, due seas relative to the high and low pressure areas that develop
to the sea incursion during the Middle Cretaceous, and to the over the continent, they seem to have greatest impact on
dynamic topography caused by the subduction of the Far- temperatures in the mid-latitudes and to induce significant
allon Plate under North America. During the Aptian increases in precipitation at low latitudes, as the shelf seas
(*20 Ma), this arm of the sea had not yet formed. Simu- act as important reservoirs of water for the atmosphere.
lated atmospheric circulation in winter over North America
is driven by a high pressure zone over the northeast of the
continent. In summer, the opposite is the case; a zone of low The Impact of Mountains on Climate
pressure develops over the southwest of North America east
of the mountains bordering the Pacific Ocean. The onset of As described in Volume 1, Chap. 2, much of the modern
the WIS completely disrupts the atmospheric circulation in topography was formed during the Cenozoic. The paleoal-
summer across the continent (Fig. 22.4). The presence of timetry proxies described earlier and the sedimentological
this sea pushes the low-pressure zone to the east, while in record suggest that the Andes, the East-African dome, as
winter temperatures increased by about 4 °C close to the well as the Himalayas and the Tibetan plateau rose mostly
WIS and precipitations, sometimes of snow, intensified over the last twenty million years. These different reliefs
along the Arctic Ocean. The warm temperate zones were form obstacles that alter climate through many mechanisms,

Fig. 22.4 Consequences of the establishment of the Western Interior (95 Ma). Pressure fields are represented by isobars (P—1000 hPa)
Seaway (according to Fluteau et al. 2006). a Atmospheric circulation in (HP = high pressure and BP = low pressure) and by winds (arrows) at
summer during the Aptian (120 Ma). b Idem during the Cenomanian 850 hPa (about 1500 m). Shaded areas: exposed land surfaces
22 Climate Evolution on the Geological Timescale and the Role … 263

at many spatial scales. Here we provide only a short over- large continental area at high altitude, are still in question. Other
view of the main interactions between mountain uplift and studies have shown (i) that the thermal contrast between a flat
climate dynamics to understand how the topographic chan- continent and the Indian Ocean was enough to initiate a lower
ges of the Cenozoic have changed climate. intensity monsoon than the modern monsoon pattern and
First, depending on its velocity, its angle with the (ii) that a prerequisite for this was the removal of the Paratethys
mountain range and the size of the obstacle, air arriving Sea which enhanced the heating of Central Asia, and therefore
towards a mountain range can undergo an orographic ascent the thermal contrast between the Indian Ocean and continental
that leads to adiabatic cooling until water condensate and Asia (Ramstein et al. 1997; Fluteau et al. 1999). More recently,
precipitate on the windward slope, leading to a rainfall the separation of air masses on either side of a mountain range
gradient between both sides of the orographic barrier. At the has been mentioned as a major factor in the initiation of con-
continental scale, this “rainshadow” produces strong vection and associated precipitation (see Boos 2015 for a
heterogeneities in rainfall patterns and associated biomes, review) and many modeling studies using various boundary
such as the ones observed today in Patagonia, for example. conditions have been undertaken to try to understand the exact
The varying temperature lapse rates, i.e. temperature role of altitude (e.g., latitudinal position (Zhang et al. 2015,
decrease with elevation, that depends on air masses char- 2018), geographic extension (Chen et al. 2014) and of vege-
acteristics, also produce thermal heterogeneity in elevated tation cover of the plateau (Hu and Boos 2017) in the estab-
regions. High mountains can also alter the surface radiative lishment and intensification of the Asian monsoon.
balance through snow-induced changes in albedo, creating Furthermore, the implications of the Himalayan-Tibetan
thermal and pressure gradients that alter air circulation. relief for the climate are not limited to the atmosphere. Rind
Lastly, mountain ranges have long been shown to create et al. (1997) used a general circulation model coupled with
different kinds of waves that propagate on the lee side and an ocean-atmosphere model (AOGCM) to show that these
vertically, ultimately disturbing winds and the spatial dis- uplifts caused a rise in sea surface temperatures (*2 °C) in
tribution of precipitation (Warner 2004). the North Atlantic (Norwegian Sea), an increase in heat
The impact of mountains on climate was considered as transport at high latitudes (a crucial parameter for under-
early as the middle of the nineteenth century, but it was not standing climate change at high latitudes and possibly the
until the 1950s that studies quantified this relief-climate freezing-over of Greenland) as well as a reduction of about
relationship. The meteorologists, Charney and Eliassen 10% in deep water production in the Norwegian Sea (linked
(1949), followed by Bolin (1950), assessed the impact of to the decrease in density of the water mass, due to the
mountain ranges on the mid-latitude westerlies, while Flohn warming of the North Atlantic Ocean). More recently, Su
(1950) suggested that the Tibetan plateau was a source of et al. (2018) suggested that plateau uplift may have con-
sensible heat significant enough to explain the establishment tributed to the establishment of the AMOC by changing the
of the Asian monsoon. Between the 1970s and 1990s, many intensity and latitude of the zonal winds, thereby altering
studies benefited from the advances in atmospheric general sea-ice formation and deep-water formation.
circulation models to quantify the impact of mountain ranges The impact of the reliefs on ocean-atmosphere dynamics
on the distribution of arid zones globally and the establish- are not limited to the Late Cenozoic orogenesis, because the
ment of the Asian monsoon. Broccoli and Manabe (1992) Earth’s history is dotted with uplifts creating mountain ranges
suggested that the aridity of Central Asia and the Great with different locations, extensions, heights, and orientations.
Plains of North America was partly due to the generation of By comparing climate simulations for a world with its current
stationary waves downwind of adjacent mountain ranges. topography and a “flat earth” world, Maffre et al. (2018)
One specific geological event has captured the attention of showed that the orographic barriers of the Andes and the
scientists: the uplift of the Tibetan plateau and the Himalayan Rockies constrain the transport of freshwater between the
range in the context of the India-Asia collision, which occurred Pacific and the Atlantic, and thus contribute to the high salinity
at the beginning of the Eocene (Molnar et al. 2010). The first of the latter, favoring the formation of deep water. The Andean
assessment studies on the impact of the uplift of the Tibetan uplift is also believed to be responsible for the establishment of
Plateau using climate models were based on sensitivity convective precipitation in tropical South America (Poulsen
experiments with varying elevations of the plateau without et al. 2010) and the strengthening of the Humboldt Current
differentiating between the Himalayan and Tibetan uplifts and (Sepulchre et al. 2009). The establishment of reliefs in south
without taking other paleogeographic changes into account and east Africa during the Mio-Pliocene probably led to the
(e.g. Kutzbach et al. 1993; Ruddiman et al. 1997). These studies aridification of East Africa (Sepulchre et al. 2006), to the
suggested that the rise of the Tibetan plateau played a crucial strengthening of the coastal upwellings of the Benguela cur-
role for atmospheric circulation in general, and for the Asian rent (Jung et al. 2014) and a change in position of the ITCZ in
monsoon in particular. However, the physical mechanisms the Atlantic (Potter et al. 2017).
involved, which use the temperature gradients induced by a
264 F. Fluteau and P. Sepulchre

The Indirect Effects of Paleogeographic Changes global average temperature (between 21.5 and 23 °C).
During the Jurassic, the drift of the Pangea to the north and
The atmospheric partial pressure of carbon dioxide (pCO2) is its break-up brought about an increase in continental sur-
driven by the carbon cycle. On the scale of geological time, faces exposed to the hot and humid climate of the equatorial
pCO2 reflects the balance between the CO2 emission fluxes band. This resulted in intensified chemical weathering of
from volcanic systems (ocean ridges and aerial volcanism), silicates thus causing an increase in the consumption of CO2.
the degree of magmatic activity from the mantle (plumes), The simulated pCO2 is lower at about 700 ppm as is the
decarbonation in subduction zones, and the CO2 fluxes average temperature of the globe at a little over 18 °C. The
consumed by the weathering of the rock silicates on the end of the Mesozoic is marked by the final break-up of
surface of the Earth and in the oceanic crust, through the Gondwana. The arid areas reduced in size, reinforcing global
burial of the organic matter. Fluctuations in pCO2 reflect the chemical weathering. This resulted in a fairly low simulated
evolution of one or both of these flows. CO2 emissions are pCO2 of between 300 and 500 ppm for the Cretaceous. The
proportional to the annual rate of production of oceanic effect of paleogeography on climate (through the regulation
crust. We have seen that the variability in this rate of pro- of chemical weathering of silicates) is therefore an effective
duction does not exceed 30% over the last 170 million years process that can be seen in long-term trends, even if it does
(Cogné and Humler 2006). To these rates should be added not explain every climate variations.
CO2 emissions from plumes of mantle volcanism and from The conditions necessary for the high latitudes to
decarbonation in subduction zones. Significant magmatic freeze-up were not present at the end of the Permian, but
events dating from the late Early Cretaceous and associated they were in place for a long period from the Lower Car-
with the establishment of some large submarine basalt pro- boniferous (340 Ma) to the Lower Permian (280 Ma),
vinces, such as the Ontong-Java Plateau, increase the pro- already in a supercontinent paleogeographic context. During
duction rate of oceanic crust by about 25% (Cogné and this glacial period, the Earth experienced a succession of
Humler 2006) and about the same increase in CO2 is injected advances and retreats of continental ice over southern
into the ocean-atmosphere system. Contributions from the Gondwana. What mechanisms would push the Earth into a
subduction zones are less constrained. Currently, this pro- different climatic state? Although the geographical configu-
cess is limited to a few subduction zones in the Pacific or ration of Pangea had not changed drastically between the
around the Indonesian archipelago, while the main deposi- Carboniferous and the Upper Permian, it had been drifting
tion zones are in the Atlantic and Indian Ocean. In the past, northwards during this period. Indeed, paleomagnetic data
the subduction of the Tethysian Ocean could have emitted a show that southern Gondwana was located at the pole during
significant CO2 flux (Hoareau et al. 2015). the Carboniferous, favoring a cooler summer in this region,
The intensity of the chemical weathering is a function of but not yet cold enough for ice to remain outside some high
climate parameters such as surface temperature and runoff. reliefs. Another more effective mechanism was needed. The
Moreover, most geochemical models considered runoff as a decrease in pCO2 was therefore necessary to explain this
function of surface temperature: the higher the temperature, glacial period. One suggestion was that colonization of
the greater the runoff, and consequently, the chemical emerged lands by plants during the Devonian increased the
weathering. The chemical weathering of silicates acts as a chemical weathering of silicates leading to a decrease in
climate regulator. We know now that this proportional pCO2 (Berner 2001). However, this vegetal colonization of
relationship between temperature and runoff is based on continents took place tens of millions of years before the
current data and is not transferable to past periods. This is beginning of the glaciation. Goddéris et al. (2017) have
easily understood by analyzing the upper Permian climate. shown that the Hercynian orogenesis could have played a
Simulation of chemical weathering is effective only in some major role. This orogenesis, resulting from the collision of
areas experiencing a tropical and humid climate. This weak Laurussia and Gondwana around 350 Ma, was at the origin
chemical weathering of silicates in the paleogeographic of the uplift of a vast chain of mountains, the Hercynian
context implies a relatively high pCO2 equilibrium of about chain, stretching for several thousands of kilometers in the
2500 ppm, and a high global average temperature of about equatorial band. With no mountains, the hot and humid
21 °C (considering that the rate of CO2 emissions from the climate of the lower latitudes causes a thick saprolite to
ridges is comparable to the current rate). Paleoclimate data form, considerably limiting the weathering of the underlying
confirm the hot and dry climate simulated over a large part bedrock. With orogenesis, the presence of relief produces
of Pangea. The paleogeography of the Triassic maintained strong mechanical erosion due to the slopes which consid-
the Earth in a relatively stable climate dominated by a high erably limit the development of thick saprolites, thus the
simulated pCO2 of around 3000 ppm and a high simulated weathering of silicate rocks is increased, leading to a
22 Climate Evolution on the Geological Timescale and the Role … 265

Fig. 22.5 Spatial positions (upper panel) and timing (lower panel) of The stable isotope curve in the bottom panel can be interpreted as
the modification of the principal mountain ranges and seaways ice-free temperatures. Adapted from Zachos et al. (2001)
discussed in the text. Bathymetry and topographic are from NOAA.
266 F. Fluteau and P. Sepulchre

decrease in the pCO2. This decline in pCO2 is further rein- calibration and the rise of paleoclimate modelling should
forced by the increased transport of organic carbon and its allow a better quantification of the links between paleo-
burial in sedimentary basins at the foot of the reliefs (God- geography and climate in the next decades.
déris et al. 2017). The threshold for freezing is defined by the
level of atmospheric CO2 concentration below which
freezing occurs. This threshold was reduced during the References
Carboniferous because of the slow, inexorable increase in
insolation produced by the Sun. Estimates, based on the Bacon, C. D., Silvestro, D., Jaramillo, C., Smith, B. T., Chakrabarty, P.,
analysis of fossil leaves and pedogenic carbonates, found & Antonelli, A. (2015). Biological evidence supports an early and
that in basins at low latitudes pCO2 fluctuated periodically complex emergence of the Isthmus of Panama. Proceedings of the
National Academy of Sciences, 112, 6110–6115.
from about 200 to 300 ppmv during the Upper Carbonifer- Bahr, A., Kolber, G., Kaboth-Bahr, S., Reinhardt, L., Friedrich, O.,
ous with a minimum concentration of 160–300 ppmv Pross, J. (2020). Mega-monsoon variability during the late Triassic:
(Montañez et al. 2016). The periodicity of these fluctuations Re-assessing the role of orbital forcing in the deposition of playa
seems to be dominated by an eccentricity signal of 405 ka. sediments in the Germanic Basin. Sedimentology, 67, 951–970.
https://doi.org/10.1111/sed.12668.
The vegetation cover of these equatorial basins also varies Berner, R. A. (2001). GEOCARB III: A revised model of atmospheric
between dry vegetation during interglacials and humid CO2 over Phanerozoic time. American Journal of Science, 301(2),
tropical vegetation during interglacials (Montañez et al. 182–204. https://doi.org/10.2475/ajs.301.2.182.
2016) which influence the organic carbon cycle, possibly Biggin, A. J., de Wit, M. J., Langereis, C. G., Zegers, T. E.,Voûte, S.,
Dekkers, M. J., & Drost, K. (2011). Palaeomagnetism of Archaean
causing fluctuations in the volume of continental ice on rocks of the Onverwacht Group, Barberton Greenstone Belt (southern
Gondwana and is superimposed on the long-term trend dri- Africa): Evidence for a stable and potentially reversing geomagnetic
ven by geodynamics. field at ca. 3.5 Ga. Earth and Planetary Science Letters, 302, 314–328.
Bolin, B. (1950). On the influence of the Earth’s orography on the
general character of the westerlies. Tellus, 2(3). http://www.tellusa.
net/index.php/tellusa/article/download/8547/9993.
Conclusion Bond, D. P. G., Hilton, J., Wignall, P. B., Ali, J. R., Stevens, L. G.,
Sun, Y., and Lai, X. (2010). The Middle Permian (Capitanian) mass
The impact of paleogeographic changes on climate and its extinction on land and in the oceans. Earth-Science Reviews, 102,
100–116, https://doi.org/10.1016/j.earscirev.2010.07.004.
evolution is fundamental. The presence of a supercontinent Bond, D. P. G. & Wignall, P. B. (2014). Large igneous provinces and
produces climates characterized by strong contrasts (conti- mass extinctions: An update. In G. Keller, & A. C. Kerr (Eds.),
nental climate at high latitudes, wide arid zone in the sub- Volcanism, impacts, and mass extinctions: Causes and effects:
tropics, as in the Permian), while a fragmented continent, Geological society of America special paper, 505 (p. 29). https://
doi.org/10.1130/2014.2505(02).
combined with vast shelf seas produced as a consequence of a Bonifacie, M., et al. (2017). Calibration of the dolomite clumped
high sea level, as in the early Upper Cretaceous, will be isotope thermometer from 25 to 350 °C, and implications for a
conducive to a climate with relatively low contrasts. The universal calibration for all (Ca, Mg, Fe) CO3 carbonates.
dispersion of landmasses therefore represents an important Geochimica et Cosmochimica Acta, 200(1), 255–279.
Bonnefille, R. (2010). Cenozoic vegetation, climate changes and hominid
climate forcing, which does not, however, offer a complete
evolution in tropical Africa. Global and Planetary Change, 72, 390–411.
explanation for the climate history of the Earth. Orogenesis is Boos, W. R. (2015). A review of recent progress on Tibet’s role in the
another major forcing of the climate at geological timescales. South Asian monsoon. CLIVAR Exchanges, 19(23), 27.
It is therefore essential to delineate the spatio-temporal evo- Bornemann, A., Norris, R. D., Friedrich, O., Beckmann, B., Schouten,
S., et al. (2008). Isotopic evidence for glaciation during the
lution of all the orogens as accurately as possible (Fig. 22.5),
Cretaceous supergreenhouse. Science, 319, 189–192.
even if the reconstructed paleoaltitudes remain, for the time Brierley, C. M., & Fedorov, A. V. (2016). Comparing the impacts of
being, subject to considerable uncertainty. While the impact Miocene–Pliocene changes in Inter-Ocean gateways on climate:
of paleogeography on atmospheric circulation has been Central American Seaway, Bering Strait, and Indonesia. Earth and
Planetary Science Letters, 444(juin), 116–130. https://doi.org/10.
widely studied through numerical modeling, the impact of
1016/j.epsl.2016.03.010.
paleogeographic changes on ocean circulation is far from Broccoli, A. J., & Manabe, S. (1992). The effects of orography on
well understood. Studies have quantified the impact of the Mid-latitude Northern Hemisphere Dry Climates. Journal of Climate,
opening of seaways (Panama or Indonesian) or large-scale 5(11), 1181–1201. https://doi.org/10.1175/1520-0442(1992)005.
Cane, M. A., & Molnar, P. (2001). Closing of the Indonesian Seaway
ocean connections (South Atlantic—Central Atlantic) on the
as a Precursor to East African aridification around 3–4 million years
climate through the redistribution of heat carried by ocean ago. Nature, 411, 157–162.
currents. The chronology of the openings and closures of the Caquineau, T., Paquette, J.-L., & Philippot, P. (2018). U-Pb detrital
ocean passages, in addition to that of the orogens, is thus zircon geochronology of the Turee Creek Group, Hamersley Basin,
Western Australia: Timing and correlation of the Paleoproterozoic
essential to a good understanding of the climate system at the
glaciations. Precambrian Research, 307, 34–50. https://doi.org/10.
scale of geological time. In this framework, the development 1016/j.precamres.2018.01.003.
of new isotopic tools, the improvements of traditional proxies
22 Climate Evolution on the Geological Timescale and the Role … 267

Chaboureau, A. C., Sepulchre, P., Donnadieu, Y., & Franc, A. (2014). Fluteau, F., Ramstein, G., Besse, J., Guiraud, R., & Masse. J. P. (2007).
Tectonic-driven climate change and the diversification of angios- Impacts of Palaeogeography and sea level changes on the Mid
perms. Proceedings of the National Academy of Sciences, 111(39), Cretaceous climate. Palaeogeography, Palaeoclimatology,
14066–14070. https://doi.org/10.1073/pnas.1324002111. Palaeoecology, 247, 357–381.
Charney, J. G., & Eliassen, A. (1949). A numerical method for Ghienne, J-F., Moreau, J., Degermann, L. et al. (2013). Lower
predicting the perturbations of the Middle Latitude Westerlies. Palaeozoic unconformities in an intracratonic platform setting:
Tellus, 1(2): 38–54. https://doi.org/10.1111/j.2153-3490.1949. Glacial erosion versus tectonics in the eastern Murzuq Basin
tb01258.x. (southern Libya). International Journal of Earth Sciences, 102(2),
Chen, G.-S., Liu, Z., & Kutzbach, J. E. (2014). Reexamining the barrier 455–482.
effect of the Tibetan Plateau on the South Asian Summer Monsoon. Goddéris, Y., Donnadieu, Y., Carretier, S., Aretz, M., Dera, G.,
Climate of the Past, 10(3),1269–1275. https://doi.org/10.5194/cp- Macouin, M., & Regard, V. (2017). Onset and ending of the late
10-1269-2014. Palaeozoic ice age triggered by tectonically paced rock weathering.
Cogné, J. P., & Humler, E. (2006). Trends and rhythms in global Nature Geoscience, 10(5): 382–386. https://doi.org/10.1038/
seafloor generation rate. Geochemistry, Geophysics, Geosystems, 7, ngeo2931.
Q03011. https://doi.org/10.1029/2005GC001148. Granot, R., & Dyment, J. (2015). The Cretaceous opening of the South
Copper, P., & Scotese, C. R. (2003). Megareefs in Middle Devonain Atlantic Ocean. Earth and Planetary Science Letters, 414(mars),
super greenhouse climates. Special Publications - Geological 156–163. https://doi.org/10.1016/j.epsl.2015.01.015.
Society of America, 370, 209–230. Haq, B. U., et al. (1987). Chronology of fluctuating sea levels since the
Daëron, M., Drysdale, R. N., Peral, M., Huyghe, D., Blamart, D., Triassic (250 million years ago to present). Science, 235,
Coplen, T. B., et al. (2019). Most Earth-surface calcites precipitate 1156–1166.
out of isotopic equilibrium. Nature Communications, 10(1), 429. Haqq-Misra, J. D., et al. (2008). A revised, Hazy methane greenhouse
https://doi.org/10.1038/s41467-019-08336-5. for the Archean Earth. Astrobiology, 8, 1127–1137.
De Wit, M. J., & Furnes, H. (2016). 3.5-Ga hydrothermal fields and Haug, G. H., & Tiedemann, R. (1998). Effect of the formation of the
diamictites in the Barberton Greenstone Belt—Paleoarchean crust in Isthmus of Panama on Atlantic Ocean Thermohaline circulation.
cold environments. Science Advances, 2, e1500368. https://doi.org/ Nature, 393, 673–676.
10.1126/sciadv.1500368. Hoareau, G., Bomou, B., van Hinsbergen, D. J. J., Carry, N., Marquer,
Dodd, M. S., Papineau, D., Grenne, T., Slack, J. F., Rittner, M., Pirajno, D. et al. (2015). Did high Neo-Tethys subduction rates contribute to
F., et al. (2017). Evidence for early life in Earth’s oldest early Cenozoic warming? Climate of the Past, 11(12), 1751–1767.
hydrothermal vent precipitates. Nature, 543(7643), 60–64. https:// Hoffman, et al. (2017). Snowball Earth climate dynamics and
doi.org/10.1038/nature21377. Cryogenian geology-geobiology. Science Advances, 3, e1600983.
Donnadieu, Y., Goddéris, Y., Ramstein, G., Nédélec, A., Meert, Hren, M. T., Sheldon, N. D., Grimes, S. T., Collinson, M. E, Hooker,
J. (2004). A ‘snowball Earth’climate triggered by continental J. J., Bugler, M., Lohmann, K. C. (2013). Terrestrial cooling in
break-up through changes in runoff.Nature, 428(6980), 303–306. Northern Europe during the Eocene–Oligocene transition, PNAS,
Donnadieu, Y., Pucéat, E., Moiroud, M., Guillocheau, F., & Deconinc,. 110, 7562–7567.
J. F. (2016). A better-ventilated ocean triggered by late Cretaceous Hu, S., & Boos, W. R. (2017). Competing effects of surface Albedo and
changes in continental configuration. Nature Communications, 7 Orographic elevated heating on regional climate: Albedo-Elevation
(janvier), 10316. https://doi.org/10.1038/ncomms10316. compensation. Geophysical Research Letters, 4(13), 6966–6973.
Dromart, G., Garcia, J.-P., Picard, S., Atrops, F., Lécuyer, C., & https://doi.org/10.1002/2016GL072441.
Sheppard, S. M. F. (2003). Ice Age at the Middle–Late Jurassic Jaramillo, C. (2018). Evolution of the Isthmus of Panama: Biological,
Transition? Earth and Planetary Science Letters, 213(3–4), 205– Paleoceanographic and Paleoclimatological implications. In C.
220. https://doi.org/10.1016/S0012-821X(03)00287-5. Hoorn, A. Perrigo, & A. Antonelli (Eds.), Mountains, climate and
Dubiel, R. F., Parrish, J. T., Parrish, J. M., & Good, S. C. (1991). The biodiversity (1st ed.). Wiley.
Pangaean megamonsoon: Evidence from the Upper Triassic Chinle Jung, G., Prange, M., & Schulz, M. (2014). Uplift of Africa as a
Formation, Colorado Plateau. Palaios, 6, 347–370. potential cause for Neogene Intensification of the Benguela
Eiler, J. M. (2007). “Clumped-isotope” geochemistry—the study of upwelling system. Nature Geoscience, 7(10), 741–747. https://doi.
naturally-occurring, multiply-substituted isotopologues. Earth and org/10.1038/ngeo2249.
Planetary Science Letters, 262, 309–327. Kasting, J., & Howard, M. T. (2006). Atmospheric composition and
Erwin, D. H. (1994). The Permo-Triassic extinction. Nature, 367, 231– climate on the early Earth. Philosophical Transactions of the Royal
236. https://doi.org/10.1038/367231a0. Society B, 361, 1733–1742.
Flohn, H. (1950). Neue Anschauungen über die allgemeine Zirkulation Kennett, J. P. (1977). Cenozoic evolution of Antarctic Glaciation, the
der Atmosphäre und ihre klimatische Bedeutung. Erdkunde, 4(3/4), Circum-Antarctic Ocean, and their impact on global Paleoceanog-
141–162. raphy. Journal of Geophysical Research, 82(27), 3843–3860.
Fluteau, F., Besse, J., Broutin, J., Ramstein, G. (2001). The Late https://doi.org/10.1029/JC082i027p03843.
Permian climate. What can be inferred from climate modelling Kiehl, J. T., & Shields, C. A. (2005). Climate simulation of the latest
concerning Pangea scenarios and Hercynian range altitude? Palaeo- Permian: Implications for mass extinction. Geology, 33(9), 757–
geography, Palaeoclimatology, Palaeoecology, 167, 39–71. 760. https://doi.org/10.1130/G21654.1.
Fluteau, F., Ramstein, G., Besse J. (1999). Simulating the evolution of Kutzbach, J. E., Prell, W. L., & Ruddiman, W. F. (1993). Sensitivity of
the Asian and African Monsoons during the past 30 Myr using an Eurasian climate to surface uplift of the Tibetan Plateau. The
atmospheric general circulation model. Journal of Geophysical Journal of Geology, 101(2), 177–190. https://doi.org/10.1086/
Research, 104(D10), 11995–12018. 648215.
Fluteau, F., et al. (2006). The impacts of the Paleogeography and sea Knauth, L. P., & Lowe, D. R. (2003). High Archean climatic
level changes on the Mid Cretaceous climate. Palaeogeography, temperature inferred from oxygen isotope geochemistry of cherts
Palaeoclimatology, Palaeoecology, 247(3–4), 357–381. https://doi. in the 3.5 Ga Swaziland Supergroup, South Africa. Geological
org/10.1016/j.palaeo.2006.11.016 Society of America Bulletin, 115, 566–580 (2003).
268 F. Fluteau and P. Sepulchre

Ladant, J. B., Donnadieu, Y. (2016). Palaeogeographic regulation of Geology, 31(2), 115–118. https://doi.org/10.1130/0091-7613(2003)
glacial events during the Cretaceous supergreenhous. Nature 031%3c0115:DTROTA%3e2.0.CO;2.
communications, 7, Article number: 12771. Ramstein, G., Fluteau, F., Besse, J., Joussaume, S. (1997). Effects of
Lefebvre, V., Donnadieu, Y., Sepulchre, P., Swingedouw, D., & orogeny, sea-level change and tectonic drift on the monsoon over
Zhang, Z.-S. (2012). Deciphering the role of Southern Gateways the past 30 millions years. Nature, 386, 788–795.
and carbon dioxide on the onset of the Antarctic circumpolar Rind, D. G., Russell, G., & Ruddiman, W. F. (1997). The effects of
current. Paleoceanography, 27(4), n/a–n/a. https://doi.org/10.1029/ uplift on ocean‐atmosphere circulation. In W. F. Ruddiman (Ed.),
2012PA002345. Tectonic uplift and climate change (pp. 123–147). New York:
Le Hir, G., Donnadieu, Y., Goddéris, Y., Meyer-Berthaud, B., Plenum.
Ramstein, G., & Blakey, R. C. (2011). The climate change caused Robert, F., & Chaussidon, M. (2006). A palaeotemperature curve for
by the land plant invasion in the Devonian. Earth and Planetary the Precambrian oceans based on silicon isotopes in cherts. Nature,
Science Letters, 310(3), 203–212. https://doi.org/10.1016/j.epsl. 443, 969–972.
2011.08.042. Ruddiman, W. F., et al. (1997). The uplift-climate connection: A
Lowe, D. R., & Tice, M. M. (2004). Geologic evidence for Archean synthesis. In W. F. Ruddiman (Ed.), Tectonic uplift and climate
atmospheric and climatic evolution: Fluctuating levels of CO2, CH4, change (pp. 471–515). New York: Plenum.
and O2 with an overriding tectonic control. Geology, 32, 493–496. Sepulchre, P., Arsouze, T., Donnadieu, Y., Dutay, J-C., et al. (2014).
Lunt, D. J., et al. (2008). Closure of the Panama Seaway during the Consequences of shoaling of the Central American Seaway
Pliocene: Implications for climate and Northern Hemisphere determined from modeling Nd isotopes. Paleoceanography, 29(3),
Glaciation. Climate Dynamics, 3, 1–18. 176–189.
Maffre, P., Ladant, J. B., Donnadieu, Y., Sepulchre, P., & Goddéris, Y. Sepulchre, P., Ramstein, G., Fluteau, F., Schuster, M., Tiercelin, J. J., &
(2018). The influence of Orography on modern ocean circulation. Brunet, M. (2006). Tectonic Uplift and Eastern Africa Aridification.
Climate Dynamics, 50(3–4), 1277–1289. https://doi.org/10.1007/ Science, 313(5792), 1419–1423. https://doi.org/10.1126/science.
s00382-017-3683-0. 1129158.
Matte, P. (1986). Tectonics and plate Tectonics model for the Variscan Sepulchre, P., Sloan, L. C., Snyder, M., & Fiechter, J. (2009). Impacts
Belt of Europe. Tectonophysics, 126, 329–374. of Andean Uplift on the Humboldt current system: A climate model
Meyer-Berthaud, B., Scheckler, S., Wendt, J. (1999). Archaeopteris is sensitivity study. Paleoceanography, 24(4). https://doi.org/10.1029/
the earliest known modern tree. Nature, 398, 700–701. 2008PA001668.
Molnar, P., Boos, W. R., & Battisti, D. S. (2010). Orographic controls on Shields, C. A., & Kiehl, J. T. (2018). Monsoonal precipitation in the
climate and Paleoclimate of Asia: Thermal and mechanical roles for the Paleo-Tethys warm pool during the latest Permian. Palaeogeogra-
Tibetan Plateau. Annual Review of Earth and Planetary Sciences, 38 phy, Palaeoclimatology, Palaeoecology, 491(février), 123–136.
(1), 77–102. https://doi.org/10.1146/annurev-earth-040809-152456. https://doi.org/10.1016/j.palaeo.2017.12.001.
Montañez, I., & Poulsen, C. (2013). The late Paleozoic ice age: An Sijp, W. P., von der Heydt, A. S. Dijkstra, H. A. Flögel, S., Douglas,
evolving paradigm. Annual Review of Earth and Planetary P. M. J., & Bijl. P. K. (2014). The role of ocean gateways on
Sciences, 41, 24.1–24.28. cooling climate on long time scales. Global and Planetary Change,
Montañez, I. P., McElwain, J. C., Poulsen, C. J., White, J. D., 119(août), 1–22. https://doi.org/10.1016/j.gloplacha.2014.04.004.
DiMichele, W. A., Wilson, J. P., Griggs, G., & Hren, M. T. (2016). Su, B., Jiang, D., Zhang, R., Sepulchre, P., & Ramstein, G. (2018).
Climate, pCO2 and terrestrial carbon cycle linkages during late Difference between the North Atlantic and Pacific Meridional
Palaeozoic glacial–interglacial cycles. Nature Geoscience, 9(11), overturning circulation in response to the uplift of the Tibetan
824–828. https://doi.org/10.1038/ngeo2822. Plateau. Climate of the Past, 14(6), 751–762. https://doi.org/10.
Montes, C., Cardona, A., Jaramillo, C., Pardo, A., Silva, J. C., 5194/cp-14-751-2018.
Valencia, V., et al. (2015). Middle Miocene closure of the Central Tan, N., Ramstein, G., Dumas, C., Contoux, C., Ladant, J. B.,
American Seaway. Science, 348(6231), 226–229. https://doi.org/10. Sepulchre, P., Zhang, Z., & De Schepper, S. (2017). Exploring the
1126/science.aaa2815. MIS M2 glaciation occurring during a warm and high atmospheric
Murphy, D. P., & Thomas, D. J. (2013). The evolution of late CO2 Pliocene background climate. Earth and Planetary Science
Cretaceous deep-ocean circulation in the Atlantic Basins: Neody- Letters, 472, 266–276.
mium isotope evidence from South Atlantic Drill Sites for Tectonic Tartèse, R., Chaussidon, M., Gurenko, A., Delarue, F., & Robert, F.
controls: Cretaceous Atlantic deep circulation. Geochemistry, (2017). Warm Archaean Oceans reconstructed from oxygen isotope
Geophysics, Geosystems, 14(12), 5323–5340. https://doi.org/10. composition of early-life Remnants. Geochemical Perspectives
1002/2013GC004889. Letters, 55–65. https://doi.org/10.7185/geochemlet.1706.
O’Dea, A., Lessios, H. A., Coates, A. G., Eytan, R. I., Valley, J. W. (2006). Early Earth. Elements 2(4), 201–204. https://doi-
Restrepo-Moreno, S. A., Cione, A. L., et al. (2016). Formation of org.insu.bib.cnrs.fr/10.2113/gselements.2.4.201.
the Isthmus of Panama. Science Advances, 2(8), e1600883. https:// Veizer, J., & Prokoph, (2015). Temperatures and oxygen isotopic
doi.org/10.1126/sciadv.1600883. composition of Phanerozoic oceans. Earth-Science Reviews, 146,
Potter, S. F., Dawson, E. J., & Frierson, D. M. W. (2017). Southern 92–104. https://doi.org/10.1016/j.earscirev.2015.03.008.
African Orography impacts on low clouds and the Atlantic ITCZ in Warner, T. (2004). Desert Meteorology. Cambridge: Cambridge
a coupled model: African Orography impacts on the ITCZ. University Press. https://doi.org/10.1017/CBO9780511535789.
Geophysical Research Letters, 44(7), 3283–389. https://doi.org/ Young, G. M., Von Brunn, V., Gold, D. J. C., & Minter, W. E. L.
10.1002/2017GL073098. (1998). Earth’s oldest reported Glaciation: Physical and chemical
Poulsen, C. J., Ehlers, T. A., & Insel, N. (2010). Onset of convective evidence from the Archean Mozaan Group (2.9 Ga) of South
rainfall during gradual late Miocene rise of the Central Andes. Africa. Journal of Geology, 106, 523–538.
Science, 328(5977): 490–493. https://doi.org/10.1126/science. Zachos, J., et al. (2001). Trends, rhythms, and aberrations in global
1185078. climate 65 Ma to present. Science, 292, 686–693.
Poulsen, C. J., Gendaszek, A. S., & Jacob, R. L. (2003). Did the rifting Zerkle, A. L., Claire, M. W., Domagal-Goldman, S. D., Farquhar J., &
of the Atlantic Ocean Cause the Cretaceous thermal maximum? Poulton, S. W. (2012). A bistable organic-rich atmosphere on the
22 Climate Evolution on the Geological Timescale and the Role … 269

Neoarchaean Earth. Nature Geoscience, 5, 359–363. https://doi.org/ climate. Palaeogeography, Palaeoclimatology, Palaeoecology, 417
10.1038/ngeo1425. (janvier), 137–150. https://doi.org/10.1016/j.palaeo.2014.10.030.
Zhang, R., Jiang, D., Ramstein, G., Zhang, Z., Lippert, P. C. & Yu, E. Zhang, X., Prange, M., Steph, S., Butzin, M., Krebs, U., Lunt, D. J.,
(2018). Changes in Tibetan Plateau Latitude as an Important factor et al. (2012). Changes in equatorial Pacific thermocline depth in
for understanding East Asian climate since the Eocene: A modeling response to Panamanian seaway closure: Insights from a
study. Earth and Planetary Science Letters, 484(février), 295–308. multi-model study. Earth and Planetary Science Letters, 317–
https://doi.org/10.1016/j.epsl.2017.12.034. 318, 76–84.
Zhang, R., Jiang, D., Zhang, Z., & Yu, E. (2015). The impact of
regional uplift of the Tibetan Plateau on the Asian Monsoon
 Biogeochemical Cycles and Aerosols Over
the Last Million Years 23
Nathaelle Bouttes, Laurent Bopp, Samuel Albani, Gilles Ramstein,
Tristan Vadsaria, and Emilie Capron

elements and particles, they sometimes record the changes

Introduction
that modified their cycle. It is possible to measure many of
these tracers, which provide a valuable insight into past
The biogeochemical cycles encompass the exchange of
climate changes.
chemical elements between reservoirs such as the atmo-
The link between the composition of the atmosphere and
sphere, ocean, land and lithosphere. Those exchanges
climate was discovered in the nineteenth century through the
involve biological, geological and chemical processes, hence
work of Jean-Baptiste Fourier, a French mathematician. He
the term “biogeochemical cycles”. A widely known cycle
showed that the Earth would be much colder than it currently
(which is not a biogeochemical cycle) is the water cycle,
is, if it was heated by incoming solar radiation alone. To
whose impact on climate is of major importance and which
explain the additional heating, he proposed, among other
has been described in Volume 1, Chap. 1. Similarly,
possibilities, that the Earth is insulated by gases present in
chemical elements such as carbon, nitrogen, oxygen and
the atmosphere. This effect of atmospheric gases blocking
sulphur are also exchanged during cycles. Some of these
some of the infrared radiation emitted by the planet is now
elements can significantly impact climate through their effect
known as the greenhouse effect. Later, in 1860, John Tyn-
on the atmospheric energy budget when they are in gaseous
dall, an English chemist, demonstrated that the two main
form (CO2, CH4, N2O). The biogeochemical cycles are also
constituents of our atmosphere, dinitrogen and dioxygen, are
affected by changes in climate, constituting a feedback in the
transparent to infrared radiation and therefore do not con-
Earth system. Other chemical compounds present in the
tribute to the greenhouse effect. On the other hand, he
atmosphere also influence the amount of energy available at
identified water vapour (H2O) and carbon dioxide (CO2) as
the surface and therefore the dynamics of the climate: these
the two main greenhouse gases in our atmosphere. This led
are aerosols, small liquid or solid particles in suspension in
the way for another chemist, Svante Arrhenius, who, in
the air. Because of the effect of climate on these chemical
1896, was the first to estimate the change in the average
temperature of the Earth’s surface triggered by a change in
N. Bouttes (&)
L. Bopp
S. Albani
G. Ramstein
T. Vadsaria
Laboratoire des Sciences du Climat et de l’Environnement,
the concentration of CO2 in the atmosphere. Lastly, it was
LSCE/IPSL, CEA-CNRS-UVSQ, Université Paris-Saclay, only in the second half of the twentieth century that
91191 Gif-sur-Yvette, France advances in measurement techniques made it possible to
e-mail: [email protected] measure the impacts of other gases such as methane (CH4),
S. Albani nitrous oxide (N2O), ozone (O3), and chlorofluorocarbons
Department of Environmental and Earth Sciences, (CFCs), on the greenhouse effect.
University of Milano-Bicocca, Milano, Italy
The direct effect of aerosols on climate was first sug-
T. Vadsaria gested by Benjamin Franklin in the eighteenth century to
Atmosphere and Ocean Research Institute, The
University of Tokyo, Kashiwa, Chiba, Japan
explain the cold winter of 1783–84. Benjamin Franklin
noted there was a “constant fog over all Europe, and a great
E. Capron
British Antarctic Survey, High Cross, Madingley Road,
part of North America” which resulted in colder conditions.
Cambridge, CB3 OET, UK He suggested volcanic eruptions in Iceland as a possible
E. Capron
explanation for this fog. Since then, the big volcanic erup-
Physics of Ice, Climate and Earth, Niels Bohr Institute, tions that have occurred in the nineteenth and twentieth
University of Copenhagen, Juliannes Maries Vej 30, centuries, such as the Krakatoa (1883), Santa Maria (1902),
2100 Copenhagen Ø, Denmark

© Springer Nature Switzerland AG 2021 271

G. Ramstein et al. (eds.), Paleoclimatology, Frontiers in Earth Sciences,
https://doi.org/10.1007/978-3-030-24982-3_23
272 N. Bouttes et al.

Katmai (1912), Agung (1963), El Chichon (1982) and track such changes are sediment cores extracted from oceans
Pinatubo (1991) have provided material to better understand or lakes, and ice cores, drilled from polar ice sheets
the effects of aerosols that have recently been incorporated (Fig. 23.1).
into climate models. The indirect effects of aerosols by Marine sediment cores, collected from the bottom of the
modifying clouds were discovered more recently. The ocean, include various organic and inorganic elements,
presence of aerosols can modify cloud characteristics by which can be used to get direct information. For example,
making them more reflective, or by extending their lifetime examining the type of plankton that lived in a region at a
before precipitation, for example. given time can tell us how cold it was. A succession of
These chemical elements and aerosols are closely linked species that thrive in warm or cold environments will indi-
to climate and climate changes. Their concentration in the cate a succession of warm and cold periods. An analysis of
atmosphere impacts the Earth’s energy budget, either the pollen present in sediment will yield information on the
directly or indirectly, while changes in climate modify the proportion of the major plants that lived on the nearby
exchanges between reservoirs of these compounds and, in continent. In addition, the material is measured to obtain the
fine, their concentration in the atmosphere. Some atmo- ratio of chemical elements such as Pa/Th, and the ratio of
spheric gases have the ability to modify the energy budget of isotopes such as oxygen and carbon isotopes (d18O, d13C,
the Earth (Volume 1, Chap. 1). The Earth receives short- D14C), which can be used as indicators of specific processes
wave radiation (ultraviolet, visible, and near-infrared) from such as ocean circulation changes, temperature changes,
the sun. Part of this radiation is reflected back by the surface, terrestrial biosphere changes, etc. (see Volume 1, Chap. 21).
clouds and the atmosphere, part of it is absorbed by the The idea that ice from ice sheets could be used to provide
atmosphere and clouds, and the last part is absorbed by the information on past changes originated in the 1950s with the
surface of the Earth. The Earth’s surface emits longwave work of Willi Dansgaard and others, who hypothesized that
radiation (infrared) because it is colder than the sun, and also the link between temperature and the number of heavy
transfers energy to the atmosphere by latent and sensible oxygen isotopes in precipitation could be applied to old ice
heat. The longwave radiation from the Earth’s surface is to reconstruct past temperature changes. Ice core drilling
partly absorbed by the atmospheric greenhouse gases, which began in Antarctica, Alaska and Greenland in the 1950s, but
then re-emit radiation in all directions, including towards the these cores were around 100 m deep and the recovery
surface of the Earth. The latter is then heated, resulting in a quality was low. Drilling to extract ice cores was spurred by
higher temperature than on an Earth without greenhouse the International Geophysical Year (1957–1958) and longer
gases. The radiation absorbed by the gases depends on their ice cores were drilled in Greenland at Site 2 (1956–1957)
properties and in which zones they absorb radiation. Among and in Antarctica at Byrd station (1957–1958) and at Little
the gases present in the atmosphere, the main greenhouse America V (1958–1959). Many more ice cores have been
gases are, in decreasing order (excluding water vapor) car- drilled since then, mainly in Greenland and Antarctica. Past
bon dioxide (CO2), methane (CH4) and nitrous oxide (N2O). climate and environmental changes are recorded both in the
Aerosols have two effects on the energy budget in the ice and in air bubbles trapped within the ice of the ice cores.
atmosphere: direct and indirect. Aerosols are tiny particles— For instance, the proportion of hydrogen and oxygen iso-
such as sea salt, dust from deserts and fires—in suspension topes in the ice provide information on past surface tem-
in the atmosphere, either in liquid or solid form. They can peratures, while the concentration of greenhouse gases can
absorb and disperse solar radiation, as well as absorb and be directly measured in the air bubbles.
emit thermal radiation. This is the direct effect. Aerosols also To understand the changes recorded in climate archives
form cloud condensation nuclei and ice nuclei: raindrops and and to test various hypotheses on feedback mechanisms,
ice develop around these nuclei. This is the indirect effect more and more climate models now include biogeochemical
because it modifies the energy budget through the modifi- cycles, and sometimes isotopes (see Volume 2, Chaps. 25
cation of the microphysics of clouds. Depending on the size and 29). Additional mechanisms and elements are added
of the drops, which is dependent on the type and size of progressively so that simulations and measured data can be
nuclei, the clouds will reflect or absorb radiation. In addition, compared directly. This continuous comparison helps to
aerosols deposited back to the surface will alter the amount increase our knowledge of the climate system resulting in
of solar radiation reflected back to space, and will disperse improved models that can be used to evaluate possible future
chemical elements that can influence various biogeochemi- changes. These coupled carbon-climate models are valuable
cal cycles. tools to help understand past changes and increase our
Some of the changes impacting on the biogeochemical confidence in future climate projections.
cycles and aerosols are recorded and preserved for thousands In this chapter we describe the main biogeochemical
of years. The main natural archives that have been used to cycles interacting with the climate: carbon (CO2 and CH4),
23 Biogeochemical Cycles and Aerosols Over the Last Million Years 273

Ice core
with air bubbles

Sediment core
with marine organism shells

Fig. 23.1 Diagram of the Earth system and schemes showing ice cores and sediment cores extraction

nitrogen, oxygen, sulphur, as well as aerosols. We review esrl.noaa.gov/gmd/ccgg/trends/monthly.html). At the begin-

existing records from marine sediment cores and ice cores ning of the pre-industrial era, it was around 280 ppm, cor-
and highlight the current knowns and unknowns. responding to a carbon stock of around 600 GtC.
On land, atmospheric CO2 is taken up by plants during
photosynthesis, turning CO2 into organic carbon. The living
Carbon Cycle biomass is then converted into dead organic carbon matter in
litter and soils. Organic carbon is progressively returned to
Two atmospheric gases containing carbon play a major role the atmosphere by autotrophic and heterotrophic respiration.
in the interactions between climate and biogeochemistry: Currently, carbon storage in the terrestrial biosphere is
carbon dioxide (CO2) and methane (CH4). While CO2 is the around 2500 GtC. If the conditions become cold enough, the
most famous and most studied element, CH4 is far from soil freezes, locking the carbon into permafrost, i.e. frozen
negligible and has a rather different cycle. We will first focus soil. During warming, the permafrost thaws, and carbon is
on CO2, then on CH4, before reviewing past changes. returned to the atmosphere. The current estimate of carbon
stored in permafrost is *1700 GtC (Tarnocai et al. 2009),
making it the single largest component of the terrestrial
Carbon Dioxide (CO2) carbon pool. The fluxes of carbon between the atmosphere
and land are around 120 GtC/year.
On the timescale of a few hundred to a few tens of thousands The ocean is the largest carbon reservoir (excluding the
of years, the main carbon reservoirs are the ocean (including lithosphere) with around 38,000 GtC both in organic and
sediments), atmosphere and land (including terrestrial bio- inorganic forms. Carbon fluxes to the ocean come from the
sphere and permafrost) (Fig. 23.2). On the longer timescale atmosphere through surface exchanges and from the conti-
of a few million years the lithosphere (rocks) starts to play a nent from riverine inputs. In the ocean, dissolved carbon
major role (see Volume 1, Chap. 2, Volume 2, Chaps. 22 (CO2(aq)) gets hydrated into H2CO3 (carbonic acid), which
and 26). Exchanges between these reservoirs involve biotic then gives bicarbonate ion (or hydrogen carbonate ion,
processes, due to biological activity such as photosynthesis HCO3−) and carbonate ion CO32−. All these dissolved spe-
by plants, and abiotic processes. cies are summed up in the term “dissolved inorganic carbon”
In the atmosphere, CO2 is the main form of carbon and is (DIC). Because the concentration of H2CO3 is very small, it
relatively well mixed over a year. Its concentration is cur- is included in CO2(aq). CO2(aq), HCO3− and CO32− are in
rently increasing due to anthropogenic activity and, at the equilibrium following the chemical equations:
beginning of 2018, the atmospheric CO2 concentration was
over 400 ppm (408 in March 2018, NOAA, https://www. CO2ðaqÞ þ H2 O  HCO
3 þH
þ
274 N. Bouttes et al.

Atmosphere : 600 GtC Terrestrial

biosphere :
2 500 GtC

Permafrost :
1700 GtC
Ocean : 38 000 GtC

Surface
sediments :
1750 GtC

Fig. 23.2 Diagram of the short term carbon cycle (up to a few tens of thousands years)

HCO 2 þ depth to which sufficient light penetrates to sustain life is

3  CO3 þ H
called the euphotic zone and extends to around 100–200 m
CO2 is exchanged between the atmosphere and the sur- deep. When plankton dies, it is partially remineralized at the
face ocean depending on the difference in partial pressure of surface, and the remainder sinks to the deeper ocean where it
CO2 (pCO2) between the atmosphere and ocean, on sea ice will be progressively remineralized. Carbon can then be
coverage and on wind. The pCO2 in the ocean is governed found in two forms: particulate organic carbon or dissolved
by temperature and salinity. The flux of carbon between organic carbon. During remineralisation, carbon and nutri-
atmosphere and dissolved inorganic carbon is around 70 ents are returned to the solution in dissolved form and can be
GtC/year in both directions. used again in the euphotic zone for photosynthesis when it is
In polar regions, CO2 is more soluble in water because of brought back to the surface. Hence, biology also transfers
the colder conditions, resulting in more dissolved inorganic carbon from the surface to the deep ocean leading to carbon
carbon in the surface ocean. These regions are also major being taken up from the atmosphere by the ocean: this is the
sites of deep convection, where surface water becomes den- biological pump. The flux of carbon between dissolved
ser due to colder and more saline conditions, and can sink to inorganic carbon and marine biology is around 50 GtC/year.
the ocean’s depths. The carbon from the surface is thus Marine biology not only produces organic carbon, but
transported to the deeper ocean. This uptake of carbon from some organisms also create a shell made of calcium car-
the atmosphere to the ocean is called the solubility pump. bonate (CaCO3). Most shells are made of calcite (coccol-
Marine phytoplankton uses dissolved CO2 during pho- ithophores and foraminifera) or aragonite (pteropods), two
tosynthesis, using nutrients and solar energy, following the forms of calcium carbonate. The equation for calcium car-
simplified equation: bonate production is:

106 CO2 þ 16 NO  þ

3 þ H2 PO4 þ 17 H3 O þ 105 H2 O Ca2 þ þ CO2
3 ! CaCO3
þ solar energy ! ðCH2 OÞ106 ðNH3 Þ16 ðH3 PO4 Þ þ 138 O2
In deep areas in the ocean where water is undersaturated
The ratios C: N: P: O2 are the Redfield ratios, and the with respect to calcium carbonate, the inverse equation takes
inorganic carbon assimilated by biology constitutes the gross place and the shells are progressively dissolved. If they reach
primary production. The difference between gross primary the bottom of the ocean, they are buried in sediments. At the
production and the carbon respired by phytoplankton is the ocean surface, the production of calcium carbonate shells
net primary production. The phytoplankton that has syn- leads to a decrease in CO23 , which increases CO2 due to the
thetized organic carbon is then grazed by zooplankton. The following equation:
23 Biogeochemical Cycles and Aerosols Over the Last Million Years 275

OH. Cl.
Atmosphere Biomass
burning

Permafrost
hydrates
Vegetation
Ocean hydrates Wetlands

Animals
(ruminants,
termites...)

Fig. 23.3 Diagram of the methane cycle

CO2 þ CO2 
3 þ H2 O  2HCO3
the concentration of CH4 is 20 times lower than CO2, it still
plays a crucial role as a greenhouse gas, with a radiative
Hence this process, called the carbonate pump, counter- forcing of around one third that of CO2. The natural sinks
acts the two carbon pumps described above. and sources of CH4 are different from those of CO2, yielding
Finally, ocean circulation impacts on the carbon cycle in a distinctly different—although often with some common
the ocean by modifying the distribution of carbon and features—evolution over time as described in Sect. “Gla-
nutrients, the latter modifying primary production and thus cial-Interglacial Cycles”.
the carbon distribution. Primary production is limited by In the atmosphere, CH4, like CO2, is well mixed over a
several nutrients such as nitrate (N), phosphorus (P) and iron year. However, while CO2 stays around 100 years in the
(Fe). The concentration of these nutrients mainly depends on atmosphere, CH4 has a shorter lifetime of around 9 years.
transport from the nutrient-rich deeper ocean layers to the This is because the main sink of CH4 is in the atmosphere:
surface which is nutrient-depleted due to nutrient absorption CH4 is oxidized by the hydroxyl radical, OH. Oxidation by
by marine biology. Hence, in upwelling zones where lots of OH, which is photochemically produced in the atmosphere,
nutrients are brought to the surface primary production is takes place mainly in the troposphere, but also, to a lesser
high. When deep convection is active, it also provides extent, in the stratosphere, and depends on several parame-
important nutrient transport to the surface favouring ters. First, it depends on the speed of the reaction with OH,
production. In the opposite, in low latitude gyres nutrients hence on temperature. Second, it depends on the quantity of
are lacking and primary production is low. Some regions in free OH, which itself, depends on other compounds reacting
the North Pacific and in the Southern Ocean display low with OH such as volatile organic compounds (VOCs) and
productivity despite high nitrate concentrations (HNLC ozone. In addition, the reaction of CH4 with OH produces
regions for high nitrate low chlorophyll). This is due to the the CH4 feedback effect: if CH4 decreases, OH increases,
lack of iron which limits production. which in turn reduces even more the concentration of CH4
(Prather 2007). Another smaller sink of CH4 in the atmo-
sphere is the reaction of CH4 with chlorine gas.
Methane (CH4) Contrary to CO2, the main natural sources of CH4 are
from the continents (Fig. 23.3). The main contributors are
Methane is the second most important greenhouse gas after wetlands, areas saturated with water such as marshes and
CO2. Discovered in 1976 by Wang et al., its capacity to swamps. CH4 is produced by microbes (methanogenic
absorb infrared radiation is around 28 times more efficient archaea) in anoxic conditions in wetlands. Locally, CH4
than CO2, over a time period of 100 years. Hence, although production strongly depends on oxygen availability,
276 N. Bouttes et al.

temperature and the composition of the soil. Currently, Glacial-Interglacial Cycles

around three quarters of CH4 production takes place in
tropical regions and a quarter in boreal regions. Animals also The main archives used to infer past changes in the carbon
produce CH4, in particular termites and ruminant animals. In cycle are ice cores and marine sediment cores (see
addition, vegetation is also involved in CH4 emission. Volume 1). While ice cores provide an invaluable direct way
Vegetation as a CH4 producer is still debated, but it can of assessing past changes in atmospheric CO2 and CH4,
impact on the transport of CH4 from the soil to the atmo- there is no direct record of past ocean and land carbon
sphere. Permafrost can also be a source of CH4 but related storage. Fortunately, the carbon element has three isotopes:
12
uncertainties are high. Finally, CH4 is also emitted by bio- C, 13C and 14C, and their ratios, which can be measured
mass burning. There is also a small continental sink of from foraminifera shells in the sediments, as well as in the
atmospheric CH4: the oxidation by methanotrophic bacteria air in ice cores, provide clues to understand changes in past
in soils. carbon reservoirs.
A smaller CH4 source is from the ocean, with CH4
coming mainly from coastal regions. However, the ocean Ice Cores
could contain large CH4 quantities trapped as methane Ice cores recording glacial-interglacial changes are extracted
clathrate in sediments, mainly on continental shelves. from the Greenland ice sheet in the Northern Hemisphere,
Methane clathrate, or methane hydrate, is a compound in and the Antarctic ice sheet in the Southern Hemisphere
which methane is trapped in a crystal of water. Clathrates are (Volume 1, Chap. 9). The oldest ice obtained from Green-
stable at low temperature and high pressure. The size of this land is *130,000 years old (NEEM Community Members
reservoir is poorly constrained and could contain 500–2500 2013), much younger than the oldest ice from Antarctica
GtC, a smaller amount than the very high quantities sug- which is *800,000 years old (EPICA Community Mem-
gested in the 1970s (Milkov 2004). bers 2004). This age could be extended further and future
While the atmospheric CH4 mixing ratio impacts on cli- expeditions are planned to find and drill Antarctic ice older
mate by modifying the atmospheric radiative balance, cli- than a million years (Dahl-Jensen 2018).
mate impacts on the sources and sinks of methane. In Ice cores provide a direct record of past atmospheric gas
particular, wetlands are very dependent on the hydrological concentrations, such as CO2 and CH4, thanks to air bubbles
cycle and microbial activity, and thus the emission of trapped in ice (Fig. 23.4). The air is trapped only at the
methane, is strongly dependent on temperature. In addition, bottom of the firn, a 60–120 m permeable layer below the
in the atmosphere, methane oxidation by OH radicals is one surface where snow progressively densifies into ice (see
of the most temperature-sensitive reactions. Volume 1, Chap. 9). It results in an age difference between

Fig. 23.4 Air bubbles trapped in

ice taken from an ice core. Credit
Sepp Kipfstuhl (Alfred Wegener
Institute)
23 Biogeochemical Cycles and Aerosols Over the Last Million Years 277

LGM

Interglacials

glacials

Fig. 23.5 Evolution of CO2 (ppm), CH4 (ppb), deuterium (‰) and Community Members (2004), Jouzel et al. (2007), Loulergue et al.
temperature (°C) over the last 800,000 years. The Last Glacial (2008), Bereiter et al. (2015)
Maximum (LGM) is indicated by the blue bar. Data from EPICA

the ice and the trapped air at a given depth level. This age measurements of past CO2 changes recorded in Antarctic ice
difference that may be as much as several hundred years and were published in the 1980s, initially covering the last
even up to several thousands of years in the case of central 30,000 years (Delmas et al. 1980), then the last
Antarctic sites (Schwander and Stauffer 1984). Estimates of 160,000 years (Barnola et al. 1987), and now as much as the
the age difference between the ice and the entrapped air are last 800,000 years (Lüthi et al. 2008; Bereiter et al. 2015,
possible but there are associated uncertainties which are Fig. 23.5). Note that only past atmospheric CO2 concentra-
detailed in Volume 1, Chap. 8. The first reliable tions determined from Antarctic ice cores are reliable, as
278 N. Bouttes et al.

Greenland ice is prone to in situ production of CO2, which

alters the atmospheric CO2 concentration already within the
13
C
ice (Tschumi and Stauffer 2000). Measurements of past CH4
12
13
C 13
C high δ13C
concentrations extracted from both Greenland and Antarctic C 12
ice cores are reliable. While the first CH4 records covered the C 12
C
last 160,000 years (Chappellaz et al. 1990), they also now
date back to 800,000 years (Loulergue et al. 2008). In
addition to greenhouse gas concentrations, measurements of
the air from ice cores now also involve d13C from CO2,
termed d13CO2 (Schneider et al. 2013; Schmitt et al. 2012;
Lourantou et al. 2010). Ice cores offer a unique possibility to 13
C 13
C
quantify the sequence of events occurring between green-
C low δ C
12 13
house gas variations and changes in ice core tracers inform C 12
on other parts of the climate system such as local surface
temperature using ice d18O (see Volume 1, Chap. 11).

Sediment Cores Fig. 23.6 Diagram of d13C in the surface and deep ocean. In the
At the bottom of oceans and lakes, sediments progressively surface waters, the value of oceanic d13C is high because of
accumulate as various particles and debris are deposited. photosynthetic activity, which preferentially uses light 12C, thus
enriching the environment with heavy 13C. In the deep ocean,
They include both organic material, such as shells, and remineralisation releases carbon with more 12C, therefore lowering d13C
inorganic material, such as clay. The invention of the first
piston corer in 1947 by Kullenberg allowed marine sediment
cores to be extracted from the ocean bottom, yielding a (Fig. 23.6). A more active biological productivity will thus
wealth of information on past ocean changes (Volume 1, increase the d13C in the surrounding environment.
Chap. 21). In particular, sediments provide information on Sediments on continents can also be extracted and stud-
past marine productivity, for example by measuring the ied, but continental sediment cores are scarcer than their
fraction of organic material, calcite or opal (biogenic silica). marine counterpart. Pollen in these sediments, for example
The proportion of organic material measured in upwelling in loess, give information on past local plant types and help
zones is large, because productivity is high. In regions where to reconstruct past vegetation changes.
organic material is not well preserved, silicate can be used as
another proxy for biological productivity. Other proxies are Past Changes
also used, such as 10Be, authigenic U, 231Pa/230Th). They Ice core data show large changes in concentrations of CO2
rely on the fact that some elements preferentially fix to and CH4 between warm interglacials and cold glacials
particles (Th) while others remain in solution (U, Pa). Their (Fig. 23.5). Concentrations of both CO2 and CH4 are higher
ratio gives an indication of past particle flux in the water during interglacials: around 280 ppm for CO2 and 780 ppbv
column, hence biological productivity. for CH4, and lower during glacials: around 190 ppm for CO2
As detailed in Volume 1, Chap. 21, carbon isotopes are and 320 ppbv for CH4 (Lüthi et al. 2008; Bereiter et al.
measured in foraminifera shells providing constraints on the 2015; Loulergue et al. 2008). While values are relatively
carbon cycle. Whenever carbon is exchanged at an interface, similar during all glacials, the interglacial CO2 values are
fractionation takes place, which modifies d13C defined as: around 20 ppm lower for the older interglacials (before
0
13  1 430 ka) compared to the more recent ones. The concomitant
C records of climate and air composition demonstrate the
12 C
B C
d13 C ¼ @
13 
sample
 1A  1000 strong link between climate and greenhouse gases, both for
C CO2 (Barnola et al. 1987; Petit et al. 1999; Siegenthaler et al.
12 C
standard
2005; Lüthi et al. 2008) and CH4 (Chappellaz et al. 1990;
The standard is the PDB (Peedee belemnite) carbon iso- Petit et al. 1999; Spahni et al. 2005; Loulergue et al. 2008).
tope standard, which corresponds approximately to average Recent research has shown that CO2 and Antarctic temper-
limestone (Craig 1957). ature changed synchronously at the start of the two last
For example, biological activity preferentially uses the deglaciations (within 200 years; Parrenin et al. 2013;
light 12C over 13C, so that plants or plankton are enriched in Landais et al. 2013).
12
C, and the environment (atmosphere for terrestrial bio- Several explanations have been put forward to explain
sphere, surface ocean for plankton) has higher d13C values the atmospheric CO2 waxing and waning during
23 Biogeochemical Cycles and Aerosols Over the Last Million Years 279

glacial-interglacial cycles. Atmospheric CO2 changes are due to sea ice coverage, changes in ocean dynamics, and
driven by changes in carbon storage in other reservoirs, carbonate compensation.
particularly on land and in the ocean. The biological pump causes more carbon to be stored
On land, the lower sea level, by 120 m, during glacial when it is stimulated, for example when more nutrients are
time increased the surface area where plants could develop, delivered to the ocean (Broecker and Peng 1982). In regions
although the larger ice sheets covering part of North of high nutrients low chlorophyll (HNLC), biological
America and Eurasia reduced the available area. Overall, due activity is limited due to the lack of iron. An influx of iron to
to the colder and drier climate, as well as reduced atmo- these zones during glacial periods would increase biological
spheric CO2 concentration, photosynthesis by land vegeta- activity. Alternatively, the biological pump could also store
tion is reduced and the terrestrial biosphere tends to more carbon if it becomes more efficient, for example with a
represent a smaller carbon reservoir during glacial period, greater carbon to nutrient ratio (Broecker and Peng 1982) or
which would increase atmospheric CO2, not lower it. This is a switch of plankton species with higher productivity
indicated by ocean d13C which decreased by around 0.03– (Archer and Maier-Reimer 1994). However, both data and
0.04‰ during the Last Glacial Maximum compared to the model simulations have shown that changes in biological
pre-industrial level. This is explained by the transfer of activity are not sufficient to sufficiently account for the
continental carbon with low d13C values (due to fractiona- decrease in atmospheric CO2 (Kohfeld et al. 2005; Bopp
tion during photosynthesis) to the ocean (Shackleton 1977; et al. 2003a; Tagliabue et al. 2009; Lambert et al. 2015).
Bird et al. 1994), causing a reduction in continental carbon Increased sea ice coverage has also been proposed, as this
of around 270–720 GtC. An understanding of the changes in could isolate the ocean, preventing carbon from getting to
the terrestrial biosphere can also be obtained from the atmosphere, hence lowering atmospheric CO2 (Stephens
paleo-biomes, using pollen from sediment cores (Adams and Keeling 2000). But such an impact has only been sim-
et al. 1990; Crowley 1995; Maslin et al. 1995) which show a ulated in very simple models, more complex models do not
carbon loss of around 750–1350 GtC. Pollen data have also show such an effect on CO2 (Archer et al. 2003).
been used to reconstruct maps of vegetation types during the Most current theories involve changes in ocean dynamics,
LGM (Fig. 23.7). and point to the Southern Ocean (Fischer et al. 2010).
However, frozen soils, i.e. permafrost, may have A larger ocean volume occupied by AABW, or slower
increased in glacial times, potentially storing more carbon, overturning, could result in more carbon stored in the deep
which could partly explain the lower CO2 (Ciais et al. 2012) ocean, reducing atmospheric CO2. Ocean circulation chan-
and help resolve the d13CO2 signal recorded in ice cores, ges are supported by data generally indicating a reduced
which strongly depends on land carbon changes (Crichton NADW and a more stratified Southern Ocean (Adkins
et al. 2016). 2013). In particular, d13C measurements show lower d13C
Although permafrost probably played a role, most of the values in the deep glacial ocean, especially around Antarc-
change is likely to have come from the ocean, which is a tica, and higher values near the surface (Curry and Oppo
much bigger carbon reservoir. In addition, d13C measure- 2005; Marchal and Curry 2008; Hesse et al. 2011). In
ments from sediment cores indicate large changes in the addition, very salty water has been measured in the deep
ocean (Curry and Oppo 2005; Marchal and Curry 2008; Southern Ocean (Adkins et al. 2002). Complementary data,
Hesse et al. 2011). In the ocean, the carbon cycle changes such as from neodymium isotopes (Basak et al. 2018), B/Ca
could originate from modifications of biological activity and ratio (Yu et al. 2016) and D14C (Skinner et al. 2010), also
physical or chemical changes. Known processes include point towards changes in the circulation in the Southern
temperature change as colder temperatures lead to more Ocean.
carbon being stored in the ocean. The sea level drop Comparison of model simulations over the last decade
of *120 m during glacial maxima results in higher con- have shown that models simulate a large range of ocean
centrations of salinity, which causes a reduction in the circulation changes, which are generally opposite to those
storage capacity of carbon in the ocean, and an increase in deduced from data. In PMIP3, most models simulate a
nutrients, which increases biological activity and thus strengthening and deepening of the NADW with LGM
increases ocean carbon storage. However, these processes boundary conditions (Muglia and Schmittner 2015). Yet
are not sufficient to fully explain the decrease in CO2 simulations have shown that better agreement with d13C and
decrease (see review by Sigman and Boyle 2000; Archer CO2 data requires lower NADW intensity and/or shoaling of
et al. 2000), and additional mechanisms are needed. NADW (Tagliabue et al. 2009; Tschumi et al. 2011; Menviel
Four main hypotheses have been proposed to explain the et al. 2017). This is also seen in terms of water mass volume
CO2 lowering by increased ocean carbon storage: increased with a smaller volume of NADW and a larger volume of
biological pump, isolation of the ocean from the atmosphere AABW filling the ocean. The latter has a larger DIC content,
280 N. Bouttes et al.

Fig. 23.7 Land vegetation maps for the modern era and the LGM, nomenclature has been downloaded from http://www.bridge.bris.ac.uk/
data from Prentice et al. (2000), Harrison et al. (2001), Bigelow et al. resources/Databases/BIOMES_data
(2003) and Pickett et al. (2004). This figure with homogenised
23 Biogeochemical Cycles and Aerosols Over the Last Million Years 281

Fig. 23.8 Diagram of the

carbonate compensation input flux
mechanism Ca2++2HCO3-

CaCO3 production 1 if more CO2

CO2+CO32-+H2O > 2HCO3-
oversaturated zone: decrease of [CO32-]
preservation 4
CaCO3 New
burial équilibrium
output flux 2 less [CO32-] 3
saturation horizon -> shoaling of saturation horizon dissolution:
increase of
undersaturated zone: [CO32-]
dissolution

resulting in more carbon in the ocean when the AABW CaCO3 dissolution, which increases [CO32−], counteracting
occupies relatively more volume than the NADW (Brovkin the initial reduction and leading to a lowering of [CO2] as
et al. 2012). This is referred to as the “standing volume the previous equation is shifted to the right, allowing the
effect” (Skinner 2009). ocean to take up more CO2 from the atmosphere. (4) When
Changes of ocean circulation could be due to changes of [CO32−] is increased, the oversaturated zone increases and
winds (Anderson et al. 2009; Toggweiler et al. 2006). But the saturation horizon is shifted down until a new equilib-
data show no clear evidence of large wind changes (Kohfeld rium is reached. Overall, the ocean takes up more carbon
et al. 2013) and model simulations have shown that this is with this mechanism through dissolution of CaCO3.
unlikely to have had a strong effect on the carbon cycle The concomitant lowering of CH4 during the glacial
(Menviel et al. 2008a). It could also be linked to ocean period could be due to either a decrease in CH4 sources,
diffusion (Bouttes et al. 2009, 2011) and particularly to mainly wetlands, or an increase in sinks, mainly more oxi-
bottom topography induced diffusion (De Boer and Hogg dation by increased OH.
2014). Alternatively, it could be linked to sea ice changes During the LGM, the colder climate, larger ice sheets and
and modifications in bottom water formation (Ferrari et al. reduced hydrological cycle all led to a reduction of wetlands
2014; Bouttes et al. 2010). Indeed, sea ice formation around and reduced emissions. The first hypothesis to explain the
Antarctica was probably enhanced, especially during winter lower CH4 concentration during the LGM has thus focused
(Gersonde et al. 2005). Yet models usually fail to represent on reduced emissions, possibly from low latitude wetlands
glacial sea ice extent (Roche et al. 2012; Goosse et al. 2013; (Chappellaz et al. 1993). Later, process-based models were
Marzocchi and Jansen 2017). Improving sea ice formation in developed and used to evaluate emissions (Valdes et al.
models during the LGM and the sinking of dense water 2005; Kaplan et al. 2006; Weber et al. 2010). But the
around Antarctica should help towards explaining the glacial resulting reduction of emissions was not enough to account
atmospheric CO2 concentration. for the low CH4 concentration detected in ice cores. It was
Finally, on longer timescales of a few thousand years, then hypothesized that the oxidizing atmospheric capacity
carbonate compensation amplifies the increase of oceanic had changed, for instance through a reduction of emissions
carbon storage, by maintaining a balance between inputs and of volatile organic compounds (VOCs) from forests (Valdes
outputs of alkalinity fluxes (Fig. 23.8). For example, (1) if et al. 2005). The VOCs react with OH in the same way as
[CO2(aq)] increases (for example due to the solubility pump), CH4, thus constituting an OH sink, which increases the
the equation below is displaced towards more [HCO3−], lifetime of CH4. If more VOCs are produced, OH concen-
reducing [CO32−]. tration is reduced and the concentration of CH4 is increased.
However, more complex chemical models show that certain
CO2 þ CO2 
3 þ H2 O  2HCO3 processes compensate for the reduced atmospheric oxidizing
capacity, such as temperature, humidity, lightening (Levine
(2) When [CO32−] is reduced, the saturation horizon, the
et al. 2011; Murray et al. 2014). In conclusion, it now
limit between oversaturated and undersaturated water, is
appears that the atmospheric oxidizing capacity is probably
shifted upwards and a larger volume of water is undersatu-
of secondary importance, as process-based models and the
rated. (3) The larger undersaturated zone results in more
282 N. Bouttes et al.

methane retroaction (less CH4 leads to more OH, hence less above the zone where these ice rafted debris (IRD) were
CH4) estimate emissions that explain almost entirely the low found in sediments (Hemming 2004).
CH4 value at LGM (Quiquet et al. 2015). Antarctic ice core records show a rapid rise in atmo-
Beyond the problem of the large changes in glacial- spheric CO2 of around 15 ppm over 2000–4000 years
interglacial GHG concentrations, another issue was raised in (Fig. 23.9), generally synchronous with the millennial-scale
the 2000s when data became available for periods older Antarctic warming, followed by a more gradual decrease
than *430,000 years BP. As shown on Fig. 23.5, older than the Antarctic temperature drop (Ahn and Brook 2008;
interglacials before *430,000 ka BP (before the “Mid Bereiter et al. 2012). Measurements have shown that the
Brunhes Event”) are characterized by a colder climate than CO2 rise was not steady, but punctuated by events with a
more recent interglacials, associated to lower GHG con- rapid increase (Ahn et al. 2012). Other data measurements
centrations. The colder interglacial climate can be attributed have resulted in several carbon sources being suggested to
to different orbital configurations and lower CO2 (Yin and explain these increases in atmospheric CO2, such as the
Berger 2010, 2012) but the reason for the lower GHG Southern Ocean (Gottschalk et al. 2016) or the North
concentrations still remains to be explained (Bouttes et al. Atlantic (Ezat et al. 2017).
2018). These rapid changes of atmospheric CO2 have been
studied only relatively recently as previously, the temporal
resolution in the records was not sufficient. Temperature
Abrupt Changes changes have been studied for longer since more high res-
olution data were available. To replicate these rapid climate
At the centennial to millennial scale, climate variability is changes in simulations, modellers have found that artificially
superimposed onto the orbital-scale glacial-interglacial adding freshwater to the North Atlantic, for example theo-
cycles and is recorded by specific expressions at different retically due to the melting of numerous icebergs, could slow
latitudes and in different climate archives (Clement and down or even stop the Atlantic meridional overturning cir-
Peterson 2008; see Volume 2, Chap. 29). Indeed, Greenland culation. This then generally leads to warming in the North
ice cores unveiled a succession of events called Hemisphere and cooling in the South Hemisphere, in line
Dansgaard-Oeschger (D-O) events during the last glacial with changes observed in the data (see Volume 2, Chap. 29).
period (Dansgaard et al. 1993; North Greenland Ice Core More recently, the impact of such hosing experiments on
Project Members 2004). Typically, a DO event is depicted as the carbon cycle and atmospheric CO2 evolution has also
an abrupt warming of 5–16 °C of the mean annual surface been tested in carbon-climate models, to evaluate the role of
temperature within a few decades toward a relatively mild the terrestrial biosphere and the ocean, in particular, in
phase. This phase is then usually characterised by a gradual coupled ocean-atmosphere-terrestrial biosphere models. The
cooling over several centuries and its end is marked by a model response to the freshwater input appears to be very
rapid cooling leading to a relatively stable cold phase per- dependent on the type, duration and amplitude of the
sisting over several centuries or even up to a thousand years. freshwater input, on the background climate (glacial vs
The signature of DO events is recorded in continental and pre-industrial) and the model. For example, the LOVECLIM
marine records in the Northern Hemisphere. In the Southern model simulates a 15 ppmv increase in the context of a
Hemisphere, there are counterparts to these DO events. Ice pre-industrial climate, but a 10 ppm CO2 decrease in a
cores indicate more gradual (millennial-scale) warming in glacial climate in response to a decrease in the AMOC dri-
Antarctica during the Greenland cold phases (EPICA ven by the same freshwater input (Menviel et al. 2008b). In
Community Members 2006; Barker et al. 2009; WAIS both cases, the ocean takes up more carbon and the terrestrial
Divide Project Members 2015). The antiphase relationship biosphere loses carbon, but the balance between the two
between the two hemispheres is attributed to the thermal outcomes results in opposite effects on the atmospheric CO2.
bipolar seesaw, a mechanism whereby heat is redistributed This balance also depends on the different time reactions of
in the Atlantic Ocean (Stocker and Johnsen 2003). the carbon reservoirs: vegetation reacts more rapidly than the
In addition to the succession of Dansgaard-Oeschger ocean.
events, another prominent feature identified in marine sedi- In general, most models simulate an overall increase in
ments from the North Atlantic is the occurrence of the atmospheric CO2 ranging from a few ppm up to more than
Heinrich events. Heinrich events are identified by the pres- 20 ppm, depending on the model and the size of the fresh-
ence of debris in sediments, and were first discovered by water flux (Obata 2007; Schmittner and Galbraith 2008;
Ruddiman in 1977. These debris are too big to be trans- Menviel et al. 2008b; Bozbiyik et al. 2011; Bouttes et al.
ported by oceanic currents, and in 1988 Heinrich proposed 2012; Matsumoto and Yokoyama 2013), but the causes are
that they could have been brought by icebergs which melted different: in some models the ocean gains carbon and the
23 Biogeochemical Cycles and Aerosols Over the Last Million Years 283

Fig. 23.9 CO2 variations

relative to abrupt warming in
Greenland as presented in Ahn
and Brook (2008)

terrestrial biosphere loses carbon (Obata 2007; Menviel et al. changes (Brook et al. 2000) and its close link with Green-
2008b; Bozbiyik et al. 2011) while in others it is the opposite land temperature is classically interpreted as reflecting
with the ocean losing carbon and terrestrial biosphere changing CH4 emissions from tropical and boreal wetlands
gaining carbon (Schmittner and Galbraith 2008; Bouttes in phase with Greenland temperature (Chappellaz et al.
et al. 2012; Matsumoto and Yokoyama 2013). Alternatively, 1993). A detailed study of the abrupt Bølling warming, the
simulations with artificially increased salinity in the South- penultimate warming in the series of abrupt climate changes
ern Ocean have been tested, resulting in a strengthening of during the last glacial, suggests that changes in Greenland
the AABW and a loss of ocean carbon, yielding an overall temperatures and atmospheric CH4 emissions occurred
CO2 increase of around 20 ppm (Menviel et al. 2015). In essentially synchronously (within 20 yr; Rosen et al. 2014).
general, simulations with atmosphere-ocean-terrestrial bio- CH4 concentrations measured in Greenland ice cores are
sphere models in glacial background climate produce a large higher than those measured in Antarctic ice cores primarily
range of CO2 changes due to the different processes, which because of enhanced CH4 emissions in the Northern Hemi-
are summed up in Fig. 23.10. sphere due to its larger land area (Chappellaz et al. 1997).
In addition, rapid CO2 changes during the last deglacia- However, rapid CH4 changes are seen in both hemispheres
tion have been recently highlighted, such as the rapid CO2 and this feature is commonly used to synchronise Antarctic
rise concomitant to the warming in the Northern Hemisphere and Greenland ice core chronologies (Blunier et al. 1998;
at the Bølling Allerød around 14,600 years ago (Marcott Buizert et al. 2015).
et al. 2014). On top of changes in the Atlantic meridional
overturning circulation, permafrost thawing, releasing large
quantities of CO2 trapped in frozen soil, has been suggested Nitrogen Cycle
as a potential driver of the CO2 rise (Köhler et al. 2014).
Changes in atmospheric methane (CH4) concentration Natural Nitrogen Cycle
measured in ice cores are linked closely to the rapid surface
temperature variations in Greenland during the last glacial Nitrogen (N) interacts with climate in two ways. First, like
period. In particular, CH4 increases of 100–200 ppbv are carbon dioxide and methane, nitrous oxide N2O is a green-
associated with the abrupt DO warming events. CH4 con- house gas. It is in fact more powerful than CO2 or CH4, but
centration is a global signal that reflects the response of the its atmospheric concentration is less, currently 325 ppb
terrestrial biosphere, mainly wetlands, to hydroclimate (NOAA, http://esrl.noaa.gov/gmd/). Second, nitrogen is also
284 N. Bouttes et al.

Abrupt events in glacial climate

Land
1 Terrestrial biosphere
change
Mechanisms

2 AMOC change
Ocean

AMOC change
3
AABW change
4

Resulting atm. CO2

Data

Ice cores

-40 -30 -20 -10 0 10 20 30 40 50 60

CO2 change (ppm)
Schmittner and Galbraith, 2008
Menviel et al., 2008b
Bouttes et al., 2012
Menviel et al., 2015
DO events between 25ka and 110ka (Ahn and Brook, 2008; Bereiter et al., 2012)

Fig. 23.10 Changes of atmospheric CO2 due to different processes in are taken from the simulations with maximum CO2 change and at the
atmosphere-ocean-terrestrial biosphere model simulations with glacial time when atmospheric CO2 for ocean and land carbon changes is at a
background climate. Figure modified from Mariotti (2013). Model data peak

a limiting nutrient for the growth of vegetation on land and efficiency of N2-fixation depends on the environment, in
in the ocean, along with other nutrients such as phosphate. particular, radiation, temperature, the presence of other
Nitrogen is essential for photosynthesis, which produces nutrients (such as phosphate and iron), and O2 concentration.
organic carbon. This connects the nitrogen cycle to the Smaller marine inputs of bioavailable nitrogen include
carbon cycle and atmospheric CO2 concentration, and ulti- atmospheric nitrogen deposition and riverine inputs.
mately connects the nitrogen cycle to climate. Nitrification by soil or marine bacteria is the oxidation of
The nitrogen cycle is governed by biochemical reactions ammonia (NH3) into nitrate (NO3−). This oxidation is done
oxidising and reducing nitrogen which can be divided into in two separate steps: first the oxidation from NH3 to nitrite
three main processes: N2-fixation, nitrification and denitri- (NO2−) by ammonia-oxidizing bacteria and archaea. Then
fication (Fig. 23.11). These processes take place both in the the oxidation of NO2− into NO3− by nitrite-oxidizing bac-
ocean and on land. teria. Nitrification in the ocean takes place at the lower
Although N2 is very abundant in the atmosphere it cannot boundary of the euphotic zone where photosynthesis is
be used in this form by most organisms. N2 has to be limited by the low penetration of light, preventing the
transformed to a bioavailable—or reactive—form (ammonia assimilation of nitrate by phytoplankton, and where rem-
NH3 or ammonium NH4+) in order to be usable, a conver- ineralisation of organic matter increases nitrate concentra-
sion, which requires a lot of energy to break the strong triple tion. In addition to this aerobic oxidation of ammonia into
bond of the N2 molecule. Most N2 fixation is done by bac- nitrate, anaerobic nitrification can also take place, called
teria called diazotrophs, which have a specific enzyme called anammox (anaerobic ammonia oxidation).
nitrogenase which combines gaseous nitrogen with hydro- Denitrification is the process that reduces NO3− to N2 gas,
gen to produce ammonia. In the ocean, the main source of releasing it back to the atmosphere. It happens during res-
bioavailable nitrogen comes from N2-fixation by marine piration by anaerobic bacteria in low O2 conditions, and
diazotrophs (cyanobacteria and proteobacteria) which are removes bioavailable nitrogen from the environment. In the
mainly present in warm waters in the low latitudes. The process, intermediate gases are produced such as N2O,
23 Biogeochemical Cycles and Aerosols Over the Last Million Years 285

Fig. 23.11 Diagram of the

nitrogen cycle
N2
nitrogen

N2O

fixation by
bacteria
ammonification
assimilation anaerobic
NH4+ oxidation:
ammonium anammox NH3

denitrification
NO2- aerobic
nitrite oxidation:
nitrification

NO3-
nitrate

which is a powerful GHG. In conditions where O2 is com- Changes in the Nitrogen Cycle During Glacial
pletely depleted, N2O instead of NO3− is respired into N2. Interglacial Cycles
Ammonification is the production of NH4+ by bacteria
and fungi from organic nitrogen originating from dead plants Atmospheric nitrous oxide (N2O) has a large glacial-
or animals, or animal waste. interglacial amplitude (Fig. 23.12), with values of *200
During nitrification and denitrification, N2O, a green- ppbv during glacial maxima and *270 ppbv during inter-
house gas, is also produced. The sources of N2O come from glacials including the Holocene and up to 280 ppb during
both ocean and land. The sink of N2O is due to photo- the interglacial *400,000 years ago (MIS 11) (Schilt et al.
chemical reaction with ozone in the stratosphere. 2010). Unlike CO2 and CH4, the concentration of N2O

Fig. 23.12 Evolution of N2O (ppb), over the last 800,000 years. Data from Schilt et al. (2010)
286 N. Bouttes et al.

during the interglacial periods before 400 ka is not lower reduced the flooded areas at the continental edges and thus
than the concentration during the more recent interglacial. In the associated source of N2O.
addition, N2O shows a millennial variability similar to that Several authors (Sowers et al. 2003; Flückiger et al. 2004;
of methane, with an amplitude very close to that of Schilt et al. 2013) also highlighted variations of around 40
glacial-interglacial transitions (Schilt et al. 2013). ppbv in the N2O concentration in the atmosphere in phase
There is little information available to explain the with Dansgaard-Oeschger events. These variations in N2O
glacial-interglacial and millennial variations in N2O. The are substantially different in amplitude from variations in
natural cycle of N2O over these time scales is governed by CH4. While methane recordings show a fairly strong corre-
two sources of approximately equal importance currently, an lation with insolation in the low and mid latitudes of the
ocean source (estimated at 4 ± 2 TgN/year) and a terrestrial Northern Hemisphere, this is not the case for N2O.
source (the natural part estimated at 7 TgN/year), and by an The concentration of N2O begins to increase before
atmospheric sink mainly linked to the photolysis of the N2O methane (during the warming phase in the Southern Hemi-
molecule in the stratosphere (12 Tg/yr). Modeling studies sphere), and then the concentrations of the two gases reach
simulating atmospheric chemistry over the last few thousand their maximum at the peak of the hot phase of the D-O
years suggest that the N2O sink during the LGM was similar event. This information is compatible with the idea that both
to now (Crutzen and Brühl 1993), resulting in a lifetime marine and terrestrial sources play an important part in the
of *120 years for N2O in the atmosphere. Therefore, it evolution of N2O: the marine source is stronger in the
appears that variations in the sources are responsible for the Southern Hemisphere, in phase with the warming period in
changes in its atmospheric concentration during the the south, and the terrestrial source is stronger in the
Quaternary. northern hemisphere, in phase with the warming period in
Isotopic data, in particular measurements of the 15N of the north.
N2O, are used to separate the ocean source from the ter- So far, very little modelling work has focused on changes
restrial source, as the former generally shows enriched iso- in the concentration of N2O in the atmosphere over the last
topic signatures compared to the latter (probably due to a hundreds of thousands of years. A simulation for the
relatively stronger denitrification process in N2O production Younger Dryas episode suggests a combination of changes
in an ocean environment). These data suggest that the ratio in the marine and terrestrial sources of N2O to explain the
of ocean to land sources did not change over the last variations measured during this event. Simulations run by
33,000 years (Sowers et al. 2003). These two sources appear Schmittner and Galbraith (2008) show that changes in ocean
to vary in phase with each other, with an increase of nearly circulation play a major role N2O variations. A reduction of
40% during warm periods compared to cold periods. the AMOC leads to decreased productivity and better ven-
The mechanisms for explaining these variations are still tilation resulting in increased subsurface oxygen concentra-
largely hypothetical. The production of N2O in the ocean is tions, which explains the decrease in N2O production.
generally linked to the presence of suboxic zones, low in
oxygen, which are found directly below some of the major
productive regions (east of the tropical Pacific, the Arabian Oxygen Cycle
Sea). In these zones deficient in dissolved oxygen, nitrates
are used by the microorganisms as a source of oxygen during The Oxygen Cycle and Its Ocean Component
the denitrification reactions leading to the remineralization of
the organic matter; nitrous oxide (N2O) is a by-product of Atmospheric free oxygen does not directly impact climate
these reactions. A decrease in the source of N2O during the since it does not absorb infrared radiations. Despite this fact,
glacial period could be the consequence of a shrinking of the cycle of oxygen has gained lots of attention primarily
these zones, which would itself be a result of changes in because of its tight relationship with life on Earth. Indeed,
ocean dynamics or local marine productivity leading to an the main source of free oxygen comes as a waste product of
increase in dissolved oxygen at the sub-surface. However, photosynthesis by plants on land and phytoplankton in the
this hypothesis is only partially supported by reconstructions ocean. The main oxygen sink is due to respiration and/or
of paleoproductivity. remineralization of organic matter by almost all living
Another theory concerning the ocean source is related to organisms, which consume di-oxygen and release carbon
the expansion of flooded surfaces on the continental shelves. dioxide. Other minor sources include the photolosyis of N2O
Recent estimates suggest that a significant portion (between and H2O in the atmosphere, whereas oxygen sinks are
0.6 and 2.7 TgN/year) of marine N2O production comes numerous and include a number of oxidation and chemical
from the continental shelves. The significant drop in sea weathering pathways (see Walker 1980, for a review of the
level (−120 m) during cold periods would have greatly global oxygen cycle).
23 Biogeochemical Cycles and Aerosols Over the Last Million Years 287

Fig. 23.13 O2 concentrations at

mid ocean depth (averaged over
200–600 m, in micromole/L)
from the World Ocean Atlas
(2009) (Garcia et al. 2010)

Many studies have examined the oxygenation of the Over the past decades, observations have shown that oxygen
Earth’s atmosphere over geological time scales, focusing for concentrations have decreased in the open ocean in many
example on the Great Oxidation Event 2.45 billion years ago ocean regions and that the tropical oxygen minimum zones
(e.g., Holland 1994) or on the variations of atmospheric O2 (OMZs) have likely expanded (Rhein et al. 2013). The
over the Phanerozoic (Berner et al. 2003). Over the last mechanisms involved are a decrease in the oxygen solubility
million years, the consensus is that the atmospheric con- due to ocean warming and the combination of reduced ocean
centration of O2 has been very stable due to the very long ventilation and increased stratification that prevents the
residence time of O2 in the atmosphere-ocean system (on the penetration of oxygen into the interior of the ocean. These
order of two million years, Catling and Claire 2005). mechanisms are very consistent with the recent global
The oceanic component of the oxygen cycle has been warming trend, suggesting that deoxygenation will continue
however much more variable over glacial-interglacial cycles with future anthropogenic climate change. Indeed, climate
(Jaccard and Galbraith 2012). This is due to the fact that the models do simulate a clear deoxygenation trend with global
oceanic reservoir of oxygen is much smaller than the warming, with an oceanic loss of oxygen of a few percent at
atmospheric one (225 Tmol O2 in the ocean vs. 3.8  107 the end of the twenty-first century (Bopp et al. 2013). At the
Tmol of O2 in the atmosphere), and that dissolved oxygen regional scale however, there is yet no consensus on the
concentrations in the ocean are very heterogeneous, with O2 evolution of subsurface oxygen levels, with very large model
concentrations ranging from 0 to almost 400 micromol/L. uncertainties.
A map of O2 concentrations at mid ocean depth illustrates
this heterogeneity (Fig. 23.13), with O2-enriched waters at
high latitudes and O2-depleted waters in the Eastern Tropical Ocean Oxygenation at the Last Glacial Maximum
Pacific and in the Northern Indian Basin (depicting the
so-called Oxygen Minimum Zones). The past record of ocean oxygenation during glacial-
The oxygen content of the ocean results from a fine interglacial cycles provides a complementary perspective
balance between the consumption of oxygen by respiring on how the oceanic oxygen content may respond to climate
organisms feeding on organic matter sinking from the sur- change or climate variability. The reconstruction of past
face, and the supply of O2-rich waters coming from the ocean oxygenation relies on sedimentary proxies of bottom
surface of the ocean through ocean ventilation. water oxygenation. The most common proxies for ocean
Because oxygen is a fundamental resource for aerobic oxygenation are based on the presence of sediment lamina-
organisms, the distribution of oxygen in the ocean has a tions (that testify very low levels of bottom water oxygen
large imprint on marine life, shaping for example the habitat levels), on redox sensitive trace metals (such as uranium and
of large fish such as tunas or billfishes (Stramma et al. 2012). molybdenum) and on benthic foraminifera assemblages.
288 N. Bouttes et al.

increase in carbon export from the surface layers that would

have increased oxygen consumption in the subsurface ocean
(Jaccard and Galbraith 2012), but also to a decreased ven-
tilation of sub-surface layers in response to changes in
oceanic circulation (Bopp et al. 2017).

Ocean Oxygenation in the Mediterranean Sea:

The Case of Sapropels

Definition and overview

Sapropel events are clearly identified in marine sediments.
They are characterized by an organic rich layer sedimenta-
tion, mainly found in the eastern basin (Fig. 23.15). The
peculiarity of this sedimentation suggests an anoxic envi-
ronment allowing the organic matter preservation-due to the
shutdown of the thermohaline ventilation (Möbius et al.
2010) and/or an enhanced biological productivity leading to
an increased oxygen consumption (Martinez-Ruiz et al.
2000).
Sapropel events have been largely described and inves-
tigated since they have been identified in the middle of the
twentieth century (Kullenberg 1952). Since the closure of
East Tethys seaway 14 million years ago (Hamon et al.
2013), the only connection between the Mediterranean basin
and the global ocean has been the Gibraltar straight. This
semi-enclosed configuration favored sapropel events that
occurred with a frequency of 21,000 years (Emeis et al.
2003).
Fig. 23.14 Changes in ocean oxygenation between the Last Glacial
Maximum and the Early Holocene for the Atlantic (top), Indian Large variability in contexts, developments and pro-
(middle) and Pacific (bottom) basins. The symbols refer to the different cesses associated with sapropels occurrences
proxies used to infer past oxygenation (squares corresponding to
laminations, circles to foraminifera species assemblages, and diamonds
There is a large variability in the imprints (strength, exten-
to redox-sensitive trace metals). Blue shadings indicate a relative sion, duration) of these sapropels (see Rohling et al. (2015)
decrease in oxygenation and orange shadings a relative increase in for a review). For example the carbon organic content levels
oxygenation from the LGM to the early Holocene. From Jaccard et al. in sapropels typically range between 1 and 10% with a large
(2014)
variability, for instance sapropel S5 occurring during the last
interglacial reached values from 7–15% (Grant et al. 2012)
Despite recent progress on the developments of new and Pliocene sapropel may reach 30% (Nijenhuis and De
proxies and the multiplication of the application of such Lange 2000).
methods to a growing number of sediment cores, the indi- Whereas pacing of sapropel is strongly correlated to
cations of past ocean oxygenation over the last precessional cycles, there are major differences in precon-
glacial-interglacial cycles portrayed by these records stay ditioning water masses that favor the occurrence of sapro-
very qualitative and mostly apply to the last deglaciation pels. Superimposed to this variability, the triggering of
(Jaccard and Galbraith 2011). Overall, marine sediments sapropel has been shown to be several thousand years after
indicate an oxygenation of the deep ocean throughout the the maximum summer insolation in the northern hemisphere.
last deglaciation and an expansion of low-oxygen waters in Ziegler et al. (2010) inferred a recurrent lag between the
the upper ocean. These changes seem quite consistent across northern hemisphere insolation maximum and sapropel
all ocean basins (Fig. 23.14). deposition.
The mechanisms responsible for such changes are still Sapropels typically lasted between *3 and *8 kyrs. For
debated (Jaccard et al. 2014; Bopp et al. 2017). In the instance, durations of *4.4, *4.0, *6.2, and *7.4 kyr
sub-surface ocean for example, the decreasing trend in for S1, S3, S4, and S5, respectively has been attributed by
oxygen levels during the deglaciation could be due to an different methods (Grant et al. 2012) Despite a large
23 Biogeochemical Cycles and Aerosols Over the Last Million Years 289

Fig. 23.15 Left, sapropels within a sediment core recovered in 2001 during RV Meteor cruise M51-3 (Hemleben et al. 2003) (photograph by
Eelco. J. Rohling and Kristian.C. Emeis). Right, late Pliocene sapropel layers outcropping at Punta Piccola (Plancq et al. 2015)

variability in time and space of benthic recolonization at the Superimposed to precession cycles, since one million
end of a sapropel. The end of enhance monsoon seems to years the 100 ky glacial-interglacial cycle has also affected
mark a sharp and widespread (basinwide) onset of deep sapropel occurrence (Köng et al. 2017). Recent modeling
water oxygenation due to resumption of strong convective studies (Grimm et al. 2015) aimed to simulate the S1 from its
deep-water formation. onset. A more specific scenario, involving a preconditioning
Sapropels are more common in the eastern Mediterranean of cold and poorly salted water coming from the last Hein-
(east of the Strait of Sicily) than in the western Mediter- rich event, was suggested as a possible cause for the S1
ranean, where they are also known as Organic Rich Layers formation.
(ORLs) (Rogerson et al. 2008). Another important issue is to reach high resolution to
As suggested by Rohling et al. (2015), the eastern capture convection patterns in the Mediterranean basin.
Mediterranean is more sensitive to development of deep-sea Using a coupled AOGCM (Atmospheric-Ocean Global
anoxia than the western Mediterranean, because of differ- Circulation Model) including a regional Mediterranean Sea
ences in the efficiency of deep-water renewal. model (1/8° much higher than previously used), Vadsaria
For a long time, the relationship between insolation et al. (2019) have revisited the impact of Nile hosing fresh
changes associated with precession cycles and impacts on water increase on triggering sapropel S1. This improvement
hydrological changes through fresh water inputs has been allows for better simulating the intermediate and deep con-
invoked as causal link to explain sapropel occurrences vection occurring in winter (Adloff et al. 2015).
(Rossignol-Strick et al. 1982). Nevertheless, there are many Moreover, the simulation of oceanic tracer as Nd allows
different contexts in, which sapropels have occurred since one to validate changes in ocean dynamics (Ayache et al.
the Miocene (Rohling et al. 2015). For instance, the role of 2016).
the cryosphere was certainly different during Pliocene and
Pleistocene. For Quaternary, it has been shown that, super-
imposed to the major effect of African monsoon and Sulphur
enhancement of freshwater from the Nile river, other forcing
factors have to be accounted for. The imprint of The sulfur cycle is of interest to climatologists because it
glacial-interglacial cycle and associated sea level changes leads to the formation of sulfuric acid (H2SO4), a submi-
has been shown to contribute strongly for sapropel occurring cronic aerosol that reflects solar radiation efficiently (direct
during deglaciation as S1 (Rohling et al. 2015; Grimm et al. effect) and which, due to its highly hygroscopic nature, has a
2015). physical influence on clouds (indirect effect). It is formed in
the atmosphere by the oxidation of SO2 whose emission
Most recent advance on S1 sapropel modeling level through the combustion of fossil reserves is 35–45 Tg
Modeling represents also a unique tool to investigate the per year. The IPCC estimates that due to the increase in
responses of the Mediterranean basin to different external sulfate aerosols in the atmosphere, anthropogenic emissions
forcing factors from insolation changes to associated hydro- of SO2 could be responsible for a radiative forcing of
logical perturbations that may produce sapropel events. −0.4 W/m2, an opposite forcing but equivalent to about one
Two important developments have been down recently third of the radiative forcing linked to the increase of CO2 in
concerning sapropel modeling: much longer simulation and the atmosphere (+1.2 W/m2). However, as noted above, the
much higher spatial resolution. comparison between these two values is of limited
290 N. Bouttes et al.

Fig. 23.16 Feedback

mechanisms linking our planet’s
climate to DMS emissions from
the ocean (adapted from Charlson
et al. 1987)

CCN Albedo

Atmospheric Solar radiation (heat, light, UV)

acidity
Sulfate and
Sulfur transport sulfonate aerosols Temperature
to continents Wind speed
DMSg

Oceanic
DMSaq planctonic production mixing layer

significance since the forcing of the aerosol may, for the MSA content of ice (Saigne and Legrand 1987). This
example, be much greater over heavily anthropized regions study of the last climate cycle in the Vostok ice core indi-
of the northern hemisphere. Moreover, the temporal evolu- cated an increase in MSA deposition fluxes by a factor of
tion of anthropogenic SO2 emissions varies greatly from one almost three between an interglacial period and the final
continent to another: emissions reached a peak in the 1980s stage of an ice age, despite the fact mentioned previously
in Western Europe and the USA, while an acceleration of that sulphate flux has shown little variation in the past
their emissions is underway in India, for example. The sulfur (Legrand et al. 1991). Since isotopic studies of sulfur
cycle in the atmosphere also involves two major natural undertaken on sulphate have subsequently clearly estab-
sources, volcanic activity and marine DMS biogenic emis- lished that DMS emissions are the major source of sulphate
sions. An important dimension of the latter source is its in Antarctica, a contradiction therefore appeared between the
potential response to climate conditions, a hypothesis pro- two proxies of DMS emissions, with the MSA suggesting an
posed by Charlson, Lovelock, Andreae and Warren, four increase in DMS emissions, but not in sulphate. The
authors of a well-known article in Nature in 1987, proposing hypothesis of an increase in DMS emissions during the
that DMS emissions produced by plankton may act as a glacial period seemed reasonable given that the phyto-
climate regulator (Fig. 23.16). Responses by the marine plankton species that emits a lot of DMS (Phaeocystis) has a
biosphere (changes in primary productivity, phytoplankton particular affinity with sea ice.
speciation, stratification of surface ocean layers, wind pat- The difference between the sulphate and MSA records
terns, etc.) to the current climate change are thus possible. Its remains unclear to date. The atmospheric studies carried out
direction and magnitude remain uncertain, but some over the last few years in Antarctica demonstrate the com-
ocean-atmosphere models incorporating an ocean biology plexity of the problem. Although the atmospheric levels of
module, coupling sulfur and carbon, suggest that a 40% the three sulfur species coincide well over time with the
increase in DMS emissions could occur in the 40 °S zone in DMS concentrations in the Southern Ocean, the interannual
response to a doubling of CO2 content and to the associated variability observed in the Southern Ocean does not show
climate change (Bopp et al. 2003b). any straightforward connection with annual sea ice cover
The idea of examining the relationship between marine (Preunkert et al. 2007). These measurements also reveal
biogenic sulfur emissions and past climate is the motivation intricate processes involving photochemistry and atmo-
behind many ice core studies. Given that sulphate in the spheric dynamics which makes the link between the two
atmosphere has several origins, the first attempt at recon- sulfur species and DMS very complex. These studies con-
stituting marine biogenic emissions was made by examining tinue to be pursued actively due to the fact that, to date, only
23 Biogeochemical Cycles and Aerosols Over the Last Million Years 291

from ice can past variability in marine biogenic emissions effect. The indirect effects, on the other hand, involve
potentially be reconstructed, since organisms like Phaeo- aerosols acting as cloud condensation (CCN) or ice nuclei
cystis do not leave any traces in marine sediments. The issue (IN), which means that water or ice aggregates around them.
is important because it contributes to our understanding of This modifies the type, extent and lifetime of clouds. For
the feedback taking place in high latitudes (regions highly example, the presence of aerosols leads to smaller but more
sensitive to global changes in climate), involving complex numerous droplets, which yields a more reflective cloud than it
processes between marine biology, sea ice, and climate. would be without aerosols. This also means that clouds formed
with aerosols will have a longer lifetime since the droplets are
smaller and won’t reach the critical size for precipitation. The
Aerosols and Dust indirect effect can result in warming or cooling depending on
the altitude where clouds are formed. Because the effect of
Aerosols are small liquid or solid particles, ranging from a aerosols on radiation and on clouds is complex and depends on
few nm to 100 µm, in suspension in the atmosphere. Natural many parameters, it remains one of the main sources of
aerosols include desert dust, sea salt, carbonaceous, sulphur uncertainties in models (Boucher et al. 2013).
and nitrogen species, largely emitted from dry and vegetated In addition, aerosols depositing back to the surface can
landscapes, the oceans, and volcanoes (Carslaw et al. 2010). also modify the albedo. This is the case of dust, and espe-
Primary aerosols are emitted directly from the surface of the cially black carbon; they can cause snow and ice to darken,
Earth, whereas secondary aerosols are formed from gaseous which reduces the albedo and leads to warming.
precursors in the atmospheric environment. Aerosols are In virtue of their composition, aerosols also act as carriers
washed out by precipitation, or removed by gravitational of specific elements, such as nitrogen, phosphorus, sulphur,
settling and dry deposition, so that their lifetime in the and iron, which are linked to important biogeochemical
atmosphere is short, a few days only—except when they cycles, including the carbon cycle (Mahowald et al. 2017).
reach the stratosphere, where they can stay for a few years, In particular, phosphorus and iron are linked to the dust
as can happen during giant volcanic eruptions. Therefore, cycle, and the peculiarity is that windblown inputs can be
unlike well-mixed GHGs, aerosols are considered to be fundamental to the mass budgets of those elements in remote
short-lived climate forcing agents, and their impacts are regions, far from the dust sources. For instance, dust-borne
characterized by a strong regional component (Boucher et al. phosphorus from North Africa replenishes the pool of this
2013). Aerosol emissions vary depending on surface climate element in the Amazon, where the loss by fluvial erosion
conditions; on the other hand, aerosols impact the climate would otherwise deplete it, with implications for the rain-
system themselves, though direct and indirect (cloud- forest. Iron, on the other hand, is a micronutrient for marine
mediated) interactions with the atmospheric radiation bud- ecosystems. Because its sources are the continents, remote
get, by changing the surface albedo, as well as by means of marine areas are depleted in this element. Where the abun-
indirect impacts on global biogeochemical cycles (Maho- dance of macronutrients such as nitrogen and phosphorus is
wald et al. 2017). accompanied by a relative scarcity of iron, which limits the
Size, shape and composition define the specific interac- primary production at the ecosystem level, i.e. in
tions of aerosols with radiation, including absorption and High-Nutrient Low-Chlorophyll (HNLC) areas, dust-borne
scattering of shortwave (solar) and longwave (terrestrial) inputs of iron become of great importance in sustaining algal
radiation. The scattering of shortwave radiation results in blooms—this is notably the case of the Southern Ocean
cooling, but absorption can lead to warming when it is above (Jickells et al. 2005).
a highly reflective surface. Through their absorption of
outgoing longwave radiation, aerosols also behave like
GHGs. For some aerosol species, one particular effect is Natural Aerosols: Overview
dominant, in other cases, opposing effects coexist. Consid-
ering the variability of aerosol spatial distribution, the In this section we will briefly describe the main natural
coexistence and mixing of different aerosol species, and the aerosol types. Note that mineral dust and sea salt are still the
diversity of aerosol-radiation interactions, it is clear that most abundant primary aerosol species by mass in the pre-
direct impacts of aerosols on climate constitute a complex sent day atmosphere.
problem (Boucher et al. 2013). Mineral (desert) dust (Fig. 23.17) is emitted into the
Aerosols also interact with clouds. Changes in relative atmosphere in response to wind erosion of the surface, in dry
humidity linked to the vertical stability of the atmospheric and semi-dry areas, with low vegetation cover. Far-travelled
column and the surface evapotranspiration balance, occur- dust particles are mostly clays and fine silts below 10 µm in
ring as a rapid adjustment to direct aerosol forcing, can diameter, and are composed mainly of silicates, along with
influence cloud formation. This is called the semi-direct carbonates, gypsum, and metal oxides. Dust aerosols interact
292 N. Bouttes et al.

SW

Direct effects Cloud

effects

LW
Snow
CO2 darkening
Impacts on
Wind erosion biogeochemical cycles

Fig. 23.17 Overview of the dust cycle and its interactions with the climate system

with both solar and terrestrial radiation, and act as ice nuclei Wind stress on the surface of the oceans drives aerosol
in mixed phase clouds. As already mentioned, dust deposi- emissions at the sea-air interface, where bubbles bursting
tion on the surface can reduce snow albedo and impact on within breaking waves eject sea sprays composed mostly of
the carbon cycle through phosphorus and iron mediated sea salt (dominating the super-micron fraction), along with
interactions with global biogeochemical cycles. organic particles (concentrated in the sub-micron fraction).
Land vegetation is a source of primary biogenic aerosol Sea salt and some organic particles are also emitted from the
particles (PBAP), in the form of fungal spores, viruses, surface of sea ice, where these form frost flowers during the
bacteria, pollens and plant debris, as well as aerosol pre- process of brine rejection that accompanies the freezing of
cursors such as isoprene and other volatile organic com- seawater. Sea salts tend to cool the atmosphere, and because
pounds (VOC), which result in the formation of secondary of their high hydroscopicity, they act as CCN (O’Dowd and
organic aerosols (SOA). Wildfires are responsible for the de Leeuw 2007). Oceans are also a major source of pre-
emissions of particulate organic matter and black carbon. cursors of sulphate aerosols, in the form of biogenic emis-
Particulate carbonaceous aerosols have a slight net cooling sions of dimethylsulfide (DMS) that is then oxidised to
effect in the atmosphere, and play a role as cloud conden- sulphuric acid and methane sulfonic acid (MSA) in the
sation and ice nuclei. Black carbon is a strongly absorbing atmospheric environment (Legrand and Mayewski 1997).
aerosol, associated with snow darkening and has a warming
effect on the atmosphere.
Wetlands emissions of ammonia, and NOx produced by Dust Variability and Impacts on Past Climates
biomass burning, biogenic soil emissions, stratospheric
injection and by the interaction of lightning with atmo- The past history of the dust cycle is imprinted in natural
spheric N2, are the main precursors to nitrogen aerosol archives such as ice, marine sediments, loess/paleosol
species, which overall have a moderate net cooling effect in sequences, and peat bogs. Dust has the greatest preserva-
the atmosphere, and may impact on biogeochemical cycles. tion potential among all aerosol species, because it is
Volcanic emissions of SO2 are one of the primary sources essentially insoluble, and traces of it are present in a variety
of precursors for the formation of natural sulphate aerosols, of environments all around the globe. In general, we can
which are characterized by a strong cooling impact on the obtain a paleodust record from natural archives when the
atmospheric radiation, and they are very efficient CCN. following conditions are met: there is preservation of the
Volcanic eruptions can also eject glass shards and ash, which deposition signal; we can establish a chronology; and we are
can leave a mark by depositing as tephra layers—very useful able to separate eolian contributions from the sedimentary
stratigraphic markers. matrix.
23 Biogeochemical Cycles and Aerosols Over the Last Million Years 293

The fifty-meter thick loess deposits of China and North the last eight glacial-interglacial cycles; colder climate states
America are probably the most spectacular evidence of how are characterized by increased dustiness, as shown by the
the dust cycle is capable of shaping immense landscapes— milestone paleoclimate records from the Vostok and EPICA
loess deposits cover 10% of the emerged landmasses. Loess Dome C (EPICA Community Members 2004) ice cores from
accumulation in China has been an ongoing process for over Antarctica (Fig. 23.18). Preservation of stratigraphy and
20 million years, since the uplift of the Tibetan plateau chronologies based on numerical (absolute) dating methods
caused widespread aridification of central-eastern Asia. allow for a more detailed reconstruction of the last
Beyond that, little is known, although isolated information glacial-interglacial cycle. In particular, global compilations
on deep paleoclimate conditions at least dating back to the of paleodust records based on dust mass accumulation rates
Paleozoic (*500 million years ago) can be derived, based provide a quantitative metric to compare paleodust records
on the analysis of geologic formations whose origin can be from different natural archives, and constitute a bench-
ultimately linked to eolian sedimentation. marking tool for models (Kohfeld and Harrison 2001).
We have a better picture of the global dust cycle on During the LGM, global dust emissions were enhanced
Quaternary time scales, especially since the late Pleistocene. by a factor of 2–4, and the increase in dust deposition in high
From polar ice core records we know that a strong latitudes was even by a factor of 10 or more. A combination
dust-climate coupling was a persistent feature at least over of changes in dust source areas and transport patterns, driven

Fig. 23.18 Overview of aerosol (nss-SO42− as a proxy for sulfate aerosols, ss-Na for sea salt, and nss-Ca for dust) deposition fluxes from the
EPICA Dome C ice core (Wolff et al. 2006), Antarctica, along with indicators of global climate, i.e. deuterium excess (Jouzel et al. 2007)
294 N. Bouttes et al.

by the large-scale atmospheric circulation, shaped the geo- very fertile soils. Ice cores do provide this kind of temporal
graphical variability in dust cycle dynamics. The sharp resolution, but we still have very little data. The top meters
decline in dust deposition to Antarctica paralleling the rise in of polar cores, corresponding to this time frame, are actually
atmospheric CO2 levels, led oceanographer John Martin made of firn, which complicates the analysis because of the
(“Give me a tanker of iron, and I will give you an ice age”), risk of contamination of the samples. Alpine ice cores, on
in the late eighties, to formulate the famous iron hypothesis: other hand, allow the analysis of dust concentration and can
the glacial increase in dust-borne iron inputs to the Southern have good chronologies for the last few decades, but reli-
Ocean may have stimulated the ocean biological pump, able, quantitative estimates of dust mass accumulation rates
resulting in an increased productivity in HNLC areas and are hampered by the extreme spatial variability of snow
subsequent carbon sequestration in the deep ocean (Jickells accumulation and post-depositional processes. Therefore, we
et al. 2005). State of the art ESMs suggest that this mech- do not have a clear pre-industrial reference state for dust
anism may have been responsible for a decrease of *20 (Carslaw et al. 2010). A few studies trying to address dust
ppmv of atmospheric CO2, out of 80/100 ppmv—the overall trends during the twentieth century yield contrasting results;
decrease in CO2 concentrations during the LGM measured in a generally increasing trend over parts of the last century
ice cores. On the other hand, the impact of the increased dust suggested by some authors may have “masked” a fraction of
burden on atmospheric radiation is estimated by models to global temperature increase, due to the net cooling effect of
be at least −1 W/m2 globally, compared to −6 W/m2 forcing dust.
by reduced GHGs atmospheric concentrations and changes
in surface albedo from decreased sea levels and the growth
of ice sheets in the Northern Hemisphere. Nonetheless, this Other Aerosol Species in Past Climates
figure might hide much larger dust impacts of the opposite
sign, with strong cooling downwind of the major dust There is much less information about other aerosol species in
sources, and warming over the bright, glaciated Arctic the past; unlike dust, solubility and volatility limit the
regions (Albani et al. 2018). preservation of most species in many environmental settings,
During the course of the entire glacial period, millennial and pose additional analytical challenges. Most of the
scale variability expressed by Dansgaard-Oeschger and information we have is from ice cores (Wolff et al. 2006;
Heinrich events in the d18O record in Greenland ice cores, is Preunkert and Legrand 2013).
almost paralleled by variations in dust deposition rates, with Na+ is a stable proxy for sea salts. Yet, we do not really
colder phases associated to dustier conditions (Rasmussen have information from lower latitudes; sea salt records from
et al. 2014). The same picture emerges from the corre- polar ice cores (Fig. 23.18) show, similar to dust, increased
sponding alternation of stadial/interstadial periods in deposition rates in glacial climates, by a factor 3–5 in
Antarctica (EPICA Community Members 2006). Antarctica and 1.5–3 in Greenland. It is not fully clear to
The deglaciation followed, characterized by a non- what extent this was due to an increase in emissions or
monotonic increase in global temperatures and CO2, mir- transport from open waters, rather than to the expanded sea
rored by decreasing dust levels in both hemispheres. While ice source, although the latter seems to be the dominant
the Holocene was initially described as a relatively flat factor. Increases in sea salts could have had a negative
period in terms of dust, based on polar records, in the last forcing, via direct and indirect effects, on the atmospheric
three decades, a few studies of North Atlantic sediment cores radiation budget.
highlighted the large variability in North African dust Sulphur and nitrogen aerosol species present more
emissions; a reduction by a factor 2–5 corresponded to the important issues with preservation, and are more difficult to
“Green Sahara” phase of the Early and mid-Holocene, interpret. The baseline sulphate records from polar ice cores
characterized by an enhanced summer monsoon, compared (Fig. 23.18), mainly derived from DMS emissions, show a
to the drier late Holocene after *5 ka BP. The possible flat signal in Antarctica and some variability in Greenland.
(positive or negative) feedbacks between the monsoon sys- This has dampened some early enthusiasm for the idea that
tem and variations in the North African dust cycle are a DMS could be a dominant feedback in driving
subject of study by the modelling community (Albani et al. glacial-interglacial variability, given the strong cooling
2015). effects associated with sulphate aerosols. Sulphate concen-
Ironically, we know relatively less about the more recent tration spikes are associated with inputs from large volcanic
past. Marine sediments and loess profiles generally can’t eruptions, and are sometimes associated with the presence of
achieve a temporal resolution fine enough to resolve the last tephra. In certain cases, a lag of one or two years was
millennia, and often surface layers of loess/paleosol deposits observed in the peaking of the two signals, suggesting a
are disturbed by agricultural practices, as they tend to be longer stratospheric residence time of sulphates. There are
23 Biogeochemical Cycles and Aerosols Over the Last Million Years 295

no clear trends in volcanic activity on late Quaternary time Adloff, F., Somot, S., Sevault, F., Jordà, G., Aznar, R., Déqué, M., et al.
scales, although several volcanic spikes marked the (2015). Mediterranean Sea response to climate change in an
ensemble of twenty-first century scenarios. Climate Dynamics, 45
deglaciation. (9–10), 2775–2802. https://doi.org/10.1007/s00382-015-2507-3.
There is very little direct information on carbonaceous Ahn, J., & Brook, E. J. (2008). Atmospheric CO2 and climate on
aerosols from ice cores on late Quaternary time scales. Some millennial time scales during the last glacial period. Science, 322,
data is available on paleofire proxies, such as ammonium or 83. https://doi.org/10.1126/science.1160832.
Ahn, J., Brook, E. J., Schmittner, A., & Kreutz, K. (2012). Abrupt
levoglucosan, which can be combined with information from change in atmospheric CO2 during the last ice age. Geophysical
charcoal, preserved in paleosols, peat bogs, and lakes—on Reseach Letters, 39, L18711. https://doi.org/10.1029/2012
the other hand, an indicator of the source of the aerosols and GL053018.
precursors, rather than the sink. Finally, the variation of Albani, S., Mahowald, N. M., Winckler, G., Anderson, R. F.,
Bradtmiller, L. I., Delmonte, B., et al. (2015). Twelve thousand
pollen assemblages archived in lakes and peat can provide years of dust: The Holocene global dust cycle constrained by natural
information on the state of past vegetation, which is a source archives. Climate of the Past, 11, 869–903. https://doi.org/10.5194/
of many aerosols and precursors. The general picture is that cp-11-869-2015.
land biogenic and fire emissions during the last glacial Albani, S., Balkanski, Y., Mahowald, N., Winckler, G., Maggi, V., &
Delmonte, B. (2018). Aerosol-climate interactions during the Last
period were lower than today, at least at high latitudes, Glacial Maximum. Current Climate Change Reports, 4, 99–114.
which is consistent with a drier and colder climate, and the https://doi.org/10.1007/s40641-018-0100-7.
reduction of vegetated areas linked to the growth of the large Anderson, R. F., Ali, S., Bradtmiller, L. I., Nielsen, S. H. H., Fleisher,
ice sheets in North America and Eurasia. More data on the M. Q., Anderson, B. E., et al. (2009). Wind-driven upwelling in the
Southern Ocean and the deglacial rise in atmospheric CO2. Science,
recent past can be found in snow and ice, showing the 323, 1443. https://doi.org/10.1126/science.1167441.
influence of anthropic activities on carbonaceous, sulphate Archer, D., & Maier-Reimer, E. (1994). Effect of deep-sea sedimentary
and nitrogen aerosols (Preunkert and Legrand 2013). calcite preservation on atmospheric CO2 concentration. Nature,
367, 260–263.
Archer, D., Winguth, A., Lea, D., & Mahowald, N. (2000). What
caused the glacial/interglacial pCO2 cycles? Reviews of Geophysics,
Conclusion 38, 159–189.
Archer, D. E., Martin, P. A., Milovich, J., Brovkin, V., Plattner, G.-K.,
In this chapter, we have looked at the major biogeochemical & Ashendel, C. (2003). Model sensitivity in the effect of Antarctic
sea ice and stratification on atmospheric pCO2. Paleoceanography,
cycles that interact with climate: carbon (CO2 and CH4), 18(1), 1012.
nitrogen (N2O), sulphur (SO2) as well as dust. These bio- Ayache, M., Dutay, J.-C., Arsouze, T., Révillon, S., Beuvier, J., &
geochemical cycles are crucial for climate mainly because of Jeandel, C. (2016), High resolution neodymium characterization
their role in the atmospheric radiative budget. In turn, climate along the Mediterranean margins and modeling of eNd distribution
in the Mediterranean basins. Biogeosciences, (April), 1–31. http://
evolution impacts these cycles by changing their sources and doi.org/10.5194/bg-2016–109.
sinks. Past changes are documented in archives such as Barker, S., Diz, P., Vautravers, M. J., Pike, J., Knorr, G., Hall, I. R.,
sediment and ice cores, which have been described in more et al. (2009). Interhemispheric Atlantic seesaw response during the
details in Volume 1. To better understand past changes of the last deglaciation. Nature, 457, 1097–1102. https://doi.org/10.1038/
nature07770.
carbon, nitrogen, sulphur and dust cycles and evaluate their Barnola, J. M., Raynaud, D., Korotkevich, Y. S., & Lorius, C. (1987).
impact on climate, they have been included in numerical Vostok ice core provides 160,000-year record of atmospheric CO2.
climate models. Going back and forth between data and Nature, 329, 408–414.
model simulations is crucial to unravel the causes of past Basak, C., Fröllje, H., Lamy, F., Gersonde, R., Benz, V., Anderson, R.
F., et al. (2018). Breakup of last glacial deep stratification in the
changes, and in the future, both more data and improved
South Pacific. Science, 359(6378), 900–904. https://doi.org/10.
modelling should help to better understand the changes in the 1126/science.aao2473.
biogeochemical cycles over the last million years. Bereiter, B., Lüthi, D., Siegrist, M., Schüpbach, S., Stocker, T. F., &
Fischer H. (2012). Mode change of millennial CO2 variability
during the last glacial cycle associated with a bipolar marine carbon
seesaw. Proceedings of the National Academy of Sciences, 109(25),
References 9755–9760. https://doi.org/10.1073/pnas.1204069109.
Bereiter, B., Eggleston, S., Schmitt, J., Nehrbass-Ahles, C., Stocker, T.
Adams, J. M., Faure, H., Faure-Denard, L., McGlade, J. M., & F., Fischer, H., et al. (2015). Revision of the EPICA Dome C CO2
Woodward, F. I. (1990). Increases in terrestrial carbon storage from record from 800 to 600 kyr before present. Geophysical Reseach
the Last Glacial Maximum to the present. Nature, 348, 711–714. Letters, 42, 542–549. https://doi.org/10.1002/2014GL061957.
Adkins, J. F., McIntyre, K., & Schrag, D. P. (2002). The salinity, Berner, R. A., Beerling, D. J., Dudley, R., Robinson, J. M., &
temperature, and d18O of the glacial deep ocean. Science, 298, Wildman, R. A. (2003). Phanerozoic atmospheric oxygen. Annual
1769–1773. https://doi.org/10.1126/science.1076252. Review of Earth and Planetary Sciences, 31, 105–134.
Adkins, J. F. (2013). The role of deep ocean circulation in setting Bigelow, N. H., Brubaker, L. B., Edwards, M. E., Harrison, S. P.,
glacial climates. Paleoceanography, 28, 539–561. https://doi.org/ Prentice, I. C., Anderson, P. M., Andreev, A. A., Bartlein, P. J.,
10.1002/palo.20046. Christensen, T. R., Cramer, W., Kaplan, J. O., Lozhkin, A. V.,
296 N. Bouttes et al.

Matveyeva, N. V., Murray, D. V., McGuire, A. D., Razzhivin, V. Broecker, W. S., & Peng, T.-H. (Eds.). (1982). Tracers in the Sea.
Y., Ritchie, J. C., Smith, B., Walker, D. A., Gajewski, K., Wolf, V., Palisades, New York: Lamont-Doherty Geological Observatory of
Holmqvist, B. H., Igarashi, Y., Kremenetskii, K., Paus, A., Pisaric, Columbia University.
M. F. J., & Vokova, V. S. (2003). Climate change and Arctic Brook, E. J., Harder, S., Severinghaus, J., Steig, E. J., & Sucher, C. M.
ecosystems I. Vegetation changes north of 55 °N between the Last (2000). On the origin and timing of rapid changes in atmospheric
Glacial Maximum, mid-Holocene and present. Journal of Geophys- methane during the last glacial period. Global Biogeochemical
ical Research, 108(D19), 8170. Cycles, 14(2), 559–572. https://doi.org/10.1029/1999GB001182.
Bird, M. I., Lloyd, J., & Farquhar, G. D. (1994). Terrestrial carbon Brovkin, V., Ganopolski, A., Archer, D., & Munhoven, G. (2012).
storage at the LGM. Nature, 371, 566. Glacial CO2 cycle as a succession of key physical and biogeo-
Blunier, T., Chappellaz, J., Schwander, J., Dällenbach, A., Stauffer, B., chemical processes. Climate of the Past, 8, 251–264. https://doi.org/
Stocker, T. F., et al. (1998). Asynchrony of Antarctic and Greenland 10.5194/cp-8-251-2012.
climate change during the last glacial period. Nature, 394, 739–743. Carslaw, K. S., Boucher, O., Spracklen, D. V., Mann, G. W., Rae,
Buizert, C., Cuffey, K. M., Severinghaus, J. P., Baggenstos, D., Fudge, J. G. L., Woodward, S., et al. (2010). A review of natural aerosol
T. J., Steig, E. J., et al. (2015). The WAIS Divide deep ice core interactions and feedbacks within the Earth system. Atmospheric
WD2014 chronology—Part 1: Methane synchronization (68–31 ka Chemistry and Physics, 10(4), 1701–1737.
BP) and the gas age–ice age difference. Climate of the Past, 11, Catling, D. C., & Claire, M. W. (2005). How Earth’s atmosphere
153–173. https://doi.org/10.5194/cp-11-153-2015. evolved to an oxic state: A status report. Earth and Planetary
Bopp, L., Kohfeld, K. E., Quéré, C. L., & Aumont, O. (2003a). Dust Science Letters, 237, 1–20.
impact on marinebiota and atmospheric CO2 during glacial periods. Chappellaz, J., Barnola, J. M., Raynaud, D., Korotkevich, Y. S., &
Paleoceanography, 18(2), 1046. Lorius, C. (1990). Ice-Core Record of Atmospheric Methane over
Bopp, L. O., Aumont, S. Belviso, & Monfray, P. (2003b). Potential the Past 160,000 Years. Nature, 345, 127–131.
impact of climate change on marine dimethyl sulfide emissions. Chappellaz, J. A., Fung, I. Y., & Thompson, A. M. (1993). The
Tellus B: Chemical and Physical Meteorology, 55(1), 11–22. atmospheric CH4 increase since the Last Glacial Maximum. Tellus,
https://doi.org/10.3402/tellusb.v55i1.16359. 45B, 228–241. 1. Source Estimates.
Bopp, L., et al. (2013). Multiple stressors of ocean ecosystems in the Chappellaz, J., Brook, E., Blunier, T., & Malaizé, B. (1997). CH4 and
21st century: Projections with CMIP5 models. Biogeosciences, 10, d18O of O2 records from Antarctic and Greenland ice: A clue for
6225–6245. https://doi.org/10.5194/bg-10-6225-2013. stratigraphic disturbance in the bottom part of the Greenland Ice
Bopp, L., Resplandy, L., Untersee, A., Mezo, P. L., & Kageyama, M. Core Project and the Greenland Ice Sheet Project 2 ice cores.
(2017). Ocean (de)oxygenation from the Last Glacial Maximum to Journal Geophysical Research, 102, 26547–26557.
the twenty-first century: insights from Earth system models. Charlson, R. J., Lovelock, J. E., Andreae, M. O., & Warren, S. G.
Philosophical Transactions of the Royal Society A, 375, 20160323. (1987). Oceanic phytoplankton, atmospheric sulphur, cloud albedo
Boucher, O., Randall, D., Artaxo, P., Bretherton, C., Feingold, G., and climate. Nature, 326, 655–661.
Forster, P., Kerminen, V.-M., Kondo, Y., Liao, H., Lohmann, U., Ciais, P., Tagliabue, A., Cuntz, M., Bopp, L., Scholze, M., Hoffmann,
Rasch, P., Satheesh, S. K., Sherwood, S., Stevens, B., & Zhang, X. G., et al. (2012). Large inert carbon pool in the terrestrial biosphere
Y. (2013). Clouds and aerosols. In T. F. Stocker, D. Qin, G.-K. during the Last Glacial Maximum. Nature Geoscience, 5, 74–79.
Plattner, M. Tignor, S. K. Allen, J. Boschung, A. Nauels, Y. Xia, V. https://doi.org/10.1038/ngeo1324.
Bex, & P. M. Midgley (Eds.), Climate Change 2013: The Physical Clement, A. C., & Peterson, L. C. (2008). Mechanisms of abrupt
Science Basis. Contribution of Working Group I to the Fifth climate change of the last glacial period. Reviews of Geophysics, 46,
Assessment Report of the Intergovernmental Panel on Climate RG4002. https://doi.org/10.1029/2006rg000204.
Change. Cambridge, UK and New York, NY, USA: Cambridge Craig, H. (1957). Isotopic standards for carbon and oxygen and
University Press. correction factors for massspectrometric analysis of carbon dioxide.
Bouttes, N., Roche, D. M., & Paillard, D. (2009). Impact of strong deep Geochimica et Cosmochimica Acta, 12, 133–149.
ocean stratification on the glacial carbon cycle. Paleoceanography, Crichton, K. A., Bouttes, N., Roche, D. R., Chappellaz, J., & Krinner,
24, PA3203. https://doi.org/10.1029/2008pa001707. G. (2016). Permafrost carbon as a missing link to explain CO2
Bouttes, N., Paillard, D., & Roche, D. M. (2010). Impact of changes during the last deglaciation. Nature Geoscience, 9, 683–
brine-induced stratification on the glacial carbon cycle. Climate of 686. https://doi.org/10.1038/ngeo2793.
the Past, 6, 575–589. https://doi.org/10.5194/cp-6-575-2010. Crowley, T. (1995). Ice age terrestrial carbon changes revisited. Global
Bouttes, N., Paillard, D., Roche, D. M., Brovkin, V., & Bopp, L. Biogeochemical Cycles, 9(3), 377–389.
(2011). Last Glacial Maximum CO2 and d13C successfully recon- Crutzen, P. J., & Brühl, C. (1993). A model study of atmospheric
ciled. Geophysical Reseach Letters, 38, L02705. https://doi.org/10. temperatures and the concentrations of ozone, hydroxyl, and some
1029/2010GL044499. other photochemically active gases during the glacial, the
Bouttes, N., Roche, D., & Paillard, D. (2012). Systematic study of the pre-industrial Holocene and the present. Geophysical Reseach
impact of fresh water fluxes on the glacial carbon cycle. Climate of Letters, 20(11), 1047–1050.
the Past, 8, 589–607. Curry, W. B., & Oppo, D. W. (2005). Glacial water mass geometry and
Bouttes, N., Swingedouw, D., Roche, D. M., Sanchez-Goni, M. F., & the distribution of d13C of RCO2 in the western Atlantic Ocean.
Crosta, X. (2018). Response of the carbon cycle in an intermediate Paleoceanography, 20, PA1017. https://doi.org/10.1029/
complexity model to the different climate configurations of the last 2004pa001021.
nine interglacials. Climate of the Past, 14, 239–253. https://doi.org/ Dahl-Jensen, D. (2018). Drilling for the oldest ice. Nature Geoscience,
10.5194/cp-14-239-2018. 11, 703–704.
Bozbiyik, A., Steinacher, M., Joos, F., Stocker, T. F., & Menviel, L. Dansgaard, W., Johnsen, S. J., Clausen, H. B., Dahl-Jensen, D.,
(2011). Fingerprints of changes in the terrestrial carbon cycle in Gundestrup, N. S., Hammer, C. U., et al. (1993). Evidence for
response to large reorganizations in ocean circulation. Climate of general instability of past climate from a 250-kyr ice-core record.
the Past, 7, 319–338. Nature, 364, 218–220.
23 Biogeochemical Cycles and Aerosols Over the Last Million Years 297

De Boer, A. M., & Hogg, A. M. C. (2014). Control of the glacial Climatic Transition (ca. 14 Ma). Climate of the Past, 9(6), 2687–
carbon budget by topographically induced mixing. Geophysical 2702. https://doi.org/10.5194/cp-9-2687-2013.
Research Letters, 41. doi:10.1002/2014GL059963. Harrison, S. P., Yu, G., Takahara, H., & Prentice, I. C. (2001).
Delmas, R., Ascencio, J. M., & Legrand, M. (1980). Polar ice evidence Palaeovegetation—Diversity of temperate plants in east Asia.
that atmospheric CO2 20,000 year BP was 50% of present. Nature, Nature, 413, 129–130.
284, 155–157. Hemleben, C., Hoernle, K., Jørgensen, B. B., & Roether, W. (2003).
Emeis, K.-C., Schulz, H., Struck, U., Rossignol-Strick, M., Erlenkeu- Ostatlantik-Mittelmeer- Schwarzes Meer, Cruise 51, 12 September–
ser, H., Howell, M. W., Kroon, D., Mackensen, A., Ishizuka, S., 28 December 2001. Meteor Ber. 03-1, 213 pp.
Oba, T., Sakamoto, T., & Koizumi, I. (2003). Eastern Mediter- Hemming, S. R. (2004). Heinrich events: Massive late Pleistocene
ranean surface water temperatures and 18 O composition during detritus layers of the North Atlantic and their global climate imprint.
deposition of sapropels in the late Quaternary. Paleoceanography, Reviews of Geophysics, 42, RG1005. https://doi.org/10.1029/
18(1), n/a–n/a. https://doi.org/10.1029/2000pa000617. 2003rg000128.
EPICA Community Members. (2004). Eight glacial cycles from an Heinrich, H. (1988). Origin and consequences of cyclic ice rafting in
Antarctic ice core. Nature, 429, 623–628. https://doi.org/10.1038/ the Northeast Atlantic Ocean during the past 130,000 years.
nature02599. Quaternary Research, 29(2), 142–152. https://doi.org/10.1016/
EPICA Community Members. (2006). One-to-one coupling of glacial 0033-5894(88)90057-9.
climate variability in Greenland and Antarctica. Nature, 444(7116), Hesse, T., Butzin, M., Bickert, T., & Lohmann, G. (2011).
195–198. https://doi.org/10.1038/nature05301. A model-data comparison of d13C in the glacial Atlantic Ocean.
Ezat, M. M., Rasmussen, T. L., Hönisch, B., Groeneveld, J., & Paleoceanography, 26, PA3220. https://doi.org/10.1029/
deMenocal, P. (2017). Episodic release of CO2 from the 2010pa002085.
high-latitude North Atlantic Ocean during the last 135 kyr. Nature Holland, H. D. (1994). Early Life on Earth. In S. Bengston (Ed.)
Communications, 8, 14498. (pp. 237–244). New York: Columbia University Press.
Ferrari, R., Jansen, M. F., Adkins, J. F., Burke, A., Stewart, A. L., & Jaccard, S. L., & Galbraith, E. D. (2011). Large climate-driven changes
Thompson, A. F. (2014). Antarctic sea ice control on ocean of oceanic oxygen concentrations during the last deglaciation.
circulation in present and glacial climates. PNAS, 111(24), 8753– Nature Geoscience, 5, 151–156.
8758. https://doi.org/10.1073/pnas.1323922111. Jaccard, S. L., & Galbraith, E. D. (2012). Large climate-driven changes
Fischer, H., Schmitt, J., Lüthi, D., Stocker, T. F., Tschumi, T., Parekh, of oceanic oxygen concentrations during the last deglaciation.
P., et al. (2010). The role of Southern Ocean processes on orbital Nature Geoscience, 5, 151–156.
and millennial CO2 variations—A synthesis. Quaternary Science Jaccard, S., Galbraith, E., Frölicher, T., & Gruber, N. (2014). Ocean
Reviews, 29(1–2), 193–205. https://doi.org/10.1016/j.quascirev. (de)oxygenation across the last deglaciation: Insights for the future.
2009.06.007. Oceanography, 27, 26–35.
Flückiger, J., Blunier, T., Stauffer, B., Chappellaz, J., Spahni, R., Jickells, T. D., An, Z. S., Andersen, K. K., Baker, A. R., Bergametti,
Kawamura, K., Schwander, J., Stocker, T. F., & Dahl‐Jensen D. G., Brooks, N., et al. (2005). Global iron connections between
(2004). N2O and CH4 variations during the last glacial epoch: desert dust, ocean biogeochemistry, and climate. Science, 308, 67.
Insight into global processes. Global Biogeochemical Cycles, 18, https://doi.org/10.1126/science.1105959.
GB1020. https://doi.org/10.1029/2003gb002122. Jouzel, J., Masson-Delmotte, V., Cattani, O., Dreyfus, G., Falourd, S.,
Garcia, H. E., Locarnini, R. A., Boyer, T. P., Antonov, J. I., Baranova, Hoffmann, G., et al. (2007). Orbital and Millennial Antarctic
O. K., Zweng, M. M., & Johnson, D. R. (2010). World Ocean Atlas climate variability over the past 800,000 years. Science, 317(5839),
2009, Volume 3: Dissolved oxygen, apparent oxygen utilization, 793–797.
and oxygen saturation. In: S. Levitus (Ed.), NOAA Atlas NESDIS Kaplan, J. O., Folberth, G., & Hauglustaine, D. A. (2006). Role of
70, Washington, D.C.: U.S. Government Printing Office, 344 pp. methane and biogenic volatile organic compound sources in late
Gersonde, R., Crosta, X., Abelmann, A., & Armand, L. (2005). glacial and Holocene fluctuations of atmospheric methane concen-
Sea-surface temperature and sea ice distribution of the Southern trations. Global Biogeochemical Cycles, 20, GB2016. https://doi.
Ocean at the EPILOG Last Glacial Maximum—A circum-Antarctic org/10.1029/2005gb002590.
view based on siliceous microfossil records. Quat. Sci. Rev., 24, Kohfeld, K. E., & Harrison, S. P. (2001). DIRTMAP: The geological
869–896. record of dust. Earth-Science Reviews, 54, 81–114. https://doi.org/
Goosse, H., Roche, D. M., Mairesse, A., & Berger, M. (2013). 10.1016/S0012-8252(01)00042-3.
Modeling past sea ice changes. Quaternary Science Reviews, 79, Kohfeld, K. E., Quéré, C. L., Harrison, S. P., & Anderson, R. F. (2005).
191–206. https://doi.org/10.1016/j.quascirev.2013.03.011. Role of marine biology in glacial-interglacial CO2 cycles. Science,
Gottschalk, J., Skinner, L. C., Lippold, J., Vogel, H., Frank, N., 308, 74–78.
Jaccard, S. L., et al. (2016). Biological and physical controls in the Kohfeld, K. E., Graham, R. M., de Boer, A. M., Sime, L. C., Wolff, E.
Southern Ocean on past millennial-scale atmospheric CO2 changes. W., Le Quéré, C., et al. (2013). Southern Hemisphere westerly wind
Nature Communications, 7, 11539. changes during the Last Glacial Maximum: Paleo-data synthesis.
Grant, K. M., Rohling, E. J., Bar-Matthews, M., Ayalon, A., Quaternary Science Reviews, 68, 76–95. https://doi.org/10.1016/j.
Medina-Elizalde, M., Bronk Ramsey, C., et al. (2012). Rapid quascirev.2013.01.017.
coupling between ice volume and polar temperature over the past Köhler, P., Knorr, G., & Bard, E. (2014). Permafrost thawing as a
150 kyr. Nature, 491, 744–747. possible source of abrupt carbon release at the onset of the
Grimm, R., Maier-Reimer, E., Mikolajewicz, U., Schmiedl, G., Bølling/Allerød. Nature Communications. https://doi.org/10.1038/
Müller-Navarra, K., Adloff, F., Grant, K. M., Ziegler, M., Lourens, ncomms6520.
L. J., & Emeis, K.-C. (2015). Late glacial initiation of Holocene Köng, E., Zaragosi, S., Schneider, J. L., Garlan, T., Bachèlery, P.,
eastern Mediterranean sapropel formation. Nature Communications, Sabine, M., et al. (2017). Gravity-driven deposits in an active
6(7099), 12 pp. https://doi.org/10.1038/ncomms8099. margin (Ionian Sea) over the last 330,000 years. Geochemistry,
Hamon, N., Sepulchre, P., Lefebvre, V., & Ramstein, G. (2013). The Geophysics, Geosystems, 18(11), 4186–4210. https://doi.org/10.
role of eastern Tethys seaway closure in the Middle Miocene 1002/2017GC006950.
298 N. Bouttes et al.

Kullenberg, B. (1952). On the salinity of the water contained in marine of the Southern Hemispheric westerlies. Paleoceanography, 23,
sediments. Meddelanden från Oceanografiska institutet i Göteborg, PA4201.
21, 1–38. Menviel, L., Timmermann, A., Mouchet, A., & Timm, O. (2008b).
Lambert, F., Tagliabue, A., Shaffer, G., Lamy, F., Winckler, G., Farias, Meridional reorganizations of marine and terrestrial productivity
L., et al. (2015). Dust fluxes and iron fertilization in Holocene and during Heinrich events. Paleoceanography, 23, PA1203. https://doi.
Last Glacial Maximum climates. Geophysical Research Letters, 42, org/10.1029/2007pa001445.
6014–6023. https://doi.org/10.1002/2015GL064250. Menviel, L., Spence, P., & England, M. H. (2015). Contribution of
Landais, A., Dreyfus, G., Capron, E., Jouzel, J., Masson-Delmotte, V., enhanced Antarctic bottom water formation to Antarctic warm
Roche, D. M., et al. (2013). Two-phase change in CO2, Antarctic events and millennial-scale atmospheric CO2 increase. Earth and
temperature and global climate during termination II. Nature Planetary Science Letters, 413, 37–50. https://doi.org/10.1016/j.
Geoscience, 6, 1062–1065. https://doi.org/10.1038/NGEO1985. epsl.2014.12.050.
Legrand, M., Feniet-Saigne, C., Sattzman, E. S., Germain, C., Barkov, Menviel, L., Yu, J., Joos, F., Mouchet, A., Meissner, K. J., & England,
N. I., & Petrov, V. N. (1991). Ice-core record of oceanic emissions M. H. (2017). Poorly ventilated deep ocean at the Last Glacial
of dimethylsulphide during the last climate cycle. Nature, 350, 144– Maximum inferred from carbon isotopes: A data-model comparison
146. study. Paleoceanography, 32, 2–17. https://doi.org/10.1002/
Legrand, M., & Mayewski, P. (1997). Glaciochemistry of polar ice 2016PA003024.
cores: A review. Reviews of Geophysics, 35(3), 219–243. Milkov, A. V. (2004). Global estimates of hydrate-bound gas in marine
Levine, J. G., Wolff, E. W., Jones, A. E., Sime, L. C., Valdes, P. J., sediments: How much is really out there? Earth-Science Reviews,
Archibald, A. T., et al. (2011). Reconciling the changes in 66(3–4), 183–197. https://doi.org/10.1016/j.earscirev.2003.11.002.
atmospheric methane sources and sinks between the Last Glacial Möbius, J., Lahajnar, N., & Emeis, K. C. (2010). Diagenetic control of
Maximum and the pre-industrial era. Geophys. Res. Lett., 38, nitrogen isotope ratios in Holocene sapropels and recent sediments
L23804. https://doi.org/10.1029/2011GL049545. from the Eastern Mediterranean Sea. Biogeosciences, 7(11), 3901–
Loulergue, L., Schilt, A., Spahni, R., Masson-Delmotte, V., Blunier, T., 3914. https://doi.org/10.5194/bg-7-3901-2010.
Lemieux, B., et al. (2008). Orbital and millennial-scale features of Muglia, J., & Schmittner, A. (2015), Glacial Atlantic overturning
atmospheric CH4 over the past 800,000 years. Nature, 453, 383– increased by wind stress in climate models. Geophysical Research
386. https://doi.org/10.1038/nature06950. Letters, 42. https://doi.org/10.1002/2015GL064583.
Lüthi, D., Le Floch, M., Bereiter, B., Blunier, T., Barnola, J.-M., Murray, L. T., Mickley, L. J., Kaplan, J. O., Sofen, E. D., Pfeiffer, M.,
Siegenthaler, U., et al. (2008). High-resolution carbon dioxide & Alexander, B. (2014). Factors controlling variability in the
concentration record 650,000–800,000 years before present. Nature, oxidative capacity of the troposphere since the Last Glacial
453, 379–382. https://doi.org/10.1038/nature06949. Maximum. Atmospheric Chemistry and Physics, 14, 3589–3622.
Lourantou, A., Lavrič, J. V., Köhler, P., Barnola, J.-M., Paillard, D., https://doi.org/10.5194/acp-14-3589-2014.
Michel, E., Raynaud, D., & Chappellaz, J. (2010). Constraint of the NEEM Community Members. (2013). Eemian interglacial recon-
CO2 rise by new atmospheric carbon isotopic measurements during structed from a Greenland folded ice core. Nature, 493, 489–494.
the last deglaciation. Global Biogeochemical Cycles, 24, GB2015. https://doi.org/10.1038/nature11789.
https://doi.org/10.1029/2009gb003545. Nijenhuis, I. A., & De Lange, G. J. (2000). Geochemical constraints on
Mahowald, N. M., Scanza, R., Brahney, J., Goodale, C. L., Hess, P. G., Pliocene sapropel formation in the eastern Mediterranean. Mar.
Moore, J. K., et al. (2017). Aerosol deposition impacts on land and Geol., 163, 41–63.
ocean carbon cycles. Current Climate Change Reports, 3(1), 16–31. North Greenland Ice Core Project Members. (2004). High-resolution
Marcott, S. A., Bauska, T. K., Buizert, C., Steig, E. J., Rosen, J. L., record of Northern Hemisphere climate extending into the last
Cuffey, K. M., et al. (2014). Centennial-scale changes in the global interglacial period. Nature, 431, 147–151.
carbon cycle during the last deglaciation. Nature, 514, 616–619. Obata, A. (2007). Climate–carbon cycle model response to freshwater
https://doi.org/10.1038/nature13799. discharge into the North Atlantic. Journal of Climate, 20, 5962–
Mariotti, V. (2013). Le cycle du carbone en climat glaciaire: état 5976. https://doi.org/10.1175/2007JCLI1808.1.
moyen et variabilité, Ph.D. thesis. 252 p. O’Dowd, C. D., & de Leeuw, G. (2007). Marine aerosol production: A
Marchal, O., & Curry, W. B. (2008). On the Abyssal circulation in the review of the current knowledge. Philosophical Transactions of the
Glacial Atlantic. Journal of Physical Oceanography, 38, 2014– Royal Society A, 365(1856), 1753–1774.
2037. https://doi.org/10.1175/2008JPO3895.1. Parrenin, F., Masson-Delmotte, V., Köhler, P., Raynaud, D., Paillard,
Martinez-Ruiz, F., Kastner, M., Paytan, A., Ortega-Huertas, M., & D., Schwander, J., Barbante, C., Landais, A., Wegner, A., & Jouzel,
Bernasconi, S. (2000). Geochemical evidence for enhanced pro- J. (2013). Synchronous change of atmospheric CO2 and Antarctic
ductivity during S1 sapropel deposition in the eastern Mediter- temperature during the last deglacial warming. Science, 339, 1060.
ranean. Paleoceanography, 15(2), 200–209. https://doi.org/10. https://doi.org/10.1126/science.1226368.
1029/1999PA000419. Petit, J. R., Jouzel, J., Raynaud, D., Barkov, N. I., Barnola, J. M.,
Marzocchi, A., & Jansen, M. F. (2017). Connecting Antarctic sea ice to Basile, I., et al. (1999). Climate and atmospheric history of the past
deep-ocean circulation in modern and glacial climate simulations. 420,000 years from the Vostok Ice Core, Antarctica. Nature, 399,
Geophysical Research Letters, 44, 6286–6295. https://doi.org/10. 429–436.
1002/2017GL073936. Pickett, E. J., Harrison, S. P., Hope, G., Harle, K., Dodson, J. R., Peter
Maslin, M., Adams, J., Thomas, E., Faure, H., & Haines-Young, R. Kershaw, A., et al. (2004). Pollen-based reconstructions of
(1995). Estimating the carbon transfer between the ocean, atmo- biome distributions for Australia, Southeast Asia and the Pacific
sphere and the terrestrial biosphere since the Last Glacial (SEAPAC region) at 0, 6000 and 18,000 14C yr BP. Journal of
Maximum. Terra Nova, 7(3), 358–366. Biogeography, 31, 1381–1444. https://doi.org/10.1111/j.1365-
Matsumoto, K., & Yokoyama, Y. (2013). Atmospheric D14C reduction 2699.2004.01001.x.
in simulations of Atlantic overturning circulation shutdown, Global Plancq, J., Grossi, V., Pittet, B., Huguet, C., Rosell-Melé, A., &
Biogeochemical Cycles, 27. https://doi.org/10.1002/gbc.20035. Mattioli, E. (2015). Multi-proxy constraints on sapropel formation
Menviel, L., Timmermann, A., Mouchet, A., & Timm, O. (2008a). during the late Pliocene of central Mediterranean (southwest Sicily).
Climate and marine carbon cycle response to changes in the strength Earth and Planetary Science Letters, 420, 30–44.
23 Biogeochemical Cycles and Aerosols Over the Last Million Years 299

Prather, M. J. (2007). Lifetimes and time scales in atmospheric Schmitt, J., Schneider, R., Elsig, J., Leuenberger, D., Lourantou, A.,
chemistry. Philosophical Transactions of the Royal Society A: Chappellaz, J., Köhler, P., Joos, F., Stocker, T. F., Leuenberger, M.,
Mathematical, Physical and Engineering Sciences, 365. http://doi. & Fischer, H. (2012). Carbon isotope constraints on the deglacial
org/10.1098/rsta.2007.2040. CO2 rise from ice cores. Science, 336, 711. https://doi.org/10.1126/
Prentice, I. C., Jolly, D., & BIOME 6000 Participants. (2000). science.1217161.
Mid-Holocene and glacial-maximum vegetation geography of the Schmittner, A., & Galbraith, E. D. (2008). Glacial greenhouse-gas
northern continents and Africa. Journal of Biogeography 27, 507– fluctuations controlled by ocean circulation changes. Nature, 456
519. (7220), 373–376. https://doi.org/10.1038/nature07531.
Preunkert, S., Legrand, M., Jourdain, B., Moulin, C., Belviso, S., Schneider, R., Schmitt, J., Köhler, P., Joos, F., & Fischer, H. (2013).
Kasamatsu, N., et al. (2007). Interannual variability of dimethyl- A reconstruction of atmospheric carbon dioxide and its stable
sulfide in air and seawater and its atmospheric oxidation carbon isotopic composition from the penultimate glacial maximum
by-products (methanesulfonate and sulfate) at Dumont d’Urville, to the last glacial inception. Climate of the Past, 9, 2507–2523.
coastal Antarctica (1999–2003). Journal of Geophysical Research, https://doi.org/10.5194/cp-9-2507-2013.
112, D06306. https://doi.org/10.1029/2006JD007585. Schwander, J., & Stauffer, B. (1984). Age difference between polar ice
Preunkert, S., & Legrand, M. (2013). Towards a quasi-complete and the air trapped in its bubbles. Nature, 311, 45–47.
reconstruction of past atmospheric aerosol load and composition Shackleton, N. J. (1977). Carbon-13 in Uvegerina: Tropical rainforest
(organic and inorganic) over Europe since 1920 inferred from history and the equatorial Pacific carbonate dissolution cycles. In N.
Alpine ice cores. Climate of the Past, 9(4), 1403–1416. R. Andersen & A. Malako (Eds.), The fate of fossil fuel CO2 in the
Quiquet, A., Archibald, A. T., Friend, A. D., Chappellaz, J., Levine, oceans (pp. 401–428). New York: Plenum.
J. G., Stone, E. J., et al. (2015). The relative importance of methane Siegenthaler, U., Stocker, T. F., Monnin, E., Lüthi, D., Schwander, J.,
sources and sinks over the Last Interglacial period and into the last Stauffer, B., et al. (2005). Stable carbon cycle-climate relationship
glaciation. Quaternary Science Reviews, 112, 1–16. https://doi.org/ during the Late Pleistocene. Science, 310(5752), 1313–1317.
10.1016/j.quascirev.2015.01.004. https://doi.org/10.1126/science.1120130.
Rasmussen, S. O., Bigler, M., Blockley, S. P., Blunier, T., Buchardt, S. Sigman, D. M., & Boyle, E. A. (2000). Glacial/interglacial variations in
L., Clausen, H. B., et al. (2014). A stratigraphic framework for atmospheric carbon dioxide. Nature, 407, 859–869.
abrupt climatic changes during the last glacial period based on three Skinner, L. C., Fallon, S., Waelbroeck, C., Michel, E., & Barker, S.
synchronized Greenland ice-core records: Refining and extending (2010). Ventilation of the deep Southern Ocean and deglacial CO2
the INTIMATE event stratigraphy. Quaternary Science Reviews, rise. Science, 328(5982), 1147–1151. https://doi.org/10.1126/
106, 14–28. science.1183627.
Rhein, M., et al. (2013). Observations: Ocean. In T. F. Stocker et al. Skinner, L. C. (2009). Glacial-interglacial atmospheric CO2 change: A
(Eds.), Climate Change 2013: The Physical Science Basis. Contri- possible “standing volume” effect on deep-ocean carbon sequestra-
bution of Working Group I to the Fifth Assessment Report of the tion. Climate of the Past, 5, 537–550. https://doi.org/10.5194/cp-5-
Intergovernmental Panel on Climate Change (pp. 255–310). 537-2009.
Cambridge, UK: Cambridge University Press. Sowers, T., Alley, R. B., & Jubenville, J. (2003). Ice core records of
Roche, D. M., Crosta, X., & Renssen, H. (2012). Evaluating Southern atmospheric N2O covering the last 106,000 years. Science, 301,
Ocean sea ice for the Last Glacial Maximum and pre-industrial 945–948.
climates: PMIP-2 models and data evidences. Quaternary Science Stephens, B. B., & Keeling, R. F. (2000). The influence of Antarctic sea
Reviews, 56, 99–106. https://doi.org/10.1016/j.quascirev.2012.09. ice on glacial-interglacial CO2 variations. Nature, 404, 171–174.
020. Stramma, L., Prince, E. D., Schmidtko, S., Luo, J., Hoolihan, J. P.,
Rogerson, M., Cacho, I., Jimenez-Espejo, F., Reguera, M. I., Sierro, F. Visbeck, M., et al. (2012). Expansion of oxygen minimum zones
J., Martinez-Ruiz, F., Frigola, J., & Canals, M. (2008). A dynamic may reduce available habitat for tropical pelagic fishes. Nature
explanation for the origin of the western Mediterranean organic rich Climate Change, 2, 33–37.
layers. Geochemistry, Geophysics, Geosystems, 9, Q07U01. https:// Spahni, R., Chappellaz, J., Stocker, T. F., Loulergue, L., Hausammann,
doi.org/10.1029/2007gc001936. G., Kawamura, K., et al. (2005). Atmospheric methane and nitrous
Rohling, E. J., Marino, G., & Grant, K. M. (2015). Mediterranean oxide of the late Pleistocene from Antarctic ice cores. Science, 310,
climate and oceanography, and the periodic development of anoxic 1317–1321.
events (sapropels). Earth-Science Reviews, 143(2015), 62–97. Stocker, T. F., & Johnsen, S. J. (2003). A minimum thermodynamic
Rosen, J. L., Brook, E. J., Severinghaus, J. P., Blunier, T., Mitchell, L. model for the bipolar seesaw. Paleoceanography, 18(4), 1087.
E., Lee, J. E., et al. (2014). An ice core record of near-synchronous https://doi.org/10.1029/2003PA000920.
global climate changes at the Bølling transition. Nature Geoscience, Tagliabue, A., Bopp, L., Roche, D. M., Bouttes, N., Dutay, J.-C.,
7, 459–463. Alkama, R., et al. (2009). Quantifying the roles of ocean circulation
Rossignol-Strick, M., Nesteroff, V., Olive, P., & Vergnaud-Grazzini, C. and biogeochemistry in governing ocean carbon-13 and atmo-
(1982). After the deluge; Mediterranean stagnation and sapropel spheric carbon dioxide at the Last Glacial Maximum. Climate of the
formation. Nature, 295, 105–110. Past, 5, 695–706.
Saigne, C., & Legrand, M. (1987). Measurements of methanesulphonic Tarnocai, C., Canadell, J. G., Schuur, E. A. G., Kuhry, P., Mazhitova,
acid in Antarctic ice. Nature, 330, 240–242. G., & Zimov, S. (2009). Soil organic carbon pools in the northern
Schilt, A., Baumgartner, M., Blunier, T., Schwander, J., Spahni, R., circumpolar permafrost region. Global Biogeochemical Cycles, 23,
Fischer, H., et al. (2010). Glacial–interglacial and millennial-scale GB2023. https://doi.org/10.1029/2008gb003327.
variations in the atmospheric nitrous oxide concentration during the Toggweiler, J. R., Russell, J. L., & Carson, S. R. (2006). Midlatitude
last 800,000 years. Quaternary Science Reviews, 29, 182–192. westerlies, atmospheric CO2, and climate change during the ice
Schilt, A., Baumgartner, M., Eicher, O., Chappellaz, J., Schwander, J., ages. Paleoceanography, 21, PA2005.
Fischer, H., et al. (2013). The response of atmospheric nitrous Tschumi, J., & Stauffer, B. (2000). Reconstructing past atmospheric
oxide to climate variations during the last glacial period. Geophys- CO2 concentration based on ice-core analyses: Open questions due
ical Research Letters, 40, 1888–1893. https://doi.org/10.1002/grl. to in situ production of CO2 in the ice. Journal of Glaciology, 46
50380. (152), 45–53. https://doi.org/10.3189/172756500781833359.
300 N. Bouttes et al.

Tschumi, T., Joos, F., Gehlen, M., & Heinze, C. (2011). Deep ocean geographic controls. Journal Geophysical Research, 115,
ventilation, carbon isotopes, marine sedimentation and the deglacial D06111. https://doi.org/10.1029/2009JD012110.
CO2 rise. Climate of the Past, 7, 771–800. https://doi.org/10.5194/ Wolff, E. W., Fischer, H., Fundel, F., Ruth, U., Twarloh, B., Littot, G.
cp-7-771-2011. C., Mulvaney, R., Röthlisberger, R., de Angelis, M., Boutron, C. F.,
Vadsaria, T., Ramstein, G., Dutay, J. C., Li, L., Ayache, M., & Richon, Hansson, M., Jonsell, U., Hutterli, M. A., Lambert, F., Kaufmann,
C. (2019). Simulating the occurrence of the last sapropel event (S1): P., Stauffer, B., Stocker, T. F., Steffensen, J. P., Bigler, M.,
Mediterranean basin ocean dynamics simulations using Nd isotopic Siggaard-Andersen, M. L., Udisti, R., Becagli, S., Castellano, E.,
composition modeling. Paleoceanography and Paleoclimatology. Severi, M., Wagenbach, D., Barbante, C., Gabrielli P., & Gaspari V.
http://doi.org/10.1029/2019PA003566. (2006). Southern Ocean sea-ice extent, productivity and iron flux
Valdes, P. J., Beerling, D. J., & Johnson, C. E. (2005). The ice age over the past eight glacial cycles. Nature 440, 23. https://doi.org/10.
methane budget. Geophysical Research Letters, 32, L02704. https:// 1038/nature04614.
doi.org/10.1029/2004GL021004. Yin, Q. Z., & Berger, A. (2010). Insolation and CO2 contribution to the
WAIS Divide Project Members. (2015). Precise interpolar phasing of interglacial climate before and after the Mid-Brunhes Event. Nature
abrupt climate change during the last ice age. Nature, 520, 661–668. Geoscience, 3, 243–246. https://doi.org/10.1038/ngeo771.
https://doi.org/10.1038/nature14401. Yin, Q. Z., & Berger, A. (2012). Individual contribution of insolation
Walker, J. C. G. (1980). The oxygen cycle. In O. Hutzinger (Ed.), The and CO2 to the interglacial climates of the past 800,000 years.
natural environment and the biogeochemical cycles (pp. 87–104). Climate Dynamics, 38(3–4), 709–724.
Berlin, Heidelberg: Springer. Yu, J., Menviel, L., Jin, Z. D., Thornalley, D. J. R., Barker, S., Marino,
Wang, W., Yung, Y., Lacis, A., Mo, T., & Hansen, J. (1976). G., Rohling, E. J., Cai, Y., Zhang, F., Wang, X., Dai, Y., Chen P., &
Greenhouse effects due to man made perturbations of trace gases. Broecker W. S. (2016). Sequestration of carbon in the deep Atlantic
Science, 194(4266), 685–690. during the last glaciation. Nature Geoscience, 9, 319–324.
Weber, S. L., Drury, A. J., Toonen, W. H. J., & van Weele, M. (2010). Ziegler, M., Tuenter, E., & Lourens, L. J. (2010). The precession phase of
Wetland methane emissions during the Last Glacial Maximum the boreal summer monsoon as viewed from the eastern Mediter-
estimated from PMIP2 simulations: Climate, vegetation, and ranean (ODP Site 968). Quaternary Science Reviews, 29, 1481–1490.
 The Cryosphere and Sea Level
24
Catherine Ritz, Vincent Peyaud, Claire Waelbroeck,
and Florence Colleoni

calcite was formed, so the benthic signal must be corrected

Introduction
in order to deduce the variations in sea level.
Another approach is to try to understand the physical
Several times during the history of the Earth extensive ice
mechanisms governing the formation and evolution of these
sheets covered part of the continents. As a result, a signifi-
ice masses. In both cases, observations of the two large
cant proportion of freshwater was stored in solid form, which
remaining ice sheets, Antarctica and Greenland are pertinent.
caused a drop in sea level.
Finally, numeric simulation uses all of the information gath-
Because of their impact on other components of the Earth
ered (mechanisms, data) to develop models to calculate the
system (atmosphere, ocean, land), the dynamics of these ice
evolution of the polar ice caps as they interact with the climate.
masses must be taken into account in order to understand the
These tools (referred to later as ‘ice sheet models’) allow us to
evolution of the climate over the time scale of the last
study, for example, the role of the ice sheets in the climate
glacial-interglacial cycles. This topic can be addressed in
system, in particular the non-linear effects that can amplify the
different ways, depending to the various scientific disciplines
forcings caused by variations in the Earth’s orbital parameters.
and tools. One approach is to characterize these ice sheets
These models are also indispensable tools to assess the rise of
according to the traces they have left behind, whether on
sea levels in the context of global warming.
land or in marine records. Marine sediments contain a record
of changes in the overall volume of ice over time through
changes in the oxygen isotopic composition of calcareous
What Is an Ice Sheet?
fossils. Due to isotopic fractionation that takes place during
the evaporation of water, the drop in sea level during cold
Some definitions of glaciological terms
periods has been accompanied by an enrichment of seawater,
An ice cap is a mass of freshwater ice which rests on the
not only in salt, but also in heavy isotopes of water (water
ground. A notable difference between an ice cap and a
molecules containing the 18O isotope of oxygen rather than
mountain glacier is that the highest point of an ice cap, usually
the most widespread isotope, 16O. See Chap. 20, Volume 1).
centered on a massif, is made of ice and called a dome, while a
This enrichment leads to variations in the isotopic compo-
glacier flows down from a mountain (or from an ice cap). Ice
sition of the calcareous shells of the foraminifera preserved
caps can be of different sizes such as mountain peaks com-
in the sediments. However, the isotopic composition of
pletely covered in ice in almost all the great mountain ranges of
foraminifera also depends on the temperature at which the
the world as well as significantly larger ice caps on the Arctic
C. Ritz (&)
V. Peyaud archipelagos (Fig. 24.1), in Iceland, on the Antarctic Penin-
Institute of Engineering, Univ. Grenoble Alpes, CNRS, IRD, sula and on some islands in the Southern Ocean. Finally, when
Grenoble INP, IGE, 38000 Grenoble, France an ice mass covers the land on a continental scale it is called an
e-mail: [email protected]
ice sheet. Two large ice sheets, in Greenland and Antarctica,
C. Waelbroeck cover virtually the whole continent on which they lie. North
Laboratoire d’Océanographie et du Climat : Expérimentation et
Approches Numériques, LOCEAN/IPSL, Sorbonne Université-
America and Northern Eurasia were partially covered by such
CNRS-IRD-MNHN, UMR7159, Paris, France ice sheets during glacial periods (Fig. 24.1). Although ice caps
F. Colleoni
and ice sheets are driven by the same processes, the discussion
OGS (National Institute of Oceanography and Applied below focuses on ice sheets that are large enough to signifi-
Geophysics), Borgo Grotta Gigante 42/c, Viale Aldo Moro 44, cantly modify sea level.
34010 Sgonico (TS), Italy

© Springer Nature Switzerland AG 2021 301

G. Ramstein et al. (eds.), Paleoclimatology, Frontiers in Earth Sciences,
https://doi.org/10.1007/978-3-030-24982-3_24
302 C. Ritz et al.

Fig. 24.1 Map of the northern hemisphere during the last glacial with the ice cap over the British islands is still disputed. It can be seen
maximum. Ice caps and ice sheets are in blue (Peltier Ice4G). The ice that a good part of the Fennoscandian ice sheet extended as far as
sheet over North America is called Laurentide ice sheet. Sometimes, the Svalbard, occupying what is currently the sea of Barents. Abbreviations
ice cap over the Rocky Mountains is referred to as the Cordilleran ice Sval—Svalbard; FJL—François Joseph Land; SZ—Severnaia Zembla;
cap. The Greenland ice sheet was not much larger than at present, but El.I.—Ellesmere Island; BI—Baffin Island. From the thesis of Peyaud
was probably connected to Ellesmere Island. The ice sheet over Eurasia (2007)
was called the Fennoscandian ice sheet. The existence of a connection

When the ice from an ice sheet flows to the coast, it process, they are still not composed of freshwater. Moreover,
begins to float on the sea, forming either tongues of ice or sea ice is only about a few meters thick.
ice-shelves. Currently, the largest of these are the Ross and Finally, the edge of a sheet (terrestrial or marine) is
Ronne-Filchner ice shelves in Antarctica (see map on generally not uniform, but rather an alternation of
Fig. 24.2). The line where the ice begins to float is called the fast-moving glaciers and relatively immobile zones. These
grounding line, and we will return to this point as it is fast-moving glaciers are called outlet glaciers. Their location
important for the evolution of an ice sheet. depends on the topography of the bedrock (their flow is
The thickness of the ice shelves can reach 2000 m at the channeled into sub-glacial valleys) and on its geological
grounding line, but they generally thin off rapidly as they properties with zones with relatively rapid flow (several
reach the ocean and are about 200 m thick at the front. At hundred meters a year) being observed in the midst of rel-
the front, the ice shelves crack and break off, producing atively stagnant ice (a few meters per year). In Antarctica,
icebergs (calving phenomenon). Ice-shelves and sea ice some of these glaciers begin several hundred kilometers
should not be confused. The latter is made of seawater and, inside the ice sheet. When they are particularly wide
although some of the salt is expelled during the freezing (*40 km), they are also called ice streams.
24 The Cryosphere and Sea Level 303

Fig. 24.2 Satellite image of Antarctica (Blue Marble, NASA) with whole of the emerged continent, with the exception of mountains
localization of the main ice shelves i.e. floating ice platforms. Siple arising above the ice, for example, in the peninsula and the
coast is a region characterized by very large ice streams. Ice covers the trans-Antarctic Mountains

What Determines Sea Level? inversion of the isotopic composition diffusion profile
of the interstitial water in the sedimentary column from
It may seem easy to determine how sea level varied over time, the water-sediment interface (Schrag et al. 1996). In this
but it is in fact a difficult characteristic to estimate. Distinct way, in cores taken from the Pacific Ocean (the world’s
mechanisms intervene in these variations, involving impor- largest ocean basin), it was possible to establish that the
tant regional specificities. Since many observations are local, average enrichment of the ocean during the LGM was
the representativeness of each measurement must be esti- 1.0 ± 0.1‰ compared to today.
mated according to the place, the time and the nature of the (2) The changes in the isotopic composition of deep water
observation. Here, we identify the various processes that can be reconstructed from isotopic analyses carried out
cause sea level to vary and assess how they impact on the on the shells of benthic foraminifera after subtracting
interpretation of observations. Since we are interested in large the influence on these values of the variations in tem-
time scales, we focus only on the level averaged over several perature of the deep waters as well as the local varia-
years and so exclude the effect of tides and currents. tions in the deep water isotopic composition
Sea level depends first of all on the mass of liquid water (Waelbroeck et al. EPILOG 2002).
available, a quantity that changes over time due to its storage
as ice on the continents during cold periods. This informa- Variations in the average isotopic composition of the
tion is for the whole globe and it can be evaluated through ocean can be translated into variations in the eustatic level of
reconstructions of the average isotopic composition of the the seas. Current knowledge indicates that the relationship
ocean. Records of the isotopic composition of seawater are between these two quantities can be approximated initially
indirect and are essentially of two types: using a constant multiplicative factor, so that an enrichment
of 1‰ during the LGM corresponds to a decrease in the
(1) The isotopic composition of deep water during the Last eustatic sea level of about 130 m (Fig. 24.3). Sea level
Glacial Maximum (LGM) can be estimated by changes have also been reconstructed from isotopic analyses
304 C. Ritz et al.

Fig. 24.3 Variations in summer

sunshine levels at 65 °N and in
the eustatic level of the sea during
the last climate cycle. Eustatic
levels calculated by Lambeck and
Chappell (2001), based on
measurements of the relative sea
level. Reconstruction of d18O
changes in the mean ocean from
d18O of benthic foraminifera
(Waelbroeck et al. EPILOG 2002)

Fig. 24.4 Glacial isostasy

of the shells of planktonic foraminifera from the Red Sea, (density depends on temperature). The impact of thermal
which record the large variations in salinity resulting from the expansion on the sea level is purely local, but its overall global
closing and opening of the Bab el Mandab Straits caused by average is sometimes added to the eustatic level of the seas.
decreases and increases in sea level (Siddall et al. 2003). In addition to the eustatic sea level, it is also important to
At the glacial-interglacial scale, the changes in global ice take into account the isostatic variations which modify the
volume constitutes the most important mechanism, and can shape of the ocean basins. Isostasy is the phenomenon of
cause variations in sea level of around one hundred meters depression of the Earth’s crust beneath the weight of the ice
(*130 m between the LGM and the present). These variations sheet (on land) and of water (under the oceans) (see Fig. 24.4).
are called changes in the ‘eustatic’ level of the seas, and this This means that both the changing quantity of liquid water and
corresponds roughly to the variations in the average level over the changing shape of the basin need to be considered when
the globe (see later for the concept of implicit ice). Alterna- attempting to measure the level of the water at the edges, which
tively, for the same mass of water, the volume changes with the themselves can be affected by a vertical movement. Although
temperature of the water because of thermal expansion it is now possible to observe the average level over the whole
24 The Cryosphere and Sea Level 305

globe by satellite, for past periods, the vast majority of

observations are taken at the coasts. Variations in relative sea defined for a totally rigid Earth without any distur-
level (RSL) is tracked at particular coastal points. It is bance of the gravity field. The glacial isostasy models
important to emphasize that isostasy has a reaction time of allow us to estimate that the eustatic level during the
several thousand years and that Scandinavia, for example, is Last Glacial Maximum was 130 ± 10 m below the
still rising (at the rate of about one meter per century) following current sea level (Lambeck, et al., EPILOG 2002).
the disappearance of the ice sheet that covered northern Europe This estimate of the eustatic level is mainly based on
during the last glacial maximum. At each point, the relative sea the measurement of coral terraces in sites not affected
level is therefore a combination of eustatic (global), isostatic, by the postglacial rebound, such as the Seychelles, but
and possibly tectonic (local) variations. There are several ways also from calculations of the glacial isostasy models
to estimate this relative sea level, using traces of ancient bea- attempting to reproduce relative variations in sea level.
ches, coral analysis, speleothems etc. The sea level at a given time and place in relation to
the current sea level is defined as the relative sea level
and can be described by the following equation:
Box 1. Glacial Isostasy


RSL x; tBP ¼ Sðx; tBP Þ  S x; tP ð24:2:2Þ
Post-glacial rebound: The relative changes in sea
level are the result of tectonic processes and climate with x being the geographical coordinates, tBP, a time
changes. These latter involve the accretion or melting Before Present, et tP, the present time. Equa-
of the ice sheets on the continents (more than 5 km tion (24.2.2) describes the relative level as the differ-
thick). During ice ages, the pressure exerted by the ence in sea level between the tBP level and the current
accumulated weight of the ice sheets on the continents level. During the formation of ice sheets on the con-
modified the shape of the surface creating large tinents, the nearby regions and those further afield
depressions in the lithosphere. These changes in shape were affected differently depending on their distance
brought about disturbances of the geoid (equipotential from the ice masses and on the movements of the
surface of the Earth’s gravity field corresponding to lithosphere and the mantle and therefore presented a
the theoretical surface of the oceans) as well as vertical sea level graph reflecting the trend of the regional
and horizontal displacements of the lithosphere and ocean variations. These trends have been grouped into
the mantle. When the shape of the Earth or the volume so-called ‘Clark zones’. Figure 24.4 presents the result
of the oceans changes, the geoid is thus modified as is of modelled sea level changes due to postglacial
the sea level. When the continental ice sheets melt, the rebound following the LGM according to the ICE-5G
lithosphere is released from the weight of the ice and ice model (Peltier 2004). The graph shows changes in
the deflection is reabsorbed at a speed which depends sea level for 18,000 years. The graphs are associated
on the viscosity of the mantle. This return is called the with their respective colors on the globe and each
post-glacial rebound, post-glacial describing the per- corresponds to a different regional trend of sea level
iod from −6000 (ka BP) to today. (6 ka BP marks the variations. The pink and purple zones correspond to
end of the melting of the ice caps of the northern the regions that were either covered with ice during the
hemisphere: the Laurentide and the Eurasian ice sheets LGM or were nearby, and so were very affected by the
having completely melted). For the Last Glacial rebound of the lithosphere after the LGM. The dark
Maximum (LGM), the relaxation time of the litho- blue zone represents regions showing subsidence of
sphere is estimated at about 3000 years. the lithosphere after the LGM. The light blue zone
represents areas influenced by the late melting of ice
Sea level: Variations in sea level are caused by from Antarctica after 9 ka BP (peak observed above
changes in the geoid surface, in the Earth’s topography the current sea level).
and in ocean mass. Sea level can be broken down into
The level of the seas is an equipotential surface of the
three contributing factors: the ice load (SICE), the
Earth’s gravity field (the geoid) which is linked to the dis-
oceanic charge (SOCE) and the glacio-eustatic level
tribution of mass. This changes in the presence of an ice
(SEUS).
sheet or as a result of displacements of the asthenosphere
S ¼ SICE þ SEUS þ SOCE ð24:2:1Þ (the ductile part of the Earth’s mantle). This phenomenon
must therefore also to be taken into account in the inter-
Glacio-eustatic changes in sea level are controlled pretation of relative sea levels. Although isostasy is a con-
by the amount of ice stored on the continents. In siderable complication for the interpretation of observations,
glacial isostasy models, the glacio-eustatic sea level is this mechanism, on the other hand, it also provides
306 C. Ritz et al.

Fig. 24.5 Mechanisms involved

in the evolution of polar ice sheets

information on the masses of the different ice sheets in the

past. The extent of the ice sheet is generally estimated Surface mass balance is the difference between
through geomorphological data. The inversion of relative sea snowfall precipitation and ablation (fusion or subli-
level data using isostatic rebound models (including the mation). Redistribution by the wind can intervene as a
geoid calculation) informs us about ice cap thicknesses in the positive or negative force, depending on the location.
past (Clark et al. 2002). The accumulation zone is where the surface mass
Finally, the ice sheets often lie on a bedrock that is below balance is positive (the central regions) and the abla-
sea level. This is true for Western Antarctica currently. When tion zone is where it is negative. We often refer to
the ice sheet melts, the sea replaces the volume of ice below accumulation as the amount of snow accumulated
the waterline and the impact on the sea level is thereby before melting occurs. To take account of the density
reduced. This volume of water is called ‘implicit ice’. of snow, all these terms are generally expressed in
water equivalent and the natural unit of time is the year
(mass balance in m/year).
Mechanisms Involved in the Evolution Flow: After snow is deposited, it becomes denser and
of an Ice Sheet transforms into firn and then into ice which flows due
to the action of gravity depending on the slope of the
The evolution of an ice sheet obeys the law of conservation surface. The displacement can therefore move across
of mass. If the supply of snow is greater than the loss via subglacial mountains or along with a rising bedrock.
melting or through the calving of icebergs, the sheet will The flow can be broken down into two factors. One is
grow or expand. In addition to this basic principle, there are the deformation of the ice which is linked to its
many processes that interact with both external (atmosphere, mechanical properties. This means that it behaves like
ocean, solid Earth) and internal (for example, flow) inter- a highly viscous fluid, the viscosity of which depends
actions. In most cases, the geometry of the ice sheet is both a on the temperature, the ‘hotter’ the ice, the more
result of and a contributor to these feedbacks. The most quickly it deforms. This deformation causes a varia-
important mechanisms are outlined in Fig. 24.5. tion in the velocity depending on depth and is gener-
ally concentrated in the underlying layers. The second
is the ability of the ice to slide over the bedrock or, in
Box 2. Mechanisms Involved in the Evolution of the case of a sedimentary base, this base may deform.
Polar Ice Sheets The two processes, the sliding and the sedimentary
The geometry of an ice sheet depends on the supply of deformation, cause a horizontal velocity of the ice at
snow, the melting at the surface and the flow of ice to the base. They can only act when the ice has reached
the edges where it may eventually drain into the ocean melting point and their efficiency depends on the
in the form of icebergs. subglacial water pressure. We see that the ice flow
24 The Cryosphere and Sea Level 307

accumulation zone, the greater the amount of solar energy

depends on the field temperature, through its effect on reflected back to space instead of being used to heat the
the viscosity and on the melting point threshold at the Earth’s surface. The climate around an area of ice-cover will
base. therefore change, facilitating the expansion of the permanent
The temperature of the ice varies from the surface snow zone.
where it takes the annual average temperature (down For a given climate, the expansion of an ice sheet will
to −60 °C in Antarctica) to the melting point at the continue either until it reaches warm regions where melting
many points of interface with the bedrock. The tem- prevails over the arriving flow of ice, or until it reaches the
perature thus increases overall with depth and this is ocean and begins to float on the sea in the form of ice
mainly due to the geothermal flux which brings con- shelves, producing icebergs (see later the impact of
tinuous heat to the base of the ice (of around 50 geography).
mW/m2). The flow affects temperature by transporting It should be noted that we emphasize the impact of
cold from the top to the bottom and from the center to atmospheric temperature which governs the melting of snow
the edges (advection). The heat produced by the and ice because this climate variable predominates. Precip-
deformation and by the sliding also has an impact and itation, which governs the amount of snow accumulated,
brings the base to melting point in active regions plays an important but lesser role and explains why, at the
where speed of movement is high. same temperature, certain regions become more easily cov-
This interaction between velocity field and flow is ered in snow and ice than others. Among these regions are
called thermo-mechanical coupling. the glaciers in Norway today, but also the windward faces of
Lastly, it is important to mention isostasy, which is reliefs (mountain or ice) where precipitation is increased by
the sinking of the bedrock under the weight of the ice the orographic effect.
(see Fig. 2.4). The same mechanisms apply to the retreat and disap-
pearance of the ice caps (deglaciation) operating in the
Ice sheets have (and have had) different shapes and life-
opposite direction within the context of global warming. The
times, but they have followed a common pattern. An ice
feedbacks mentioned are still active and they tend to amplify
sheet begins to develop when, as a result of a cooling cli-
the retreat of the ice front. There are, however, two important
mate, snow becomes permanent in a region, i.e. when
differences. On the one hand, the growth rate of an ice sheet
summer melting fails to eliminate all of the winter snow. It is
is limited by the supply of snow by precipitation, whereas
usually in areas of high altitude where this process occurs
the rate of decrease is limited by the amount of heat available
first. In a mountain region, we see the emergence of glaciers
to melt the ice. It is easy to understand that the latter limi-
in places where the slope is not too steep, that is to say,
tation will not work well when the ice sheet reaches low
mainly in the high valleys. If the cooling continues, these
latitudes and that deglaciation in this case can occur much
glaciers will grow and descend until they fill up the lower
faster than glaciation. This was the case for the two large ice
valleys. When the region concerned is a high plateau,
sheets that existed during the Ice Age over America and
freezing is more abrupt with a threshold effect when the
Eurasia, and this had a strong impact on sea level. This
snow becomes perennial on the plateau. From this point on,
explains the typical saw-tooth pattern of sea level rise and
the flow of ice plays a role by transporting ice to areas where
fall during a glacial period, with a slow drop followed by an
snow is not permanent, whether by valley glaciers or by
abrupt rise (Fig. 24.3). On the other hand, the role of ice
glaciers forming on the periphery of the plateau which bring
flow is not symmetrical between freezing and deglaciation.
ice to the surrounding plain.
Although it helps the advance of the ice front as described,
From this moment on, two positive feedbacks supporting
flow tends to thin the ice layer, so that its impact on the
the ice expansion and cap formation are established. The first
volume of ice is not uniform. In particular, it may happen
of these is the altitude-temperature feedback, which is based
that this flow ‘runs away’ and that the ice sheet then
on the fact that atmospheric temperature drops with altitude.
decreases abruptly (with the help of the altitude-temperature
Areas where snow would normally not be permanent, freeze
feedback), which can also lead to a rapid deglaciation phase.
over due to the movement of ice, and this increases the
Flow is a fundamental mechanism in the evolution of an
altitude of their surface. As a result, the temperature at the
ice sheet, and we explain below that it results from inter-
surface of the ice drops and summer melting is reduced, so
action between several processes which are themselves
that eventually the snow becomes perennial. Over the years,
governed either by the major laws of physics or by laws
this mechanism gradually causes the ice front to advance
(often empirical) describing the properties of the materials
causing the accumulation zone (see Fig. 24.5) to extend
(ice of course but also the bedrock). We won’t go into the
further. The second feedback is the albedo. Because snow
detail of these equations which call upon continuum
has a very high albedo (of around 0.9), the larger the
308 C. Ritz et al.

mechanics, but we shall nevertheless indicate what these and would look like a thin sheet of ice. Capitalizing on this
laws are, and what implications they have on the flow of the aspect ratio, two separate approximations have been pro-
ice caps. In the following, we shall only deal with the case of posed: one for the part of the ice sheet resting on land
ice assumed to be incompressible. This hypothesis does not (Shallow Ice Approximation, SIA), the other for the floating
hold true for snow, whose density increases from the surface part (Shallow Shelf Approximation, see later). These
down to a maximum depth of 100 m, but in most cases, the approximations are used in ice sheet models (Ritz 2001).
mechanical effect of this layer of snow and firn is the same as Moreover, they allow a qualitative understanding of the
a layer of ice of the same weight would be, which amounts interaction between the geometry of an ice sheet and its flow.
to removing about twenty meters from the total thickness of For the resting part for example, the SIA shows that the
ice at the chosen spot (the thickness is of the order of a few velocity of the ice (averaged over its thickness) is propor-
kilometers). tional to the thickness to the power of four and to the slope
If we consider the balance of forces applied to an ice of the surface to the power of three. Although the thickness
particle, the only body force is gravity. The other forces are shows little variation over the entire ice sheet, the slope
surface forces arising from contact with other ice particles, or varies from 10−3 in the central regions to nearly 10−2 at the
with the base or water (when it is floating). The ice flows edges, thus implying a speed 1000 times greater (the size of
sufficiently slowly so that we can ignore accelerations and variation observed in reality). This also explains why the
inertial forces (Coriolis), and the balance of forces applied to thickness of an ice sheets is strongly related to their expanse
ice sheets is often referred to as a quasi-static equilibrium. In with the amount of snow accumulation being of only minor
terms of mechanical behavior and at the scale of the strain importance. When an ice sheet grows, its slope at the surface
rate occurring in ice sheets, ice is considered as a viscous increases and its drainage increases greatly, creating a neg-
fluid, i.e. the strain rate (deformation per unit of time, directly ative feedback which limits thickening. Another result of the
expressed as function of the spatial derivative of the veloci- SIA, used in the interpretation of ice cores, concerns the fact
ties) is connected to the stress. Water, for example, is also a that most of the deformation is concentrated in the layers
viscous fluid, but its viscosity (lower than that of ice) does not near the bottom, and that higher up, horizontal speed chan-
depend on the stress (nor on the strain rate). This is called ges little with depth and in a first approximation, the thin-
linear viscosity (also known as Newtonian). Ice, on the other ning of the ice layers decreases linearly with depth.
hand, is characterized by non-linear viscosity which decrea- The ice flows through deformation but its velocity at the
ses with stress according to a power law (with an exponent of interface with the bedrock (basal velocity) also contributes to
approximately 2): the more the ice deforms, the easier it is to the flow. Two processes intervene, the sliding over the
deform. This type of behavioral law is not exceptional, it is bedrock and the deformation of the underlying sediment.
also found for lava, mud and even chocolate. As with most Based on Antarctic observations, it appears that it is the
viscous materials (maintaining the analogy with chocolate), latter mechanism which is the most effective, leading to
the viscosity of ice decreases as its temperature increases (in speeds of several hundred meters per year. In both cases, the
an exponential relationship). Depending on the location of basal velocity is negligible as long as the temperature at the
the ice sheet, the temperature can vary from −50 °C at the interface is below melting point. On the other hand, at the
surface to melting point at the base, and this can influence the melting point, not only is sliding possible but water is pro-
viscosity by a factor of up to 500. duced. This brings about another mechanism, the higher the
Combining the quasi-static equilibrium, the law of vis- water pressure, the greater the basal velocity, due to both a
cous behavior, incompressibility and the various boundary lubrication effect (less friction) and the fact that the
conditions, we arrive at a system of equations which rigor- water-saturated sediment is easier to deform.
ously takes into account the mechanical equations. This The temperature field in the ice thus affects the flow in at
system (called ‘full stokes’) can be solved numerically, but least three ways: through the viscosity, through the threshold
the cost in terms of computation time is such that it is not (melting point) from which basal movement is possible and
conceivable at this time to apply this method to the entire ice through the subglacial water pressure. The temperature in the
sheet, especially when the aim is to also study its temporal ice can be estimated reasonably well by solving the heat
evolution. This approach is therefore limited to localized equation and taking into account any changes in surface
research. Fortunately, an approximation exists which takes a temperature over time. In general, temperature increases
‘thin layer’ approach making it possible to treat the ice sheet with depth. It is very cold at the surface. At around 10 m
overall. Indeed, a characteristic of ice sheets is their very depth, where seasonal variations are mitigated, it has the
small aspect ratio, i.e. the ratio of thickness to expanse. For value of the mean annual temperature. In Antarctica, for
Antarctica, for example, the thickness is around 3 km and example, the temperature at 10 m varies from about −20 °C
the expanse is 3000 km (ratio of 1/1 000). If it were made at the coast to −60 °C in the center. At the base of the ice,
into a scale model 3 m wide, it would only be 3 mm thick the temperature is often close to melting point because
24 The Cryosphere and Sea Level 309

geothermal energy (coming from the Earth) brings heat The geometry and dynamics of ice streams are balanced by
continuously to the base of the ice and the ice acts as an this buttress but if an ice shelf disintegrates (for example,
insulating material preventing this heat from escaping into through the effect of oceanic heat melting it from below), the
the atmosphere. The thicker the ice, the higher the basal drainage of all the tributary ice streams can be greatly
temperature. This temperature also depends on the amount increased. This phenomenon occurred a few years ago in the
of geothermal energy, which unfortunately is not well Antarctic Peninsula after the disintegration of the Larsen ice
understood underneath the current ice sheets where it is shelf. The question arises of how much it might have con-
difficult to measure directly. The variation in temperature tributed to glacial variations in the past and whether or not it
with depth is not linear (this would be pure diffusion), offers an alternative explanation to the IRD layers found in
because the flow transports cold from the top to the bottom marine cores and which testify to the arrival of icebergs.
and from upstream to downstream (advection). Moreover, To mechanically model the floating part, we can use the
the deformation of the ice produces heat (as it does with all shallow shelf approximation to determine a relationship
materials). The same is true for the sliding on the bedrock between the extension rate (variations in speed along the
and for the deformation of the sediment. Areas with rapid length of the flow) and the thickness of the ice shelf (at a
flow have a relatively warm base. If we ignore the phe- specific power determined by the authors). If the ice-shelf is
nomena related to flow (diffusive case), temperature would not confined (if it does not encounter resistance from the
increase linearly with depth. Because of the flow, the ice is coasts or an island), then its speed increases from the
actually colder overall at the top and warmer at the base. grounding line to the front where it can exceed one kilometer
Finally, if the base is at melting point, the additional heat is per year. Above all, this approximation makes it possible to
used for melting. It should be noted that the melting point demonstrate that the grounding line cannot be stable if the
drops with the ice pressure and therefore the depth. Below base is below sea level and with a reverse slope (if it goes
3000 m of ice, the melting point is around −2.2 °C. upwards as one goes from the center to the edge of the ice
It can be seen that the temperature and velocity variables sheet, Schoof 2007). This instability can be explained
are connected by several terms. This interdependence is qualitatively as follows: if the ice flow at the grounding line
called ‘thermomechanical coupling’, and this coupling leads increases strongly as the ice thickens, any retreat of the
to a positive feedback which can have important effects on grounding line, caused, for example, by the disintegration of
the evolution of the ice sheet. Suppose, for example, that an the corresponding ice shelf, will intensify drainage, leading
ice sheet is growing. Its basal temperature increases due to to a further retreat of the grounding line. This result is par-
the insulating effect of the thickness of the ice, which ticularly important because it indicates that some ice sheets
reduces its viscosity and favors deformation. This produces are inherently unstable because the location of the grounding
heat which in turn increases the temperature. This positive line determines the extent of the cap and its volume. This is
feedback loop will quickly bring the base to the melting especially the case for Western Antarctica, but also for some
point where the water produced introduces a second feed- parts of East Antarctica. For ice sheets in previous ice ages,
back (the more water there is, the faster the basal velocity it is possible that this instability could have contributed to
and the more heat is produced which in turn melts the ice). rapid deglaciation, a process which is not yet fully
These feedbacks have been suggested as explanations for understood.
armadas of icebergs recorded as layers of ice rafted debris Finally, subglacial isostasy (see Fig. 24.4) is a mecha-
(IRD) in marine sediments during glacial periods (see nism which plays an important role because it modulates
Chap. 20). It is clear that this mechanism is involved in many of the others. For example, subsidence of the Earth’s
triggering a rapid flow. On the other hand, the most realistic crust under the weight of the ice does not directly change its
numerical models have difficulties to simulate the return to a thickness but it alters the altitude of the surface allowing
slow phase suggested by observations of rapid climate more intense melting since the surface is lower. Moreover,
variability during glacial periods. the slope of the surface is also decreased, which slows down
When the ice sheet ends with a marine edge, the ice the flow and tends to make the ice sheet larger. Since iso-
streams enter the sea and form ice shelves. If the bay is stasy occurs with a lag, this also leads to highly non-linear
sufficiently closed, these ice shelves limit the flow of the effects. The position of the grounding line is another
streams from which they originate or which flow into the example of an element that is very sensitive to isostasy
same bay. This is called a buttressing effect. This effect because it is defined by floating (on the marine side), and
occurs at all levels, from small bays of a few kilometers to this is determined by the relative sea level, itself affected by
large ice shelves such as the Ross and Ronne (see Fig. 24.2). isostasy.
310 C. Ritz et al.

Reconstruction of Sea Levels and Ice Sheets Ocean. All these marks indicate the direction and route of
in the Past the ice flow locally. In some cases, a change in direction
over time has been observed and this information makes it
Over the past several millions of years, the Earth has seen a possible to infer variations in the geometry of the ice sheet,
succession of glacial periods during which huge ice sheets in particular the displacement of the domes.
covered North America, Eurasia and many mountain ranges The deposition of transported material leads to the for-
around the world (including the Alps). These long glacial mation of a wide variety of moraines. We note the frontal
periods were interspersed with interglacials, as is the case in moraines which mark the maximum extent of each advance
the present day, when a warm climate prevailed, confining of a glacier. Ground moraines are not very thick (a few meters
freezing to the polar regions and to the highest mountains. on average) and may be flat or irregular. Drumlins, often
These evolutions of the cryosphere have been studied grouped in fields, are elongated ovoidal hills. All these
since the nineteenth century, but a global understanding of deposits are structured by the flow of ice, and thus indicate its
these phenomena has only become possible in the past few direction. During periods of retreat, a sub-glacial hydrologi-
decades, thanks to direct observation of the current ice sheets cal network is formed if the base is temperate. The deposits
and to the discovery of the paleoclimatic information con- associated with these phases are formed when the materials
tained in numerous sedimentary deposits. transported are abandoned, e.g. eskers at the site of former
sub-glacial canals. An esker occurs in the form of an elon-
gated ridge sometimes over hundreds of meters in length.
Data that Enable the Reconstruction These ridges are formed by materials being deposited in the
of the Geography of Ancient Ice Sheets tunnels of the subglacial rivers located at the base of the
glaciers. Their often winding shape follows that of the tun-
Reconstruction of sea levels using the 18O/16O ratio of the nels that created them. All these traces left behind on the
ocean provides a representation of how the amount of ice ground have been observed and compiled since the nine-
evolved over time (Fig. 24.3), which acts as a good indicator teenth century. For about twenty years, satellites have helped
of the overall volume of all of the ice sheets. However, to to give a large scale view of these lineations. Traces some-
know where these ice sheets were located (America, Eurasia, times indicate a multitude of contradictory directions that
etc.) and how far they extended, we must rely on geomor- reflect changes in the flow of the ice over time. The traces
phological information which we shall briefly review. Fig- must therefore be classified chronologically and then inter-
ure 24.1 (Last Glacial Maximum in the northern preted in large coherent sets. Based on this synthesis, it is
hemisphere) and 2.2 (Antarctic) show the different regions possible to recreate the geometry of the caps at different
mentioned. times. A major difficulty arises from the fact that an
The expanse and dynamics of ancient ice sheets can be advancing glacier moves the deposits of previous glaciers
estimated directly from the deposits and traces that they left and, through erosion, can erase previous traces. Therefore,
behind on the ground. In general, pre-glacial bedrocks are we can only reconstruct the glaciers from previous glacia-
preserved in areas where the base of the ice remained cold; tions if their traces have not been erased, so only if another
while in regions where the base of the ice reached melting more extensive glaciation has not occurred since. This is the
point (temperate base), the flow of ice and of basal water case for Fennoscandia, where the moraines of the penultimate
reshaped the bedrock. We can then distinguish between glaciation (Saalien or MIS 6, *180–140 ka BP) are still
formations due to glacial erosion and those due to sediment visible much further south than those of the Last Glacial
deposition. Glacial erosion abrades (polishes) the rock out- Maximum (see Fig. 24.6). Another difficulty with formulat-
crops and creates incisions (streaks caused by the scraping of ing an overview comes from the dating of geomorphological
transported debris). The drop in pressure downstream of the traces. For recent periods, radiocarbon dating allows very
obstacles causes freezing of the basal water and the frac- precise estimates for the past 30,000 years approximately,
turing and plucking of rocks. This passage creates ‘roches although it is essential to use organic matter. Prior to this
moutonnées’ (or ‘sheepback rocks’). On a larger scale, gla- period, which only covers from the end of the last glaciation
cial erosion hollows out valleys into troughs (U-shaped), to the present, or when there is no organic material, other
especially on coasts as fjords are formed. At sea, it is pos- techniques are necessary. Optically stimulated luminescence
sible, using multi-frequency sonars, to observe underwater (OSL) indicates how long a rock has been exposed to solar
channels that have obviously been dug out by the flow of radiation and has been used, for example, to date quartz
ice, highlighting the location of ‘paleo ice streams’ crystals in moraines throughout the last glaciation. More
(Anderson et al. EPILOG 2002). Observations of this type accurate datings have radically changed our notions of the
have been detected around the Antarctic and the Arctic history of Fennoscandia (Svendsen et al. 2004).
24 The Cryosphere and Sea Level 311

Fig. 24.6 The Eurasian ice sheet at various times in the past. Each glaciers or ice streams is particularly rapid. A tongue of ice then forms
map, centered on the Urals, represents the same region of northern before disappearing, because its low altitude makes it prone to ablation.
Eurasia but at different times. The extent of the cap was determined It can be seen that at the beginning of the Ice Age (90 ka), the ice sheet
from geomorphological data. During the Saalien (the penultimate is prominent in the East but limited towards the West over Scandinavia.
glacial maximum, age >140 ka), the ice sheet extended particularly far During the Last Glacial Maximum, the situation is reversed. Note that
towards the south reaching almost its maximum size during the between each period represented here, there were periods when it was
Quaternary. The presence of lobes (the Dnieper during the Saalien and relatively clear of ice and snow, with the disappearance of the ice sheet
the Don for the maximum expansion) should be noted. These lobes may located over the Sea of Barents and the separation into two sheets.
be the result of specific events called surges, during which the flow of Adapted from Svendsen et al. (2004)

The first major consolidated work to reconstruct the ice with the global ice volume estimates from other sources
sheets at the end of the last glaciation was published by (marine sediments, Mix and Ruddiman 1984).
Denton and Hughes (1981). Their method of reconstruction Afterwards, understanding quickly evolved into a more
involved the assumption that the profile of the ice sheets was dynamic view of ice sheets. For example, Boulton and Clark
parabolic (a hypothesis which is more or less valid for current (1990) proposed a history of the Laurentide and in particular
ice sheets but which is based on the assumption of a sta- the migration and junction of the various domes (we will
tionary state) and the use of geomorphological data to con- return to this in the section on the Laurentide). As for
strain the extent. Since moraine dating was not as accurate at Fennoscandia, Lambeck et al. (2006) present its evolution
the time as it is now, in many areas it was assumed that the throughout the last glaciation and Hughes et al. (2016)
moraines were from the Last Glacial Maximum even though propose a chronology of the deglaciation. This type of
some had been formed during previous glaciations. Their overview often requires input from several groups of
reconstructions (even the ‘minimal’ one) are therefore con- researchers and large-scale projects were established to
siderably overestimated compared to the values acknowl- estimate the extent and dynamics of ancient ice sheets
edged today. These reconstructions were also in contradiction (EPILOG 2002; QUEEN, Svendsen et al. 2004).
312 C. Ritz et al.

From the inversion of relative sea level data (RSL) and Rocky Mountains modifying the planetary waves. Ice
observations of the isostatic rebound which is still occurring, appeared between 2.75 and 2.55 Ma in Eurasia, then in
it is possible to estimate the ice load that produced these Alaska and Canada, as evidenced by debris transported by
isostatic variations, and therefore the thickness of the ice icebergs which then appeared in North Atlantic sediments
sheets. This was the approach mainly used for the Last (Shackleton et al. 1984).
Glacial Maximum and for the last deglaciation (Peltier 2004;
Clark et al. 2002; Lambeck et al. 2006).
Using all this information, it is possible to envisage a The Last Three Million Years
scenario representing the life of the ice sheets. This is the
story outlined in the next section, with a focus on the last For the past three million years, the Earth’s climate has
glacial-interglacial cycle for which there is more information oscillated between glacial and interglacial periods and most
available. It should be borne in mind, however, that although of the variations are due to the formation and subsequent
the main points are now well understood, there are still many melting of ice sheets in the northern hemisphere. In
regions and periods for which information is lacking. Antarctica they only fluctuated, with significant variations in
Western Antarctica and in the peninsula (Anderson et al.,
EPILOG 2002; Ritz et al. 2001).
The Last 50 Million Years Until *1 Ma ago, the variations in the extent and vol-
ume of the ice sheets remained moderate, with variations of
The Antarctic continent is located in the high latitudes since 60 m in sea level. The oscillations subsequently slowed
from the Late Cretaceous era (*240 Ma). However, it down and gained amplitude. About 900 ka ago, the domi-
remained free of ice until the Eocene-Oligocene transition, nant periodicity of glaciations went from 40,000 to
34 Ma ago (Chap. 6, Volume 2). Marine sediment record- 100,000 years and the amplitude of the oscillations doubled,
ings indicate an abrupt increase in benthic d18O at this date, with variations of more than 100 m in the sea level. This
suggesting that water depleted in 18O was trapped in the event is known as the Mid-Pleistocene Revolution
form of ice. The appearance of the Antarctic ice sheet is (MPR) (Raymo et al. 2006).
attributed to the opening of an oceanic corridor between The freezing-up of the northern hemisphere has varied
South America and the Antarctic Peninsula: the Drake greatly over time: not all glaciations are alike, nor are all
Passage. The organization of the Antarctic circumpolar interglacials. The recent maximum expansion of the conti-
current would have led to the thermal isolation of Antarctica. nental ice sheets is fairly well documented. However, many
However, more recently, based on modeling work, other questions remain, in particular on the marine sides of the ice
authors believe that a reduction in CO2 concentration was sheets. Below. we highlight some irregularities that are of
the cause of the cooling in Antarctica (De Conto and Pollard interest.
2003).
In Chap. 6, a figure shows a new threshold around The ice shelves in the Arctic Ocean
15 Ma, to which the permanency of the East Antarctic ice The Lomonosov Ridge is an underwater mountain range
sheet is attributed. There are indicators, however, that the ice located at a depth of 1000 m. It stretches across the Arctic
may have retreated since 3 Ma, so that at 3 Ma the ice sheet Ocean from Greenland to Siberia through the North Pole.
could have been thicker than during the Last Glacial Several striations aligned at the top of this ridge suggest the
Maximum. possibility of a gigantic ice-shelf that might have covered the
As for the northern hemisphere, scientists estimate that center of the Arctic Ocean. These undated striations are
the first freeze-ups occurred 7 Ma ago on southern Green- unlikely to have been caused by icebergs which leave more
land. Amplification of these glaciations then took random marks. They are also corroborated by marks in other
place *3 Ma ago. Several hypotheses have been proposed parts of the Arctic Ocean (Jakobsson et al. 2008).
to explain this amplification: (i) the closure of the Panama
Strait occurred at this time and modified the oceanic circu- The Stage 11 Interglacial
lation, causing warm water to travel up to the high latitudes During the Marine Isotope Stage 11, 400,000 years ago, the
of the North Atlantic, providing a considerable source of Earth experienced a long interglacial. Estimates (currently
water vapor and therefore precipitation; (ii) the reduction in under discussion) of sea level vary between 20 m above and
CO2 concentration leading to cooling; (iii) an uplift of the 8 m below the current level. If the high value is correct, it
24 The Cryosphere and Sea Level 313

would require not only a massive deglaciation of Greenland The Laurentide Ice Sheet
and Western Antarctica but also a moderate loss of ice in The Laurentide ice sheet over North America was the first to
East Antarctica. Because the orbital parameters were close to form, with ice appearing on the heights in the form of two
those of today, this interglacial is often considered to be one separate caps, one over Keewatin (itself derived from the
of the best analogues of the present time. junction of the Keewatin and Baffin caps) and the other on
Labrador (Fig. 24.7). Throughout the ice age, it underwent
Stage 6: Saale glaciation fluctuations, with the two caps joining during the colder
The penultimate ice age, called Saale glaciation in Eurasia, periods followed by growth (both in volume and area) of the
reached its maximum about 140,000 years ago. The Eur- resulting ice sheet, generally with two distinct domes.
asian ice sheet covered a large part of Eastern Europe, Conversely, during interstadials (warmer intervals during
Russia and Siberia, practically equivalent to the maximum the glaciation) it shrank and in some cases separated into two
expansion during the Quaternary in these regions (Svendsen (or even three) ice sheets. Another ice cap also developed on
et al. 2004). The cooling causing the development of this cap the Rocky Mountains, the Cordilleran ice sheet, but it took
must have been much greater than during the LGM. Its at least 60 ka BP before there was a junction between the
expansion towards the south is particularly impressive Laurentide sheet and the still undeveloped Cordilleran sheet.
because one of its lobes (the Diepner lobe, probably a The Cordilleran ice sheet only became heavily glaciated at a
transitory phenomenon) came within 500 km of the Black late stage (after 30 ka BP). This phenomenon was attributed
Sea. It appears in the southeast, that signs of maximum to the influence of the Laurentide on atmospheric circulation.
extension are found even further south than for the Saale The Laurentide needed to be high enough to affect the jet
glaciation, but it is difficult to know if these correspond to stream, which had the effect of increasing the transport of
the same ice sheet and what its age might be (see Fig. 24.6). moisture to the Cordillera.
It should be noted that despite fluctuations, a significant
amount of ice persisted over North America throughout the
Scenario of the Last Glacial-Interglacial Cycle ice age.
During the Last Glacial Maximum (21 ka BP), the area
The temporal pattern of the last glacial-interglacial cycle is covered is shown in Fig. 24.1. During the deglaciation, the
shown in Fig. 24.3. The external forcing associated with retreat of the Laurentide accelerated from 15 ka
variations in the Earth’s orbit is represented by the evolution BP. A proglacial lake (Lake Agassiz) was formed south of
of summer insolation at 65 °N latitude. The sea level indi- the ice sheet. It emptied around 8200 years by breaking its
cates the overall volume of ice. ice dam and rapidly injected a large quantity of fresh water
It is clear that the volume of ice reacts systematically to into the North Atlantic, with repercussions on the thermo-
variations in summer insolation. As it is mainly the summer haline circulation. The deglaciation in North America con-
climate that is responsible for the ablation, an orbital con- tinued up to 6 ka BP.
figuration with strong summer insolation is unfavorable to
the ice sheets. In addition, the slow freeze-ups and the rapid The Fennoscandian Ice Sheet
deglaciations are not symmetrical. The volume over the Ice appeared later on Eurasia and initially in a limited way
whole cycle is a sawtooth trend because the retreats are not on the Arctic archipelagos (Svalbard, François Joseph) and
complete and each advance (around 120, 90, 80, 60 and on the mountains of Scandinavia. It was only towards 90 ka
40 ka BP) resumes from a still glaciated situation, even more BP that a major glaciation appeared in Western Siberia.
so as the Ice Age progresses. Then, during each cold period a large sheet developed with a
However, the geographical distribution reveals great junction between the Svalbard archipelago, northern Scan-
diversity (the geographical aspects are summarized in dinavia and New Zemble (Barents and Kara seas). During
Fig. 24.6 for Fennoscandia and 2.7 for Laurentide). each interstadial, the Barents and Kara seas were free of ice
During the Eemian (125 ka BP), the eustatic sea level and often ice only remained in arctic archipelagos and on
was up to 6 m higher than currently, testifying to a strong high points in Scandinavia (Svendsen et al. 2004). This
retreat of the ice. The southern part of Greenland and behavior, which suggests a strong sensitivity to the climate,
Western Antarctica were probably much less ice-covered can be explained by several positive feedbacks:
than they are at present. However, ice core drilling indicates
that there was an ice cap in the central part of Greenland at – the terrestrial part (lying on a bedrock above sea level)
an altitude comparable to the current altitude (NGRIP was more limited in area and thickness than the Lau-
Members 2004). rentide. As a result, it is more subject to the
314 C. Ritz et al.

Fig. 24.7 Reconstruction of the

Laurentide ice sheet during the
glaciation. In general, there were
three glacial centers: the first
icecap developed on the Arctic
archipelago (Baffin), followed
quickly by another on Keewatin,
and then these two caps grew
together (then generally called the
‘Keewatin cap’). At the same
time, another cap developed on
Labrador. Around 106 ka BP, the
Labrador and Keewatin ice sheets
were still independent, separated
by the Hudson Bay. Later, they
met to form a single mass of ice
with two domes (Labrador and
Keewatin/Baffin). This ice sheet
again broke up into several small
ones around 85–74 ka BP before
finally reforming and growing
until the LGM. Image adapted
from Boulton and Clark (1990).
According to Peyaud (2006)

altitude-surface temperature feedback. This instability is Scandinavian sheet prevented the transport of moisture
often referred to as the ‘instability of small ice sheets’; towards the east.
– the Barents and Kara caps are ocean ice sheets whose The Fennoscandian ice sheet appears to have reached its
base is below sea level (as is the case for West Antarctica) maximum extent to the south between 20 and 18 ka BP. The
and this makes them sensitive to the instability of the deglaciation of the continental shelf began around 15 ka BP
marine ice sheets, which is linked to the dynamics of the and from 13 ka BP onwards the ice retreated northward to
ice flow. In addition, these ice sheets formed a glacial the archipelagoes of the Arctic, while in the south, the
barrier which retained huge lakes. These lakes tempered Scandinavian cap receded to the Gulf of Bothnia and the
the Siberian summers and reduced the melting of the Finnish border. After 10,000 years BP, the ice was confined
glacial ice sheets, another positive feedback, but occur- to the Norwegian mountains and at the climate optimum of
ring during the cold period unlike the previous two. the Holocene, the ice cover was probably smaller than at
present.
Another interesting point concerns the location of the ice
sheet during the glacial period. During the LGM, the Antarctica
Fennoscandian ice sheet was similar in size as it was at There is much less information available on the evolution of
90 ka BP, but located further to the west. This phenomenon the Antarctic ice sheet partly because of the harsh conditions
is not explained but an interaction between the topography (logistics, climate), partly because the variations were
of the ice sheets and atmospheric circulation is suspected. smaller and partly because most traces of the Last Glacial
For example, at the end of glaciation, the presence of the Maximum are to be found at sea. Information comes from
24 The Cryosphere and Sea Level 315

ice core drilling (mostly located on the Antarctic plateau (see leading to thinning of the ice sheet; (iv) the movements of
Chap. 10, Volume 1), marine drilling at the edges, the grounding line govern the size of the ice sheet, and the
multi-beam sonar data at sea and marks of glaciations on the larger the sheet, the thicker it becomes. The grounding line is
rocks and mountains beyond the current cap (trimlines or the sensitive to sea level (if this lowers, ice shelves are able to
limit between the glacial erosion model and the model of ground) by a purely geometric effect. In addition, the flow of
erosion by atmospheric processes). Digital modeling of the ice streams and the shift in the grounding line are linked, as a
ice cores helps to connect the various data to explain the retreat occurs in association with an acceleration. For
mechanisms involved. example, a disintegration of the ice shelves through an
For the past 3 million years, Antarctica has oscillated acceleration of the flow may force a retreat of the grounding
between two (or even three) states, with the East Antarctic line. Conversely, a retreat of the grounding line reduces the
and West Antarctic ice sheets behaving differently (oppo- basal friction of the ice stream (because it begins to float)
sitely in terms of volume). Four processes govern this evo- and allows it to accelerate. It should be noted that the pre-
lution: (i) the surface temperature has little immediate cipitation effect acts in opposition to the others and tends to
influence. Indeed, since the maximum temperature on the produce a smaller ice sheet during the glacial period.
coasts is about −10 °C, ablation is not a significant mech- In the interglacial period, conditions are roughly similar
anism except on ice shelves. In the long term, climate to the current ones. Compared with the present, the glacial
temperature variations propagate through the ice and even- period has lower temperatures, less precipitation, a lower sea
tually reach the base of the glacier, where deformation and level, and a colder ocean which help to maintain the ice
sliding are concentrated. The flow is then intensified but it shelves. The grounding line then advances, now limited only
takes about ten thousand years before this process is felt; by the continental slope as shown by the grounding line map
(ii) precipitation has an immediate effect: the more it snows, proposed by Bentley et al. (RAISED 2014) (Fig. 24.8). It is
the more the ice sheet grows. In general, precipitation is the amplitude of the variation in the grounding line that
deemed to be related to atmospheric temperature. Therefore, makes the difference between the East and West ice sheets.
it snows more during warmer periods because the air can Recent modeling results (Pollard and DeConto 2009)
then hold more moisture and this is borne out (at least on the indicate that a third state, even more free of ice than the
Antarctic plateau) by ice core analysis; (iii) flow brings the current one, may occur during particularly long and warm
ice towards the coast, is sensitive to basal conditions and to interglacials. In this third state, West Antarctica virtually
the thermomechanical coupling mentioned above, but the disappeared and this could explain the high sea level at some
main feature of Antarctica is the presence of ice shelves that times in the past (Eemien, MIS11).
act as a buttress limiting the speed of the ice streams. If these Around East Antarctica, the continental slope is located a
ice shelves disappear, the upstream glaciers will accelerate few tens of kilometers from the current coast (see Fig. 24.8).

Fig. 24.8 Position of the

grounding line in Antarctica.
Current (red line) and during the
last glacial (white lines). Note that
during the glacial period this line
is located near the continental
slope and that the ice sheet has
spread over most of the
continental shelf (in light blue).
Figure Quantarctica map, from
Bentley et al. RAISED (2014)
316 C. Ritz et al.

This is the only margin of advance of the sheet. During progressively throughout the glacial period. As for the last
glaciation, the East Antarctic ice sheet is therefore slightly deglaciation, it appears to have occurred late (15 ka BP), but
thicker at the edges but this effect (confirmed by glaciation its speed depends on the model used (Huybrechts EPILOG
tracks on the mountains) remains confined to the edges. The 2002; Ritz et al. 2001), as well as on the volume of
central regions are thinner due to lower accumulation. It is Antarctica ice during the glacial period. The speed of
estimated that the Antarctic shelf was about 100 m lower deglaciation depends on the model, but all agree that it has
during the Last Glacial Maximum (Ritz et al. 2001). As for only just finished, which is in accordance with observations
volume, the impact of the central regions is stronger and showing that the retreat of the grounding line in the Ross Sea
overall the volume of East Antarctica is less in glacial period continued until 3000 years BP (Conway et al. 1999).
than in interglacial.
West Antarctica has much more room to spread out Greenland
(Fig. 24.8). Indeed, the two large embayments in which the In terms of process, Greenland is an intermediate ice sheet.
ice shelves of Ross and Ronne-Filchner are located are During the glacial periods, ablation was negligible there and
shallow. its expanse was limited by the movement of the grounding
During glacial periods, these ice-shelves are grounded, line, in other words, it was limited by the continental slope.
advancing the grounding line by about one thousand kilo- During interglacials, ablation plays an important role, as
meters. Here the effect of the grounding line far outweighs evidenced by the fact that its current edge is mostly terres-
that of precipitation. This is why West Antarctica is more trial (there can be no coastline without ablation). This makes
voluminous during glacial periods. However, there is still Greenland sensitive to warming climates and explains why it
much debate about the exact volume. Estimates for the total is assumed that this cap was significantly smaller during the
contribution of the Antarctic to sea-level between the glacial Eemian. In terms of shape, Greenland changed from domed
maximum and now range from 7 m to over 20 m (sea level during interglacial (high and narrow) to flat and expanded
equivalent). These extremes correspond to two possible during glacial periods.
scenarios in West Antarctica during glaciation. For the same
grounding line position, it is possible to have relatively
stagnant ice streams and a very thick ice sheet at the edges, Conclusions
similar to the edges of East Antarctica currently. However, it
could be that regions where there are ice shelves currently The climate system is complex and the long-term component
were very active, formed from huge ice streams comparable which includes sea level and the evolution of the polar ice
to the Sipple Coast in West Antarctica. These regions must sheets is no exception to this rule. In this chapter, we have
have been relatively flat, which had repercussions as far as seen that the mechanisms involved can be internal to the ice
the center of West Antarctica. The first hypothesis was long sheets such as thermomechanical coupling and feedbacks
supported by the trimlines which indicated altitudes at least related to subglacial hydrology. However, most of the pro-
1000 m above the current level, but recent datings of these cesses are related to interactions with other elements of the
trimlines have in many places indicated that they are more system:
than one million years old and therefore do not concern the
Last Glacial Maximum (Bentley et al. 2010). Alternatively, – the atmosphere, due to the altitude-surface temperature
the second hypothesis is supported by the ice core analysis link which causes the instability of small ice sheets. The
from the Sipple Dome (the Ross ice-shelf slope) and Berkner influence of the ice sheets on the general circulation is
Island (middle of the Ronne-Filchner ice-shelf). also noted and we have seen that it could explain the
For the time being, numerical modeling is the only tool to interactions between the ice sheets;
evaluate the evolution of Antarctic geometry during the – the ocean, since the ice sheets determine sea level, but
recent glacial-interglacial cycles. Only a few polar cap also the oceanic general circulation. In the other direction,
models have tackled this problem because simulation of the the movement of the grounding line, which is the key
movements of the grounding line remains a major difficulty. process of the Antarctic evolution, results from interac-
It should be noted that the mechanisms presented (which are tion between glacial dynamics and the local ocean;
included in the models) accurately reproduce the evolutions – the solid Earth, through the mechanism of isostasy.
described above, that not all interglacials were similar (some
were less frozen, like stage 11 or Eemian, others were The data available for the past, in terms of sea level or
intermediate between glacial and interglacial periods) and through glacio-geomorphology reconstructions, indicate that
that the freezing-up process is slow, the volume increasing all the mechanisms mentioned above are indeed active.
24 The Cryosphere and Sea Level 317

Moreover, numerical modeling makes it possible to simulate Denton, G. H., & Hughes, T. J. (1981). The last great ice sheets. Wiley.
most of the recorded evolutions. However, the speed and EPILOG. (2002). Quaternary Science Reviews, volume 21 (tous les
articles dont Anderson, J. B. et al., pp. 49–70, Huybrechts, P.,
amplitude of certain events remain hard to explain: pp. 203–231, Lambeck, K. et al., pp. 343–360, Waelbroeck, C.
et al., pp. 295–305).
– During the deglaciation, there were periods of very rapid Hughes, A. L. C., Gyllencreutz, R., Lohne, S., Mangerud, J., &
rise in sea level, of around 5 cm per year, for several Svendsen, J. I. (2016). The last Eurasian ice sheets—A chronolog-
ical database and time-slice reconstruction, DATED-1. Boreas, 45,
centuries. This occurred in particular around 14,200 ka 1–45. https://doi.org/10.1111/bor.12142.
BP with an event called ‘melt water pulse 1 A’, the cause Jakobsson, M., Polyak, L., Edwards, M., Kleman, J., & Coakley, B.
of which (Laurentide or Antarctica) is still controversial. (2008). Glacial geomorphology of the Central Arctic Ocean: The
– The Heinrich events during which armadas of icebergs chukchi borderland and the Lomorosov Ridge. Earth Surface
Processes and Landforms, 33, 526–545.
invaded the North Atlantic. Purely glacial mechanisms Lambeck, K., & Chappell, J. (2001). Sea level change through the last
(thermo-mechanical coupling) have been suggested but glacial cycle. Science, 292, 679–686.
this is difficult to reproduce correctly by the 3D ice sheet Lambeck, K., Purcell, A., Funder, S., Kjaer, K. H., Larsen, E., &
models. An interaction with the ocean that would melt an Möller, P. (2006). Constraints on the Late Saalian to early Middle
Weichselian ice sheet of Eurasia from field data and rebound
ice shelf at the mouth of the Hudson Strait has recently modelling, Boreas, 35. https://doi.org/10.1080/
been suggested and would allow a better agreement with 03009480600781875.
the ocean recordings than the previous hypothesis. Mix, A. C., & Ruddiman, W. F. (1984). Oxygen-Isotope Analyses and
Pleistocene Ice Volume. Quaternary Research, 21, 1–20.
NGRIP Members. (2004). High resolution record of Northern Hemi-
These two examples show that glacial dynamics may sphere climate extending into last interglacial period. Nature, 431,
have played a more important role than was previously 147–151.
assumed, especially when the ice sheets are subjected to Pollard, P., & DeConto, R. M. (2009). Modelling West Antarctic ice
climate forcings (ocean or atmosphere), a hypothesis which sheet growth and collapse through the past five million years.
Nature, 458, 329–332. https://doi.org/10.1038/nature07809.
is supported by the current observations of acceleration of Raymo, M. E., Lisiecki, L. E., & Nisancioglu, K. H. (2006).
outlet glaciers (in Greenland and Antarctica. For this reason, Plio-Pleistocene ice volume, Antarctic climate, and the global delta
18
research in this area of glacial dynamics is being actively O record. Science, 313(786), 492–495. https://doi.org/10.1126/
pursued in order to better assess the future behavior of the science.1123296.
Peltier, W. (2004). Global glacial isostasy and the surface of the ice-age
two remaining ice sheets, Greenland and Antarctica, in the Earth: The ICE-5G(VM2) model and GRAC. Annual Review of
context of climate change. Earth and Planetary Sciences, 32, 111–149.
Peyaud, V. (2006). Rôle de la dynamique des calottes glaciaires dans
les grands changements climatiques des périodes
glaciaires-interglaciaires. Thèse de doctorat, université Joseph
References Fourier, Grenoble 1.
Ritz, C., Rommeleare, V., & Dumas, C. (2001). Modeling the evolution
of Antarctic ice sheet over the last 420,000 years: Implications for
Bentley, M. J., Fogwill, C. J., Le Brocq, A. M., Hubbard, A. L., altitude changes in the Vostok Region. Journal of Geophysical
Sugden, D. E., Dunai, T. J., et al. (2010). Deglacial history of the Research, 106, 31, 943–31, 964.
West Antarctic ice sheet in the Weddell Sea embayment: Con- Schoof, C. (2007). Ice sheet grounding line dynamics: Steady states,
straints on past ice volume change. Geology, 38, 411–414. stability, and hysteresis. Journal of Geophysical Research, 112,
The RAISED Consortium, Bentley, M. J., et al. (2014). A commu- F03S28, https://doi.org/10.1029/2006jf000664.
nity-based geological reconstruction of Antarctic ice sheet deglacia- Schrag, D. P., Hampt, G., & Murray, D. W. (1996). Pore fluid
tion since the Last Glacial Maximum. Quaternary Science Reviews, constraints on the temperature and oxygen isotopic composition of
100, 1–9. the glacial ocean. Science, 272, 1930–1932.
Boulton, G. S., & Clark, C. D. (1990). A highly mobile Laurentide ice Siddall, M., Rohling, E. J., Almogi-Labin, A., Hemleben, C.,
sheet revealed by satellite images of glacial lineations. Nature, 346, Meischner, D., Schmelzer, I., et al. (2003). Sea-level fluctuations
813–817. during the last glacial cycle. Nature, 423, 853–858.
Clark, P. U., Mitrovica, J. X., Milne, G. A., & Tamisiea, M. E. (2002). Shackleton, N. J., et al. (1984). Oxygen isotope calibration of the onset
Sea-level fingerprinting as a direct test for the source of global of ice-rafting and history of glaciation in the North Atlantic Region.
meltwater pulse I. Science, 295, 438–441. Nature, 307, 620–623.
Conway, H., Hall, B. L., Denton, G. H., Gades, A. M., & Waddington, Svendsen, J. I., et al. (2004). Late quaternary ice sheet history of
E. D. (1999). Past and future grounding-line retreat of the West Northern Eurasia. Quaternary Science Reviews, 23, 1229–1271.
Antarctic ice sheet. Science, 286, 280–283.
De Conto, R. M., & Pollard, D. (2003). Rapid Cenozoic glaciation of
Antarctica induced by a declining atmospheric CO2. Nature, 421,
245–249.
 Modeling and Paleoclimatology
25
Masa Kageyama and Didier Paillard

development phase of a model, the first step in evaluation is

Why Develop Paleoclimate Models?
to compare it to the available observations for the current
climate and its short-term variability over time scales rang-
Today, climate models are widely discussed because of the
ing from a few years to a few decades. In recent decades,
climate predictions they produce for the next century (see
anthropogenic disruption has created previously unknown
Chap. 31), particularly when the IPCC (Intergovernmental
levels of atmospheric greenhouse gas concentrations, so
Panel on Climate Change) assessment reports are published.
models need to be tested in climate configurations different
The models used for climate prediction are developed first
to the current climate. There is no perfect paleoclimate
and foremost based on the data from observations of recent
equivalent for the forcing of anthropogenic disturbance, but
decades. Before being applied to forecasting, these models
paleoclimates, even if they are not recorded with the same
are evaluated to assess their ability to reproduce the present
precision as the present climate, offer examples of climates
climate and its recent variations over the last decades. They
very different from the current one and of transitions of
represent our understanding of the current climate and of the
varying rapidity between states. If we have correctly
mechanisms that play important roles in recent variations,
understood the climate system and if we want to use this
but also our ability to translate this understanding into digital
understanding to predict climate in the future, we must be
codes which is necessarily limited by computing capabili-
able to reproduce the variations in past climates.
ties. Thus, even though the most powerful computers are
An example of an evaluation based on paleoclimates is
currently used to produce simulations with the most complex
one dealing with entry into the last glacial period, also called
climate models, they are limited by computing power in
last glacial inception. This occurred about 115,000 years
terms of resolution and system complexity. We will return
ago, and the major cause of this disturbance, external to the
later to this crucial question of finding the best compromise
climate system, is the difference in insolation received by the
between the desired timespan for a climate simulation and
Earth. In principle, a climate model, even if it does not
the complexity of the climate model used, a question that is
include an ice sheet model, should be able to simulate
even more critical in paleoclimatology, where the time scales
perennial snow cover at the formation sites of the first ice
are much longer than for the IPCC predictions (which last a
caps located north of present-day Canada. Following various
few centuries at most).
trials, it was found that very few atmospheric general cir-
It is easy to understand why models are used for the
culation models were capable of simulating this perennial
forecasting of future climates: this approach, based on our
cover. Furthermore, those that did manage to do so were
understanding of the physics of the climate system, provides
often ones that simulated a current climate that was much
the only means of obtaining these forecasts. But modeling
colder than observations. These trials were therefore far from
can also contribute a lot to our understanding of current and
satisfactory. However, as components (ocean, vegetation)
past climates which are characterized through observations
were added one by one to the climate models, it was found
and reconstructions. A first justification is the evaluation of
that these components could play an important role in
the models used for future forecasts. During the
amplifying the initial insolation signal. Thus, over a longer
period, the disappearance of forests or the appearance of sea
M. Kageyama (&)
D. Paillard
Laboratoire des Sciences du Climat et de l’Environnement,
ice, initiated by changes in insolation, support the develop-
LSCE/IPSL, CEA-CNRS-UVSQ, Université Paris-Saclay, ment of perennial snow cover by modifying the albedo. It
91190 Gif-Sur-Yvette, France can therefore be concluded that it would be difficult to
e-mail: [email protected]

© Springer Nature Switzerland AG 2021 319

G. Ramstein et al. (eds.), Paleoclimatology, Frontiers in Earth Sciences,
https://doi.org/10.1007/978-3-030-24982-3_25
320 M. Kageyama and D. Paillard

simulate permanent snow cover using only a simple atmo- conceptual models. For each of these families of models, we
spheric model because of the absence of positive feedback give examples of their use in paleoclimatology.
mechanisms. It can be seen here that while a simple evalu-
ation can lead to a better understanding of the system, it is
also only the start of an extensive exploration of increasingly
Some Basic Modeling Concepts
complex models to find the one that best fits with the data.
Paleoclimate simulation is not only used to evaluate the
Vocabulary
models used for climate predictions over the next century.
This would be an extremely narrow application, especially
Before showing how climate modeling contributes to the
given the time scales that can be handled by these models.
study of past climates, it is useful to define the concept of a
The starting point for the development of paleoclimate
model. Indeed, this word has quite different meanings in the
modeling is the assumptions applied when interpreting the
various scientific disciplines. In general, a ‘model’ is a
data. A paleoclimate model seeks to formalize the assump-
representation of a set of scientific ideas, formulated within
tions based on the physical principles of the climate system
as rigorous a framework as possible, which explains a
and to test whether these principles can explain the observed
complete set of phenomena. The model is judged to be even
climate variations. It is clear then that the models can be
more effective when it is simple and concise, and when it
extremely varied depending on the data they are trying to
offers a maximum number of solutions. It becomes quanti-
interpret. Indeed, even though the components of the climate
tative when it is based on mathematical relationships. When
system are all interdependent, which is a defining feature of
we talk about climate modeling, we mean ‘physical’ models
this system, it is not always necessary to represent them all
of the climate system incorporating a set of mathematical
in detail in order to reproduce an observed phenomenon. In
equations that trace the evolution of the system from a
fact, it is more interesting to isolate the processes or key
starting position within boundary conditions. It is therefore a
components responsible for a phenomenon. This is one of
system of first-order differential equations, which can gen-
the approaches to paleoclimate modeling which tries to build
erally be written in the following form:
a ‘minimal’ model to explain a phenomenon. This is very
different from the models used to predict the climate of the dX ðtÞ
next century, but these approaches are important to provide a ¼ f ðX ðtÞ; tÞ ð25:1Þ
dt
better understanding of the climate system and its evolution.
Modeling can also highlight the importance of a particular where X(t) is a vector dimension N which provides an
forcing or process. By comparing experiments which include a overall description of the state of the model at each instant t,
certain process or forcing with experiments which exclude it, it and f(X, t) is a function of X and of time t which describes
is possible to study its impact and identify which mechanisms the evolution of the system. The dimension N of the vector X
explain this impact. These ‘sensitivity experiments’ are not (t) thus represents the ‘size’ of the model, which is some-
necessarily very realistic but they complement the more real- times called the ‘number of degrees of freedom’ of the
istic simulations of paleoclimates, by helping to better under- system, and the space of dimension N of all the vectors X is
stand them. One example, in Section “General Circulation called ‘the phase space’. If the state of the system is known
Models, Complex Models of the Earth System”, an attempt is at a given instant t0, denoted by X0 = X(t0) and called the
made to understand the impact of ice caps versus the impact of a ‘initial condition’, then Eq. (25.1) makes it possible to know
lower atmospheric CO2 concentration on the climate of the Last the state of the system at all times.
Glacial Maximum (LGM). In order to understand this, simu- When referring to a climate model, we may imagine a
lations are created where the ice caps of the LGM are placed in very ‘complex’ system, with a large number of degrees of
the context of the current CO2 concentration, and also where the freedom. This is often the case, but not always. Indeed, it is
CO2 concentration of the LGM is positioned with the current important to highlight two contradictory aspects of climate
ice caps. Although these simulations do not correspond to real modeling. On the one hand, modeling aims to improve our
situations, they provide a better understanding of the simulated understanding of how the system functions, and on the other
glacial climate by imposing both ice cap and CO2 concentration hand, it is trying to provide the best possible representation
forcings from the LGM. of it. In order to explore and understand what is happening
This chapter starts by presenting the basic concepts of within a system of equations, it is preferable that the number
modeling, definitions essential to our understanding of of degrees of freedom N be small. Conversely, to achieve a
models and the digital experiments used in climatology and good representation of a system as complex as the climate,
paleoclimatology. We then focus on the three main families the number of degrees of freedom N needs to be large and
of paleoclimate models: the most complex general circula- will be limited only by computing power. Although both of
tion models, climate models of intermediate complexity, and these qualify as modeling, the second case is more
25 Modeling and Paleoclimatology 321

accurately referred to as ‘climate simulation’ where the values are the model parameters. When these parameters are
primary objective is to achieve a maximum of realism, at the spatialized, i.e., dependent on their geographical location,
expense of an in-depth understanding of how the system they are then considered to be boundary conditions. When
operates. In the most sophisticated climate models, there are the parameters are time dependent, they may be referred to as
several million degrees of freedom thus making it difficult to model forcings. For example, for an atmospheric model, the
understand and analyze the detail of the sequence of pro- surface temperature of the oceans is a boundary condition
cesses involved in the simulations carried out by these (and also a forcing, if it depends on time), and the atmo-
models. A typical strategy is therefore to multiply the spheric concentration of CO2 is a parameter (and also a
number of simulations, as described later, by carrying out forcing, if it depends on time). For a coupled ocean-
sensitivity experiments. Conversely, much simpler models, atmosphere model, this same sea surface temperature is a
which may produce less realistic results, can provide insight prognostic variable while pCO2 remains a parameter. For a
into the root causes of certain mechanisms that underlie the climate-carbon coupled model, pCO2 is explicitly calculated
phenomena being represented. If the aim of modeling is and thus becomes a model variable as well. It is often inter-
summarized in the maxim ‘understand so as to better sim- esting to explore how the model outcomes change when the
ulate’, it is obvious that a whole spectrum of models of values of certain parameters change. These are called sensi-
varying complexities is necessary in order to tackle the tivity experiments, because the objective is not to perform
different aspects of a problem. realistic climate simulations, but to see how sensitive the
Before further describing climate modeling in general and model is to certain parameters (examples of experiments of
the problems encountered in paleoclimatology in particular, this type are shown in Sections “General Circulation Models,
we will revisit Eq. (25.1) in more detail in order to explain Complex Models of the Earth System” and “Examples of
some concepts that are widely used either implicitly or Long-Term Simulations and Studies of Sensitivity to Forc-
explicitly. The vector X(t) which describes the whole system ings”). When this type of study systematically includes many
is also called the prognostic variable set of the model. This parameter values and many parameters, this is called the
refers to all quantities Xi(t) in the system (25.1) possessing exploration of the parameter space of the model, and is
an equation of evolution. Moreover, it is often useful to sometimes imprecisely referred to as the exploration of ‘the
include additional variables to represent the physical quan- phase space’ (although, strictly speaking, it is the space of the
tities used in the equations, quantities which depend directly prognostic variables and not of the parameters).
on the prognostic variables Xi(t) without recourse to an
associated evolution equation. For example, the quantity y
(t) = X1(t) + X2(t) is deduced from the quantities Xi(t) and so Dynamic Systems
the evolution equation for the derivative dy(t)/dt is redundant
in the system of Eq. (25.1). These additional variables are It is also important to briefly outline the general results that
called the diagnostic variables of the model, because they can be obtained from an equation system such as system
are mainly used to provide a better understanding of the 25.1. First, the choice of functions f(X, t) must be restricted
model in terms of the customary physical values. Thus, to cases likely to have a physical meaning. Instead of starting
typically, in an atmospheric circulation model, the only from a single initial condition X0, we start with a set of
prognostic variables at each point of the grid of the model proximate initial conditions, which fill an initial volume V0
are temperature, humidity, and wind velocity on the hori- in the phase space. For the ‘physical’ cases, the second
zontal plane, with evolution equations representing the principle of thermodynamics implies that, at time t, the
conservation of energy and water (transport equations) and corresponding states X(t) fill a volume V(t) which decreases
the conservation of momentum (i.e. the Navier-Stokes with time (in the case of dissipative systems) or remains
equation) on the horizontal plane. All other values (vertical constant (in the case of conservative systems). While the
velocities, energy fluxes, precipitation, clouds etc.) are conservative systems retain the memory of the initial con-
deduced more or less directly. These are merely diagnostic dition, since the volume V(t) remains constant, this infor-
or secondary variables, but they are nevertheless very useful mation is gradually lost in dissipative systems. Indeed, in
at all the stages of modeling, from the design of the model to general, this volume tends towards zero as time t approaches
the analysis of the results. These diagnostic variables, often infinity. Climate (like many other physical systems) is a
more numerous than the prognostic variables, do not mean dissipative system. Figure 25.1 gives examples of typical
additional degrees of freedom. behaviors of a system for two different initial conditions.
Moreover, the notation of the system of Eq. (25.1) always As dissipative systems gradually forget their initial con-
involves values which are established at the outset, deeming dition, this may turn out to be positive: because this initial
these to be either physical values external to the model under information is in any case lost after a certain time, this
consideration, or more or less well defined constants. These information is not relevant to the long-term behavior of the
322 M. Kageyama and D. Paillard

Fig. 25.1 Examples of behaviors of simple dissipative dynamic equilibrium. c Convergence towards a limit cycle. After a transitional
systems. The departure points are represented by the black dots. phase, the system has a periodic oscillation. d Here, the trajectories
a Convergence of trajectories towards a single equilibrium point. converge towards a more complex object than a simple point or cycle.
b Convergence to a point of equilibrium dependent on the initial This object is called a ‘strange attractor’ (a well-known example of this
position. A ‘catchment area’ can thus be defined for each point of case is the Lorenz system, 1963)

system. We can therefore focus on the asymptotic behavior importance, since it will be quickly forgotten. It should be
alone in the model, that is to say its behavior from moment t1 noted, however, that this may not be the case for the initial
onwards, after this initial condition is forgotten. Conversely, state of other physical components of the system, such as the
the transitional phase between the beginning of the simula- ocean, which have much longer time constants.
tion at t0 and time t1, is highly dependent on the initial Nevertheless, this initial state of the atmosphere is not
condition selected, and the results will only be relevant if completely irrelevant. The discovery of chaotic systems in
this condition is correctly understood. Knowing that the the 1970s demonstrated that the ‘convergence’ of the tra-
duration t1 − t0 is a few weeks for atmospheric dynamics, jectories in a dissipative system does not necessarily mean
we see here an essential difference between a climate model that the model ‘converges’ towards a point of equilibrium
and a meteorological model. For the latter, this transitory (Fig. 25.1a), or even towards a simple trajectory such as a
phase is the most interesting. A major difficulty in weather limit cycle (e.g. periodic oscillation, Fig. 25.1c). In fact, if the
forecasting is to provide, in real time, an initial condition that volume V(t) tends to zero, this does not imply that it is a limit
represents the state of the atmosphere ‘now’ in order to be point, nor even a simple line (such as a closed curve, for an
able to anticipate its state in the hours and days to come. For oscillator). The system can eventually ‘converge’ towards
the climatologist, the initial state of the atmosphere is of little much more complex objects, known as ‘strange attractors’.
25 Modeling and Paleoclimatology 323

A classic example is the Lorenz attractor, in the shape of nature of the atmosphere, which becomes inherently unpre-
butterfly wings, which has a fractal structure (Fig. 25.1d). dictable in a short period of time. If all the boundary con-
This is the attractor of a very simple model, with only three ditions of the atmosphere (and all parameters) are considered
degrees of freedom, which was formulated by Lorenz (1963), to be either constant, or having a simple annual cyclical
to illustrate the chaotic behavior of the atmosphere. forcing, then the system will be stationary, in other words
The result is that although the initial condition is effec- the statistical variables that define the climate will be stable
tively forgotten, it is nevertheless critical to the determina- over time. Although the terms of the system of Eq. (25.1) are
tion of the ‘true trajectory’ of the system. It is so critical that never zero and the atmosphere changes endlessly, we refer to
it makes it absurd for ‘long-term’ simulations, that it, beyond the climate model as being in equilibrium. Conversely, the
time t1, to focus on a ‘single real trajectory’ for the model climate will change only when the boundary conditions of
originating from a given initial state. Chaotic systems, like the atmosphere, or some of its parameters, change over a
the atmosphere, are in fact characterized by a high depen- time period of a decade or more, in other words, a timeframe
dence on the initial conditions: a tiny difference between two compatible with the concept of climate. In a way, although a
different starting points becomes exponentially greater. It climate model aims to represent primarily atmospheric
can be shown that they are also highly dependent on the variables, it is only the slower physical components, other
parameters used. For these systems, long-term deterministic than the atmosphere, that cause the ‘climate’ of the model to
predictions simply do not make sense. It is therefore nec- evolve, i.e. to change the statistical distribution of the results.
essary to focus on a set of trajectories, not to define a single This is the case, for example, when there is a change in the
result, but instead to assemble a set of possible results. ocean, carbon cycle, ice caps. The evolution of the climate is
Indeed, only the ‘average’ trajectory is significant as it therefore only predictable if the ‘non-atmospheric’ compo-
represents a particular statistic of the attractor of the system. nents are predictable. The chaotic nature of the atmosphere
In practice, it is therefore meaningless to try to calculate does not imply that climate is unpredictable.
what the weather will be like on a given day at a given
location, beyond a few weeks into the future. Only climate
magnitudes (averages, differences, etc.) have some meaning. The Framework of a Climate Model
To calculate these climate averages, two solutions are pos-
sible. If we consider that the system is stationary, that is to Selecting Components of the Climate System:
say that its statistical features do not change over time (this Model and Boundary Conditions
would be the case, for example, in the absence of forcings),
then it would simply be a matter of averaging the model In practice, it is not possible (or even desirable) to have a
results over several decades, as is done by geographers who mathematical model that simulates all the phenomena that
use an average of weather variables over thirty years to can interact with and modify the climate, including not only
define a ‘climate’. We consider that the trajectory followed the atmosphere and the oceans, but also the terrestrial and
by the model represents the system that can be interpreted marine biosphere, biogeochemical cycles, ice cap dynamics,
statistically, as is the case for the real climate. However, hydrology and continental erosion. The first difficulty is to
when the system is subjected to a forcing, such as the current choose a relevant subsystem to define the variables of the
anthropogenic disturbance, the system can no longer be model, and that can be the object of a temporal evolution
considered ‘stationary’, and its statistical characteristics (i.e. given by a system of evolution equations, as above in (25.1).
climate values) will evolve over time. A temporal average is The other factors have to be imposed, in other words fixed as
therefore no longer relevant and several simulations need to boundary conditions or forcings. For example, when coupled
be carried out, differing only in their initial condition. This ocean-atmosphere models are used to simulate the past, the
ensemble simulation establishes the range of different tra- following will typically be imposed: (1) changes in coastli-
jectories covering the range of possibilities. nes, everything related to continental surfaces, particularly
changes in topography, ice sheet extent and height, changes
in ocean bathymetry; (2) everything related to atmospheric
Climate and Determinism concentrations, in particular greenhouse gases, but also, in
some cases, dust and other aerosols; (3) changes in the
Before further describing (paleo)climate models, certain orbital parameters of the Earth. We then examine the
paradoxes surrounding the idea of climate should be high- response of the ocean-atmosphere system to these boundary
lighted. As we have seen above, climatology conveys a conditions and forcings, either taken together in order to
statistical approach, as opposed to meteorology, which has, obtain a realistic simulation of the climate or taken sepa-
above all, a deterministic perspective. The fundamental rately to study the role of each of them individually. This is
reason for this distinction stems mainly from the chaotic referred to as a sensitivity experiment.
324 M. Kageyama and D. Paillard

Coupling of Several Components development of models known as ‘intermediate complexity’

models (see Section “Earth system models of intermediate
In general, depending on the scientific question posed and complexity (EMICS)”).
the computing power available, the first step is to identify the
part of the system which will be explicitly represented by the
model. If, for example, the intention is to represent the Comparison with Paleoclimate Data
evolution of the ice caps under the influence of the orbital
parameters of the Earth, it is essential to concentrate on a If our objective is to explain climate variations as recon-
model which explicitly represents the dynamic of the ice structed from paleoclimate records via modeling based as
caps. However, when the focus is on fast-reacting compo- much as possible on the physics, it is important to set up the
nents (ocean, atmosphere), it is acceptable to fix the model and experimental design so that this comparison is
slow-moving components by establishing these as boundary easiest. This gives rise to a second major difficulty in paleo-
limits. The converse is not true. Thus, the evolution of the climatology simulations: paleoclimate indicators (proxies)
ice caps only makes sense when its interaction with a are never clear-cut in terms of the physical variables simu-
changing climate system is considered. Herein lies one of the lated by models. These indicators are dependent on particular
main difficulties with modeling the climate over long periods climate parameters, but are these the ones being simulated? It
of time (both past and future). How can the interactions is therefore risky to rely only on the same physical models as
(exchanges of energy and matter) between physical objects those specially developed for comparison with current
with very different time constants be calculated in a mean- observations, whose important parameters can be measured
ingful way? Several strategies are possible. by oceanographers, glaciologists and meteorologists. The
A—If only a state of equilibrium in the system is of ideal situation is to be able to quantitatively compare paleo-
interest, then the equilibrium of one of the systems (for climate simulations with paleoclimate indicators. The most
example, the ice caps) can be found by fixing another (e.g. promising strategy is to explicitly simulate these indicators in
giving the atmosphere fixed boundary conditions), then by the models so that the multiple factors likely to influence them
performing the inverse operation (the atmosphere is calcu- are taken into account. For example, it is useful to explicitly
lated while the ice caps are fixed), then reiterating the process simulate the water isotopes (d18O and dD) in atmospheric and
until the results converge. This is called asynchronous cou- ocean models, and the carbon isotopes (d13C) in biogeo-
pling. In essence, the two physical components are coupled, chemical models, in order to have a more direct comparison
but in a way that does not reflect (or very inexactly) the real between the output and the measurements. An example of the
flow of time. This makes it possible to achieve equilibrium of application of one of these models is given in Chap. 29.
the coupled system over a relatively short period of time (for Another important difficulty concerns the chronology of
example, a few decades for the atmosphere), since the events. Climate models need boundary conditions and
fast-reacting component (the atmosphere is the factor which forcings (e.g. variations of insolation, concentrations of
demands the most computing time due to the fact that it has atmospheric CO2, sea level) in order to produce results such
inherent small-scale variations) is always calculated by as temperatures or precipitation. It is difficult to enter all of
assuming an equilibrium with the slow component. the forcing parameters at the same chronological scale into
B—Although asynchronous coupling can be a good the model. It is also often problematic to compare the results
method of calculating climate equilibria, it is not a very of these simulations with paleoclimate data whose time scale
rigorous way to calculate a meaningful evolution of the is not clearly defined. This is especially true when using
system. This may provide a useful approximation if the time short-term simulations, such as the results of general circu-
constants of the physical systems are very different (e.g., lation models (atmospheric or coupled ocean-atmosphere
atmosphere and ice caps), in which case it amounts to models). For example, to better understand how a glaciation
considering that the slow component is the sole driver of the starts, simulations of the atmosphere, or of the
evolution of the system with the fast component merely ocean-atmosphere system, are performed. But when is the
adapting to the slow component. If physical components start of a glaciation? If marine isotopes can be relied upon,
with ‘intermediate’ times are included (such as the ocean, the ice caps began to grow at around 120 or 122 ka BP. This
which reacts more slowly than the atmosphere but faster than moment could then be simulated by imposing to the model
the ice caps), this strategy may fail. Therefore, it is some- the insolation, greenhouse gas forcings etc. which best fit
times desirable to simplify the physics of fast-reacting this time period. In the case of an ‘equilibrium’ simulation
components (especially the atmosphere) in an attempt to that numerically integrates the atmosphere over a few dec-
calculate only the long-term variations that can be used ades, or the ocean-atmosphere system over 1000–
directly by the slower components. This has led to the 2000 years, with constant forcings, this strategy is likely to
25 Modeling and Paleoclimatology 325

be disappointing, since the forcing is ‘barely’ adequate to General Circulation Models, Complex Models
simulate the desired objective i.e. an accumulation of of the Earth System
perennial snow on continents at high northern latitudes. The
modeling strategy is then to start at 115 ka BP, when the Equations, Discretization and Parametrization:
astronomical forcing seems most favorable, to maximize Example of Atmospheric General Circulation
the response of the system. Since it is difficult to modify all Models
the boundary conditions of the model in a coherent way so
that it aligns with the situation that existed 115 ka BP ago, A natural approach to simulate the characteristics of the
the best approach is often to maintain conditions close to Earth’s climate is to look at the basic equations describing
the control situation, the ‘current’ state, or rather the the behavior of the atmosphere and the ocean. First, we will
‘pre-industrial’ state, such as sea level and greenhouse gases. describe how atmospheric general circulation models
Often, when modelers talk about experimenting with entry (AGCM) are constructed in order to represent the evolution
into glaciation using OAGCMs, this is in fact a simulation of atmospheric characteristics (temperatures, winds, precip-
with the same boundary conditions and forcings as for the itation, etc.) on a global scale. We know (see Chap. 1) that
pre-industrial period except for the insolation, which is at this scale, the atmospheric circulation is driven by the
changed to correspond to the astronomical forcing of 115 ka differential in insolation between the equator and the poles.
BP ago. The objective is not to make an ice cap ‘grow’ The fundamental equations are therefore energy conserva-
(which would require a comprehensive ice cap model and tion, supplemented by mass conservation (of air and water),
thousands of years of integration) but simply to check that momentum conservation and the law of perfect gases.
when insolation is modified, snow may accumulate in certain
locations. There is only a very distant connection between Conservation of energy:
this and the available paleoclimate observations, and there-

fore comparison with the data is not easy since the condi- DI=Dt ¼ p Dq1 =Dt þ Q ð25:2Þ
tions imposed on the numerical experiment are idealized.
where
In general, models can only represent a small part of the
global climate system with the other parts being imposed or
I is the internal energy per unit of mass (I = cpT, cp being the
ideally represented. Although their aim is to describe certain
specific heat of air at constant pressure), p is the pressure,
aspects of the problem in the most realistic way possible,
q is the density of the atmosphere,
they cannot claim to be exhaustive. Before embarking on
Q is the heating rate of the atmosphere per unit of mass,
any modeling exercise, it is essential to formulate a precise
D/Dt is the Lagrangian (Material) derivative: D=Dt ¼
hypothesis corresponding to the selected configuration.
 @t@ þ u @x
@ @
þ v @y @
þ w @z
Taking the example above of entry into glaciation, it is not
possible to ‘simulate the start of an ice age’ in all of its u, v, w being the wind components with the dimensions
aspects. However, some questions can be formulated and an x (longitude), y (latitude) and z (altitude).
attempt made to answer them. For example, for a general
atmospheric circulation model: ‘Taking a control situation Conservation of momentum:
i.e. a pre-industrial climate as a starting point, does simply Dv 1
¼ 2X  v gradð pÞ þ g þ F ð25:3Þ
changing the radiative forcing at the top of the atmosphere Dt q
without changing ocean surface temperatures (which would
require an ocean model), or vegetation (requiring a vegeta- where
tion model), or the expansion of the ice sheet (requiring an
ice sheet model), or anything else, bring about persistent v = (u, v, w) is the velocity of the wind relative to the surface
snow cover in some northern regions?’ Formulated in this of the Earth,
way, it is easier to understand the gap between this and a X is the rotational angular velocity of the Earth,
broad-ranging simulation of a glacial inception. More com- p is the atmospheric pressure,
prehensive models can answer more general questions, but g is the acceleration due to gravity,
there is no ‘all-encompassing’ model. It is therefore impor- F is the force exerted per unit of mass.
tant to correctly define the hypothesis to be tested, and to
choose the relevant model configurations to do this. Conservation of mass (of air and water):
The following sections describe the major families of Dq
climate models, from the most complex to the conceptual. ¼ q  divðvÞ þ C  E ð25:4Þ
Dt
Each section gives examples of the application of these
models to paleoclimate questions. where C is the creation rate of the species under consider-
ation, and E is its destruction rate.
326 M. Kageyama and D. Paillard

The law of perfect gases: These equations are solved for the boundary conditions
and forcings chosen by the modeler to answer the posited
p ¼ qrt ð25:5Þ questions. For the atmosphere, these are greenhouse gas
As described above, these equations are very general and concentrations, insolation (the amount of energy entering the
are valid for both small and global spatial scales. The whole art atmosphere at its summit) and surface conditions: distribu-
of the modeler involves simplifying these equations for a given tion of the different surface types (oceans, land, ice caps,
problem and expressing them in a form so that they can be different types of vegetation), orography, ocean surface
solved numerically for this problem. It is the choice of the conditions (temperature and sea ice coverage). An initial
simplifications and of the expression of the equations that state for all the prognostic variables of the model is also
makes the differences between the models. These are always chosen. From this initial state, the evolution of the atmo-
based on a set of assumptions deemed important for the sphere is calculated over the time necessary for its charac-
problem being studied. Numerical simulations are then a test teristics (temperature, precipitation, wind etc.) to be in
of our understanding of the system, expressed as a set of equilibrium with the imposed boundary conditions. It should
equations which define the numerical model. be noted here that many climate models of the ‘general
A first simplification of Eqs. (25.2)–(25.5) is often done in circulation model’ type have been developed from meteo-
current atmospheric general circulation models: the hydrostatic rological forecasting models, at least for their atmospheric
approximation. The objective of these models is to represent part. However, this does not mean that these models can
the characteristics of the troposphere, the lowest layer of the predict the weather (meteorology) on a specific date in the
atmosphere which determines the climate on the Earth’s sur- past or the future. Given the chaotic nature of the atmo-
face. This layer, which reaches altitudes from 10 km (at the sphere, it is impossible to make weather predictions further
poles) to 20 km (at the equator), is extremely thin compared to out than ten days. What we are trying to establish is a sta-
the radius of the Earth (*6400 km) and is a fine layer in tistical equilibrium for boundary conditions and for specific
which particles of air travel much further and faster in a hor- forcings, not the weather on a particular date.
izontal direction than a vertical one. From these considerations How is this done in practice? It is not possible to solve the
of scale, it can be deduced that when we consider atmospheric equations analytically, that is to say, it is not possible to
circulations with a horizontal scale much greater than the obtain general formulae describing the temporal evolution of
thickness of the troposphere, the atmosphere is close to the the prognostic variables of the system for a particular point
hydrostatic equilibrium, as described by the equation: of the troposphere. The equations are solved using numerical
methods which involve discretizing them. The state of the
Dpatm ¼ qg Dz ð25:6Þ atmosphere is described using a finite number of values
which is nevertheless large for general circulation models
where Dpatm is the difference in atmospheric pressure (around 105–106). There are many methods of discretization
between two levels separated by altitude Dz, q is the density and we will come back to this. One of the simplest ways is to
of air, g is the acceleration due to gravity. describe the state of the atmosphere using the prognostic
This direct relationship between pressure and altitude leads variables of the equations on a three-dimensional grid cov-
atmospheric specialists to often present variations in vertical ering the globe. Let X(t) be the set of these values describing
atmospheric properties as a function of pressure: for example, a the state of the atmosphere at time t. The basic unit of
pressure of 1000 hPa indicates a level close to the surface, a temporal discretization is called the ‘time step’. Starting
pressure of 500 hPa indicates the mid-troposphere and a from the initial state, describing the state of the atmosphere
pressure of 200 hPa indicates the altitude where the subtropical X0 at time t = t0, the equations enable the state of the
jet streams are most intense. It is just above this level of pressure atmosphere X to be calculated with the following t = t0 + Dt
that the transition between troposphere and stratosphere is time step:
found. The hydrostatic approximation considerably simplifies
the solution of the system of Eqs. (25.2)–(25.5), because by X ðt0 þ DtÞ ¼ X0 þ DX
judiciously choosing the vertical coordinate, the vertical speed
is diagnostically deduced from the horizontal components of DX can be obtained directly through the differential
the wind (thanks to the continuity equation). The prognostic equations chosen to describe the evolution of the atmo-
variables of the system of equations are therefore temperature sphere, which allow us to calculate DX/Dt and then, once
and humidity, and the two components of the horizontal wind. Dt is fixed, DX and X(t0 + Dt). Thus, progressing time step
All other characteristics of the atmosphere can be deduced from by time step, the evolution of the atmosphere can be cal-
these four variables. It is therefore the evolution of these four culated over a period long enough to obtain robust statistics,
variables that have to be calculated, using the fundamental allowing a simulated climate to be defined based on the
equations, simplified by the hydrostatic approximation. results of the model.
25 Modeling and Paleoclimatology 327

The time step Dt cannot be freely chosen. Obviously, the explicitly represented in current models, whose spatial res-
smaller the Dt, the longer it will take to get to the result. olution is of the order of a hundred kilometers. Therefore,
However, there is a maximum time step, equal to c Dx, the modeler will not attempt to represent the process in
where Dx is the chosen spatial resolution and c is the char- detail, but rather to represent its impact on the atmospheric
acteristic speed of propagation of the information from one characteristics at the resolution of the model. Thus, for
point to another. This is called the Courant, Friedrichs and example, each cloud is not represented individually; rather
Lewy criterion (or CFL criterion), named after the mathe- the impact of clouds on the radiative balance and on pre-
maticians who formulated it. Thus, obtaining a simulation cipitations is formulated. This is called parametrization
with a fine spatial resolution takes a long time because it of a subgrid process. These parameterizations represent
requires not only calculations to be made on more points, but simplifications of reality in the sense that we have an
also a smaller time step. A compromise must therefore be incomplete knowledge of it and the process itself is not
made between spatial resolution and time taken to produce represented but rather its impact at the relevant scale. The
the simulation. Paleoclimate studies which require long parameterizations, as well as the methods of discretization of
simulations over several hundreds or even thousands of the equations used, along with the spatial and temporal
years often used models of coarser resolution rather than resolutions, constitute the main characteristics of a model.
those used for climate forecasts into the next century. Returning to the methods of discretization of the equa-
However, nowadays, there are specific projects in which the tions governing the evolution of the state of the atmosphere,
same models are used to compare the mechanisms of past two approaches can be identified. The first is a description of
and future climate changes. the atmosphere in a finite number of points, generally
Within the atmospheric general circulation models, two organized into a three-dimensional grid. These are called
types of processes are often differentiated: dynamic pro- ‘grid-point models’ or ‘grid-box models’, with the ‘box’
cesses and physical processes. The first type deals with the referring to the smallest unit volumes of the grid. The res-
evolution of the circulation and can only be calculated from olution of the model is defined by the size of this box, or by
the three-dimensional spatial distribution of other variables, the number of points used to describe the longitudes, lati-
such as temperature. It is through the use of the dynamic laws tudes, and the number of vertical levels. There are many
[Eqs. (25.2)–(25.5)] that we can run the simulation forward, examples of grids, among which grids whose points are
time step by time step. The second type are calculated for regularly spaced in terms of longitude and latitude, and
each vertical column separately for a given time step. These grids whose points are regularly spaced in terms of longitude
are mainly radiation, clouds, precipitation and surface and the cosine of latitude. In general, the vertical levels are
exchanges. The distribution of the three-dimensional vari- not evenly distributed. In particular, they need to be closer
ables used at the dynamic stage of integration is obviously together in the boundary layer of the atmosphere, the layer
closely dependent on the evaluation of the physical processes closest to the surface.
for each vertical column. The dynamic and physical calcu- A second type of approach involves using spherical
lations are therefore carried out alternately, sometimes using harmonics to describe the variations in the atmosphere on the
different time steps. Taking the example of the atmospheric horizontal plane. The grid point method is retained for the
model included in the IPSL_CM6 model used in the vertical dimension. These ‘spectral’ methods are particu-
Sixth IPCC Assessment Report (publication planned for larly suited to the atmosphere, which forms a continuum on
2021) the time steps are from 430 s (high-resolution version the surface of a sphere. The calculations for this method are
with 50 km and 79 vertical levels) to 2 min (low-resolution faster, in particular due to the fact that the first and second
version with 300 km and 39 vertical levels) for the dynamic derivatives on the horizontal can be easily expressed for this
processes and 15 min for the physical ones. type of decomposition. The spectral models are well suited
In a general circulation model, we try to achieve the best for the representation of waves in the atmosphere with a
representation of both types of processes. Circulation is smaller number of degrees of freedom than in the grid point
calculated based on the basic laws of fluid mechanics, models. The advantages of the spectral models are, however,
expressed for the particular case of a thin atmospheric layer less significant for fine resolutions, as there are many cal-
surrounding a rotating planet. We have seen how these culations, especially for physical processes, which still have
equations can be simplified for this specific context, based to be carried out on a grid model. In general, the number of
on the characteristic scales of global atmospheric circulation. points in the grid exceeds the number of degrees of freedom
However, there is a second type of simplification inherent to in the spectral method so as to avoid problems with aliasing.
the construction of a model, and this is related to the physical These grids therefore give the impression of a finer resolu-
processes defined above. The fine details of these processes tion than the real number of degrees of freedom of the
are not always well understood. Moreover, their character- model. This is why the description of the resolution of these
istic spatial scale is often much too small for them to be models refers to the number and type of harmonics chosen.
328 M. Kageyama and D. Paillard

The oceanic general circulation models are also con- 2°), about 70 years per day; at low resolution (IPSL-CM6-
structed based on fluid dynamics equations with the addi- LR, atmosphere 144  133  79, ocean 1°) 16 years per
tional constraint of salt conservation. The discretization used day, and at medium resolution (IPSL-CM6-MR, atmosphere
is in grid point because of the geometry of the basins. The 280  280  79, ocean 1°) about 6 years per day.
specificities of oceanic general circulation models are not Thanks to the improvements in supercomputers over
detailed further here. recent decades, the development of coupled atmosphere-
ocean models, followed by atmosphere-ocean-vegetation
models has become possible. These models require a longer
Towards an ‘Integrated’ Model of the Earth computing time, not because there are many additional cal-
System culations to be performed for a given duration, but because
vegetation, and even more, the ocean, are components of the
Historically, the first climate simulations carried out with climate system whose response time is far greater that of the
general circulation models employed ‘only’ atmospheric atmosphere. While we consider that simulations using an
models. Interactions with the surface, especially with the atmospheric model alone, forced by boundary conditions
ocean, were very limited because the majority of the surface which repeat each year, must be integrated over a period of
characteristics were imposed (surface temperature of the 20 to 50 years to obtain a response from the atmosphere
oceans, presence of sea ice, surface albedo, roughness of in equilibrium with these boundary conditions, a coupled
terrain etc.). As a result, the atmospheric circulation obtained atmosphere-ocean model, in principle, needs to be integrated
was in equilibrium with these surface conditions and other over one or even several thousand years. The biggest chal-
forcings. In particular, it was then possible to evaluate the lenge then is to close-off the water and energy balances in
response of the atmosphere to changes in the ocean surface. the model to avoid a gradual drift related not to the imposed
It is clear, however, that the ocean, like vegetation and land forcing but to the model itself.
surfaces, does not remain unaffected when faced with cli-
mate change. Modelers therefore quickly sought to estimate
the impact of feedbacks from the other components of the ‘Realistic’ Modelling of Paleoclimates
climate system on the atmosphere, which in turn defines the
climate at the surface. Figure 25.2 shows the evolution of Boundary Conditions and Initial Conditions
climate models since their inception. It shows the coupling Many paleoclimate simulations aim to ‘recreate’ past cli-
first with ocean surface models, then with complete ocean mates as accurately as possible. The models can then be
circulation models. In parallel, land surface models have evaluated in the context of climates documented by paleo-
progressed from simple hydrological models, with fixed climate indicators which are different from the current cli-
albedo and surface terrain, to models including interactive mate. It also provides a better understanding of the possible
vegetation, allowing the surface characteristics to be calcu- connections between differences in climate between distant
lated according to changes in vegetation caused by changes regions and supports reconstructions by providing better
in climate or by man. Finally, models increasingly include a spatial and temporal coverage or by including a regional
representation of atmospheric chemistry and aerosols, which phenomenon not covered by the reconstructions, thereby
have a significant influence on radiation, as well as the improving our understanding of them. We will return to the
biogeochemical cycles such as the carbon cycle. In this type comparison between models and data and the value of this
of model, the atmospheric concentration of CO2 is no longer exercise at the end of the section.
imposed and is instead calculated from emissions. How can a ‘realistic’ simulation of a paleoclimate be
Climate models developed in this way require enormous achieved? First, depending on the question at stake, the part
computing power. Simulations are generally carried out on of the climate system being assessed needs to be defined. For
supercomputers adapted to this type of coding. These com- example, if we want to study the continental climate in the
puters are scarce, which explains the limited number of context of given ocean conditions, it is best to use an
general circulation models in the world. The models, their atmospheric model, possibly coupled with a dynamic land
resolution, as well as the components of the climate system surface model or vegetation model. Once the subsystem is
and the processes to be included are chosen at the outset selected, a paleoclimate simulation is carried out by
according to the issue to be addressed, but also in keeping imposing the most realistic forcings and boundary condi-
with the current limits in computing power. The perfor- tions possible on this subsystem for the simulation period.
mance of the models will vary depending on the model and Thus, the more the subsystem is constrained, the more
the computer used, but for example, the approximate com- conditions there are to be imposed, conditions that need to
putation times for the IPSL model are: run at very low res- be known for the study period. Continuing with the example
olution (IPSL-CM5A2: atmosphere 96  95  39, ocean of the simulation using only a general atmospheric
25 Modeling and Paleoclimatology 329

Fig. 25.2 History of climate

models used in successive IPCC
reports. From the 4th IPCC report
(2007), Solomon et al. (Ed.),
Cambridge University Press,
http://www.ipcc.ch/publications_
and_data/ar4/wg1/en/contents.
html

circulation model, the ocean surface temperatures and the the model itself and those related to the fixing of boundary
extent of sea ice have to be defined for all ocean-type points conditions for a specific experiment. So, this simulation
in the atmospheric model. In practice, this is extremely summarizes both our knowledge of climate characterization
restrictive since at the very least, reconstructions of ocean (incorporated into the climate model), and our understanding
surface conditions with a spatial resolution similar to the of the forcings of this climate (expressed in the forcing and
model for a typical seasonal cycle would be required. These boundary conditions). It is also constrained by the limita-
numerical experiments are almost always based on strong tions of computers and technology of its time.
assumptions as to the ocean surface conditions as well as the Take the example of the climate simulations for the Last
land-ocean distribution, topography, and extent and altitude Glacial Maximum (LGM, about 21,000 years ago). The first
of the ice sheets. The other forcings are better known, at least simulations of the climate of this period (Gates 1976) were
for the relatively recent periods of the Quaternary: insolation carried out shortly after the first reconstructions of the ocean
(Berger 1978; Laskar 2004) and atmospheric concentration surfaces (sea surface temperature, extent of sea ice) and of
of greenhouse gases, with measurements from Antarctic ice land (ice caps) were produced (CLIMAP 1976). At this time,
cores now dating as far back as 800,000 years. the LGM atmospheric concentration of CO2 was not known.
Thus, a ‘realistic’ simulation of a paleoclimate is based on It was only in the 1980s (Manabe and Broccoli 1985) that
several series of assumptions: those related to the design of the first numerical simulation was carried out taking into
330 M. Kageyama and D. Paillard

account the main forcings for the climate of the Last Glacial time frame compared to the response time of the deep ocean!
Maximum: the expanse and altitude of the ice sheets, ocean Within the international PMIP2 project, eight groups have
surface conditions and atmospheric concentration of CO2. It carried out multi-centennial coupled ocean-atmosphere
was a simulation derived from one of the most sophisticated simulations, which shows how difficult this type of experi-
stand-alone atmospheric models of that time. The simulated ment is to perform. This was confirmed in the 3rd phase of
duration was short (three months for the first simulation!) the PMIP, contemporary with CMIP5, for which 9 models
compared with current norms: theoretically, a few decades of have finally yielded results for the LGM. For PMIP4, there
simulations are required to obtain statistically robust results, are about fifteen modeling groups planning to undertake this
depending on the region and the phenomenon in question. It simulation, the future will tell us how many succeed.
must be recognized that at the time of the first paleoclimate This example of modeling of the LGM climate shows that
simulations, the duration of the simulations carried out with ‘realistic’ modeling of this climate has evolved in line with
general circulation models was greatly restricted by the cost the forcings and tools available.
of computing time. To save time, experiments were carried We have seen that the uncertainties in a ‘realistic’ sim-
out under unchanging January or July sunshine conditions, ulation of a climate are due to two types of factors: those
producing significant results for relatively short durations of related to the formulation of the model and those inherent in
simulations (typically 90 days). These results, which were a the selection of boundary conditions. The next two sections
technical feat at the time of their publication, would now show how to quantify these uncertainties, both through
probably be viewed very critically, mainly due to their short rigorous comparisons between the results of models forced
duration, even in conditions of fixed insolation. Models are by identical boundary conditions, and through sensitivity
evolving as our understanding of the climate system and studies with respect to these boundary conditions.
computer capabilities improves. Simulations therefore need
to be revised periodically in the light of these advances. The Comparing results from different models:
simulations of the first phase of the PMIP project (Paleo- Modeling Intercomparison Projects
climate Modeling Intercomparison Project, Joussaume and How can the results of different models be compared? These
Taylor 1995) using general circulation models, ran for at differences may be due to the models themselves, or to the
least ten years after an adjustment for the boundary condi- boundary conditions and forcings imposed on these models.
tions of at least one year. The results of several models can only be rigorously com-
These first simulations of the climate of the Last Glacial pared by assigning them the same boundary conditions/
Maximum used the ocean surface conditions reconstructed forcings. Such exercises have been proposed for the modeling
by the CLIMAP project (1976, 1981). These reconstructions of current climates using atmospheric general circulation
were the result of a major work of data synthesis, but models (AMIP project, Atmospheric Model Intercomparison
problems were quickly identified, particularly for the sub- Project, http://www-pcmdi.llnl.gov/projects/amip/), followed
tropical regions, where a higher temperature than is the case by coupled models, both for current and future climates
currently was reconstructed, and for the North Atlantic, (CMIP project, Coupled Model Intercomparison Project,
where the winter sea ice cover was overestimated. These http://www-pcmdi.llnl.gov/projects/cmip/). The CMIP5 exer-
problems were partially amplified by the methods used to cise corresponds to the results produced for the 5th IPCC
extrapolate the reconstructions for each site to cover the report and CMIP6, currently underway, will provide its first
globe with an even grid. The CLIMAP project, at the results for the 6th IPCC report, which will be published in
specific request of modelers, provided reconstructions for the 2022. In the same vein, PMIP (Paleoclimate Modeling
months of February and August. However, it is entirely Intercomparison Project), the project to compare paleoclimate
possible that certain species used for reconstructing SSTs models came into being in the 1990s (http://pmip.lsce.ipsl.fr).
are not particularly sensitive to these specific months, but to At first, this project involved atmospheric general circulation
other factors. Thus, manipulating data to construct boundary models (PMIP1 project, http://pmip1.lsce.ipsl.fr/) for the
conditions for models, especially atmospheric ones, can Middle Holocene (6000 years ago) and the Last Glacial
prove to be extremely restrictive for the interpretation of data Maximum (21,000 years ago). It was then extended to cou-
records. Furthermore, since they are used to establish the pled atmosphere-ocean and atmosphere-ocean-vegetation
boundary conditions, they cannot also be used to validate the models (PMIP2 project, Braconnot et al. (2007a, b), http://
model. Therefore, as soon as they became available, it was pmip2.lsce.ipsl.fr/). A new feature of PMIP3 was to use cli-
very useful to use coupled ocean-atmosphere models to mate models strictly identical to those used for CMIP5.
simulate paleoclimates. It is worth highlighting again the PMIP4 coordinates both CMIP6 simulations, which will
challenge represented by the first coupled simulations of the therefore use the same models as those used for climate
climate of the LGM. Again, in this case, the first published projections, and simulations based on other models, usually
simulation was only about thirty years long, a very short longer ones or for older climates. The PMIP4-CMIP6
25 Modeling and Paleoclimatology 331

simulations concern the following climates: the last millen- ECHAM5.3-MPIOM-127-LPJ model simulates a drop in
nium, the Middle Holocene, the Last Glacial Maximum, the temperature of between 3 and 6 °C.
last interglacial and the Middle Pliocene (Kageyama et al. This shows how models developed to first represent the
2018). current climate can diverge in their representation of cli-
The PMIP project first focused on defining precise mates different from the current one. This discrepancy is
boundary conditions and forcings for the Middle Holocene found in the forecasts of future climates, but only paleocli-
and the Last Glacial Maximum. This made it possible to mate simulations allow climate simulations different from
rigorously compare the results of the models participating in today’s climate to be compared with the data.
the project with the paleoclimate reconstructions. Below is
an example of a comparison of PMIP2 model results for Comparisons Against Paleoclimate
Europe during the Last Glacial Maximum. Figure 25.3 Reconstructions
shows the temperature of the coldest month in an average In Fig. 25.3, reconstructions of the temperature of the
seasonal cycle simulated for this period by the eight models coldest month by Wu et al. (2007) are indicated by diamond
whose results were available in the database in November shapes with the same color code as the output of the models.
2009. The differences with the current climate are shown. It should be noted that for Western Europe, all of the models
The color of the diamonds indicates the average temperature simulated temperatures that are warmer than the recon-
reconstructed from pollen data by Wu et al, (2007), on the structed temperatures. However, it is important to take into
same color scale as for the one used for the model results. In account both the dispersion of the results of the models,
this figure, it can be seen that the most noticeable cooling of which is done to a certain degree in Fig. 25.3 by including
at least 12 °C is simulated by the models in the northern part the results of all the models as well as the level of uncer-
of the area under study, on the Fennoscandinavian cap and tainty of the reconstructions, which cannot be shown on the
on the sea ice off the coast of Scandinavia. This cooling maps. Figure 25.4 compares the same model and recon-
lessens towards the south, where it is about 3 °C. Even if the struction results from another perspective. Here, we have
same boundary condition forcings are applied to the models, chosen to show the average temperature by longitude for
the climates obtained differ from one model to another. For Western (10° W–15° E) and Central Europe (15–50° E) and
example, the cooling simulated over the ice cap is between the reconstructions with their uncertainty range. This time
12 and 18 °C for the ECBILTCLIO model, whereas it is we see that the temperatures simulated by the models are
greater than 30 °C in the HadCM3M2 model. Around the compatible with the reconstructions, if we take into account
Mediterranean, the CCSM3 model simulates practically no the uncertainty characterizing the reconstructions, including
temperature change for the coldest month, whereas the for Western Europe.

Fig. 25.3 Maps: temperature of 70

the coldest month in an average 60
CCSM3 HADCM3M2
seasonal cycle, as simulated by 50
the coupled ocean-atmosphere 40
models participating in the
30
PMIP2 project (November 70
2009 PMIP2 database); 60
Diamonds: same variable, as CNRM-CM33 FGOALS-1.0g
50
reconstructed by Wu et al. (2007)
40
30
70
ECBILTCLIO 60 IPSL_CM4_MR
50
40
30
70
60
ECH-MPI-LPJ MIROC3.2.2
50
40
30
-10 0 10 20 30 40 50 -10 0 10 20 30 40 50

-36 -30 -24 -18 -12 -6 -3 0 3

332 M. Kageyama and D. Paillard

Fig. 25.4 Comparison between Mean Temperature of the Coldest Month - LGM - CTRL
the simulated temperatures of the 10
coldest month (during an average reconstructions(Wu et al, 2007)
CNRM-CM33
seasonal cycle) for the LGM in

temperature anomaly (° C)
0 HadCM3M2
Europe (continuous lines, given MIROC3.2
for different models of the PMIP2 FGOALS-1.0g
database), and reconstructed -10 IPSL-CM4-V1-MR
temperatures by Wu et al. (2007) ECHAM53-MPIOM127-LPJ
(diamond shapes and the -20 CCSM3
ECBILTCLIO
uncertainty bars associated with
them) for the Atlantic (top) and -30
Eastern (bottom) regions. The longitudes [-10:15], latitudes [30:70]
values shown are differences from -40
the reference (pre-industrial) 30 35 40 45 50 55 60 65 70
climate 10
temperature anomaly (° C)

-10

-20

-30
longitudes [15:50], latitudes [30:70]
-40
30 35 40 45 50 55 60 65 70
latitude

Ramstein et al. (2007) summarized the efforts of the years Sensitivity Experiments
2000’s to reach a congruent comparison between recon-
structions and simulations for winter temperatures during the We have seen that the uncertainties in the results of
Last Glacial Maximum. The simulations of the PMIP1 numerical models stem from the formulation of the
project, using standalone atmospheric models forced by the models themselves, since these models are built on
CLIMAP (1981) ocean surface conditions, resulted in overly assumptions considered relevant to the given problem, and
high winter temperatures compared to the initial on the conditions imposed on the model, which are
pollen-based reconstructions of Peyron et al. (1998). By themselves based on assumptions because we lack the
working on the boundary conditions (expansion of the ice necessary level of precision and spatial and temporal
caps), on the models (transition to coupled atmosphere- coverage. How can these uncertainties be calculated? One
ocean models) and on reconstructions (new reconstructions method, discussed above, is to increase the number of
by Wu et al. (2007), based on the same pollen records as models used. Similarly, if we are unsure of the boundary
used by Peyron et al. (1998), but taking into account the conditions to be imposed on the model, or if there are
effect of low levels of CO2 on vegetation) it became possible several sets of boundary conditions possible, we can carry
to reduce the large differences between simulated and out several experiments with different sets of boundary
reconstructed temperatures. It should be pointed out that for conditions so that the climate responses to these condi-
other variables (in particular the summer temperatures), the tions can be compared and we can determine whether
comparison between models and data was much more pos- these differences in boundary conditions cause differences
itive from the start. The example of the coldest month in the simulated climate. For example, for climate simu-
temperatures was taken specifically because it illustrated lations using an atmospheric general circulation model, the
what can be learned from the models and the reconstructions surface temperatures of the oceans are generally not
through the sometimes tedious exercise of comparing mod- known with great certainty for all months of a given
els with data. This example shows that it is important to period and for all the grid points of the model.
consider all the possible factors contributing to the differ- Assumptions are then made so as to reconstruct the sea-
ences between simulations and reconstructions in order to sonal cycle of ocean surface temperatures based on the
reduce these discrepancies: the models, boundary conditions, points that are available and about which there is also
but also the reconstructions themselves. It is also important some uncertainty. It is possible for the model to perform
to have results from many models to overcome the uncer- several simulations based on different ocean surface tem-
tainty associated with the use of a single model. perature scenarios. We can then analyze the one that
25 Modeling and Paleoclimatology 333

produces the land temperatures/precipitation values most 3. the altitude and extent of the ice caps and the change in
compatible with the reconstructed data. These are known land-ocean distribution caused by changes in sea level.
as sensitivity experiments to sea surface temperatures.
In the case discussed above, we examined several sce- To better understand why the glacial period climate is
narios with different ocean surface temperatures to find out different from the current climate, simulations can be carried
which was most realistic for the period in question. More out in which only one or two of these boundary conditions are
generally, it is also instructive to analyze the signature or imposed and the simulated climate is compared with a more
influence of each forcing among several forcings and ‘realistic’ simulation where all of the boundary conditions are
changes in boundary conditions. To continue with the applied. By carrying out these simulations where one or two
example of the climate of the Last Glacial Maximum, if a boundary conditions or forcings are imposed with glaciation
coupled atmosphere-ocean model is used, the forcings values, the aim is not to try to achieve a realistic simulation.
applied to the model to obtain a simulation of the climate of Instead, the purpose is to gain a better understanding of the
this period are: response to each type of forcing. These simulations are also
called sensitivity experiments with 1, 2 or 3 forcings.
1. the insolation; This approach is illustrated in Fig. 25.5 where the
2. the greenhouse gas concentration (CO2, CH4, N2O); response of the IPSL climate model to LGM conditions is

Fig. 25.5 Average annual air

temperature at 2 m simulated by
the coupled ocean-atmosphere
IPSL model: a for pre-industrial
conditions; b, c and d anomalies
compared to this pre-industrial
climate: b for a pre-industrial
simulation, where the
atmospheric greenhouse gas
concentration has been replaced
by ice age values; c same for ice
caps, d simulation with all LGM
conditions; e difference
d − (b + c), allowing the
quantification of the proportion of
the difference between
pre-industrial and the LGM
related to the interactions between
the impacts of the ice caps and the
reduction in greenhouse gas
concentrations. For figure (a),
isolines every 10 °C, dashed lines
for negative values, long dashes
for 0 °C, and continuous lines for
positive values
334 M. Kageyama and D. Paillard

analyzed by separating the impact of the ice caps from the Outlook
impact of the reduction of atmospheric greenhouse gas
concentration. This figure refers to the annual average tem- In this section, we mainly describe atmospheric general
perature of air 2 m above the surface. The values for this circulation models, including some coupled with oceanic
temperature in the pre-industrial climate simulation are general circulation models. These models, now comple-
shown in Fig. 25.5a and the differences between the simu- mented by vegetation models, carbon cycle models and
lated LGM and pre-industrial climates are shown in atmospheric chemistry models, are becoming increasingly
Fig. 25.5d. This map shows an overall global cooling, complex, with more components of the climate system, more
moderate (by a few degrees approximately) over the oceans processes and more associated feedbacks being included.
but very strong (more than 30 °C) over the ice sheets of the The complex models of the Earth system have sometimes
northern hemisphere. Figure 25.5b shows that the contribu- been described as the ‘biggest’ models that can run on the
tion of greenhouse gases to this response is less extreme over ‘biggest’ computers, a sort of ‘maximum’ model. These limit
the continents and a little weaker over the oceans than the the number of numerical experiments that can be run for
response to all of the LGM conditions. Figure 25.5c shows each given problem and the number of sensitivity experi-
the impact of the ice caps alone. It shows that these are ments that can be conducted to better understand the influ-
responsible for a significant cooling over the continents of ence of a particular process or mechanism. This situation is
the northern hemisphere, but also for a cooling of between 0 changing, as advances in computing now allow modelers to
and 2 °C over most of the oceans, with the notable exception carry out more experiments for a given period. These
of the Southern Ocean. It is worth noting that the sum of the experiments are essential to improve our understanding of
anomalies shown in Fig. 25.5b and d is not equal to the the importance of forcings or processes within a change in
difference between the LGM and pre-industrial climates climate. New developments in complex climate system
shown in Fig. 25.5d. This difference is shown in Fig. 25.5e. models, the improvement in their resolution, the inclusion of
This shows that in many regions the impact of the two new processes or new components must consider the nec-
factors taken together is greater than the sum of the impacts essary compromise between the computational time required
of each factor considered separately. This is referred to as the for a simulation and the number of simulations that can be
synergy between the various forcings. In other areas, such as performed with the computer available. With increasing
north of the Nordic Seas, the impact of the two forcings computing capabilities, it is possible to improve the resolu-
together is lower than the sum of the impacts of the indi- tion of the models and increase the number of processes
vidual forcings. This shows that to quantify the impact of a included. The models used for the IPCC assessment exer-
specific forcing within a group of two forcings (as in this cises provide a good indication of the progress of climate
case, the impact of the ice caps and the reduction in green- modeling over the past two decades (Fig. 25.2). In the
house gases), four simulations must be carried out: a control future, we will see new components of the climate system
simulation (in this case, the pre-industrial climate), one being integrated as for instance ice caps. The aim of these
including all the forcings (in this case, the LGM) and sim- developments is mainly to provide a better prediction of the
ulations with each factor taken individually. This method is future climate, but they also contribute greatly to the study of
called the ‘factor separation’ method developed for the paleoclimates, which in turn makes it possible to evaluate
atmospheric sciences by Stein and Alpert (1993). these models under climate conditions different to current
So far, we have studied sensitivity experiments on the ones.
forcings and boundary conditions imposed on models.
Sensitivity experiments may also be applied to internal
processes of the climate system. To examine the importance Earth System Models of Intermediate
of this processes in the response by the climate system to a Complexity (EMICS)
disturbance, its formulation in a model can be modified. For
example, if deep convection at the equator is suspected to Basic Principles and History
have an important influence on some aspect of the climate in
the mid-latitudes, the formulation for deep convection in the We have seen that climate models developed from general
model can be altered and a sensitivity experiment for this circulation models were intended to be as comprehensive as
process can be performed with a modified model, under the possible in their representation of the climate system. This
same boundary conditions. This type of sensitivity experi- requires considerable computing power and calculation times,
ment also makes it possible to evaluate the importance of a which in practice forces the modeler to limit the number of
feedback by excluding or activating it. simulations performed. These simulations are also quite short
25 Modeling and Paleoclimatology 335

compared to the typical timeframe in climate evolution. Thus, type being useful for its ‘efficiency’, the other for the spatial
in parallel with the development of general circulation models, and temporal detail in its representation.
simpler models, more adapted to the study of paleoclimates,
have been developed. The aim was to represent from the
outset the slow-moving components of the climate system, the Examples of Long-Term Simulations and Studies
ocean and the ice caps, in order to study long-term climate of Sensitivity to Forcings
change (i.e. for time scales in excess of a thousand years). To
develop these models, representation of the rapid components One of the first models of intermediate complexity of the
of the climate system, particularly of the atmosphere, has to be climate system is the one developed by the Catholic
simplified. In fact, the term ‘simple model’ is misleading University of Louvain-la-Neuve. This model includes sim-
because it refers above all to models which are more efficient plified representations (by latitude and vertically, for the
in their use of computing time. Developing a model of this northern hemisphere) of the atmosphere, the ocean, sea ice
type is not necessarily ‘simple’ because one cannot simply and the polar ice sheet. It was developed specifically to study
retain the basic equations of atmospheric dynamics. The the glacial-interglacial cycles, as demonstrated by the first
saving of computation time is generally achieved by estab- simulations of Gallée et al. (1992). Since then, Berger et al.
lishing parameters for the transport of heat and moisture by the (1998) and Loutre and Berger (2000) have taken up this
stationary and/or transient waves (such as, for example, by the model and carried out sensitivity experiments to identify the
depressions of the mid-latitudes), in other words, by trying to respective roles of orbital variations and greenhouse gases in
represent the effect of these phenomena without explicitly the last glacial-interglacial cycle.
calculating them. This makes it possible to extend the duration Figure 25.6 shows a selection of the results of these two
of a time step and to use a coarser spatial resolution. The articles in terms of volume of ice (top) and temperature in
number of degrees of freedom of these models lies between the northern hemisphere (bottom). The continuous lines
conceptual models (around ten) and general circulation mod- represent the results of the model forced by both the varia-
els (*105–106) and they are called ‘EMICs’, ‘Earth system tions in CO2 recorded in the ice cores (Jouzel et al. 1993)
Models of Intermediate Complexity’ (Claussen et al. 2002). In and by the variations in the orbital parameters as calculated
fact, the terminology was created long after the development by Berger (1978). The initial state of the model is an inter-
of the first models in this category. It emerged at a time when glacial state, with no ice sheet in the northern hemisphere.
the modelers who specialized in these models decided to join Over the last 200,000 years, the model simulates two major
together to define the specificity of their models compared glaciations, with a complete freeze-up occurring in steps and
with others. These models are characterized by a more com- a complete deglaciation following the glacial maximum. The
plete representation of the climate system than the ‘simple’ coldest temperatures are of course simulated for these glacial
ocean-atmosphere models and by a relatively short computing maxima. During the last interglacial and the last glaciation
time compared to the general circulation models, characteris- (between 126,000 and 80,000 years), despite a high level of
tics which allow the evolution of the climate system to be recorded variability, temperatures remain sufficiently high so
studied over long time scales and many different scenarios to that the frozen-over periods last no longer than 15,000 years.
be explored. There are many EMICs, corresponding to the The red lines correspond to a simulation where the
many different ways the representation of the climate system insolation is constant and equal to the current insolation, and
can be ‘simplified’. It should be noted that some models have where CO2 levels vary in a similar way to the previous
been developed by ‘downgrading’ a general circulation model, simulation. It can be seen that the volume of ice increases to
i.e. by reducing its vertical and horizontal resolutions. These about 35  1015 m3 and remains at around 30  1015 m3
are the most complex models in the EMIC category. during the rest of the simulation. Although the average
As with general circulation models, climate models of temperature of the northern hemisphere varies in line with
intermediate complexity can be used to obtain realistic cli- the greenhouse gas forcing, it remains too cold to bring
mate simulations or to study the sensitivity to certain forc- about deglaciation. The other curves are the result of simu-
ings, processes or feedbacks of the represented system. In lations where CO2 remains constant (fixed at 210 ppm for
the following sections, we give examples of the use of the alternating dash-dot line and at 250 ppm for the dashed
EMICs, both for long-term simulations and for studies curve) and where variations of the orbital parameters are
requiring numerous experiments. Experiments of this type taken into account. This time, alternation between glacial
could not have been carried out using general circulation and interglacial periods is obtained, with maximum ice
models given current capacity of computing power. This volumes reached for the same periods as in the reference
shows the complementarity of the two types of models, one simulation at values inversely related to the level of imposed
336 M. Kageyama and D. Paillard

northern hemisphere ice

volume in 1015m3
mean annual temperature averaged over
the northern hemisphere, in °C

time in kyrs, 0 being the present day

and negative sign for past years

Fig. 25.6 Ice volume (top, in 1015 m3) and average temperature of the The red line represents the results for constant insolation, equal to
northern hemisphere (bottom, °C), as a function of time in kyrs, current levels, with variable CO2. The light and dark blue lines
simulated by the Louvain-la-Neuve coupled model ice sheet-northern represent the results for variable insolation and constant CO2, at 210
hemisphere climate (Gallée et al. 1992) over the last 200,000 years. In and 250 ppm respectively. According to Berger et al. (1998), and
black, the simulation where CO2 and insolation vary in line with the Loutre and Berger (2000). The authors thank M.-F. Loutre for
Vostok data for CO2 and according to the forcing by Berger (1978). providing the results of the Louvain-la-Neuve model

CO2. These sensitivity experiments show that variations in scales able to reflect the variations in climate during the
insolation are an essential forcing to explain Quaternary (glacial-interglacial cycles, abrupt events from
glacial-interglacial cycles, because forcing the model with the last glacial period). This efficiency also means that it is
CO2 variations alone does not produce these cycles. On the possible to carry out numerous simulations to explore the
other hand, experiments with constant levels of CO2 show sensitivity to the forcings used of a result or to choices made
that the extent of a glaciation is highly dependent on CO2. during the construction of the model. An example of this
Examples of other long-range simulations are given in type of use of an intermediate complexity model is given
other chapters of this book. In Chap. 29, an example is given in Chap. 29. The CLIMBER-2 model, incorporating a rep-
of modeling interactions between northern hemisphere ice resentation of oxygen isotopes in the ocean, is used to
sheets, the ocean and the atmosphere over a period of explore the response to freshwater discharges from the North
50,000 years (Calov et al. 2002), as well as an example of American ice cap in multiple scenarios. The response is then
modeling of 18O variations in the glacial ocean in response compared to marine records to determine the most probable
to freshwater inputs from the North American ice cap scenarios in terms of duration and amplitude of the fresh-
(Roche and Paillard 2005). water input.
Another example is given by the work of Schneider
von Deimling et al. (2006). This study analyzes ensembles
Example of the Use of Intermediate Complexity of simulations for the climates of the present day, the future
Models to Explore a Multitude of Forcings and the Last Glacial Maximum. These ensembles are formed
or Parameters: Exploration of a ‘Phase Space’ by varying eleven parameters of the model within acceptable
ranges, based on our current knowledge of the climate sys-
In the previous section, we showed that, thanks to its effi- tem. The main effect of changes in these parameters is to
cient use of computing time, a climate model of intermediate vary the amplitude of the feedbacks in the climate system
complexity allows us to perform long simulations over time and the climate sensitivity of the model (defined as the
25 Modeling and Paleoclimatology 337

difference in global temperature due to a doubling of CO2, Edward Lorenz proposed a very simple model, based on an
see Chap. 31). The current climate, as defined by the idealized thermal convection, which for the first time illus-
observations, makes it possible to make a first choice of trated that the complexity of the behavior of a dynamic
parameters so that an acceptable simulated climate can be system was absolutely unrelated to the number of degrees of
achieved. The authors show that reconstructions of tropical freedom of this system, as many previously imagined. He
temperatures during the Last Glacial Maximum help to showed that a very simple system (in this case with only three
constrain even further the selection of values of these degrees of freedom) can produce unpredictable behavior,
parameters and so reduce the uncertainties associated with called ‘deterministic chaos’. This conceptual model still
future climate change. This study shows that modeling of plays an important educational role, and its mathematical
past climates followed by comparison with reconstructions properties are still a subject of active research. Below, some
can help in the evaluation of climate models used to forecast examples directly relevant to the climate system are descri-
future climate. This conclusion is also advanced by Harg- bed in more detail. Other examples, shown in Chap. 28, aim
reaves et al. (2007) using a general circulation model. to achieve a better understanding of the glacial-interglacial
dynamics (Calder, Imbrie models etc.).

Outlook
The Budyko/Sellers Model
It might be expected that with the advances in computing,
climate models of intermediate complexity would no longer The Earth’s climate is determined above all by its radiative
have any reason to exist. In fact, this is not the case because balance. By simulating simplified balances, it is possible to
they will always be less time consuming than general circu- estimate the magnitude of a change in temperature caused
lation models, which incorporate more and more mechanisms by, for example, changes in the incident solar radiation
and use an increasingly fine resolution. In some ways, general (volcanic dust, changes in the solar constant, nuclear winter,
circulation models, such as those used for IPCC simulations, etc.). In 1969, two publications (Budyko 1969, Sellers 1969)
are defined by the capabilities of the most powerful com- came to a somewhat surprising conclusion: if we take
puters. We need to be able to use these models to produce account of the feedback between temperature and albedo, a
simulations of several hundred years within a reasonable time relatively small decrease in the solar constant (−1.5% or
on available computers. Long simulations, necessary to −2%) is enough for the Earth to completely freeze over.
understand past climate changes reconstructed using multiple A similar result also occurs when the greenhouse effect is
indicators and the need to explore different scenarios and modified. This indicates that there is a critical threshold
model parameters, require faster models. Today’s general towards cooling which causes the climate system to move
circulation models will no doubt become the intermediate into a very different state. The meaning of these results has
complexity models of tomorrow, but this concept will con- now become even more pertinent with the theory of Snow-
tinue to exist. In addition, it is important to retain this hier- ball Earth (Chap. 26).
archy of models because each type of model is established on In their original versions, Budyko’s and Sellers’ models
different assumptions. By comparing the results of different are explicitly dependent on latitude and predict a tempera-
models, it is possible to highlight the relative importance of a ture T(y) where y is the latitude. A much simpler version can
particular process which is included in one model but not in be formulated to represent the phenomenon of runaway
the other or which is represented differently in each model. albedo-temperature feedback, with a model with no geo-
graphical dimensions. Writing the radiative balance of the
Earth as a global average:
Conceptual Models
ð1  aÞ  S=4 ¼ ð1  eÞrT4 ð25:7Þ
The main objective of the models described above is to try to
reproduce the observations we have for the climate system where a is the albedo of the Earth, S is the solar constant; e is
and its variations in the past. As has been highlighted, a corrective term to represent the greenhouse effect; r is the
modeling also aims to improve our understanding of these Stefan-Bolztman constant; then the overall global tempera-
variations, and it is therefore useful to describe some aspects ture of the planet is easy to calculate.
of the system using extremely simple models, which are The problem becomes more interesting with the
intended to illustrate some key processes. These are called albedo-temperature feedback. Indeed, if we assume that a is
conceptual models. There are many varied examples. One a decreasing function of T, with, for example a constant
example is the Lorenz model (Fig. 25.1d), which often serves a(T) (*0.3) at high temperatures for a ‘blue’ planet, a
as an archetype of the chaotic system. The meteorologist constant a(T) (*0.7) at very cold temperatures for a ‘white’
338 M. Kageyama and D. Paillard

planet, and a linear a(T) in-between, then we obtain the The Stommel Model (1961)
diagram shown in Fig. 25.7.
The possibility of multiple equilibria leads to the exis- The existence of multiple equilibria concerns other compo-
tence of thresholds beyond which the climate system sud- nents of the climate system and an important example in
denly shifts to a new state of equilibrium. Moreover, this paleoclimatology is the Stommel model (Fig. 25.8).
leads to a phenomenon of hysteresis, since it is not possible The model is composed of two well-mixed boxes, of the
to easily return to the original state by reversing the distur- same volume, representing a mass of cold water with low
bance. Thus, to return to the initial state, an inverse pertur- salinity for high latitudes (with temperature T1 and salinity S1)
bation of much greater amplitude is necessary. This is one of and a mass of warm water with high salinity for low latitudes
the difficulties with the ‘snowball’ theory: although it is (with temperature T2 and of salinity S2). The difference in
relatively ‘easy’ for the planet to freeze over completely, as density Dq = q2 − q1 between these two boxes is obtained as
was shown by Budyko and Sellers, it is much more difficult a function of the positive coefficients of thermal expansion a
to get out of this cold state. and saline contraction b assumed to be constant:

0.4
Radiative Budget

0.5
0.3

ε=
W.m-2

ε=
250 ε=

6
0.
200 ε=

150

100 Temperature °C
-40 -20 0 20 40

60
Temperature Temperature
°C °C
40 40

20 20

0 0
S = 1368
ε = 0.4

-20 -20

-40
ε
-40

Solar constant W.m -2

0.3 0.4 0.5 0.6 0.7 1200 1400 1600 1800 2000

Fig. 25.7 Simplified Budyko and Sellers model. Top: dashed line, the corresponding points of radiative equilibrium as a function of the
solar term of the radiative balance, i.e. (1 − a(T))  S/4, with a(T) linear greenhouse effect e (left), or as a function of the solar constant S (right).
between −15 and +15 °C, and constant beyond this range. Solid line, The equilibrium shown in dotted lines is unstable. For the current
infrared term, for different values of the greenhouse effect. Balance is parameters, there are therefore two possible stable equilibria, corre-
achieved when the curves intersect. Note that there are several points of sponding either to our climate (temperature of around +15 °C) or to a
equilibrium, especially for the current situation (e = 0.4). Bottom: the completely frozen planet (temperature of around −40 °C)
25 Modeling and Paleoclimatology 339

The equilibrium (or equilibria) of the system is then

-Σ Σ easily obtained:
m
R ¼ ljaDT  bDSjDS; ð25:11Þ
T1, S1 T2, S2
which gives a second-degree equation in DS, with an abso-
lute value, which can be rewritten as:
Pole Equator
bDS Rb
F ¼ xj1  xj by setting: x ¼ ;F¼ ð25:12Þ
Fig. 25.8 Configuration of the Stommel model aDT lðaDT Þ2
The function F(x) is plotted in Fig. 25.9.
Dq ¼ aðT2  T1 Þ þ bðS2  S1 Þ ð25:8Þ The definition of x shows that x measures the intensity of
the salinity gradient relative to the temperature gradient. The
A thermohaline circulation m, proportional to the differ- sign of m, and consequently, the direction of the thermo-
ence in density between the two boxes, mixes the two cor- haline circulation is positive when x is less than one. The
responding bodies of water. This circulation will be positive dominant effect is then that of the temperature gradient
(Fig. 25.8) when the cold water (box 1) is sufficiently dense DT. This is a mode of ‘thermal’ circulation. This is the case
to sink below the warm water (box 2), in other words when for the unique solution xT when F is negative. Conversely, if
Dq < 0, and hence: the imposed flux of salt R is sufficiently strong, and conse-
m ¼ lDq ¼ l½aðT2  T1 Þ þ bðS2  S1 Þ ð25:9Þ quently, if F is large enough (greater than 0.25), the solution
xS of the problem is greater than 1, and the thermohaline
where l is an arbitrary constant. circulation is reversed because the only equilibrium possible
Conversely, if the saline water in box 2 is denser (Dq of the system is of the ‘saline’ type. On the other hand, for an
> 0), the thermohaline circulation m will be negative. intermediate value of the forcing F, the ‘thermal’ and ‘sal-
If we impose the temperature gradient DT = T2 − T1 and ine’ modes are both possible solutions of the problem.
force the salinities by applying a constant flow of salt R into A third equilibrium point also becomes possible with
box 2, and −R into the box 1, we can infer the evolution in solution xI. However, a rapid analysis of the stability of the
salinity, which are the only variables in the problem. For equilibria obtained shows that this is an unstable equilib-
example, for box 1 (the strict opposite is in box 2): rium. In fact:

dS1 dDS
¼ R þ jmjðS2  S1 Þ ¼ R þ ljaDT  bDSjDS: ¼ 2ðR  jmjDSÞ ð25:13Þ
dt dt
ð25:10Þ

Fig. 25.9 Diagram of stability

for the Stommel model (function
F(x) defined by Eq. (25.12)). The
dotted section corresponds to an
unstable equilibrium
340 M. Kageyama and D. Paillard

The equilibrium is stable when any infinitesimal increase This model is composed of two superimposed boxes, one
in the salinity gradient DS leads to an increase in the term | representing a mass of cold, low-salinity surface waters
m|DS, so that the derivative of DS relative to time becomes (temperature T1 and salinity S1), and the other a slightly
slightly negative. The direction of variation of |m|DS as a warmer, saltier water mass for the depths (temperature T0
function of DS is in fact given directly by the figure repre- and salinity S0). The deep-water box is much larger than the
senting |1 − x|x as a function of x. The descending inter- surface water box, and we assume that T0 and S0 are con-
mediate branch (dashed line in the figure) indicates a stant. The difference in density Dq = q1 − q0 between these
decrease in |m|DS when DS increases, leading to an unstable two boxes is obtained as a function of the positive coeffi-
equilibrium. The ‘thermal’ and ‘saline’ branches, on the cients of thermal expansion a and saline contraction b,
other hand, are perfectly stable. For the same value of the assumed to be constant:
temperatures and the salt flux, there are therefore two stable
equilibria possible in this system. Dq ¼ aðT1  T0 Þ þ bðS1  S0 Þ ð25:14Þ
Interestingly, these multiple equilibria are found in much Vertical mixing (convection c) is very small if the column
more complex models of the ocean, and even in some is well stratified (if Dq < −e < 0). In the opposite case, it
ocean-atmosphere coupled models (Rahmstorf 1996; will be large:
Rahmstorf et al. 2005). It therefore seems that this very
simple model captures an important aspect of thermohaline c ¼ c0 ; c0 small if Dq\  e;
circulation, which explains the existence of sudden varia- c ¼ c1 ; c1 large if Dq [  e:
tions in the deep ocean circulation. These variations are most
likely involved in the sudden climate changes observed The variables of the problem this time are the temperature
during the ice ages (Heinrich events and Dansgaard- and the salinity of the surface water box, and the equations
Oeschger events, see Chap. 29). for the corresponding evolution are formulated as follows:

dT1 dS1
¼ kðTA  T1 Þ þ cðT0  T1 Þ; ¼ R þ cðS0  S1 Þ:
dt dt
The Welander Model ð25:15Þ
The multiplicity of equilibria does not explain everything, At equilibrium, the time derivatives in both equations are
and of course there are other types of possible behaviors. equal to zero, which yields:
Another oceanographic example (Fig. 25.10), similar to the
kTA þ ce TO R
Stommel model, concerns convective-advective oscillations T1e ¼ ; Se1 ¼ S0 þ e ð25:16Þ
kþc e c
(Welander 1982). Although their relevance to climate vari-
ations is not established, some have suggested that these where ce is the value of the vertical mixing for a temperature
oscillations may play a role in the recurrence of T1e and a salinity Se1 , that is, for a density difference of Dqe
Dansgaard-Oeschger events. (ce= c0 if Dqe < −e; ce= c1 otherwise). This leads to a
solution:
kðTA  TO Þ R
k (TA -T1 ) Σ Dqe ¼ a
k þ ce ðDqe Þ
þb e
c ðDqe Þ
; i.e. Dqe ¼ F ðDqe Þ

ð25:17Þ
Surface T1 , S1 where F is a constant depending on the sign of Dq (F = F0 if
Dqe < −e; F = F1 otherwise). So, there are zero, one or two
solutions, depending on the parameter values, as shown in
Fig. 25.11.
In general, if c is a continuous function of Dq then F(Dq)
will also be a continuous function and there will be an odd
c number of solutions, alternately stable and unstable, as in the
Stommel model. The case of ‘zero solution’ in Fig. 25.11
(discontinuous case) would in fact correspond to an ‘unsta-
Deep T0,S 0 ble equilibrium’ for a continuous model. There is then a
boundary cycle (an oscillation): in the absence of intense
convection (c = c0), there is a tendency towards an equi-
Fig. 25.10 Configuration of the Welander model (1982) librium in the other domain (with a strong convection,
25 Modeling and Paleoclimatology 341

Fig. 25.11 Examples of equilibria given by F = Dq, where F is fact, the system oscillates); b there are two equilibria which coexist, one
discontinuous and can only have two constant values. Depending on convective and the other diffusive; c there is only one equilibrium
these values, there are three possible cases: a there is no solution (in (convective or diffusive)

necessary to understand the issue at hand. Using models

based on different assumptions, it is possible to compare
their importance and to better understand the role of a
specific process. In principle then, one model is not more
reliable than another. A model is always based on a set of
assumptions and its relevance must be judged in relation to
the question that the modeler has chosen to study. A model
generally provides a result, without any uncertainty explic-
itly associated with this result. The uncertainty is concealed
within the assumptions on which the model is based and in
the imposed boundary conditions, which, although as real-
istic as possible, are never as well defined as we would
Fig. 25.12 Relaxed oscillation in the Welander model. On a TS ideally like. It is therefore interesting to compare the results
diagram, a convective zone (when the density q1 is large) and a
diffusive zone, separated by an iso-density curve Dq = −e, are defined.
of several models with the same boundary conditions, just as
If the values T, S are located in the convective zone (c = c1), the model it is interesting to study the sensitivity of the results of a
tends to bring them back to the point of attraction E1 (located in the model to certain conditions with poorly constrained bound-
diffusive zone). If they are in the diffusive domain (c = c0), the model aries. This may help to determine if a better understanding of
tends to bring them back to the point of attraction E0 (located in the
convective zone). The model will thus oscillate from one mode of
these conditions is required or if this is of little importance
operation to another, without ever reaching equilibrium for the climate being studied.
Here, we mostly described the use of global models. We
c = c1), as indicated in Fig. 25.12. There is then an oscil- showed that the resolution of these models, even of the
lation between a ‘diffusive’ state (c small = c0) and a ‘con- most complex ones, does not always allow comparison
vective’ state (c large = c1). These type of oscillations have with reconstructions which are often representative of
also been observed, under certain conditions, in regions much smaller than the ‘boxes’ of a model. We saw
three-dimensional ocean models. that the modeling of paleoclimate indicators such as iso-
topes permits a more precise comparison and also an
analysis of the recorded signal as a function of climate
Conclusions and Outlook parameters. There are also models with a finer resolution
that can be used on a regional scale. The use of this type of
In this chapter, we explained the basic principles of climate model for paleoclimates, still quite limited at the present
modeling and illustrated their application through examples time, is bound to develop in the future, in parallel with their
using models of different levels of complexity to show the increasing use for climate forecasts. An example of
strengths of each model type. This hierarchy of climate downscaling methods is given for the Last Glacial Maxi-
models is important not only for practical reasons related to mum by Jost et al. (2005), which highlights the significant
the computing time required to study a particular scale of differences between the simulated climates at the European
space or time, but above all, because it represents a range of scale using these methods. There is still a lot of work to be
grouped assumptions in each model type, which are done to generate these models. It should be noted that there
342 M. Kageyama and D. Paillard

are also regional models which include a representation of Hargreaves, J. C., et al. (2007). Linking Glacial and future Climates
climate indicators such as water isotopes (e.g., REMO-ISO, through an ensemble of GCM simulations. Climate of the Past, 3,
77–87.
Sturm et al. 2007). IPCC. (2007). 4e rapport. http://www.ipcc.ch.
The evolution of climate models in the near future will Jouzel, J., et al. (1993). Vostok ice cores: Extending the climatic
involve not only an increasingly fine resolution as comput- records over the penultimate Glacial Period. Nature, 364, 407–412.
ing power increases and an explicit resolution of phenomena Joussaume, S. & Taylor, K. E. (1995). Status of the paleoclimate
modeling intercomparison project (PMIP). In Proceedings of the
previously parametrized or ignored, but also the construction First International AMIP Scientific Conference. WCRP Report,
of increasingly comprehensive models of the climate system. (pp. 425–430).
Not only will the physical and dynamic aspects of the cli- Jost, A., Lunt, D., Kageyama, M., Abe-Ouchi, A., Peyron, O., Valdes,
mate system be addressed, but also the biogeochemical P. J., Ramstein, G. (2005). High resolution simulations of the Last
Glacial Maximum climate over Europe: A solution to discrepancies
cycles, in particular the carbon cycle. This trend is already with continental paleoclimatic reconstructions? Climate Dynamics,
emerging and is important not only to understand the evo- 24, 577–590. https://doi.org/10.1007/s00382-005-0009-4.
lution of past climates but also to forecast the climates of the Kageyama, M., Braconnot, P., Harrison, S. P., Haywood, A. M.,
future (see Chap. 31). Jungclaus, J. H., Otto-Bliesner, B. L., et al. (2018). The PMIP4
contribution to CMIP6—Part 1: Overview and over-arching anal-
ysis plan. Geoscientific Model Development, 11, 1033–1057.
https://doi.org/10.5194/gmd-11-1033-2018.
For Further Information Laskar, J., Robutel, P., Joutel, F., Gastineau, M., Correia, A., Levrard,
B. A. (2004). A long-term numerical solution for the insolation
quantities of the earth. 428. https://doi.org/10.1051/0004-6361:
McGuffie, K., & Henderson-Sellers, A. (2005). A climate modelling 20041335.
primer (3rd ed., 280 p). Wiley. Lorenz, E. (1963). Deterministic nonperiodic flow. Journal of the
Atmospheric Sciences, 20, 130–141.
Loutre, M.-F., & Berger, A. (2000). No Glacial-Interglacial cycle in the
References ice volume simulated under a constant astronomical forcing and a
variable CO2. Geophysical Research Letters, 27, 783–786.
Manabe, S., & Broccoli, A. J. (1985). A comparison of climate model
Berger, A., (1978). Long-term variations of daily insolation and Sensitivity with data from the Last Glacial Maximum. Journal of
quaternary climatic changes. Journal of the Atmospheric Sciences, the Atmospheric Sciences, 42, 2643–2651.
35, 2362–2367. Peyron, O., et al. (1998). Climatic reconstruction in Europe for 18,000
Berger, A., et al. (1998). Sensitivity of the LLN Climate model to the YR B.P. from Pollen Data. Quaternary Research, 49, 183–196.
astronomical and CO2 forcings over the last 200 Ky. Climate Rahmstorf, S. (1996). On the freshwater forcing and transport of the
Dynamics, 14, 615–629. Atlantic Thermohaline circulation. Climate Dynamics, 12, 799–811.
Braconnot, P., et al. (2007a). Results of PMIP2 coupled simulations of Rahmstorf, S., et al. (2005). Thermohaline circulation hysteresis: A
the Mid-Holocene and Last Glacial Maximum—Part 1: Experi- model intercomparison. Geophysical Research Letters, 32, L23605.
ments and large-scale features. Climate of the Past, 3, 261–277. Ramstein, G., et al. (2007). How cold was Europe at the Last Glacial
Braconnot, P., et al. (2007b). Results of PMIP2 coupled simulations of Maximum? A synthesis of the progress achieved since the first
the Mid-Holocene and Last Glacial Maximum—Part 2: Feedbacks PMIP model-data comparison. Climate of the Past, 3, 331–339.
with emphasis on the location of the ITCZ and mid- and high Roche, D., & Paillard, D. (2005). Modelling the oxygen-18 and rapid
latitudes heat budget. Climate of the Past, 3, 279–296. Glacial climatic events: A data–model comparison. Comptes
Budyko, M. (1969). The effect of solar radiation variations on the Rendus Geoscience, 337, 928–934.
climate of the earth. Tellus, 21, 611–619. Schneider von Deimling, T., et al. (2006). Climate sensitivity estimated
Calov, R., et al. (2002). Large-scale instabilities of the Laurentide ice from ensemble simulations of Glacial Climate. Climate Dynamics,
sheet simulated in a fully coupled climate-system model. Geophys- 27, 149–163.
ical Research Letters, 29, 2216. Sellers, W. (1969). A global climatic model based on the energy
Claussen, M., et al. (2002). Earth system models of Intermediate balance of the earth-atmosphere system. Journal of Applied
complexity: Closing the gap in the spectrum of climate system Meteorology, 8, 392–400.
models. Climate Dynamics, 18, 579–586. Stein, U., & Alpert, P. (1993). Factor separation in numerical
CLIMAP. (1976). The surface of the Ice-Age earth. Science, 191, simulations. Journal of the Atmospheric Sciences, 50, 2107–2115.
1138–1141. Sturm, C. et al. (2007). Simulation of the stable water isotopes in
CLIMAP. (1981). Seasonal reconstructions of the earth’s Surface at precipitation over South America: Comparing regional to global
the Last Glacial Maximum. Geological Society of America, Map circulation models. Journal of Climate, 20, 3730–3750.
Chart Series MC-36, Boulder, Colorado. Stommel, H. (1961). Thermohaline convection with two stable regimes
Gates, W. L. (1976). Modeling the Ice-Age climate. Science, 191, of flow. Tellus, 13, 224–230.
1138–1144. Welander, P. (1982). A simple heat salt oscillator. Dynamics of
Gallée, H., van Ypersele, J.-P., Fichefet, T., Marsiat, I., Tricot, C., Atmosphere and Oceans, 6, 233–242.
Berger, A. (1992). Simulation of the Last Glacial cycle by a Wu, H., et al. (2007). Climatic changes in Eurasia and Africa at the Last
coupled, sectorially averaged climate-ice sheet model. 2. Response Glacial Maximum and Mid-Holocene: Reconstruction from Pollen
to insolation and CO2 variations. Journal of Geophysical Research, Data using inverse Vegetation modelling. Climate Dynamics, 29,
971, 15713–15740. https://doi.org/10.1029/92JD01256. 211–229.
 The Precambrian Climate
26
Yves Goddéris, Gilles Ramstein, and Guillaume Le Hir

More than 88% of the history of the Earth occurred in the expansion of land surfaces between 3.2 and 2.6 billion years
Precambrian. The Precambrian began with the formation of ago and the growth of the partial pressure of oxygen in the
the Earth 4.6 billion years ago (Ga) and ended 542 million atmosphere around 2.3 Ga.
years ago (International Stratigraphic Chart, www.strati-
graphy.org). It is subdivided into two large eons: the
Archean (between 4 and 2.5 Ga) and the Proterozoic (from Climate Indicators
2.5 to 0.542 Ga). The International Commission on
Stratigraphy is proposing to add an extra eon, the Hadean, Little is known about the evolution of the Earth’s climate
covering the first 600 million years of the history of our during the Precambrian. The number of indicators available
planet. Notwithstanding, this eon is described as having an is very limited. Firstly, sedimentological data are difficult to
informal status since no pre-Archean rock has been observed interpret, given the age of these sediments which have
today. In fact, the oldest rocks date back to 4 billion years generally been disarranged. Secondly, paleontological data
ago (U/Pb dating on zircon crystals). These are the Acasta are virtually unusable in terms of climate reconstruction.
gneisses in the Slave Province of Canada. The two formal They are fragmentary and represent only a very simple
eons of the Precambrian are subdivided into eras. In par- monocellular biosphere, which is difficult to relate to any
ticular, the Proterozoic contains three eras: Paleoproterozoic environmental evolution. Finally, the isotopic data measured
(2.5–1.6 Ga), Mesoproterozoic (1.6–1.0 Ga) and Neopro- on sediments generally have poorly preserved the original
terozoic (1.0–0.542 Ga). climate signal, having been very often subjected to
Today, we find outcrops from the Archean on all the post-deposit perturbations (diagenesis in particular). From a
continents. Among the largest, two fragments of continents quantitative perspective, two isotopic signals have been used
larger than 0.5
106 km2 were identified: the craters of with varying degrees of success: d18O measured on siliceous
Kaapvaal (South Africa) and Pilbara (Australia). They are (cherts) sediments since the 1970s (Knauth and Epstein
dated at about 3.6–2.9 Ga. Finally, the Precambrian has 1976) and, more recently, the d30Si ratio measured on the
witnessed several major events in the history of the Earth. same cherts (Robert and Chaussidon 2006).
These include the onset of plate tectonics, the emergence of Opal (SiO2) was precipitated massively during the
the biosphere at least 3.5 billion years ago, the rapid Archean. The reasons for this level of precipitation are
unknown. It may have been caused by biological activity or
directly abiotically from the silica-saturated ocean. Finally,
this opal may have been produced during the stabilization of
Y. Goddéris (&) clay minerals on the seabed or may be derived from the
Géoscience Environnement Toulouse, CNRS, Université de weathering of volcanic glass. It was subsequently subjected
Toulouse III, UMR 5563, Toulouse, France to diagenesis and today appears as siliceous sedimentary
e-mail: [email protected] rocks called cherts. The isotopic oxygen (d18O) composi-
G. Ramstein tions of these cherts show them to be increasingly depleted
Laboratoire des Sciences du Climat et de l’Environnement, in heavy isotopes as we go back in time, reaching a value of
LSCE/IPSL, CEA-CNRS-UVSQ, Université Paris-Saclay, 91190
Gif-Sur-Yvette, France 16‰ compared to the international standard SMOW, 3 Ga
ago (Fig. 26.1).
G. Le Hir
Institut de Physique du Globe, CNRS, Université Pierre et Marie
Curie, UMR 7154, Paris, France

© Springer Nature Switzerland AG 2021 343

G. Ramstein et al. (eds.), Paleoclimatology, Frontiers in Earth Sciences,
https://doi.org/10.1007/978-3-030-24982-3_26
344 Y. Goddéris et al.

Table 26.1 U-Pb and Re-Os geochronological constraints on Cryogenian glacial onsets and terminations
Paleocontinent Age (Ma) Methoda Reference
Marinoan deglaciation/cap carbonate: 636.0–634.7 Ma
Laurentia >632.3 ± 5.9 Re-Os (63)
South China 635.2 ± 0.5 U-Pb ID-TIMS (57)
Southern Australia 636.41 ± 0.45 U-Pb CA-ID-TIMS (59)
Swakop 635.21 ± 0.59/0.61/0.92 U-Pb CA-ID-TIMS (56, 83)
Marinoan glacial onset: 649.9–639.0 Ma
Congo >639.29 ± 0.26/0.31/0.75 U-Pb CA-ID-TIMS (83)
Southern Australia <645.1 ± 4.3 Re-Os (137)
South China <654.2 ± 2.7 U-Pb SIMS (134)
South China <654.5 ± 3.8 U-Pb SIMS (58)
Sturtian deglaciation/cap carbonate: 659.3–658.5 Ma
Southern Australia >657.2 ± 2.4 Re-Os (137)
Tuva-Mongol ia 659.0 ± 4.5 Re-Os (63)
Southern Australia <659.7 ± 5.3 U-Pb SIMS (366)
Laurentia 662.4 ± 3.9 Re-Os (60)
South China >662.7 ± 6.2 U-Pb SIMS (65)
Sturtian glacial onset: 717.5–716.3 Ma
Oman >713.7 ± 0.5 U-Pb ID-TIMS (365)
South China <714.6 ± 5.2 U-Pb SIMS (64)
South China <715.9 ± 2.3 U-Pb SIMS (62)
South China <716.1 ± 3.4 U-Pb SIMS (62)
Laurentia >716.5 ± 0.2 U-Pb CA-ID-TIMS (32)
Laurentia <717.4 ± 0.1 U-Pb CA-ID-TIMs (32)
Laurentia <719.47 ± 0.29 U-Pb CA-ID-TIMS (61)
Re-Os isochron ages from sedimentary organic matter. Errors are quoted at the 2r level of uncertainty. Where multiple uncertainties are given,
they represent analytical/analytical + tracer solution/analytical + tracer solution + decay-constant uncertainties
Ca chemical abrasion; ID6TIMS isotope-dilution and termal-ionization mass spectrometry; SIMS secondary-ion mass spectrometry
Table from Hoffman et al., Sci. Adv. (2017)

Fig. 26.1 The d18 O of the

Precambrian ocean measured in
the cherts
26 The Precambrian Climate 345

As for carbonates, the d18O of silica is a function of the values than is currently the case. If this scenario proves to be
temperature and of the d18O of the water in which the silica correct, the temperature of the water as inferred from the
precipitated: d18O data on cherts, could be significantly lower. Kasting
! and Howard (2006) argue in favor of ‘moderate’ climates at
106 the end of the Archean and during the Proterozoic.
1000 ln a ¼ 3:09
ð26:1Þ
T ðK Þ2 An additional element was added to this debate by Robert
and Chaussidon (2006), who measured the isotopic com-
where the fractionation factor position of silicon (d30Si) in Precambrian cherts which is
distinctly less sensitive to diagenesis than d18O. These data,
1000 þ d18 Ochert ðSMOW Þ when translated into temperatures, (requiring the use of a
a¼ ð26:2Þ
1000 þ d18 Owater ðSMOW Þ silicon cycle model and therefore additional assumptions),
suggest temperatures of around 70 °C, 3 Ga ago, and 20 °C,
ln denotes its natural logarithm, and d18 Owater is the d18 O of 800 million years ago, thus confirming very high tempera-
the water from which the cherts precipitated. tures in the distant past.
Applied to the cherts, assuming that the isotopic com- There remains a potentially major problem: the formation
position of seawater was quite similar to that of today, this conditions of Precambrian cherts are unknown. Neverthe-
paleothermometer predicts the temperatures T of the water less, these results showing a gradual cooling of sea water
from which the cherts precipitated to be close to 85 °C 3 Ga from very high values seem to have been confirmed recently
ago, and 50 °C at the Precambrian-Cambrian boundary. by a totally independent method, based on the resurrection of
These temperatures are either the sign of very hot oceans proteins of unicellular Archean organisms using phyloge-
or the consequence of an alteration of the cherts after netic and statistical methods of analysis (Gaucher et al.
deposition by meteoric or hydrothermal fluids, in which case 2008).
they provide no information on climate. The question Finally, irrespective of any debate on the terrestrial tem-
remains open because the d18O isotopic signal is particularly perature during the Archean/Proterozoic, the d18O of the
sensitive to diagenesis. A third hypothesis has been formu- cherts show a significant increase at the end of the Archean,
lated more recently: it assumes that d18O of seawater was between 2.7 and 2.5 Ga (of about 10‰), suggesting a rapid
much lower than its present value, due to tectonic processes. cooling of the oceans of about 20 °C (Fig. 26.2).
Under these conditions, the temperature of the sea water
could have been similar to the current one. Unfortunately,
there is no constraint on the d18O of seawater during the The Theory of the Paleothermostat
Archean. Authors generally assume that this ratio remained
constant over time at −1‰ in comparison with the SMOW, In 1981, Walker, Hays and Kasting published a
calculated from the level of the current ocean which would ground-breaking article explaining why the climate
have received the melt water from all the ice caps on land. remained relatively stable (within a temperature range
This value is the result of the equilibrium that is supposed to allowing water to remain in the liquid state) for between one
exist between the 18O depletion of sea water resulting from million to one billion years (Walker et al. 1981). This study
the interactions between water and lithosphere at low tem- was carried out in order to solve the faint young sun para-
peratures and its 18O enrichment during water/lithosphere dox. The models of stellar evolution predict the evolution of
interactions at high temperature in the hydrothermal systems. the solar constant during the history of the Earth and make it
Recent work (Kasting et al. 2006) suggests that the d18O possible to calculate that during the Archean, it would have
ratio may have been significantly lower (−9‰ compared to been 20–30% weaker than today. Under these conditions,
SMOW) during the Archean and Proterozoic periods. This the Earth should have totally frozen over although this
argument is based on the fact that the Archean oceans were contradicts the isotopic data which provide an estimate of the
probably shallower than the oceans are currently. This temperature of the fluid envelopes of the Earth.
results in a shallower water column over the oceanic ridges The residence time of the exosphere carbon content (i.e.
before 800 Ma, implying a reduction in hydrostatic pressure all the carbon contained in the ocean, the biosphere and the
in the hydrothermal systems. This decrease in pressure atmosphere) is around 200,000 years which is very short
limited the penetration of seawater into the ridges in the compared to the geological processes of sedimentary carbon
depths, thus reducing the gain in 18O by sea water through burial and continental weathering (François and Goddéris
alteration at high-temperature of the oceanic crust 1998). This measure gives an indication of the average time
(T > 350 °C). This reduction in flux at high temperature spent by a carbon atom entering the ocean-atmosphere sys-
results in an imbalance in the 18O cycle in the tem via volcanic degassing, for example, before exiting via
ocean-atmosphere system and its stabilization at lower sedimentary deposits. This response time is very short in the
346 Y. Goddéris et al.

Fig. 26.2 Ocean temperatures

reconstructed from d30Si and
from d18O of Precambrian cherts

context of geological time. If the history of the Earth was FMOR ; the dissolution of the continental carbonates which
condensed into one day, the entire carbon content of the transfers carbon from the continental crust to the ocean in the
ocean and atmosphere would be recycled about every 3 s. dissolved form HCO 3 Fcw ; and the oxidation of old sedi-
This short response time is obviously linked to the small size mentary organic compounds exposed to the atmosphere Fow
of the exosphere carbon reservoir (3
1018 mol) and to the (black shales, for example, sediments rich in organic matter
fluxes in exchanges of matter (of around 1012–1013 mol of deposited during episodes of large-scale ocean anoxia and
carbon per year) between this reservoir and the geological which are exposed to the atmosphere by tectonic activity):
reservoirs. The implications of this short residence time are
important for the climate. Any imbalance between the car- CaCO3 þ CO2 þ H2 O ! Ca2 þ þ 2HCO
3 ð26:3Þ
bon inflows (e.g. solid soil degassing) and outflows (e.g.
dissolution of continental carbonates by carbonic acid:
CO2 consumption due to silicate weathering) that would
have occurred over several million years (a short timescale CH2 O þ O2 ! 2HCO
3 þH
þ
ð26:4Þ
compared to the history of the Earth) would cause gigantic
fluctuations (several orders of magnitude) of the carbon oxidation of sedimentary organic carbon.
dioxide content in the atmosphere, with dramatic conse- The sinks are represented by the precipitation of car-
quences for the climate. Fluctuations of this scale have not bonate minerals Fcd on the ocean floor or on the continental
occurred in the history of the Earth, apart from some isolated shelves, through the mediation or not of biocalcification, and
episodes of global glaciations at the end of the Proterozoic. the burial of organic carbon within sediments Fod , both on
The carbon input and output flows must therefore be close to land and in the ocean.
equilibrium over the geological time scale, from 106 to
109 years (see François and Goddéris (1998) for a complete Ca2 þ þ 2HCO
3 ! CaCO3 þ CO2 þ H2 O ð26:5Þ
mathematical assessment).
precipitation of carbonates
The carbon cycle at the million-year scale is described in
Fig. 26.3. Only geological flows are taken into account, all CO2 þ H2 O ! CH2 O þ O2 ð26:6Þ
rapid recycling (biosphere fluxes and ocean-atmosphere
interface) are ignored. autotrophic productivity and burial of organic carbon.
The inflows to the ocean-atmosphere system are soil The long-term equilibrium of the carbon cycle, mathe-
degassing from volcanoes, Fvol and from the oceanic ridges matically required because of its low residence time in the
26 The Precambrian Climate 347

Fig. 26.3 The exosphere carbon

cycle at the scale of a million
years

exosphere system, results in the following proximate equa- weathering and carbonates, whereas the loss of alkalinity is
tion, which needs to be maintained on time scales of a related to precipitation of the carbonates. The response time
million years to ensure relative stability of the climate: of the alkalinity cycle to any geological perturbation is
around 3000 years, and is directly related to the ocean’s
Fvol þ FMOR þ Fcw þ Fow  Fcd þ Fod ð26:7Þ mixing time which constrains the response time of the
Small deviations from this near equality are likely to alkalinity of the world’s oceans to any disturbance (François
explain the climate fluctuations, which may be as significant and Goddéris 1998). So, we again have the quasi-equality:
as the establishment of the Permo-Carboniferous glaciation
Fsw þ Fcw  Fcd ð26:9Þ
or the climate optimum in the mid-Cretaceous. In general,
under the effect of geological forcings, these imbalances are The combination of the Eqs. (26.7) and (26.9) give the
two orders of magnitude smaller than the fluxes themselves following proximate equation:
(François and Goddéris 1998). However, this
quasi-equilibrium requires a physical basis, which the theory Fvol þ FMOR þ Fow  Fsw þ Fod ð26:10Þ
of the paleothermostat described below provides. If we ignore the existence of imbalances in the organic
The flux of silicate weathering Fsw does not appear in the sub-cycle of carbon (Fow = Fod, which is a strong assump-
carbon balance because the silicates are not present in the tion), we obtain:
continental crust in significant amounts. Yet, this flow
actually consumes carbon from the exosphere: Fvol þ FMOR  Fsw ð26:11Þ

CaSiO3 þ 2CO2 þ 2H2 O ! Ca2 þ þ 2HCO

3 þ SiO2
The paleothermostat theory explains the physical reasons
for this equality on the basis of two assumptions: firstly, that
ð26:8Þ
the weathering of continental silicate rocks depends on the
This reaction is a generic reaction, the silicate mineral climate and, in particular, on the temperature and the con-
(here the Wollastonite) is scarce on Earth, but it accounts for tinental runoff. Secondly, excluding water vapor, atmo-
the alteration budget and therefore illustrates the point. The spheric CO2 is assumed to be the primary greenhouse gas
Ca2+ and HCO 3 ions are carried to the ocean by the rivers. If
whose variability controls the climate.
the ocean is saturated with carbonate minerals, this increased Concerning silicate weathering, the intuition of Walker
alkalinity will cause the precipitation of calcium carbonates et al. (1981), based on measurements of the dissolution rate
(Eq. 26.5), and thus the storage of exosphere carbon in the of silicate minerals in the laboratory, has been largely veri-
sedimentary envelope of the Earth. The contribution of fied in the natural environment. The consumption flux of
alkalinity to the ocean is thus represented by silicate atmospheric CO2 by silicate weathering was measured on a
348 Y. Goddéris et al.

large number of monolithological silicate drainage basins,

  Major Climate Events in the Precambrian
both granite Fgra and basaltic ðFbas Þ. It increases with
temperature T and with continental runoff R according to the From 4.5 to 2.4 Ga
following laws:
   In the absence of indicators with the necessary level of
  48200 1 1 resolution, the climate history of the Precambrian is essen-
Fgra ¼ kgra R exp   ð26:12Þ
R T T0 tially reconstructed through numerical modeling.
   Three to four Ga ago, the solar constant was 30% lower
42300 1 1
ðFbas Þ ¼ kbas R exp   ð26:13Þ than it is currently. In this context, extremely high levels of
R T T0 greenhouse gases would have been required to prevent
prolonged global glaciation, which has not been observed in
where the fluxes are expressed in moles of CO2 consumed
the bedrock from that time or in preserved sediments. NH3
per m2 of land surface area and per year. The approximate
was suggested initially as a cause but this is not a likely
Eq. (26.11) can thus be expressed as:
candidate because it is rapidly destroyed by photolysis. This
  
48200 1 1 is not true for CO2. At least 0.3 bar of CO2 is required in the
Fvol þ FMAR ¼ Agra kgra R exp   atmosphere to counteract the low solar constant (Pavlov
R T T0
   et al. 2000). The reasons for the presence of very high levels
42300 1 1
þ Abas kbas R exp   of CO2 are related to the paleothermostat theory involving
R T T0
the erosion of continental silicates (Walker et al. 1981). The
ð26:14Þ weaker solar constant forces the climate towards colder
where Agra and Abas are the land areas where granite and conditions. This limits the consumption of CO2 by silicate
basalt areas touch. If, for some reason related to the internal erosion and allows the CO2 concentration in the air to grow
geology, soil degassing increases, increases in temperature until the climate becomes sufficiently hot and humid to
and runoff are required to maintain the carbon cycle balance. stabilize the weathering of the silicates so that it reaches the
This increase occurs naturally due to the increase in the point where it compensates for the soil degassing, leading to
concentration of CO2 in the atmosphere, in response to the a balance in the exosphere cycle of carbon. Moreover, since
degassing. Due to its dependence on climate, and because this degassing is suspected to have been much more intense
CO2 is a greenhouse gas, the land silicate weathering on land in this distant past, following the dissipation of the internal
will therefore track and compensate for the fluctuations in heat of the Earth, the CO2 concentration in the air will be set
volcanic degassing of CO2 over time and thus prevent at a high value, essential to ensure a high level of chemical
massive fluctuations in the Earth’s climate. erosion of the silicates.
The paleothermostat theory is particularly useful to However, the absence of siderite (FeCO3) in Archean
explain the absence of long-term global glaciations during paleosols suggests an upper ceiling of 0.015 bar of CO2 in
the Archean, when the solar constant was 20–30% lower the atmosphere (see references in Catling and Claire 2005).
than it is currently. The colder climate resulting from the Another greenhouse gas was therefore necessary and
lower solar energy consumption led to a slowdown in the methane may have played this role. Methanogenic bacteria
consumption of CO2 by continental silicate weathering and are probably among the first organisms to have evolved on
therefore an increase in the CO2 pressure in the air. The the surface of the Earth, and they must have constituted an
induced warming continued until the silicate weathering important part of the primitive biosphere as early as 3.5 Ga.
balanced out the soil degassing again. However, CO2 is Before 3.5 Ga, probably more methane degassed from the
probably not the only greenhouse gas involved, which ocean ridges into the prebiotic atmosphere than nowadays,
complicates the paleothermostat theory. given the lower mantle at this time (Kump et al. 2001). The
Methane, for example, is a less abundant but more effi- virtual absence of oxygen in the atmosphere before 2.2 –
cient gas than CO2 in terms of its greenhouse effect. 2.3 Ga allowed the methane to accumulate in it, probably
Methane pressure results from the balance between its pro- with mixing ratios of close to 1.6 ppmv, a value one thousand
duction time (methanogenesis) and its destruction time times higher than at present. The residence time of methane
(oxidation by OH radials). In fact, any excess of methane in air was also probably a thousand times greater than at
over this balance is not very stable and connects with the present, because, in this weakly oxidized atmosphere, the OH
CO2 paleothermostat because a sudden injection of methane radicals produced by photolysis of water reacted with H2. As
oxidizes quite quickly in CO2. a result, the surface temperature of the Earth could have been
as high as 85 °C (Kasting and Howard 2006).
26 The Precambrian Climate 349

Fig. 26.4 Qualitative curve

showing the evolution in the
concentration of CO2 and CH4 in
the atmosphere at the end of the
Archean

The CH4/CO2 ratios probably remained well below 1 up The Great Oxidation Event (GOE)
to 3.0 Ga, preventing the formation of a haze of organic
compounds in the atmosphere (Lowe and Tice 2004). The geochemical and climatological event that marked the
However, in the 3.0–2.7 Ga range, the level of atmospheric beginning of the Proterozoic is the oxygenation of the
CO2 could have been considerably reduced, following the atmosphere. Biomarkers indicate that the first photosynthetic
growth of the particularly active continental crust at that organisms appeared as early as 2.7 Ga. Atmospheric oxygen
time. The continental crust had probably reached 60% of its probably started to grow around 2.3 Ga. Over 100 or 200
present size, 3–2.9 Ga ago, whereas before 3.2 Ga it was million years, oxygen pressure increased from 10−5 bar to
only at 10%. In response to this growth in the surface area of 2
10−2 bar (Catling and Claire 2005). The d13C of sedi-
the continents, silicate erosion increased, forcing the level of mentary carbonates (Fig. 26.5) shows a major surge of more
atmospheric CO2 to decrease (Goddéris and Veizer 2000; than 10‰ at this time (see references in Catling and Claire
Lowe and Tice 2004). The average global surface tempera- 2005).
ture dropped by 10 °C (Goddéris and Veizer 2000), at the This increase in d13C is generally thought to result from
same time as the CH4/CO2 ratio increased, allowing a haze the burial of a large amount of organic carbon; this burial
of organic compounds to form around the Earth (Lowe and caused an imbalance in the carbon cycle: organic matter, low
Tice 2004). This resulted in the cooling being reinforced in 13C, was no longer depleted, while photosynthetic
before the paleothermostat slowly compensated. Traces of organisms continued to pump out carbon dioxide depleted in
13
glaciations were indeed observed in the supergroups of C, causing an increase in the d13C of atmospheric CO2
Pongola and Witwatersrand in South Africa and in the green (reflected in the d13C of carbonates) and an increase in the
chists of Berlingue in Zimbabwe, all of these formations O2 content of the atmosphere (the oxygen created by pho-
dating back to *2.9 Ga (Fig. 26.4). tosynthesis not being fully consumed during the decay of

Fig. 26.5 The evolution of d13C

in Precambrian carbonates
350 Y. Goddéris et al.

organic matter because of the increased level of conserva- The consequences for the GOE climate were significant.
tion). The issue of the oxygen balance during the Paleo- The residence time of methane in the atmosphere decreased
proterozoic is still debated to this day, and in particular, the strongly and the concentration of CH4 in the atmosphere
possible role a decrease in the quantity of reduced gas probably fell to around 300–100 ppmv, a decrease by a
degassed by the mantle is questioned (see references in factor of 5–16 compared to Archean values. This caused
Catling and Claire 2005). However, there is no doubt about significant cooling, which could not be immediately com-
the accumulation of O2 between 2.3 and 2.1 Ga (Catling and pensated for by the low level of CO2. This triggered the
Claire 2005). The GOE is also recorded in the variations in Huronian glaciations during the time it took for the pale-
the isotopic composition of iron in sedimentary pyrites othermostat to allow the partial pressure of CO2 to rise. This
between 2.6 and 1.8 Ga, which is suspected to be related to major glacial phase may have included at least one episode
the isotopic composition of seawater (Fig. 26.6). The d56Fe of total glaciation, but paleomagnetic data on the position of
of these sediments shows widely dispersed values before the continental masses remain scanty and difficult to inter-
2.3 Ga, but this dispersion decreases considerably after pret. However, recent paleomagnetic studies show that the
2.3 Ga. These changes are explained by the existence of an Huron glaciation was probably similar to the ‘standard’
iron rich ocean, fed by hydrothermal springs (containing glaciations of the Quaternary. However, the GOE context is
dissolved iron Fe2 +) before 2.3 Ga. This ocean periodically completely different in terms of atmospheric composition,
purges itself of a certain amount of its iron during the pre- paleogeography, solar insolation… and therefore the rhythm
cipitation of iron oxides during upwelling episodes. Such of this glaciation is still an open question.
purges brought water to more or less oxygenated zones. This
precipitation extracts the isotope 56 preferentially from the
iron; this process of fractional precipitation of iron is thus The Proterozoic
capable of modifying the d56Fe isotopic composition of the
deep ocean so long as all the iron carried by the upwellings Following the Huronian glaciations, the Earth seems to have
has not precipitated and a part of it returns to the deep ocean. been subjected to a warm climate persisting over most of the
After 2.3 Ga and the GOE, the Proterozoic ocean became Proterozoic, lasting for about 1 billion years. To date, no
stratified, characterized by a permanently and thoroughly trace of glaciation has been found in the timespan 2–0.8 Ga.
oxygenated surface zone and an anoxic deep ocean. This The measurement of d13C from acrytarches (microfossil
configuration allows the precipitation of all of the remains of cysts of photosynthetic eukaryotic organisms)
hydrothermal Fe2 + brought by the upwellings (see refer- allows a rough estimation of the atmospheric CO2 level.
ences in Catling and Claire 2005). In this way, the isotopic Indeed, the difference between these ratios and the mean
composition of the deep ocean is no longer affected, since all ratio of oceanic carbonates (close to 0‰) defines, with some
the iron carried by upwellings precipitated, and stabilized at assumptions, the isotopic fractionation involved in photo-
between 0 and 1‰. Finally, recent data using the indepen- synthesis that occurred during the Calvin cycle. This frac-
dent fractionation of the mass of isotopes of sulfur D33S tionation depends, among other things, on the CO2 pressure
confirm progressive oxidation of the surface layers from 2.4 of the water and can therefore be linked to atmospheric CO2
to 2.3 Ga onwards (Papineau et al. 2007). pressure. At 1.4 Ga, the level of CO2 was about 10–200

Fig. 26.6 The d56Fe of

Precambrian sediments
26 The Precambrian Climate 351

times the pre-industrial level i.e. 2800–56,000 cm3 of CO2 deposits was located mostly in the intertropical zone, which
per m3 of air (Kaufman and Xiao 2003). implies a major glaciation, since the ice reached the equator
The level of Proterozoic methane is uncertain. The oxy- (Evans 2000). The histogram showing the presence of gla-
gen level (Fig. 26.7) is estimated to be about 5–18% of its cial deposits as a function of paleolatitude, is totally atypical
current level, based on the low sulfate concentration of for snowball glaciations. Although this histogram shows a
Proterozoic marine carbonates (Pavlov et al. 2003). Since peak at high latitudes for all of the Phanerozoic glaciations,
these sulfates are produced by the oxidation of sulphides, this peak is displaced to the lower latitudes for snowball
their low abundance reflects a low partial pressure of oxy- glaciations (Evans 2000). (2) Glacial deposits are directly
gen. Based on these values, Pavlov et al. (2003) calculated a overlaid with atypical carbonate deposits (cap carbonates)
methane pressure in the range of 100–300 ppmv, assuming a with no break, suggesting a transition of extraordinary
CH4 production in the deep anoxic ocean basins 20-fold rapidity on a geological scale from a very cold climate to a
higher than current levels. The surface temperature is very hot climate (Hoffman et al. 1998). If the Earth was
unknown, the d18O of the cherts suggest 50 °C at the end of covered with ice, the hydrological cycle would have almost
the Precambrian. As seen above, these high values remain completely shut down, thereby allowing CO2 to accumulate
questionable. In order to maintain an average global tem- in the atmosphere as a result of volcanic degassing. When
perature of 15 °C on the Earth’s surface, 300 PAL of CO2 more than 0.29 bar of CO2 has accumulated in the atmo-
were needed at 2 Ga, and only 10 PAL at 0.6 Ga (1 sphere (Pierrehumbert 2004), the greenhouse effect intensi-
PAL = present atmospheric level). Log10. fies and deglaciation is suddenly initiated. A climate
characterized by a strong greenhouse effect followed the
very cold climate. Continental weathering recommenced and
The End of the Proterozoic: Global Glaciations quickly reached a high level. The massive surge of alkalinity
in the ocean became predominant over carbonates, which
After more than a billion years of no glaciations, the end of explains the presence of carbonate deposits directly on top of
the Proterozoic (the Neoproterozoic, from 900 to 543 Ma) glacial deposits. (3) Banded iron formations (BIF) reap-
was marked by the strongest glaciations in the history of the peared during and just after glaciation, although these had
Earth. These are suspected to have been global, hence the disappeared during the Proterozoic, around 1.8 Ga. The
name ‘snowball’ glaciations. Two events are acknowledged: return of the BIFs is qualitatively compatible with the
the first between 723 and 667 Ma (the Sturtian glaciation) installation of sea ice over the whole of the oceans, greatly
and the second between 667 and 634 million years (the reducing the vertical mixing of the ocean and favoring the
Marinoan glaciation). They were followed by a glacial epi- development of anoxic conditions in the deep ocean. The
sode of lower intensity around 583 Ma, comparable to the Fe2+ emitted at the ridges can therefore be transported by
glaciations of the Phanerozoic (the Gaskiers glaciation). upwellings to the surface waters, where it precipitates as BIF
The Sturtian and Marinoan glaciations have a particular in contact with oxygen. (4) The presence of an iridium peak
set of characteristics. (1) The paleolatitude of the glacial in the basal cap carbonates suggests an accumulation of

Fig. 26.7 History of the level of

oxygen in the atmosphere
352 Y. Goddéris et al.

iridium (brought by cosmic materials falling to the ground) Glaciation Onset

on pack ice for 3–12 million years and its inclusion in
sediments at the time of an ice flood corresponding to the Using models coupling climate and the carbon cycle, Don-
end of Marinoan glaciation (Bodiselitsch et al. 2005). These nadieu et al. (2004) showed that glacial triggering was clo-
last decades the chronology of Neoproterozoic glaciations sely linked to the configuration of the continents during the
has been considerably improved. The chronology is now late Proterozoic. From 800 Ma onwards, the supercontinent
well established with two dissymmetric glaciations: a long Rodinia (Fig. 26.9), located at the equator, began to drift.
lasting Sturtian episode for more that 50 Ma and a shorter Numerous basaltic effusions mark the beginning of rift-
Marinoan episode lasting around 15 Ma separated by a short ing. The formation of these highly weathered basaltic sur-
interglacial period only lasting 10 Ma (Hoffman et al. 2017). faces led to an increase in the consumption of atmospheric
(5) The d13C isotopic ratio of the cap carbonates is well CO2 (Goddéris et al. 2003). Moreover, the continental
documented, especially for the most recent snowball blocks began to drift, but remained between the latitudes 60°
glaciation (Fig. 26.8). It is particularly low at around S and 60° N. The dislocation of Rodinia increased the supply
−3‰ towards the end of the glaciation, reaching very low of moisture to the continents, which favored continental
values (−5‰) at the top of the cap carbonates (Halverson weathering. This weathering occurred even more rapidly on
et al. 2005). large continental areas located in the hot and humid
As early as 1998, Hoffman et al. interpreted this low intertropical convergence zone. CO2 consumption increased
value as a sign of shallow burial of organic carbon, in turn, and atmospheric pressure of CO2 plunged. The dislocation
an indicator of significantly slower biological productivity in of Rodinia alone (Fig. 26.10) and the consequent increase in
the oceans. By taking the isotopic fractionation of carbon weathering (Fig. 26.11) explain a decrease in radiative
during photosynthesis at −20‰, the proportion of organic forcing of 6.85 W/m2 and an overall cooling of 8 °C. To this
carbon buried relative to total carbon sedimentation increa- effect is added the intensified weathering of the fresh basaltic
ses from 10% at the base of the cap carbonates to 0% at the surfaces brought into the hot and humid climatic zones,
top (compared to 25% today). These low values suggest an conditions favoring their weathering (Goddéris et al. 2003).
oceanic biosphere very affected by the glacial episode, This is particularly true of the magmatic Laurentian
which is indicative of its magnitude. Nevertheless, there are province (Donnadieu et al. 2004). The combined effects of
still many uncertainties concerning the interpretation of the lowering continental masses and weathered basaltic surfaces
cap carbonates. They are largely related to the fact that cap caused the system to tip into global glaciation, bringing sea
carbonates are in fact dolomites and that the kinetics of their ice as far south as 30°, at which point the positive feedback
precipitation is still not fully understood although it appears between albedo, ice cover and cooling accelerated. Atmo-
to be associated with general conditions of anoxia and spheric CO2 increased from over 1800 ppmv, 800 Ma ago at
probably with a bacterial activity reducing the amount of the time of Rodinia, to a level below the threshold for global
sulphates at the water-sediment interface. glaciation of 250 ppmv.

Fig. 26.8 Evolution of the d13C

of sedimentary carbonates at the
end of the Proterozoic. The two
gray bands indicate the
glaciations thought to be global
26 The Precambrian Climate 353

Fig. 26.9 The supercontinent

Rodinia, about 800 million years
ago

The tectonic theory for glacial inception has two The methane valve then closes for reasons that have still to
advantages: be discovered and the remaining CO2 level is no longer
sufficient to prevent tipping into global glaciation. Finally,
1. It explains the persistence of cold climates, even outside Pavlov et al. (2003) suggest that atmospheric methane levels
the episodes of ‘snowball’ glaciation for the duration of remained high throughout the Proterozoic (100–300 ppmv)
the Neoproterozoic, as long as the continents are dis- and that an additional episode of oxygenation of the surface
persed along the equator. This configuration is observed layers created the conditions favorable to glaciation in the
in a window from 750 to 600 Ma (Torsvik et al. 2001) Neoproterozoic. This scenario is similar to the one elabo-
and thus includes the two extreme episodes. rated to explain the Huronian glaciations. Rapid oxidation of
2. It also explains why such global glaciations did not occur methane to CO2 considerably reduces the greenhouse effect
in more recent times. The equatorial configuration of the in a time window too short for the paleothermostat to
continents, the driver of global glaciation, never occurred rebalance the temperature of the Earth’s surface.
during the Phanerozoic.

The possibility that clathrates (complexes formed from During Glaciation

several molecules of methane and present today in certain
marine sediments) played a destabilizing role has been During the main glacial period, the near absence of sedi-
proposed to explain the triggering of the Marinoan glaciation mentation does not allow data-based reconstruction. As a
and the decrease of the d13C of the carbonate sediments just result, all theories are based on climate modeling. The
before glaciation (Fig. 26.8). This hypothesis suggests an ad average annual temperature over land during a complete
hoc increase in CH4 degassing from sediments just before glaciation is about −25 °C in the equatorial zone, with the
glaciation. The result is a decrease of d13C in the sea water coldest point being reached on the West African Craton with
and an increase in the level of atmospheric CH4. The −110 °C (Donnadieu et al. 2003). The oceans become
resulting global warming disturbs the paleothermostat and covered over with ice very quickly. The thickness of the sea
CO2 is consumed faster by the weathering of the silicates. ice is subject of debate. Model-based simulations suggest
354 Y. Goddéris et al.

Fig. 26.10 The Rodinia

dislocated about 750 million
years ago

that this thickness must have rapidly exceeded a kilometer light snowfall would have continued on the continents due to
(Goodman 2006). However, for photosynthetic activity to the sublimation of ice from the sea ice. In addition, the ice
persist during glaciation, the existence of thinner ice sheet would have remained dynamic as its base was wet. Ice
allowing light to filter through, at least in some areas, would flow rates of 5–10 m/year have been calculated for the most
be required. In contrast to the simulation suggesting thick ice continental parts of the ice sheet and a rate of 50 m/year for
everywhere, Hyde et al. (2000) presented a possibility where coastal areas (Donnadieu et al. 2003). An active water cycle
open water areas persisted in the tropical zone, while the rest is therefore maintained during total glaciation, even if it is
of the ocean was completely covered with ice. This simu- greatly reduced.
lation was performed by coupling an energy balance model
with an ice model. Nevertheless, simulations carried out with
climate models taking into account the general circulation of Exiting from Glaciation
the atmosphere and the ocean show that this solution is not
stable. The ocean was probably entirely covered with ice In order to exit from glaciation, very high levels of CO2 are
with no real “oasis zones” where life could exist. By con- required. Caldeira and Kasting (1992) calculated that a fully
trast, the presence of thin ice (of around 10 m thick) in the engulfed Earth would require 0.12 bar of CO2 to initiate
equatorial zone may have been possible. This was suggested deglaciation. According to these authors, the greenhouse
by McKay (2000) and then modeled by Pollard and Kasting effect then becomes sufficiently powerful to counteract the
(2005). As for land masses, ice sheets became quickly high albedo of a totally frozen Earth. Unfortunately, the
established. Donnadieu et al. (2003) showed that by setting energy balance model used to produce this estimate was
the CO2 pressure at the pre-industrial value of 280 ppmv and probably too simple for this particular environment but, even
by reducing the solar constant by 6%, the continental ice more importantly, the solar constant was fixed at its present
sheet would have taken 400,000 years to reach its equilib- value. Taking a solar constant of 6%, the deglaciation
rium size of 190 million km3 of ice covering 90% of con- threshold becomes 0.29 bar. Pierrehumbert (2004) has also
tinental surfaces (Fig. 26.12). Its maximum thickness would shown that the use of a GCM further pushes back this
have been 5000 m. Even when the ice cover became total, threshold for deglaciation and that 0.29 bar must be
26 The Precambrian Climate 355

Fig. 26.11 Weathering rate of

continental silicate rock
(104 molCO2/km2/year) during
the Neoproterozoic for the
Rodinia configuration and for the
dislocated configuration at 1800
ppmv of CO2

Fig. 26.12 Numerical

simulation of the thickness of
continental ice under
Neoproterozoic conditions, at 280
ppmv of CO2
356 Y. Goddéris et al.

considered to be a low limit. The reasons it is difficult to melt ensure a massive diffusion of CO2 into the ocean, ensuring
the ‘snowball’ Earth are multiple. One of these is that the the balance between the ocean and the atmosphere (Le Hir
dryness of the atmosphere resulting from the very low et al. 2007). Under these conditions, the ocean undergoes
temperatures limits the formation of high-level clouds and major acidification (the pH drops to 6) and the weathering of
thus the greenhouse effect. Similarly, the virtual absence of the oceanic crust becomes an efficient carbon sink, pro-
water vapor limits warming of the planet. longing glaciation by counteracting the growth of atmo-
During the glaciation period, CO2 accumulates in the spheric CO2. If this process is included, it appears that even
atmosphere. Indeed, the virtual disappearance of the over 30 million years, atmospheric CO2 does not reach the
hydrological cycle means that atmospheric CO2 is no longer threshold for deglaciation (Fig. 26.13)
absorbed by continental weathering (Hoffman et al. 1998). The most recent studies show that an important problem
Carbon therefore accumulates in the atmosphere through with the hypothesis of ‘Snowball’ Earth is understanding the
volcanic degassing. By taking the current degassing rate of conditions under which it melts. Nevertheless, processes
6.8
1012 mol of CO2 per year, it takes 8 million years for able to reduce the duration of the glaciation have never been
0.29 bar of CO2 to accumulate in the atmosphere. This tested. For example, prolonged volcanic activity during
timeframe is compatible with the estimates of the duration of glaciation probably led to the accumulation of ash on the ice,
‘Snowball’ Earth (Bodiselitsch et al. 2005). However, this decreasing its albedo, and therefore, the CO2 threshold
only applies if all of the carbon sinks are stopped during required to initiate deglaciation.
glaciation and this is not the case, since the existence of thin
ice allows a fracture at the ocean-atmosphere interface.
A surface area of 3000 km2 of open water is sufficient to Conclusion

Twenty years ago, the Precambrian was a large unknown

gap in our understanding of the geochemical and climate
history of our planet. This period, lasting approximately 4
billion years, saw the emergence of the major regulatory
processes in our environment (the emergence of the conti-
nents, the establishment of plate tectonics, the appearance of
life and of the modern atmosphere). Our understanding
gradually unfolded as the experimental techniques (in par-
ticular isotopic techniques) advanced and as modeling
improved. This has been a major leap forward that has
occurred in the past ten years. Very significant progress has
been made in describing the evolution of the environment at
the Archean-Proterozoic boundary. Today, thanks to the
ever more refined study of the isotopic ratios of sedimentary
rocks of this period (d13C, d56Fe, D33S), the Great Oxy-
genation Event is one of the best documented events of the
Precambrian. Similarly, the description of the major, possi-
bly global, glaciations that marked the end of this very long
period has improved dramatically in recent years. We now
know the number, age and probable extent of the glaciations
that preceded the Cambrian explosion of life, and we have
an increasingly coherent image of their modalities.
However, four billion years is a long time and many areas
of uncertainty still remain. We have progressed to the point
where certain key periods in the Precambrian have been
clearly illuminated. For instance, the description of Neo-
proterozoic glaciation has improved on many issues. The
most striking one of the new Cryogenian chronology is the
grossly unequal duration of the cryochrons (Fig. 5.14a). The
Sturtian lasted four time longer than the Marinoan. Another
Fig. 26.13 Evolution of CO2 concentration in the atmosphere and pH
of the ocean during a global glaciation, assuming contact between the surprising caracteristic of the new chronology is the brevity
ocean and the atmosphere via fractures in sea ice of the nonglacial interlude between the cryochrons. When
26 The Precambrian Climate 357

Fig. 26.14 Glacial epochs on Earth since 3.0 Ga. a Black bands gray) since 3.0 Ga. Ellipse GOE is centered on the Great Oxidation
indicate duration of the Sturtian and Marinoan cyrochrons (Table 26.1). Event, as recorded by the disappearance of mass-independent S isotope
The graded start to the Marinoan cryochron denotes chronometric fractionations  0.3 per mil (‰) in sedimentary sulfide and sulfate
uncertainty, not gradual onset. B Snowball Earth chrons (black), mineral. The dashed gray line indicates questionable glaciation.
regional-scale ice ages (medium gray), and nonglacial intervals (light Figure from Hoffman et al., Sci. Adv. 2017

the Marinoan began, a Snowball Earth had terminated less numerical modeling. Earth and Planetary Science Letters, 08, 101–
than 20 Ma earlier. When the Sturtian began, no low-latitude 112.
Donnadieu, Y., Goddéris, Y., Ramstein, G., Nédelec, A., & Meert, J. G.
glaciation had occurred for 1.7 Ga (Fig. 5.14b). Neverthe- (2004). Snowball EARTH triggered by continental break-up
less, more obscure periods, marked by slower, more pro- through changes in runoff. Nature, 428, 303–306.
gressive changes and yet which cover very long portions of Evans, D. A. D. (2000). Stratigraphic, Geochronological, and paleo-
the Precambrian, have still to be elucidated. In particular, the magnetic constraints upon the Neoproterozoic climatic paradox.
American Journal of Science, 300, 347–433.
reasons why most of the Proterozoic did not experience any François, L. M., & Goddéris, Y. (1998). Isotopic constraints on the
glaciation need to be explored. Is this truly the case or is cenozoic evolution of the carbon cycle. Chemical Geology, 145,
there a bias in our observation of Precambrian formations? If 177–212.
this is true, then this would be the longest persistent hot Gaucher, E. A., Govindarajan, S., & Ganesh, O. K. (2008). Paleotem-
perature trend for precambrian life inferred from resurrected
period (1.2 billion years!) in the history of our planet. Why proteins. Nature, 451, 704–708.
are the great quasi-periodic glacial advances observed since Goddéris, Y., et al. (2003). The Sturtian Glaciation: Fire and ice. Earth
the emergence of multicellular life not active in the Paleo and Planetary Science Letters, 211, 1–12.
and Meso-Proterozoic? The climate history of the Precam- Goddéris, Y., & Veizer, J. (2000). Tectonic control of chemical and
isotopic composition of ancient oceans: The impact of continental
brian that we have so far reconstructed concerns essentially growth. American Journal of Science, 300, 434–461.
that of its glacial crises (the GOE, the Neoproterozoic), but Goodman, J. C. (2006). Through thick and thin: Marine and meteoric
these represent only a small fraction of the immensely long ice in a ‘Snowball Earth’ climate. Geophysical Research Letters, 33.
Precambrian. Reconstructing a less disjointed image of this https://doi.org/10.1029/2006gl026840.
Halverson, G. P., Hoffman, P. F., Schrag, D. P., Maloof, A. C., & Rice,
period is a major challenge for the coming decades. A. H. N. (2005). Towards a Neoproterozoic composite carbon
isotope record. Geological Society of America Bulletin, 117, 1181–
1207.
References Hoffman, P. F., Abbot D. S., Ashkenazy, Y., Benn, D. I., Brocks, J. J.,
Cohen, P. A., et al. (2017). Snowball Earth climate dynamics and
Cryogenian geology geobiology. Science Advances.
Bodiselitsch, B., Koeberl, C., Master, S., & Reimold, W. U. (2005). Hoffman, P. F., Kaufman, A. J., Halverson, G. P., & Schrag, D.
Estimating duration and intensity of Neoproterozoic snowball P. (1998). A Neoproterozoic Snowball Earth. Science, 281, 1342–
Glaciations from Ir Anomalies. Science, 308(5719), 239–242. 1346.
Caldeira, K., & Kasting, J. F. (1992). Susceptibility of the early earth to Hyde, W. T., Crowley, T. J., Baum, S. K., & Peltier, W. R. (2000).
irreversible Glaciation caused by carbon dioxide clouds. Nature, Neoproterozoic ‘Snowball Earth’ simulations with a coupled
359, 226–228. climate/ice sheet model. Nature, 405, 425–429.
Catling, D. C., & Claire, M. W. (2005). How Earth’s atmosphere Kasting, J. F., & Howard, M. T. (2006). Atmospheric composition and
evolved to an oxic state: A status report. Earth and Planetary climate on the early earth. Philosophical Transaction Royal Society
Science Letters, 237, 1–20. London B, 361, 1733–1742.
Donnadieu, Y., Fluteau, F., Ramstein, G., Ritz, C., & Besse, J. (2003). Kasting, J. F., et al. (2006). Paleoclimates, ocean depth, and the oxygen
Is there a conflict between the Neoproterozoic Glacial deposits and isotopic composition of seawater. Earth and Planetary Science
the Snowball Earth interpretation: An improved understanding with Letters, 252, 82–93.
358 Y. Goddéris et al.

Kaufman, A. J., & Xiao, S. H. (2003). High CO2 levels in the the rise of atmospheric oxygen. Earth and Planetary Science
proterozoic atmosphere estimated from analyses of individual Letters, 255, 188–212.
microfossils. Nature, 425, 279–282. Pavlov, A. A., Hurtgen, M. T., Kasting, J. F., & Arthur, M. A. (2003).
Knauth, L. P., & Epstein, S. (1976). Hydrogen and oxygen isotope Methane-rich proterozoic atmosphere. Geology, 31, 87–90.
ratios in nodular and bedded cherts. Geochimica Cosmochimica Pavlov, A. A., Kasting, J. F., Brown, L. L., Rages, K. A., & Freedman,
Acta, 40, 1095–1108. R. (2000). Greenhouse warming by CH4 in the atmosphere of Early
Kump, L. R., Kasting, J. F., & Barley, M. E. (2001). Rise of Earth. Journal of Geophysical Research, 105(E5), 11981–11990.
atmospheric oxygen and the ‘upside-down’ Archean Mantle. Pierrehumbert, R. T. (2004). High levels of atmospheric carbon dioxide
Geochemistry Geophysics Geosystems, 2(1). https://doi.org/10. necessary for the termination of global Glaciation. Nature, 429,
1029/2000gc000114. 646–649.
Le Hir, G., Goddéris, Y., Ramstein, G., & Donnadieu, Y. (2007). Pollard, D., & Kasting, J. F. (2005). Snowball Earth: A thin-ice solution
A scenario for the evolution of the atmospheric pCO2 during a with flowing sea glaciers. Journal of Geophysical Research, 111.
Snowball Earth. Geology (in press). Robert, F., & Chaussidon, M. (2006). A palaeotemperature curve for
Lowe, D. R., & Tice, M. M. (2004). Geologic evidence for Archean the precambrian oceans based on silicon isotopes in cherts. Nature,
atmospheric and climatic evolution: Fluctuating levels of CO2, CH4, 443, 969–972.
and O2 with an overriding tectonic control. Geology, 32(6), 493– Torsvik, T. H., et al. (2001). Rodinia refined or obscured: Paleomag-
496. netism of the Malani igneous suite (NW India). Precambrian
McKay, C. P. (2000). Thickness of tropical ice and photosynthesis on a Research, 108, 319–333.
Snowball Earth. Geophysical Research Letters, 27, 2153–2156. Walker, J. C. G., Hays, P. B., & Kasting, J. F. (1981). A negative
Papineau, D., Mojzsis, S. J., & Schmitt, A. K. (2007). Multiple sulfur feedback mechanism for the long-term stabilization of earth’s
isotopes from paleoproterozoic Huronian interglacial sediments and surface temperature. Journal of Geophysical Research, 86, 9776–
9782.
 The Phanerozoic Climate
27
Yves Goddéris, Yannick Donnadieu, and Alexandre Pohl

The Phanerozoic period covers the last 542 million years of

The Proxies for the Phanerozoic Climate
Earth’s history, about 12% of the history of our planet. With
regard to the evolution of life, the Phanerozoic experienced
There are no direct indicators of climate conditions in the
major events such as the rapid diversification of multicellular
geological past. However, qualitative reconstructions can be
organisms which first appeared in the Cambrian (541–
produced from sedimentological and paleontological data,
485 Ma), the colonization of continental surfaces by living
and, in general, are more reliable than they are for the Pre-
organisms during the Ordovician (485–444 Ma) and the
cambrian. Geochemical data, in particular the isotopic indi-
appearance of the first hominids about 8 million years ago.
cators measured in marine sediments, provide some
Over this same period, the appearance of the Earth’s surface
quantification, but interpretation of them is rarely straight-
changed considerably: the continental drift during the
forward. Below we provide a non-exhaustive list of exam-
Phanerozoic moved all of the continental masses situated in
ples of indicators. Finally, we will highlight the indicators
the southern hemisphere and along the equator since the
that allow the reconstruction of CO2 concentration in the
Cambrian to join together around 280 million years ago to
atmosphere over geological time.
form a supercontinent: the Pangea. This would then disin-
tegrate during the Jurassic around 180 million years ago
with the emergence of the Atlantic Ocean. Sedimentological Indicators
Our understanding of the Phanerozoic climate, even if it
is still incomplete, has been rapidly evolving in recent years. A compilation of sedimentological data indicative of glacial
Data accumulated over the last few decades has improved climate was carried out in 1992 by Frakes et al. (1992). It
consistently in quality and the numerical models used to consists of an inventory of tillite-type glacial deposits (clays
reconstruct past climates have evolved considerably: they formed from erosion products resulting from the friction of
have a better spatial resolution and include an increasing glaciers on their bedrock) and, on the other hand, a recon-
number of processes. In general, it is true to say our con- struction of the minimum paleolatitude attained by rock
ception of how and why the Earth’s climate evolved during debris carried by sea ice. The result shows a fluctuation of
this key period in the history of life on Earth is now hot and cold modes over a period of approximately 135 Ma.
undergoing a true revolution. The coldest climate mode was identified during the
Permo-Carboniferous glaciation.
A more recent study provides results in agreement with
those of Frakes et al. (1992). Boucot et al. (2004) compiled
data on continental coal deposits, indicators of an arid cli-
mate, as a function of paleogeography and time. They con-
structed a qualitative curve of variations on the equator-pole
climate gradient for the entire Phanerozoic. They interpret
Y. Goddéris (&)
Géosciences Environnement Toulouse, CNRS-Université de the existence of weak gradients as the sign of a warm
Toulouse III, UMR 5563, Toulouse, France climate.
e-mail: [email protected]
Y. Donnadieu
A. Pohl
Aix-Marseille Université, CNRS, IRD, Coll France, CEREGE,
Aix-en-Provence, France

© Springer Nature Switzerland AG 2021 359

G. Ramstein et al. (eds.), Paleoclimatology, Frontiers in Earth Sciences,
https://doi.org/10.1007/978-3-030-24982-3_27
360 Y. Goddéris et al.

Isotopic Indicators processes on the d18O of seawater is not yet clearly under-
stood. There is as yet no consensus on this issue especially
The d18O of Carbonates since recent studies suggest that the value of the d18O of the
A detailed study of the isotopic composition of oxygen ocean has remained constant since 760 Ma (Bergmann et al.
(d18O) in carbonate sediments was carried out mainly on 2018; Hodel et al. 2018).
fossilized brachiopod shells (Veizer et al. 1999). It shows The second trend highlighted in the long-term recordings
two trends (Fig. 27.1). The first is a slow, almost linear, of the d18O of the brachiopods is the periodic oscillations
increase of d18O from the Cambrian, from values of around superimposed on the long-term linear trend described pre-
−10‰ (relative to the standard Pee Dee Belemnite) up to viously. If this is subtracted, the oscillations have an
current values close to 0‰. This increase is still difficult to amplitude of 3–5‰ (Veizer et al. 2000). The most surprising
interpret. If the d18O of the ocean has remained close to its aspect is that the period of these oscillations is in agreement
present value and if the evolution of the d18O of the bra- with the periodicity of the hot and cold modes determined by
chiopods is interpreted in terms of temperature over the last Frakes et al. (1992), suggesting the presence of a true climate
540 million years, then the temperature of seawater must signal. The use of a paleothermometer, linking the isotopic
have reached 70 °C in the Cambrian, a level which is lethal fractionation between calcite and seawater to the precipita-
to most marine organisms and therefore difficult to reconcile tion temperature of the carbonate, makes it possible to
with the very large phase of diversification of marine reconstruct the temperature variations of the water in which
organisms documented at this time (Zhuravlev and Riding the brachiopods lived, provided that the d18O of the seawater
2001). Two possibilities have been proposed to solve this is known, a fact dependent on the volume of continental ice.
paradox: either the decrease of d18O in the past reflects a It should be noted, however, that examples of diagenetic
diagenetic alteration of the brachiopod shells, in which case alteration have been identified in which isotopic exchange
the signal is irrelevant, or the d18O of seawater was lower in with runoff leads to values for d18O very different from the
the past. Seawater is, in fact, influenced by the tectonic original values, but in which seasonal oscillations seem to be
processes: as silicate rocks are transformed into d18O-de- preserved. This is merely an artifact. Finding a
pleted clay sediments, continental and hydrothermal alter- pseudo-climatic periodicity in a diagenesis signal is not
ation at low temperatures tend to increase the d18O of the impossible and does not constitute proper evidence of the
water in contact with the minerals. A fractionation of 20% is preservation and the consistency of the isotopic signal.
observed for the low-temperature alteration at the ridges, The fractionation a between calcite and water is expres-
12.5% for the continental alteration, while the sed by the relationship:
high-temperature hydrothermal alteration decreases the d18O
of seawater by enriching the alteration production with a 18:03  103
T ðKÞ ¼ ð1Þ
fractionation of −18‰. The role of these geological 1000 ln a þ 32:42

Fig. 27.1 d18O of carbonate

sediments (calcite) measured over
the whole Phanerozoic. The two
light and dark shaded ranges
contain 68% and 95%
respectively of the data points.
The lozenges represent
measurements made on aragonitic
fossils
27 The Phanerozoic Climate 361

where temperature was approximately 10 °C at the end of the

Cretaceous and could have reached 15 °C around 100 Ma
1000 þ r18 OPDB
sample (Friedrich et al. 2012). Measurements for the Mesozoic are
a¼ ð2Þ
1000 þ r18 OPDB
solution
also covered by d18O on phosphates (see next section).
Finally, the Cenozoic is covered by a high-resolution
is the d18O (PDB) of the carbonate sample, and d18O is the database (Zachos et al. 2008), which is the most significant
d18O of the solution from which the carbonate precipitated. step forward in terms of climate reconstruction (Fig. 27.3).
A d18O of seawater of 0‰ compared to SMOW is equiva- The d18O measurements carried out in forty ODP and DSDP
lent to −30‰ compared to PDB. A variation of −1‰ in the drill sites, were conducted on benthic foraminifera that once
d18Osample corresponds to an increase of +4 °C in the tem- lived in the deep ocean. They are generally considered to be
perature at which precipitation occurred. indicators of changes in surface water temperature at high
The volume of continental ice and the d18O of seawater latitudes (where dense surface waters sink to produce the
are far from being well known through the geological past of deep waters of the global ocean) and of changes in the
the Earth. To simplify, Veizer et al. (2000) assumed the isotopic composition of the ocean on average, which are a
volume of continental ice to be twice the current volume at function of the ice cap volume. They are particularly good at
the glacial period peaks and to be zero during a warming recording the phases of rapid growth of the Antarctic ice
period. An oscillation in the ice volume of this scale could sheet.
account for a 2‰ change in the d18O of seawater, which
presumes that the remaining 1–3‰ are attributable to tem- The d18O of Phosphates
perature changes. Making these assumptions, Veizer et al. Another particularly promising approach is based on the
(2000) propose that seawater in the equatorial zone (where study of d18O measured in phosphates, particularly in fish
all fossil brachiopods have been found) was up to 3.5 °C teeth or conodonts, small tooth-shaped structures of 0.25–
colder during the Ordovician glacial maximum than now, 2 mm, consisting of apatite and having belonged to vermi-
3 °C colder during the Permo-Carboniferous and 2 °C form animals that disappeared at the end of the Triassic. The
colder during the Jurassic, without taking into account rapid paleothermometer is expressed as follows:
shifts which could reach amplitudes of 9 °C (Fig. 27.2).

It should be noted that the majority of the brachiopod Tð CÞ ¼ 112:2  4:2 r18 OSMOW
sample  r 18 SMOW
O water ð3Þ
fossils used are from the Paleozoic (Veizer et al. 1999) and
that the resolution for the Mesozoic and Cenozoic ages is r18 OSMOW 18
sample is the d O (SMOW) of the phosphate sample,
weak within this Phanerozoic database. The Mesozoic was
and r18 OSMOW 18
water is the d O of the solution from which the
covered more precisely by measurements of d18O from
phosphate precipitated.
benthic and planktonic foraminiferal shells (e.g., Bice et
The advantage of phosphates, especially the enamel of
Norris 2002) and on belemnite rostra (Dera et al. 2011). One
fossil teeth and conodonts, is their greater resistance to
of the most remarkable results is the estimation of the
diagenetic alteration than carbonates. In general, the d18O
deep-water temperature during the Cretaceous. This
measured on phosphate does not appear to show significant

Fig. 27.2 Temperature

anomalies of tropical waters
reconstructed from d18O
carbonate data
362 Y. Goddéris et al.

Fig. 27.3 d18O of benthic

foraminifera during the Cenozoic

decay as one goes back in time, suggesting a lower sensi- The ‘Clumped’ Carbonate Isotope Method
tivity to diagenesis and therefore better reliability. This has or the D47 Method
been clearly demonstrated for the Devonian. The d18O The major problem with using oxygen isotopes for the
measured on apatite from conodonts indicates a seawater reconstruction of seawater temperatures in the past is the
temperature of about 25 °C for the end of Givetien and lack of knowledge of the d18O ratio of the seawater in which
Frasnian (391–374 Ma), taking the d18O of the seawater to the carbonates and phosphates formed. A new technique has
be −1‰ due to the probable absence of ice caps, as these recently been proposed, which makes it possible to over-
only developed during the Famennian (Caputo et al. 2008). come this limitation. This involves essentially counting the
The d18O measured on the calcite of the brachiopod shells number of bonds between rare isotopes in the CaCO3
from the same period indicates significantly higher temper- molecules, in particular, the 13C–18O bonds. The difference
atures of between 30 and 40 °C (Veizer et al. 1999). Similar between the actual number of rare bonds and the number of
differences are observed in the amplitudes of the temperature bonds there would be if the bonds were stochastically dis-
changes between the two methods. For example, measure- tributed depends entirely on the temperature and not at all on
ments of d18O on phosphates suggest a drop in tropical water the isotopic composition of the water in which the carbonate
temperature from 40 to 32 °C between 490 and 465 million was formed. It is measured with the assistance of D47:
years which seems to be correlated with a major acceleration  47 
in the expansion of biodiversity (Trotter et al. 2008). The Rmeasured
D47 ¼  1  1000 ð4Þ
d18O data from brachiopods suggest a temperature drop of R47
stochastic
only 4 °C over the same period. It appears that the isotopic
composition of brachiopod shells depends largely on kinetic where R47measured is the ratio of the mass of
18 13 16
O C O
16 12 16
fractionation processes (typical of diagenesis) and, to a lesser molecules to the mass of light O C O molecules mea-
extent, on the metabolism of these animals. Consequently, sured in the CO2 emitted from the attack on carbonate by
the d18O measured on the brachiopods could be a weak phosphoric acid. R47 stochasitc is the same as the ratio for a
reflection of the environmental conditions that prevailed at stochastic distribution of the molecules. The D47 depends on
the time of the formation of the shell. Nevertheless, the the temperature of the medium in which the carbonate
debate on the validity of the brachiopod data is still ongoing, formed (Ghosh et al. 2006) according to the formula:
especially since the recent publication of a new paleother-
D47 ¼ 0:0592ð106  T 2 Þ  0:02 ð5Þ
mometer which revises upwards the temperatures recon-
structed from phosphates (Pucéat et al. 2010). This technique also has the advantage of being impervi-
The use of fish teeth from various parts of the world ous to diagenesis within a temperature range of 0–200 °C.
supports the reconstruction of latitudinal gradients of water The first use of this technique was devoted to the study of
temperatures, which provides essential clues to climates in samples from the Lower Silurian (around 435 million years)
the distant past. Finally, data from the teeth of fossil verte- and from the Middle Pennsylvanian (Carboniferous, around
brates offer immense opportunities in terms of the mea- 310 million years) (Came et al. 2007). It produces contra-
surement of temperatures and their latitudinal gradients in dictory results: for the Carboniferous samples, they are, for
continental environments. example, in agreement with the d18O measurements on
27 The Phanerozoic Climate 363

Fig. 27.4 Phanerozoic

temperature anomalies. The black
line tracks the output of a
numerical model and represents
the mean global temperature
anomalies (Berner 1994). The
gray line represents the d18O data
on calcite (Veizer et al. 2000),
and the two diamonds show the
D47 data (Came et al. 2007)

brachiopods, but for the Silurien samples, they are com- The simplified budget of the d13C of the oceanic DIC doc is
pletely opposite in that they indicate temperatures around 8 ° written as:
C higher than the current ones, whereas the d18O data
ddoc
indicates temperatures around 2.5 °C lower than currently Coc ¼ Fcw ðdcw  doc Þ þ Fow ðdow  doc Þ þ Fcw ðdcw  doc Þ
(Fig. 27.4). dt
þ FMOR ðdMOR  doc Þ
By combining the d18O and D47 data from Ordovician
sedimentary calcites, a recent study showed for the first time  Fcd ðdoc  ecarb  doc Þ  Fod ðdoc  eMO  doc Þ
that it was possible to calculate the d18O ratio of seawater ð6Þ
over this period of time and thus to trace back the volume of
ice present on the continents during the glacial peak at the where Coc is the DIC content of the ocean and Fcw, Fow, Fvol
end of the Ordovician (Finnegan et al. 2011) the estimate of and FMOR are the carbon fluxes transferred from the litho-
which is in agreement with the reconstruction of sea level sphere to the ocean by dissolving continental carbonates, by
variations for that time (Loi et al. 2010). the oxidation of sedimentary organic carbon, by degassing
linked to the volcanic activity and by the oceanic ridges,
respectively. Fcd and Fod are the fluxes of carbonate deposits
Indirect Isotopic Indicators from all environments, and the burial of organic carbon
respectively. The d are the d13C corresponding to each of
The d13C of Carbonate Sediments these fluxes: dcw is close to 0‰, dow to −25‰; dMOR is
The 13C/12C ratio (d13C expressed with respect to the PDB estimated at −5 or even −6‰. dvol is less well known, but its
standard) of the carbonate sediments recorded the isotopic value is certainly located between the mantle value and that
composition of the total carbon dissolved in seawater (in the of the carbonates deposited on the abyssal sea floor, that
form of dissolved CO2, bicarbonate and carbonate ions, is ±0‰ on average over a long-time scale. ecarb is the iso-
P
denoted by CO2 or by the acronym DIC—Dissolved topic fractionation between the DIC of seawater and the
Inorganic Carbon) at the time of deposition. The main trend carbonate minerals. This fractionation is low (around 1.2‰)
of this isotopic indicator is a general increase during the (Hayes et al. 1999), indicating that carbonate deposits cannot
Paleozoic, from −2‰ during the Cambrian to +4‰ at the be responsible for the temporal evolution of doc. However,
end of the Carboniferous. Veizer et al. (2000) (Fig. 27.5). the fractionation between the buried organic matter and the
This geological stage presents the highest value for this oceanic DIC eMO is very high (±20‰ today). The organic
signal for the entire Phanerozoic. After a rapid decrease flows Fow and Fod dominate the budget because this frac-
during the Permian, the d13C of carbonates registered min- tionation means that their combined flow is multiplied by
imal fluctuations around the present value of +1.5‰. This d13C and so is an order of magnitude greater than the other
d13C is an indicator of the behavior of the carbon cycle, but terms. The flux variations most influencing the temporal
unfortunately it is not very clear how exactly to interpret it. evolution of the d13C of the ocean are therefore those that
364 Y. Goddéris et al.

Fig. 27.5 d13C of calcite

sediment from the Phanerozoic.
% of data points

affect the processes with a d13C signature furthest away from CO2 outgassing from the mantle when the Karoo-Ferrar
the oceanic value, in other words, the oxidation of the traps were established in South Africa and could therefore be
organic sedimentary carbon exposed on land or the burial of linked to a warming of the climate by the greenhouse effect.
organic carbon in sediment. However, it is impossible to It should also be pointed out that, at shorter intervals during
discriminate between them using the d13C signal alone. glacial-interglacial Quaternary oscillations, the d13C of the
However, geochemical interpretation is possible for some ocean decreased during glacial periods due to a reduction in
major excursions. In particular, the positive and very large the biosphere, a consequence of aridity on the continents
Carboniferous excursion (a positive excursion of 4‰ (Part III, Chaps. 1 and 2).
throughout the Carboniferous (Veizer et al. 1999) is inter- Finally, some very rapid and pronounced negative
preted as the recording of the burial of a very large amount excursions could be attributed to the destabilization of
of organic carbon (this was the time large coal beds were methane hydrates contained in marine sediments. The
being deposited in Europe, Russia and North America), methane released by the sediments is characterized by a d13C
which extracted preferentially the 12C from the of −60‰ and these negative values allow very pronounced
ocean-atmosphere system. This resulted in a higher con- excursions of d13C. In general, it is assumed that the CH4
sumption of atmospheric CO2 and a cooling of the climate contribution to the exosphere is short (104 to 105 years) and
which coincides reasonably well with the Permo- intense, resulting in rapid negative excursions (e.g. the
Carboniferous ice age. This observation encouraged the thermal maximum of the Palaeocene-Eocene transition
association of any positive excursion of d13C with a cooling (McInerney and Wing 2011).
of the climate, especially since many positive excursions It is even possible to imagine a combination of effects: the
coincide with episodes of extensive burial of organic matter Karoo traps were established in the coal-rich sediments of
(episodes of global anoxia and black shale formation), such Gondwana 183 million years ago. As a result, a massive
as the end of the Devonian (Frasnian-Famennian episode, degassing of reduced carbon, low in 13C, towards the
circa 375 Ma) or the global anoxic events of the Cretaceous atmosphere, causing the negative excursion in d13C
(Aptian circa 120 Ma and Turonian circa 90 Ma). This observed during the Toarcian (McElwain et al. 2005).
interpretation, however, is not conclusive. To illustrate the Finally, a database of all the d13C values measured on
difficulty of interpreting the d13C signal in geochemical and benthic foraminifera from 40 ODP and DSDP drillings
climate terms, it should be noted that certain anoxic epi- provides high-resolution coverage of the entire Cenozoic
sodes, during the formation of gray or black shales, are (Zachos et al. 2008). The dominant signal from this curve is
accompanied by negative excursions of the oceanic d13C, the decrease in d13C of about 2% from mid-Miocene
such as the Toarcian anoxic episode. These events probably (15 Ma) onwards. The reasons for this reduction remain
indicate favorable conditions (anoxic environment) for the obscure. It could be a sign of a global decrease in the burial
preservation of organic matter, but the total amounts of of organic carbon over the last 15 million years. However,
organic matter ultimately buried are low. The Toarcian over the same period, isotopic fractionation eMO decreased
negative excursion can be interpreted as the result of a large by approximately 8‰, whereas it remained relatively
27 The Phanerozoic Climate 365

constant over the rest of the Phanerozoic. This fractionation with the paleothermostat theory. In fact, weathering becomes
was reconstructed over the whole Phanerozoic by measuring a function of climate in a positive feedback loop. The colder
the d13C of organic carbon in marine sediments over many it gets, the more erosion increases, forcing an uptake in CO2
different periods and comparing it to the d13C of sedimentary consumption by weathering, in turn forcing increased cool-
carbonates (Hayes et al. 1999). Since this fractionation is ing. The silicate weathering and volcanic degassing are
dependent on the concentration of H2CO3 in the waters, it uncoupled, and the carbon content of the ocean and the
can be inferred that its decline is linked to a drop in atmo- atmosphere is consumed in less than a few million years
spheric CO2 since the Miocene. Yet this simple interpreta- (Goddéris and François 1996), which would lead to unreg-
tion is challenged by independent estimates of the CO2 level istered climate disasters during the Tertiary. It should be
suggesting pressures below 300 ppmv during the Miocene underlined, however, that it is possible that orogens pump
(see section on atmospheric CO2). Nevertheless, the com- CO2 while respecting the paleothermostat theory. The con-
bination of the drop in eMO fractionation and the drop in sequences of the uplift of a mountain range are much more
oceanic d13C suggests an increase in CO2 being buried in the complex than a simple uplift causing increased weathering.
form of organic carbon in sediments since 15 Ma, probably We shall see later that they lead to a number of geological
linked to the establishment of the Himalayan orogeny. phenomena which ultimately link orogenesis to cooling.
Moreover, the evolution of the 87Sr/86Sr isotopic ratio of
The 87Sr/86Sr Isotope Ratio of Carbonate seawater cannot be interpreted solely in terms of changes in
Sediments the relative importance of mantellic and continental flows.
Marine carbonates record the strontium isotopic ratio, The isotopic ratio of the source rocks of the weathering may
87
Sr/86Sr, of seawater without fractionation. The residence also have changed over time, and, particularly in the oro-
time of Sr in seawater (2–5 million years) ensures an even genic zones, which considerably complicates the interpre-
value throughout the ocean and therefore low dispersion of tation of the isotopic sign of strontium. Nevertheless, the
the data. A high-resolution curve (1 million years) has been quality of the Phanerozoic signal should motivate further
published by Veizer et al. (1999). This signal has been widely analysis of this indicator in the future.
used to constrain the extent to which CO2 is consumed by Along with the 87Sr/86Sr isotopic signal, the 187Os/188Os
alteration of continental silicates during the Cenozoic, par- osmium isotopic ratio of seawater measured from sediments
ticularly in response to the Himalayan orogeny (Raymo taken during ocean drilling programs is also used to con-
1991). Two main types of Sr intake to the ocean are identified: strain the evolution of continental and hydrothermal
an exchange flux at the level of the ocean ridges, which weathering fluxes. The methodology is very similar to that of
doesn’t affect the Sr concentration of the water but modifies its Sr, but the major advantage of osmium is its short residence
isotopic ratio. Today, the water enters the ocean ridges with a time in the ocean, around 10–30 kyr, although it is suffi-
87
Sr/86Sr ratio of 0.709 and exits after contact with mantellic ciently longer than the mixing time of the water masses to
rocks with a typical value of 0.703. This process therefore ensure values that are representative of the global ocean.
tends to reduce the 87Sr/86Sr ratio of seawater and bring it River contributions are the dominant factor, with a
closer to the mantle value. Conversely, the isotopic ratio of the 187
Os/188Os ratio of 1.3 and a flux of 1800 mol per year−1.
rivers, inherited from the weathering of continental rocks is Hydrothermal inputs have an isotopic signature of 0.13 and a
now equal to 0.712. There is some correlation between peri- flux of around 100 mol per year−1. A flow linked to cosmic
ods with high 87Sr/86Sr seawater ratios and glaciation epi- dust, with a ratio of 0.13 and a flux of 80 mol per year−1,
sodes. This has been interpreted as a sign of greater must also be added. The main sinks are ocean sediment
weathering during cold climate periods, in response to deposits. The ratio can therefore be indirectly linked to the
intensified physical erosion, which in turn promotes chemical evolution of the Earth’s climate and of greenhouse gases,
weathering. The signal is particularly clear for the last through the characterization of the geological flows of the
40 million years, and the rapid increase in the 87Sr/86Sr ratio carbon cycle. Nevertheless, osmium is a very scarce element,
of seawater has been interpreted as the signature of increased which makes it difficult to measure. The mean concentration
continental weathering during the uplift of the Himalayas. It of Os is 10 fg/g (fentogram gram−1, 1 fg = 10−15 g) in
has been suggested that there is a correlation between this seawater and 9 fg g−1 in rivers.
increase in the 87Sr/86Sr ratio and a decrease in the CO2
concentration in the atmosphere. In this model, the Himalayas
are considered to have triggered the cooling in the Cenozoic. The Level of Atmospheric CO2
This hypothesis has been extensively developed. It is
found again in the more recent literature, linking orogeny There is clearly no direct measurement of this level for the
with global cooling of the climate in response to intensified distant past beyond 900,000 years, the period covered by ice
weathering. Nevertheless, this hypothesis is in contradiction cores drilled in the Antarctic ice. All methods are therefore
366 Y. Goddéris et al.

indirect and depend on a number of assumptions. This method consists of studying the trace d13C of pedogenic
inevitably produces significant uncertainties. In addition, the carbonates contained in goethite, a mineral formed during
low residence time of carbon in the ocean-atmosphere sys- soil alteration reactions (Royer et al. 2001).
tem (200,000 years) is an inherent cause of scattering of data A third approach uses the measurement of isotopic frac-
points. Temporal resolution for the geological past very tionation of carbon by phytoplankton and its relationship
rarely reaches this level of precision. It follows that two with the dissolved CO2 content in seawater. Initially, the
points, attributed to the same geological moment, may have difference between the d13C of carbonates and that of total
very different values. organic carbon was used. It was subsequently found to be
An excellent overview was carried out by Royer et al. error-prone, in particular due to the presence of organic
(2001) using several reconstruction methods. For the very matter of various origins in both continental and marine
distant past, the counting of the stomata of the fossil leaves is sediments. This fractionation is now measured by directly
commonly used. For modern species, there is a positive using biomarkers in the organic matter, such as alkenones
correlation between the number of stomata and the ambient (Pagani et al. 2005). The link between isotopic fractionation
CO2 level. These correlations are applied to old geological and the level of CO2 dissolved in water is based on corre-
samples. This is probably the least precise method, but it has lations established for the present. For example, this one is
the advantage of being able to go back very far in the past (as based on a compilation of GEOSECS campaign data:
far back as the Devonian) and does not encounter the prob-  
lems experienced by isotopic systems, such as diagenesis. eP ¼ 12:03 CO2aq  3:56
  ð8Þ
A second method allows measurements to be traced back 10  CO2aq  90 lM
to the Paleozoic. It involves measuring the d13C of pedo-
where ep is the photosynthetic fractionation of phytoplank-
genic carbonates in paleosols and is based on the fact that the  
level of CO2 in modern soils results from a mixing of the ton, and CO2aq is the concentration of gaseous CO2 dis-
atmospheric CO2 in the atmosphere and the CO2 in the soil solved in water.
through respiration. The d13C of this C2atm mixture is regis- The use of correlations established for current conditions
tered in the pedogenic carbonates: is the main weakness of this method, since they are
extrapolated to CO2 ranges that are significantly higher than
dsample  1:0044dresp  4:4 the current level, using compounds made by organisms with
COatm
2 ¼ Sð z Þ ð7Þ
datm  dsample an unknown metabolism. In addition, isotopic fractionation
is also a function of the growth rate of these organisms,
where dsample is the d13C of the pedogenic carbonate. The which complicates reconstruction.
method requires setting the d13C of the breathed CO2, dresp, A final method is based on the measurement of the ratio
at typical values. It therefore depends on the proportion of of boron isotopes 11B/10B (d11B) in carbonate sediments
C4 plants to C3 plants, for which the isotopic fractionations (Royer et al. 2001). The relative abundance of the two dis-
are very different. In practice, it is not applicable after the solved borate species (H4 BO4 and H3 BO 4 ) depends on the
emergence of C4 plants, 15 Ma ago. The method also pH of the sea water. There is an isotopic fractionation of
requires knowledge of the d13C of the atmosphere datm and S about 19‰ between the two species. The carbonates are
(z), the amount of breathed CO2 at the estimated depth z of mainly made up of the H4BO4 species and the isotopic d11B
the pedogenic carbonate in the paleosol. The d13C of the composition of the carbonates will therefore depend on the
atmosphere is estimated by measuring the d13C of marine pH. Nevertheless, the link with atmospheric CO2 is not clear.
carbonates of the same age and by imposing the fractiona- First, this requires assumptions that d11B of seawater
tion value between the carbonates and the atmospheric CO2. remains constant over time, that the isotopes are shared
As for the fraction of breathed CO2, it is calculated by between the two species and that the relative abundances of
11
making major assumptions about soil temperature, porosity B and 10B remain the same. It was shown that this was
and biological productivity. In fact, if CO2 production in probably not the case and that d11B of total borate probably
soils is a function of productivity and temperature (which changed in the past. Finally, we must make strong
partially controls the degradation of organic matter by bac- assumptions about the alkalinity of seawater, to bring the pH
teria), its diffusion to the atmosphere depends largely on the up to the pressure of atmospheric CO2.
physical structure of soils. The level of CO2 at a given depth The compilation of all these reconstructions inevitably
is therefore dependent on the relative importance of the shows a large dispersion of points (Royer 2006; Foster et al.
production and loss by diffusion. This is by far the most 2017) (Fig. 27.6). Nevertheless, some trends may emerge.
uncertain method. A recent recalibration of the method has The level of atmospheric CO2 seems to have been high
led to a considerable reduction in past reconstructed CO2 before the Devonian (with values generally in excess of
pressures (Breecker et al. 2010; Foster et al. 2017). A similar 2000 ppmv). This period is followed by a time interval
27 The Phanerozoic Climate 367

covering the end of the Carboniferous and the beginning of 1992). However, these climate oscillations are also observed
the Permian, during which time atmospheric CO2 remained in isotopic data, such as in the d18O of carbonates deposited
between 200 and 500 ppmv. The data for the Mesozoic are on the seabed, which reflect, at least partially, the d18O of
more confusing and show no clear trend. For the same time seawater at the time of deposition (Veizer et al. 2000). These
period, it is common to have atmospheric CO2 estimates isotopic data remain difficult to interpret because they
varying by a factor of 5–10. Finally, the Cenozoic appears to combine not only climate indicators (seawater temperature,
be marked by a general decrease in atmospheric CO2 pres- continental ice volume), but also geochemical data such as
sure, reflected above all in a general reduction in the maxi- salinity of the seawater, the speciation of carbonates, the
mum reconstructed values. The most precise record (i.e., d18O of seawater which is itself influenced by continental
with the best temporal resolution and lowest dispersion) is and hydrothermal alteration flows. Nevertheless, the d18O
the one obtained by reconstructing the isotopic fractionation during the Phanerozoic shows oscillations with a periodicity
in carbon of the oceanic biosphere based on the measure- of 135 million years, in line with sedimentological recon-
ment of the d13C of the alkenones (Pagani et al. 2005). It structions, thereby reinforcing its validity as a good climate
shows a rapid decrease in atmospheric CO2 from the indicator. To date, this periodicity of 135 million years
beginning of the Eocene until the end of the Oligocene: remains largely unexplained, but its length indicates that it
around 50 million years ago, the CO2 content is estimated to might have to do with the tectonic processes that shaped the
have been 1500 ppmv and fell to between 200 and Earth’s surface or with astronomical movements (Shaviv and
300 ppmv 23 million years ago. CO2 levels then remained Veizer 2003). The accumulation of recently obtained iso-
constant throughout the Miocene at values slightly below topic data on phosphates (including fossil fish teeth and
250 ppmv. Finally, CO2 levels during the Pliocene were conodonts) has greatly improved the resolution of the
explored using two methods: through isotopic fractionation alternation of hot and cold modes, especially during the
in carbon and by the counting the stomata of fossil leaves. Mesozoic and Devonian periods (Dromart et al. 2003;
Both methods suggest that CO2 levels have risen: between Pucéat et al. 2003; Joachimski et al. 2004).
350 and 450 ppmv from 2.9 to 3.3 million years for the first The cold climate modes of the Phanerozoic occur during
method, and between 370 and 250 ppmv from 5.3 to the Ordovician (from about 470–440 Ma), the
2.6 million years for the second. Permo-Carboniferous (from about 330–270 Ma), the Juras-
sic and the Cretaceous. This period is marked by a succes-
sion of short cold events: at the end of the Toarcian around
The Great Climate Modes of the Phanerozoic 176 Ma, at the Callovian-Oxfordian boundary around
and Their Possible Causes 161 Ma, at the transition from the Lower Valanginian to the
Middle Valanginian towards 140 Ma, at the beginning of the
The climate reconstructions of the Phanerozoic show a Aptian around 125 Ma and at the Cenomanian-Turonian
succession of modes warmer than currently and of cold boundary around 94 Ma. Finally, the end of the Cenozoic,
modes similar to currently, with the emergence of ice caps. when Antarctica first started to freeze over 34 million years
This succession is observed in sedimentological records of ago up to the current period, was in cold mode with ice first
glacial sedimentary deposits, including tillites, and in ice appearing in the southern polar regions and later in the
rafted debris (IRD), debris carried by sea ice (Frakes et al. northern hemisphere.

The Causes of Cold Climate Modes

These are generally subjected to more study than the causes

of warm modes. The following processes have been
suggested:

1. Orogenesis
The establishment of mountain chains causes an increase
in physical weathering following the establishment of
glaciers, steep slopes and alternating freeze-thaw regime
(Raymo 1991). This results in a greater fracturation of the
rock and thus greater sensitivity to chemical weathering
Fig. 27.6 Levels of phanerozoic CO2 reconstructed by various which consumes CO2. Therefore, locally, this process
methods based on proxies increases the vulnerability of the continental surfaces to
368 Y. Goddéris et al.

weathering. The result would be a drop in atmospheric clouds, increasing the albedo of the atmosphere and
CO2 and an overall cooling. This theory is based on cooling the climate (Shaviv and Veizer 2003). To date,
measurements of current weathering rates in mountain there is no experimental evidence of the validity of this
ranges, on the strong correlation between physical ero- mechanism, which remains purely speculative.
sion and chemical weathering in modern watersheds, and 5. The fragmentation of a supercontinent. The resulting
on the 87Sr/86Sr isotopic ratio of the water run-off from increase in rainfall activates the consumption of CO2 by
the Himalayas. A second effect should also be taken into silicate weathering and thus cools the climate. This effect
account: the very high sedimentation rates in the seas at is particularly important if the supercontinent breaks up
the foot of the orogens lead to very efficient burial of along the equator, the site of intense rainfall (Goddéris
organic carbon and thus to increased consumption of et al. 2014).
atmospheric CO2 thanks to the action of the biological 6. The migration of continents towards the low latitudes,
pump. This process is responsible for trapping two to characterized by climatic conditions favoring the
three times more CO2 than the silicate weathering in the weathering of continental silicates and thus an increased
Himalayan orogeny (Galy et al. 2007). consumption of atmospheric CO2 (Nardin et al. 2011).
2. The development of vascular vegetation on land. Tall 7. Any reduction in degassing of greenhouse gases from the
vegetation with a developed root system acts at three mantle or sediments towards the atmosphere.
levels on weathering rates. The roots stabilize the soils 8. The establishment and subsequent weathering of basaltic
and thus increase the contact time between inland waters provinces on the continents. Basalts weather much more
and the silicate minerals. In addition, root and microbial efficiently than the average continental crust on which
respiration in soils increases the partial pressure of CO2, they spread (in equivalent conditions, basalt weathers
and thus acidifies the water which percolates towards the eight times faster than granite). This finding was estab-
bedrock. Finally, the plants secrete organic acids which lished from a study of weathering in basaltic watersheds
also contribute to the acidification of the waters. As a (Dessert et al. 2001). The weathering of new basalt thus
result of these three effects, there is an increase in the produces a long-term decrease in the partial pressure of
consumption of atmospheric CO2 through dissolution of atmospheric CO2. The question remains as to the
the continental silicates. This hypothesis is based on weathering of submarine basaltic plateaus. Do they
studies carried out in particular in Iceland on lava flows contribute to the cooling of the climate system or not?
on slopes covered and uncovered with stemmed vegeta- The pH buffer imposed by carbonate speciation in sea-
tion. It appears that weathering rates are eight to ten times water nevertheless suggests that weathering of oceanic
larger under dense vegetation cover (Berner 2004). basalts is a minor phenomenon, with basalt dissolution
A small-scale laboratory study suggests that non-vascular being minimal at around pH 8, a value for seawater
plants (lichens and mosses) could have a similarly which probably didn’t change much over the course of
accelerate chemical weathering of continental surfaces the Phanerozoic.
(Lenton et al. 2012).
3. Increased burial of organic carbon during the anoxic
phase of the ocean. This hypothesis is often proposed to
explain positive excursions in the d13C ratio of oceanic The Causes of Warm Climate Modes
carbonates correlated with climate cooling. It requires
particular environmental conditions: either conditions Curiously, cold modes have always been considered to be
favorable to maintaining water stratification in large accidents in a prolonged warm state. This is probably the
ocean basins and preventing the ventilation of the deep reason why the suggested causes of warm modes are fewer
waters in these basins, or conditions of oceanic hyper- and less discussed in the literature, with the exception of the
productivity leading to the absorption of oxygen in the thermal event of the Palaeocene-Eocene transition.
deep waters through the recycling of organic matter The following mechanisms have been proposed:
produced in the euphotic zone. This burial may also
occur on land-based environments, as has happened 1. Any increase in degassing of greenhouse gases from the
during the Carboniferous period. mantle or sediments to the atmosphere. This could be due
4. The movement of the solar system into a galactic arm. to increased volcanic activity releasing massive amounts
This recent hypothesis attempts to explain the periodicity of CO2, basaltic effusion events over land (Dessert et al.
of 135 million years in cold modes. The galactic arm is 2001) or methane degassing from gas hydrates accumu-
an area of formation of intense stars and of emission of lated in sediments (McInerney and Wing 2011).
galactic cosmic rays. Reaching the atmosphere, these are 2. The creation of a supercontinent, reducing rainfall and
thought to participate in the nucleation of low-level thus weathering of the continental silicates, allowing an
27 The Phanerozoic Climate 369

increase in pressure of CO2 in the air and a warming of Finally, f5 expresses the level of dependence on the litho-
the climate (Goddéris et al. 2014). logical type. It can be expressed as a constant factor equal to
8 or 10 for new (rapidly deteriorating) basaltic surfaces and
equal to 1 for granite surfaces (Dessert et al. 2001).
For example, this simple formalism shows that the estab-
The Phanerozoic Terrestrial Paleothermostat lishment of an orogen leads to an increase in the consumption
of atmospheric CO2 through silicate weathering (f3 increases),
The short residence time (200,000 years) of carbon in the but that the conditions of the paleothermostat are always
exosphere, as well as the reaction time of ocean alkalinity verified: the climate cools globally, and the decrease in the f1
(3000 years) impose the following near-equal relationship and f2 factors compensates for the increase in f3. It can then be
between the fluxes of inorganic carbon, ignoring any pos- said that the vulnerability of continental surfaces to weath-
sible imbalances in the organic carbon cycle (see Chap. 5): ering has changed. Indeed, if the degassing of the solid Earth
does not change, Eq. (9) dictates that CO2 consumption by
Fvol þ FMOR  Fsw ð9Þ silicate weathering remains virtually constant on the scale of
where Fvol is volcanic degassing, FMOR ocean ridge degas- several million years. However, Eq. (10) dictates a decrease
sing and Fsw half the CO2 consumption by silicate weath- in f1 and f2 to compensate for the increase in f3. We can say that
ering. This factor of a half stems from the fact that two moles the weathering of continental surfaces has increased, whereas
of atmospheric carbon are consumed for two equivalents of the total silicate weathering flux has remained unchanged.
alkalinity produced by the dissolution reaction of the con- However, to allow f1 and f2 to adapt to the new conditions, the
tinental silicates. Only one of these two moles will finally be equilibrium level of CO2 is lower, and the climate is colder
buried in the form of ocean carbonate. The other mole of and dryer. Similarly, the establishment of a basaltic province
carbon remains in the ocean-atmosphere system (see increases the factor f5 and the climate will cool in compen-
Chap. 5). The flux of CO2 consumption by silicate weath- sation. The same applies to the colonization of the continental
ering is a function of temperature and of continental runoff. surfaces which are described below.
Generally, it increases as the CO2 content increases. But its It should nevertheless be noted that the relation 10 is a
response to an increase in CO2 is also a function of the simplification. The relationship between CO2, temperature,
continental plant cover, the presence of orogens and of and continental runoff is complex and is largely dependent
intense physical weathering, the configuration of the conti- on the paleogeographic configuration, which complicates the
nents, the modification of the superficial lithology, follow- problem considerably.
ing, for example, the establishment of basaltic surfaces on Finally, if the possibility of imbalance in the organic
land during major magma events. Fsw can therefore be carbon cycle is taken into account, the thermostat equation is
expressed in the following way: written as:

Fvol þ FMOR þ Fow  Fsw þ Fod ð12Þ

Fsw af1 ðT Þ  f2 ðRÞ  f3 ðerosionÞ
ð10Þ
 f4 ðvegetationÞ  f5 ðlithoÞ where Fow is the oxidation of sedimentary organic carbon
and Fod is the overall burial of organic carbon. If deposits
where T is the continental temperature, and R is the runoff. were to increase, for example, due to the development of
The functions f1 and f2 are known: the first is an exponential exceptional conditions for the preservation of organic matter
function of the temperature, the second a linear function of (such as the appearance of large-scale anoxia), while the
the runoff. f1 and f2 have been determined for granites and oxidation of exposed sedimentary organic carbon on land
basalts. The function f3 is unknown. The only indicator remained constant, this would give the following inequality:
available is that there is a very strong positive correlation
between physical erosion fluxes and chemical weathering ðFvol þ FMOR Þ  Fsw  Fod  Fow  0: ð13Þ
fluxes for both large and small watersheds. We can deduce
In this case, silicate weathering must be less than the total
from this that f3 is an increasing function of the rate of
degassing of the solid Earth to maintain the paleothermostat
erosion, but its precise mathematical expression had yet to
balance. This condition will be verified because the increase
be defined. As for f4, studies of lava flows in Iceland suggest
in buried organic carbon reduces the CO2 pressure in the air
that the rate of weathering increases by a factor of 8 when
which, in the first order, causes a decrease in the f1 and f2
vascular vegetation develops (Berner 2004). f4 increases
factors.
with vegetation cover but also when mosses and lichens cede
Finally, it should be noted that the paleothermostat con-
to vascular plants with a well-developed root system (see
stitutes a very powerful stabilizing force of the Earth’s
discussion on the Devonian in the following section).
370 Y. Goddéris et al.

climate. If causes external to the CO2 cycle disturb the cli- CO2 levels. Berner (2004) estimates that the average global
mate (cosmic rays and cloud nucleation, the passage of the temperature was ±6 °C higher than it is currently, based on
Earth into a cloud of galactic dust, methane degassing), the a numerical modeling study. These very high values are
resulting change in f1 and f2 factors would cause an imbal- confirmed by measurements using the paleothermometer
ance in the carbon cycle, which would find a new balance made up of the number of rare molecules Ca18O13C16O2 in
over a few million years, by adjusting the pressure of the carbonates of the Lower Silurian (Came et al. 2007).
atmospheric CO2 to re-establish climate conditions verifying Whatever the causes, this warm climate state was inter-
the paleothermostat. rupted by several glacial events of very different durations
and amplitudes. The Atlas Fig. chapter 3.9 shows the
paleogeographic maps of the Earth during the main periods
The Paleozoic Climate: The Chronology of the Phanerozoic.
of Major Trends and Their Causes

In general, the Paleozoic climate is described as warmer than The Ordovician Glaciation
the current one, except for two glacial events with very
different characteristics. The d18O data on apatite show a long-term cooling trend
The reasons for this warm climate state are not clearly during the Lower and Middle Ordovician and a sudden drop
understood, but several hypotheses have been put forward. in temperatures during the Hirnantian in the Late Ordovician
On the one hand, the degassing of the solid Earth seems to (Trotter et al. 2008), which is independently confirmed by
have been generally greater by about 60% than it is at pre- the Δ47CO2 analysis (Finnegan et al. 2011, Fig. 27.7). Gla-
sent. This assertion is based on the fact that the sea level was cial sediments, which are the only direct evidence of
generally higher in the geological past than it is today, Ordovician glaciation, are only documented during this very
except for the Permo-Carboniferous transition. This high sea short cool interval, which has long suggested that glaciation
level can be explained firstly by the larger volume occupied is a short-term cold accident punctuating an otherwise very
by the ocean ridges and therefore a supposedly greater hot period of geological time. Geochemical studies recon-
degassing of the solid Earth. This result has never been structing the composition of oceanic d18O (Finnegan et al.
confirmed by other methods apart from sea level and remains 2011) and glacio-eustatic variations (Loi et al. 2011) suggest
questionable. On the other hand, the absence of abundant that the ice cap at the South Pole would have reached a
vascular vegetation until the end of the Devonian prevented volume almost twice as large as during the Last Glacial
the development of modern soils on land surfaces. This Maximum. Indirect indices such as variations in sea level
absence of soil reduced the contact time between the inland (Dabard et al. 2015) or d18O excursions (Rasmussen et al.
water and minerals, thus limiting their weathering. Similarly, 2016) today suggest that the first ice caps could have been in
the absence of a root system reduced the acidity of soil place since the Middle Darriwilien (about 470 Ma) in the
solutions and thus the consumption of atmospheric CO2 by Ordovician. In addition, it appears that glacial events also
continental silicate weathering, which in turn promotes high punctuated the Lower Silurian. The Ordovician glaciation is

Fig. 27.7 Tropical sea-surface

temperatures reconstructed based
on d18O data measured on apatite
(Trotter et al. 2008) and based on
Δ47CO2 proxy data (Finnegan
et al. 2011)
27 The Phanerozoic Climate 371

therefore increasingly considered to have been a long cold Eifelian (390 Ma; Stein et al. 2007). Trees, 8–10 m high,
period (about 470–425 Ma), sometimes referred to as ‘Early began to colonize the land towards the end of the Givetien
Paleozoic Ice Age’ (Page et al. 2007), and within which the (385 Ma), with, among others, giant ferns such as Arche-
Hirnantien only represents a glacial maximum. This vision is opteris and Cladoxyopsides (Anderson et al. 1995). True
supported by the most recent climate models (Pohl et al. large forests are likely to have become established towards
2016). the end of the Devonian (Frasnian; 380 Ma, Scott and
The causes of this glaciation are still poorly understood. Glaspool, 2006). As trees appeared and flourished, weath-
Nardin et al. (2011) showed that the long-term cooling of the ering of the continental silicates rapidly accelerated, while
climate can be explained by the paleogeographic evolution degassing from the solid Earth remained almost constant
occurring throughout the Ordovician, and in particular the (Berner 2004). Soils developed along with root systems,
migration of continents in the intertropical zone conducive to increasing the acidification of the water in contact with the
weathering, which brings about a fall in the atmospheric minerals as well as increasing the contact time between
concentration of CO2. Regarding the Hirnantien glacial inland water and silicate rocks. This resulted in a rapid
peak, the best explanations also suggest a fall in atmospheric decrease in the partial pressure of atmospheric CO2, from
CO2, but the mechanisms to achieve this are subject to 2000 ppmv in the early Devonian to 1000 ppmv at the end
debate. Kump et al. (1999) proposed an interesting (Foster et al. 2017). The extent to which the climate cooled
hypothesis: the fall in CO2 level could have been a conse- as a result of this colonization is uncertain however because
quence of the establishment of New Caledonian and Appa- the change of the albedo of continental surfaces, following
lachian orogens during the Middle and Upper Ordovician, the replacement of bare soils by forests, compensates at least
which would have increased the vulnerability of the conti- partially for the fall in atmospheric CO2 (Le Hir et al. 2011).
nental surfaces to weathering. Other mechanisms have also It is also interesting to note that the overall cooling of the
been proposed, including the establishment of the first plants climate may have been beneficial to the development of
on land (Lenton et al. 2012). The difficulty in explaining this modern leaves, which are large in size and have many
event lies in the magnitude of the cooling, which is around stomata, encouraging primary production on land to the
−7 °C at tropical latitudes, whereas sea surface temperatures detriment of more primitive plants. We can thus infer the
appear to have varied by only 1–2 °C at the same latitudes establishment of a positive feedback between cooling and
during the last glacial-interglacial cycle (CLIMAP Project the colonization of land surfaces by ever more efficient
1981). However, numerical modeling of the plants.
ocean-atmosphere coupled system during the Ordovician Another mechanism that could explain the drop in CO2
revealed climatic instability associated with the sudden during the Devonian is an increase in the amount of CO2
development of sea ice, which explains a sharp fall in tem- trapped in sediment, also as a result of colonization of land
peratures in response to a moderate decrease in atmospheric by continental plants. Indeed, the appearance of lignin in
CO2 concentration, thus loosening the constraints that would plant tissues from 410 Ma onwards increased the amount of
be placed on CO2 sinks to explain the geochemical data organic carbon preserved in continental environments and
(Pohl et al. 2016). To explain the emergence from the Hir- on the margins. Lignin is indeed much more resistant to
nantian glacial maximum, Kump et al. (1999) proposed the mineralization than marine organic matter. It was first
following mechanism: as the ice cover on the supercontinent thought that lignin appeared before the development of
Gondwana increased, the available surface of continental organisms capable of decomposing it, causing an increase in
silicates exposed to weathering falls, thus causing an accu- the burial of carbon and the reduction of CO2, which could
mulation of CO2 in the atmosphere. This persuasive scenario have contributed to the establishment of the glaciation of the
was tested with a simple climate model. Late Paleozoic (Nelsen et al. 2016). Nevertheless, recent
studies have shown that decomposers evolved in parallel to
lignin, which calls into question an ‘organic’ trigger for the
The Devonian Climate Permo-Carboniferous glaciation (Nelsen et al. 2016).
Finally, the end of the Devonian is characterized by a
The Devonian (419–359 Ma) is marked by numerous bio- mass extinction event, affecting tropical marine environ-
logical disturbances. Although the first traces of vegetation ments in particular. This event lasted approximately 1–
appeared during the Middle Ordovician (Rubinstein et al. 3 million years, culminating at the Frasnian-Famennian
2010) and the existence of vegetation fires during the Sil- boundary. It is accompanied by the deposition of anoxic
urian are suggested based on the presence of charcoal in the sediments (black shales), accompanied by two positive
sedimentary record, the development of a long-stemmed excursions of the d13C of carbonate sediments. It has been
biosphere begins in the Lower Devonian. Plants reaching proposed that these events are the consequence of the
heights of up to 2 to 3 meters were identified during the emergence of pulses as vascular plants colonized the land
372 Y. Goddéris et al.

while in the Devonian, this carbon sink was thirty times

lower (Berner 2004). This is nearly double the estimated
burial rate for the Cenozoic. The reason for the efficiency of
this burial during the Carboniferous remains partly obscure,
but it is probably linked to two interconnected factors: (i) the
abundance of vegetation in certain zones, (ii) a low sea level
during the Carboniferous (although it was high during the
Devonian), creating low-lying coastal lands and extensive
swamps. This efficient burial on the continents of reduced
carbon depleted in 13C, instigated the longest and most
extensive positive excursion of the d13C in carbonate sedi-
Fig. 27.8 Calculated levels of CO2 in the atmosphere at the end of the
ments (5‰, spread out over almost all of the Carboniferous
Devonian
and Permian).
Although this model of the forcing of the Carboniferous-
and of increased weathering bringing additional nutrients to Permian glaciation by biotic factors is commonly accepted,
the ocean and causing intense blooms in the oceanic bio- several elements today are causing us to question it. First,
sphere. These would have led to increased oxygen con- the colonization of the continents by vascular plants of large
sumption in the surface waters and the establishment of growth form was completed at the end of the Devonian,
large-scale anoxia. Moreover, these anoxic events are cor- several tens of millions of years before the establishment of
related with second-order fluctuations of the sea level (a few the Carboniferous-Permian glaciation (Davies and Gibling
tens of meters), thereby disturbing the weathering of car- 2013). In addition, the increase in continental organic carbon
bonate platforms which alternate between being emerged burial seems relatively uncorrelated both with the d13C
and flooded. A numerical study, coupling a 1D energy bal- signal, which shows uniformly high values between 360 and
ance climate model and a geochemical model of ocean and 260 Ma (Fig. 27.5), and with the chronology of the colo-
continental surfaces predicts a fall of 1500 ppmv across the nization of continental surfaces by plants. A recent study
Frasnian-Famennian boundary, over an interval of only coupling a climate model with a long-term carbon cycle
3 million years, as well as a cooling reaching more than 2 ° model, highlighted the key role of tectonic changes in the
C in the equatorial zones (Fig. 27.8; Goddéris et Joachimski entry into as well as the exit from the Carboniferous-Permian
2004). These results are in agreement with the development glaciation (Goddéris et al. 2017). The Hercynian orogen, by
of the first glaciers at the end of the Devonian. creating steep topographic slopes, would have allowed
physical erosion to increase causing the destruction of the
superficial soil formations, which had previously protected
The Permo-Carboniferous Glaciation the bedrock from chemical weathering. The weathering of
the newly exposed continental rocks would then have
The cooling trend which started in the Devonian continued induced a drop in atmospheric CO2 to levels allowing entry
and reached its peak during the Permo-Carboniferous into glaciation (Fig. 27.9 carb). This scenario of tectonic
glaciation (330–270 Ma) also known as the ‘Late Paleo- forcing leading to glaciation is not in opposition to the biotic
zoic Ice Age’ (Montañez and Poulsen 2013). It is the most hypothesis, since the increased weathering of the continents
important glacial event of the Phanerozoic. Debris carried by would increase the nutrient flows to the ocean and thus
drifting icebergs reached the paleolatitude of 30° (Frakes necessarily favor an increase in primary productivity in the
et al. 1992). The temperature of the tropical waters was ocean.
probably 2 °C below present values (Veizer et al. 2000). At The sequestration of organic carbon in continental sedi-
the same time, the continents were covered with dense for- ments may have forced a decrease in CO2 but it also led to a
ests in low latitude areas (in the latitude band 15°N–15°S). considerable increase in O2 pressure in the atmosphere, from
This major and prolonged cooling is traditionally con- 15% at the end of the Devonian to 32% of the air content
sidered to have been the final consequence of the coloniza- around 290 Ma (Berner 2004). This increase is also recorded
tion of continental surfaces by vascular plants, favoring in the d13C of fossil remains of continental vegetation. The
continental weathering. A second major effect of this fractionation of carbon isotopes reached 23‰ 290 Ma ago
accelerated growth also occurred: the burial of organic car- and was therefore about 5‰ higher than the average of the
bon in continental environments reached a level during the estimated values for the rest of the Paleozoic and Mesozoic.
Carboniferous never before attained during the whole In modern plants, this fractionation is a function of the O2/
Phanerozoic. Approximately 31  1015 mol Ma−1 of car- CO2 ratio in the atmosphere. A value of 23‰ indicates a
bon were buried on the continents during the Carboniferous, ratio of 1000. While the CO2 pressure was around 300 ppmv
27 The Phanerozoic Climate 373

Fig. 27.9 carb—Paleogeographic pattern of CO2 consumption by slopes and high runoff maintain a high flow of erosion (308 Ma). The
silicate weathering immediately before, during and after the gray continental surfaces represent arid areas with no runoff and no
Carboniferous-Permian glaciation. Weathering is inhibited by the weathering
development of a thick saprolith along the equator, except when steep

(close to the current value), O2 reached a pressure close to Fig. 27.5). This is interpreted as indicating a drastic reduc-
30% of the total pressure of the atmosphere at ground level. tion in the sequestration of continental organic carbon in
response to the decline in productivity by the biosphere, due
to a major reduction in precipitation (Berner 2004). More-
The End of the Paleozoic over, the end of the Permian is marked by intense volcanic
activity, with the establishment of large fissural eruptions
The end of the Paleozoic is marked by the exit from the (traps) in Siberia. These three factors reinforced the global
Permo-Carboniferous glaciation. The climate of the end of warming trend from the end of Permian onwards.
Permian is a much drier and warmer climate.
The transition from a cold mode to a warm mode is
triggered during the final aggregation of the Pangea in the The Mesozoic
middle of the Permian. The formation of a supercontinent
reduced the amount of precipitation on the continents, which The three geological stages of the Mesozoic have long been
partially inhibited the consumption of CO2 by silicate considered to be typical examples of hot climates, especially
weathering. In response to this imbalance between the vol- the Cretaceous. Long-term carbon cycle models such as
canic source of CO2 and the sinks through weathering and GEOCARB and all subsequent generations estimate very
deposition of carbonates, atmospheric CO2 accumulated, high values for atmospheric CO2 pressure over the entire
heating the system until the increased weathering due to Mesoozoic, between 4 and 10 times the current value, sup-
increasing temperature compensated for the lack of precip- ported by strong degassing of the solid Earth (Berner 2004).
itation, restoring the balance between sources and sinks of The only significant event was the appearance of flowering
CO2. This era is explored by much more powerful numerical plants (angiosperms) in the Cretaceous, thought to further
models than the simple models used for the Paleozoic. The increase the efficiency of the consumption of atmospheric
use of atmospheric general circulation models coupled with CO2 through silicate weathering. This brought about a sig-
biogeochemical cycle models allows a more detailed nificant drop in CO2 after 130 million years ago.
investigation of the role of paleogeography. New perspectives on the climate trends of the Mesozoic
Simulations carried out by atmospheric general circula- have emerged recently and suggest other important
tion models coupled with biogeochemical cycle models long-term changes. The first notable feature (Fig. 27.10) is
calculate that 250 Ma ago, CO2 pressure was only the steady increase in the d13C of carbonate sediments by
2400 ppmv in response to the coming together of Pangea about 1‰ from its lowest point in the Jurassic to the middle
(Donnadieu et al. 2006) and that continental temperatures Miocene in the Cenozoic (Katz et al. 2005). This increase is
reached 19 °C on average. By comparison, today, a similar typically explained by an increase in the ratio of organic
average, strongly influenced by the very cold Antarctic carbon to total carbon buried in marine sediments, reach-
continent, would be only 6.7 °C. The end of the Permian is ing ±20%, as the dislocation of Pangea increased the sur-
marked by a major negative excursion of the d13C of marine face area available for the accumulation of organic carbon.
carbonates (−3‰ over about ten million years. See Many sedimentological studies show a significant increase in
374 Y. Goddéris et al.

Fig. 27.10 d13C of carbonate

sediments over the last
200 million years. (according to
Katz et al. 2005)

the amount of organic carbon preserved on the shores of the model of global biogeochemical cycles, Donnadieu et al.
Atlantic Ocean as it opened up during the Jurassic and (2006) explored the climate and biogeochemical conse-
Cretaceous periods. quences of this dislocation. A configuration such as that of
This slow increase in the amount of buried organic carbon Pangea implies a weak continental runoff, due to its large
is also responsible for a net supply of oxygen to the continental nature. This causes a partial inhibition of conti-
ocean-atmosphere system (3  106 Gt O2). This process nental silicate weathering. According to the theory of the
should also consume atmospheric CO2 by storing an paleothermostat, atmospheric CO2 will increase, forcing the
increasing share of photosynthesized carbon in sediments. temperature to rise until the consumption of CO2 by silicate
The paleothermostat (see Chap. 5, Volume 2) also allows weathering compensates once more for the degassing of the
the carbon cycle to be maintained close to equilibrium. solid Earth. On the other hand, a configuration where the
Indeed, the cooling initiated by the increase in sequestration continents are dispersed leads to an increase in runoff and
of organic carbon will be compensated for by a reduction in thus greater efficiency of silicate weathering. This results in
the consumption of CO2 by the weathering of continental increased CO2 consumption due to weathering (Fig. 27.11),
silicates. The response of CO2 pressure in the atmosphere to and the climate cools down until the paleothermostat is again
this long-term evolution still needs to be quantified. balanced. At a constant rate of degassing of the Earth, the
A second notable feature of the Mesozoic is the frag- average annual temperature of the continents would have
mentation of Pangea, which began as early as 250 Ma. An decreased from 19 °C at the beginning of the Triassic to
event of this magnitude has a major impact on the Earth’s 10 °C at the end of the Cretaceous (Donnadieu et al. 2006).
climate and the carbon cycle. Using a numerical model An interesting finding is that this simulated global cooling
coupling an atmospheric general circulation model and a is not linear over the whole Mesozoic. The main episode of
27 The Phanerozoic Climate 375

Fig. 27.11 Levels of

atmospheric CO2, calculated for
the last 250 million years (red
line), in response to continental
drift. The vertical lines represent
the available data

cooling and humidification of continental surfaces occurred became colder, but more humid, allowing the paleother-
during the Upper Triassic, between the Carnian and the mostat equilibrium to be maintained (equilibrium degassing
Rhetian (237–201 Ma). The Lower and Middle Triassic are of the solid Earth—silicate weathering), but at a lower level
characterized by very high continental temperatures and of CO2 than in the middle of the Triassic. It is remarkable
extensive aridity (average annual continental runoff of only that these cooler conditions (but nevertheless up to 4 °C
23.5 cm/year), consistent with sedimentological data from warmer than is currently the case on the continents) persisted
the establishment of redbeds and massive deposits of evap- after the Triassic, driven by the break-up of the Pangea
orites. The corresponding atmospheric CO2 pressures are rather than by its general latitudinal movement.
close to 3000 ppmv, in agreement with the current recon-
structions (Royer 2006), suggesting values between 2000
and 4000 ppmv for the same period. The Cenozoic
An abrupt change occurred in the last stage of the Triassic
(Rhetian 209–201 Ma), during which time 50% of the The overall climate evolution of the Cenozoic is better
Mesozoic cooling occurred in the model. Goddéris et al. understood than that of the preceding epochs. Nevertheless,
(2008) calculated CO2 pressures of around 900 ppmv and the causes of this evolution are still widely disputed. The
global average temperatures lower by 4.6 °C. The CO2 climate history of the last 65 million years is that of a
levels estimated based on the count of stomata on fossil transition from the warmer Cretaceous climate, characterized
leaves confirm these low levels of CO2 between 500 and by little or no polar ice caps, towards the current glacial
1000 ppmv (Royer et al. 2004). The d18O measured on climate.
brachiopod shells show an increase of 0.8‰ between the The oldest stage of the Cenozoic, the Paleocene, is
Carnian and the Rhaetian, i.e. an overall cooling of more characterized by a climate similar to that of the late Creta-
than 3 °C, (Korte et al. 2005) in line with modeling results. ceous. The first break with the Mesozoic is at the
Similarly, sedimentological data clearly show an increase in Paleocene-Eocene transition (56 Ma, Fig. 27.3). This tran-
moisture and a decrease in temperature during the Rhaetian sition is marked by an extremely intense global warming.
(Fig. 27.12). Changes in clay mineralogy and in the condi- The deep waters of the ocean warmed up to about 5–7 °C in
tions for pedogenesis are signs of the installation of cooler response to global warming and to a reorganization of ocean
and wetter climate regimes around 209 Ma ago, during the circulation. Similarly, the surface waters heated up by 8 °C
Norian-Rhaetian transition (Ahlberg et al. 2002). (Thomas et al. 1999; Zachos et al. 2003; Sluijs et al. 2006).
The causes of this rapid cooling may be found in the This warming, probably reinforced by the destabilization of
general drift of the Pangea towards the north. During the methane hydrates in the sediments (McInerney and Wing
Middle Triassic, large continental areas were located in the 2011) was of short duration, spanning just 200,000 years.
southern zone of the inter-tropical divergence, a very arid This brief episode was followed by the Eocene climate,
area and therefore not conducive to weathering. The shift of which lasted about 5 million years. (Fig. 27.3).
the Pangea to the north brought these large areas into the From the time of entry into the Middle Eocene, the cli-
humid equatorial zone, allowing increased atmospheric CO2 mate began to cool down globally, leading to the appearance
consumption through increased runoff. Thus, the world of small temporary ice sheets that developed on the Antarctic
376 Y. Goddéris et al.

Fig. 27.12 Continental runoff

calculated for the Carnian and
Rhaetian, two stages of the
Triassic, at 3122 ppmv of CO2.
The red rectangle shows the
position of Europe and indicates
increased humidification of the
area between the Carnian and the
Rhaetian

in the last phase of the Eocene (Zachos et al. 2008). At the for this cooling. From 50 Ma onwards, ocean basins began
Eocene-Oligocene transition (34 Ma) a distinct, cold climate to become established in the Drake Passage area, allowing
pulse was experienced. The d18O data obtained from benthic shallow water exchanges between the Atlantic and Pacific
foraminifera show a deep-water cooling of about 4 °C Oceans. Isotopic analyses of neodymium in deep sediments
(Fig. 27.3). A permanent Antarctic cap began to emerge indicate that around 41 Ma, the flow of exchanges between
(Zachos et al. 2008). ocean basins intensified around Antarctica (Scher and Martin
The second step towards global cooling occurred at the 2006). The intensification of the rate of expansion of the
Oligocene-Miocene boundary (23 Ma). But this cold epi- seabed in the Drake and the Tasmanian Passages 34 Ma ago,
sode is followed by the Miocene climate optimum between allowed the Antarctic circumpolar current to become estab-
23 and 15 Ma. The latter is particularly problematic because, lished, definitively isolating the South Pole continent. This
according to the d11B and d13C data from alkenones, date coincides with a major development phase of the
atmospheric CO2 pressure should have been low, at around Antarctic ice cap.
the present value, or even lower, between 200 and 300 ppmv The role of ocean passages in global climate change has
(Pagani et al. 1999). The reasons for this climate optimum been called into question by modeling studies, which tend to
are still unknown. show that glaciation and the appearance of an ice cap over
From 15 Ma onwards, the climate cooled rapidly and the Antarctica are mainly associated with a decrease in the level
East Antarctic sheet developed. The last stage of cooling, of CO2 Lefebvre et al. (2012), see also Chap. 3. It should be
5 Ma ago, was marked by the establishment of the West noted, however, that this premise is very poorly documented.
Antarctic ice sheet (Fig. 27.3). Indeed, this is a period for which d11B data are non-existent.
The reasons for this cooling are still in dispute. There are At best, we know that CO2 levels were around 1000 ppmv
two opposing theories. According to one, the climate evo- 40 Ma ago, and about 300 ppmv 24 Ma ago. These data
lution of the Cenozoic was largely driven by the opening of suggest decreasing levels of CO2 although it is not possible
key ocean passages and the closure of others. According to to document the evolution precisely. Nevertheless, DeConto
the other, it was the drop in CO2 level that was responsible and Pollard (2003) have shown in a simulation that even
27 The Phanerozoic Climate 377

where the Drake Passage is kept artificially closed, the the Himalayas and that the discharge of the debris produced
decrease in atmospheric CO2 concentration causes glaciation by mechanical erosion is too fast to allow the progress of
to occur over the Antarctic. Glaciation is simply delayed by effective chemical weathering. This would produce a
about 2 to 3 million years. weathering system which would be very limited by the very
Changes in the climate caused by the opening or slow kinetics of the dissolution of minerals. In addition, the
non-opening of the Drake Passage will also affect the global lithology is such that calcium silicates are scarce in the
carbon cycle, potentially increasing the climate response of Himalayas and the weathering fluxes are mostly of sodium
the system. One possible consequence of the opening of the and potassium silicates. Since these chemical reactions do
Drake Passage is an upheaval of the thermohaline circulation not lead to precipitation of carbonates, their effect on the
facilitating the formation of deep waters in the Antarctic and carbon cycle is minimal over the long term (France-Lanord
triggering the plunging of waters in the North Atlantic and Derry 1997).
(Huber and Nof 2006). The result is a warming of the However, the rate of sedimentation, which is extremely
northern hemisphere of approximately 3 °C and a severe high in the Bay of Bengal, is responsible for the preservation
cooling of the Antarctic. As most of the continental area is of very large quantities of organic matter, of both continental
located in the northern hemisphere, an increase in global and marine origin. It is estimated that the sequestration of
consumption of CO2 by continental silicate weathering is to carbon at the foot of the orogen is two to three times higher
be expected and thus a reduction in the amount of CO2 in the than the consumption of CO2 by weathering of the Hima-
atmosphere, reinforcing the cold climate mode being estab- layan silicates. A recent study shows that 100% of the
lished (Elsworth et al. 2017). In conclusion, although the organic carbon of continental origin transported by the
overall climate effect of the opening of the Drake Passage Himalayan rivers is preserved in the sediments of the Gulf of
remains weak, there may have been positive feedbacks in the Bengal (Galy et al. 2007). France-Lanord and Derry (1997)
carbon cycle which substantially amplified the response. estimated that the sedimentary organic carbon reservoir grew
These have yet to be documented with precision. to 0.6  1012 mol yr−1. This value is of a similar order of
A second driver of the evolution of climate also took magnitude to estimates from numerical simulations, carried
place during the Cenozoic: the Himalayan orogen. out using a carbon cycle model reversing the records of d13C
How orogeny affects the carbon cycle is complex. We in carbonates during the Cenozoic period (Goddéris and
have identified two effects: one is the chemical weathering of François 1996). The Himalayas consume carbon
exposed silicates in the mountain chains, the other is the (Fig. 27.13), but in organic form, and therefore, they are, at
sequestration of organic carbon at the foot of the mountains. least partially, responsible for the cooling of the climate
Take first the increase in weathering of continental surfaces during the Cenozoic. The quantification of the impact of this
through increased erosion. The development of glaciers, the mechanism on atmospheric CO2 has yet to be completed.
alternating freezing and thawing patterns at high altitudes
and steep slopes all favor the break-up of rocks and increase
the area of contact with solutions. This results in increased Abrupt Climate Events During
weathering and increased consumption of CO2. This the Phanerozoic
increased weathering is seen in an increase in the erosion
factor f3 in Eq. (10), and the level of CO2 is lowered until the As well as defining the major climate modes of the
weathering of the silicates again compensates for the Phanerozoic, recent efforts have defined episodes of rapid
degassing of the solid Earth (Goddéris and François 1996). climate change that have punctuated the history of the Earth
Currently, 4  1012 kg yr−1 of suspended solids are trans- at a 100,000−year scale. We outline below three of these
ported to the ocean from the Himalayan zone, representing events, discussing their causes.
17% of the world’s erosion flow, whereas the ratio of the
Himalayan surface to the total continental area is only 4%. It
is therefore to be expected that a major orogen would con- The Callovian-Oxfordian Transition (Middle
siderably increase the consumption of atmospheric CO2 Jurassic-Upper Jurassic)
through chemical weathering of the exposed rocks (increase
in factor f3). However, this result is not confirmed by current This was a brief cooling episode during the Jurassic. Such
data of fluxes of dissolved elements in the rivers from the events occurred several times during the Jurassic and Cre-
Himalayas. They suggest a modest consumption of taceous periods.
0.7  1012 mol yr−1 of CO2 by weathering of Himalayan The d18O values measured in fish teeth and belemnites
silicates, only 6% of the world total of 11.7  1012 mol suggest an abrupt drop in temperature of 8 °C starting in the
yr−1. One of the reasons for this low rate of chemical Upper Callovian and remaining until the middle Oxfordian.
weathering may be that the erosion motor is too efficient in At the same time, boreal ammonite fauna invaded the
378 Y. Goddéris et al.

Another hypothesis advanced recently connects the

massive reduction in carbonate deposits on the continental
shelves to the cooling episode. To date, it was assumed that
the arrival of colder climate conditions resulted in a reduc-
tion in bioconstruction activity in the reef areas. Conversely,
Donnadieu et al. (2011) suggest that the strong decrease in
carbonate reef activity due to external causes (tectonic rea-
sons for example) caused an accumulation of alkalinity in
the oceans. In fact, this alkalinity continued to be supplied
by rivers (continental weathering), whereas the alkalinity
sink by deposition of carbonates was greatly reduced. The
result was a massive dissolution of atmospheric CO2 in the
oceans. This caused CO2 pressure to drop from 800 to
200 ppmv during crises in the carbonate production (lasting
a few hundred thousand years), resulting in a global average
cooling of 9 °C.

Fig. 27.13 Atmospheric CO2 consumption by the Himalayas, per kg The Cretaceous-Tertiary Boundary, Meteorite
of sediment exported. On the left, the contribution of silicate and the Deccan Traps
weathering, reconstructed from the balance of each cation exported
by the mountain range. On the right, the contribution associated with The Cretaceous-Tertiary boundary (K–T), dating back to
the sequestration of organic carbon in the Bay of Bengal
(France-Lanord and Derry 1997) 66 Ma, has been studied in detail because it corresponds to a
mass extinction event, which eliminated, among other spe-
cies, the dinosaurs. Two major events occurred at the K–T
Tethyan domain, suggesting widespread cooling at all lati- boundary: the collision of a meteorite with the Earth and the
tudes (Dromart et al. 2003). The total duration of the event establishment of the Deccan traps. This latter is a major
was about 3 million years. This episode is also marked by magmatic event which may have had a major impact on the
the reduction, by a factor of about 10, of carbonate deposits. biosphere, but certainly had on the Earth’s climate from 105
The d13C values also suggest a positive excursion of about to 106 years.
0.5‰ over the same period. Finally, a drop in the sea level of Dessert et al. (2001) have simulated the impact of the
several tens of meters is supported by evidence (Dromart Deccan Traps on the geochemistry and climate of the Earth.
et al. 2003), suggesting the establishment of temporary ice The total volume of lava put in place is 3  106 km3, cor-
caps (Fig. 27.14). responding to the emission of 1.6  1018 mol of CO2, or
The reasons for this cooling are not completely under- half the current carbon content of the exosphere. This
stood, nor, indeed are those of all the abrupt cooling epi- emission could have occurred within a timeframe of about
sodes of the Jurassic and the Cretaceous. However, several 105 years. This time scale is shorter than the response time
avenues of enquiry have been opened. In particular, for this of the geological carbon cycle. This is thus far beyond the
specific event, the amount of organic matter buried in marine capacity of the Earth’s paleothermostat to respond. This
sediments increased greatly during the middle Callovian just produced a very rapid increase in the partial pressure of CO2
before the cooling. The percentage of organic matter which was increased to more than 3.5 times its initial level in
increased from less than 1% during the early Callovian to 100 000 years (i.e. 1000 ppmv, assuming that the
5%, and even 10% in the middle Callovian. This increased pre-disturbance CO2 levels were at the pre-industrial level of
rate of burial of organic matter may have led to an increase 280 ppmv: Dessert et al. 2001) The global average tem-
in consumption of atmospheric CO2 on a temporal scale perature was thus increased by 4 °C a hundred thousand
sufficiently short, relative to the response time of the ter- years after the establishment of the Traps.
restrial paleothermostat so that it was unable to intervene as Once the eruption ended and time passed, the Earth’s
a stabilizer. This would have resulted in a reduction in CO2 paleothermostat could then take on its stabilizing role. The
pressure which could have initiated the subsequent cooling. surplus CO2 was slowly consumed by silicate weathering,
This cooling, accompanied by a drop in sea level caused by which was itself accelerated by the increased greenhouse
the development of glaciers on the coldest continents, was effect. Over 2 million years, the level of CO2 returned to a
responsible for the near halting of carbonate sedimentation stable level, one that was lower than the pre-disturbance
on the continental shelves. level by 60 ppmv, corresponding to an overall cooling
27 The Phanerozoic Climate 379

Fig. 27.14 Summary of events

at the Middle Jurassic-Upper
Jurassic transition. On the left, the
d13C of the carbonates, measured
at different locations. On the
right, layers rich in organic
carbon, and the accumulation rate
of carbonates

of ±0.5 °C compared to the temperature prior to the estab- level stabilized a few million years after the end of the event
lishment of the Deccan Traps (Fig. 27.15). This cooling was at 750 ppmv below its pre-disturbance level of 4500 ppmv,
the result of continental surfaces (continental shelf) that were which caused a global cooling of more than 1 °C.
very resistant to weathering, being replaced by 500,000 km2 At the time of the K–T limit, another major phenomenon
of fresh basaltic surfaces, eight to ten times more prone to occurred: a large meteorite with a diameter estimated at
weathering. Mathematically, it is the factor f5 (Eq. 10) of the about ten kilometers collided with the Earth and fell into the
paleothermostat which has increased globally (the whole of Yucatan Peninsula. The impact created a large crater, iden-
the continental surfaces being slightly more vulnerable to tified by geophysics, which is currently buried under a
weathering) and, for a degassing which returned to its thousand meters of sediment (the Chicxulub crater). This
pre-disturbance level, the CO2 must stabilize at a lower level event was catastrophic, much shorter than the great fissure
in order to correct the imbalance due to continental eruptions of the Deccan, which date from the same period,
weathering. but which have had a prolonged impact for several hundreds
The result of a magma episode, such as the establishment of thousands of years.
of continental traps, is initially a short-lived warming epi- The impact of the meteorite is easily identified because
sode (105 years), followed by a global cooling that persists sediments from the K–T boundary are composed of a thin
for several million years, as long as the basaltic surfaces layer, rich in iridium, a very rare metal on Earth, a sign of
exposed to the atmosphere are not entirely destroyed by contribution of cosmic origin. This layer also contains
weathering. A similar study carried out on Siberian traps minerals (spinels) whose chemical composition indicates
(Permo-Triassic boundary) shows that the atmospheric CO2 that they could not have been formed on Earth. Analysis of
380 Y. Goddéris et al.

• an enormous quantity of water vapor is emitted into the

atmosphere, which becomes charged with nitrates and
sulfates, and falls back in the form of highly acid rain,
toxic to plants;
• finally, chondrites contain many toxic heavy metals,
particularly nickel, which inhibits chlorophyll activity.

We are obviously far from a full comprehension of all the

events that marked the end of the Cretaceous, with the dis-
appearance of many animal and vegetal species. Continental
sediments testify to the appearance of widespread fires and
to the pioneering return of the ferns, the most resistant of
plants and the first to colonize the areas devastated by fire. In
the ocean, sedimentological, biological and geochemical
data show a considerable decrease in primary production by
algae; this only returned to its former level after about five or
six million years.
It is likely that the collision with the meteorite and fumes
from the Deccan fissures both contributed to the major
changes in the environment that marked the KT boundary,
the first event through considerable sudden effects lasting
several years and the second through geochemical effects
that persisted over long periods relative to the time constants
Fig. 27.15 Quantification of the impact of the establishment of the
of the biosphere.
Deccan traps on the global average temperature and the level of
atmospheric CO2
The Paleocene–Eocene Thermal Maximum (PETM)

the spinels showed that the meteorite had the chemical The destabilization of methane hydrates (very active green-
composition of carbon chondrites, with the particularity of house gases) in sediments can cause climate fluctuations
being very rich in sulfur. The impact of such a collision was over short timescales (105 years). Several events of this type
considerable. This impact, out of all proportion with current have been identified during the Phanerozoic, but the best
observations, is difficult to simulate because of the extent of documented is located at the Paleocene-Eocene transition. In
the disturbance to the chemistry of the atmosphere. Thus, the the space of 20 000 years (Fig. 27.16), the d13C of the ocean
simulations are based on assumptions made in the context of decreased by 3‰, before returning to its initial value 240
studies on a ‘nuclear winter’, where climatologists have 000 years later (McInerney and Wing 2011). Over the same
calculated the impact on the global climate of a large-scale time, the temperature of the deep ocean waters increased
nuclear conflict. This can only be considered to be a very from 5 to 7 °C. Similarly, a warming of 8 °C of surface
simplified approach. waters was recorded. This warming is attributed to a sudden
The collision of a meteorite with the Earth has many destabilization of methane hydrates in ocean sediments
consequences, although it is difficult to quantify them (Dickens 2003). The methane released by sediments is
precisely: characterized by a d13C of −60‰. As a result, a flow of 2500
Gt of carbon spread over 20,000 years is sufficient to explain
• it releases an enormous quantity of aerosols (sulfates, the observed isotopic excursion. This event caused a sig-
nitrates) which reach the upper atmosphere where they nificant but temporary warming of the atmosphere. The d13C
can remain for several years; excursion is then reabsorbed over 200,000 years by the
• the aerosols cause an attenuation of about 50% of the ‘conventional’ processes of the carbon cycle: continental
solar radiation, resulting in a cooling of about ten degrees weathering and sequestration in sediments.
on the ground for a decade. Agronomists estimate that The reason for the destabilization of the gas hydrates has
half the vegetation of the northern hemisphere could have yet to be explained. These can be released into the ocean and
been killed in the first years; atmosphere if the water temperature rises or the pressure
• the disruption is greatest if the impact occurs in the spring decreases. For example, regional eruptions occurred in the
when the vegetation most needs solar radiation; North Atlantic 55 Ma ago, shortly before the PETM
27 The Phanerozoic Climate 381

Fig. 27.16 Temporal evolution

of d13C of benthic foraminifera at
the Paleocene-Eocene transition
at three distant oceanic sites. The
arrow represents a period of
240,000 years

(McLennan and Jones 2006). The reduction in the height of not sure that complex models, such as atmospheric general
the water column could have led to a drop in pressure and a circulation models coupled with ocean-atmospheric models,
destabilization of the methane hydrates in sediments. There react correctly when boundary conditions are changed in
may be a link between these regional eruptions and the such a drastic manner.
PETM, although this assumption needs to be tested by more Nevertheless, the refining of analytical techniques and
precise dating. Another hypothesis is that a significant models, the process of trial and error, the successes and the
change in ocean circulation could have significantly warmed failures allow us to discover a general history of the Earth’s
the deep waters (by 4–5 °C), causing the destabilization of climate at the same time as multicellular organisms evolved.
gas hydrates stored in sediments and initiating the PETM. One of the major debates of recent years has been around the
link between the level of atmospheric CO2 and the evolution
of climate during the Phanerozoic, which appear to have
Conclusions been decoupled during certain major events. Given the
enormous uncertainties that exist in the reconstruction of
The reconstruction of climates on the scale of geological CO2 levels using isotopic methods or based on paleonto-
time is still open to discussion. The main difficulty lies in the logical data (Royer 2006) and taking into account the
fact that climate and biogeochemical cycles cannot be dis- uncertainties around isotope-based climate reconstructions,
sociated. There are numerous indicators of geochemical and it is not possible to claim the existence or non-existence of a
climatic changes, but they are always difficult to interpret decorrelation between CO2 level on one side and climate on
because of their indirect nature. The numerical models used the other. However, the use of a new generation of models
are often very simple, taking a global average approach in that closely couples the carbon cycle with climate by taking
most studies, which does not explicitly take into consider- into account the spatial variability of the processes suggests
ation the many parameters of the climate system. Yet ancient a coherence in the joint history of CO2 and climate, in line
climates also represent an amazing testing-ground where with the major climate trends of the Mesozoic (Donnadieu
new techniques can be developed and innovative ideas can et al. 2006). Similarly, the emergence of isotopic techniques
be explored. In this field of study, climate models, initially allowing the reconstruction of the climate with increasingly
developed to understand the evolution of current climate, are fine latitudinal resolution makes it possible to reconsider the
applied to extreme conditions, and the extent to which they commonly studied events in the climate history of our pla-
are suitable is questionable. One limitation is the simplistic net. New consolidated images appear, in which atmospheric
way the changing geographical configuration of the past is CO2 is a key driver of climate change but is modulated by
taken into account due to the lack of precise information. first order factors, such as the paleogeographic configuration,
This is a climate factor of the highest importance and we are largely ignored for a long time, or the more or less periodic
382 Y. Goddéris et al.

variations in the Earth’s orbit, the role of which is explained Geophysics Geosystems, 7(11). https://doi.org/10.1029/
in Chap. 7. 2006gc001278.
Dromart, G., et al. (2003). Ice age at the middle-late jurassic transition?
One of the major challenges for the near future is Earth and Planetary Science Letters, 213, 205–220.
understanding the relationship between these climate and Elsworth, G. et al. (2017). Enhanced weathering and CO2 drawdown
geochemical evolutions on the one hand and biological caused by the latest Eocene strengthening of the Atlantic meridional
evolution on the other. overturning circulation. Nature Geoscience, 10(3), 213–216.
Finnegan, S., et al. (2011). The magnitude and duration of late
ordovician-early silurian glaciation. Science, 331, 903–906.
Foster, G. L., Royer, D. L., Lunt, D. J. (2017) Future climate forcing
References potentially without precedent in the last 420 million years. Nature
Communications. https://doi.org/10.1038/ncomms14845.
Frakes, L. A., Francis, J. E., & Syktus, J. I. (1992). Climate modes of
Ahlberg, A., Arndorff, L., & Guy-Ohlson, D. (2002). Onshore climate the phanerozoic. Cambridge: Cambridge University Press.
change during the Late Triassic marine inundation of the Central France-Lanord, C., & Derry, L. A. (1997). Organic carbon burial
European Basin. Terra Nova, 14(4), 241–248. forcing of the carbon cycle from Himalaya erosion. Nature, 390,
Anderson, H. M., Hiller, N., & Gess, R. W. (1995). Archeopteris 65–67.
(progymnospermopsida) from the Devonian of Southern Africa. Friedrich, O. et al. (2012). Evolution of middle to Late Cretaceous
Journal of the Linnean Society, 117(4), 305–320. oceans - a 55 m.y. record of Earth’s temperature and carbon cycle.
Bergmann, K. D. et al. (2018). A 600-million-year carbonate Geology, 40(2), 107–110.
clumped-isotope record from the Sultanate of Oman. Journal of Galy, V., et al. (2007). Efficient organic carbon burial in the Bengal fan
Sedimentary Research, 88, 960–979 sustained by the Himalayan erosional system. Nature, 450, 407–
Berner, R. A. (2004). The phanerozoic carbon cycle (p. 150p). New 410.
York: Oxford University Press. Ghosh, P., Garzione, C. N., & Eiler, J. M. (2006). Rapid uplift of the
Bice, K. L., & Norris, R. D. (2002). Possible atmospheric CO2 Altiplano revealed through C-13-O-18 bonds in paleosol carbon-
extremes of the Middle Cretaceous (late Albian-Turonian). Paleo- ates. Science, 311(5760), 511–515.
ceanography, 17, 1070. Goddéris, Y., et al. (2008). Causal of casual link between the rise of
Boucot, A. J., Xu, C., & Scotese, C. R. (2004). Phanerozoic climatic nannoplankton calcification and a tectonically-driven massive
zones and paleogeography with a consideration of atmospheric CO2 decrease in the Late Triassic Atmospheric CO2? Earth and
level. Paleontological Journal, 38, 115–122. Planetary Science Letters, 267, 247–255.
Breecker, D. O., Sharp, Z. D., & McFadden, L. D. (2010). Atmospheric Goddéris, Y., & François, L. M. (1996). Balancing the cenozoic Carbon
CO2 concentrations during ancient greenhouse climates were and Alkalinity Cycles: Constraints from isotopic records. Geophys-
similar to those predicted for AD 2100. Proceedings of the ical research letters, 23(25), 3743–3746.
National Academy of Science, 107, 576–580. Goddéris, Y. et al. (2014). The role of palaeogeography in the
Came, R. E., et al. (2007). Coupling of surface temperatures and Phanerozoic history of atmospheric CO2 and climate. Earth-Science
atmospheric CO2 concentrations during the paleozoic era. Nature, Reviews, 128, 122–138.
449, 198–201. Goddéris, Y. et al. (2017). Onset and ending of the late Palaeozoic ice
Caputo M. V. et al. (2008). Late devonian and early carboniferous age triggered by tectonically paced rock weathering. Nature
glacial records of South America. In Resolving the late Paleozoic Geoscience, 10(5), 382–385.
ice age in time and space (Vol. 441). Geological Society of Hayes, J. M., Strauss, H., & Kaufman, A. J. (1999). The abundance of
13
America Special Papers, (pp. 161–173). C in marine organic matter and isotopic fractionation in the global
Dabard M. P. et al. (2015). Sea-level curve for the Middle to early Late biogeochemical cycle of carbon during the past 800 Ma. Chemical
Ordovician in the Armorican Massif (western France): icehouse Geology, 161, 103–125.
third-order glacio-eustatic cycles.Palaeogeography Palaeoclimatol- Hodel, F. et al. (2018). Fossil black smoker yields oxygen isotopic
ogy Palaeoecology, 436, 96–111. composition of Neoproterozoic seawater. Nature Communications,
Davies N. S., & Gibling, M. R. (2013). The sedimentary record of 9, Article Number: 1453.
Carboniferous rivers: continuing influence of land plant evolution Huber, M., & Nof, D. (2006). The ocean circulation in the Southern
on alluvial processes and Palaeozoic ecosystems. Earth-Science Hemisphere and its climatic impacts in the eocene. Palaeogeogra-
Reviews, 120, 40–79. phy, Palaeoclimatology, Palaeoecology, 231, 9–28.
DeConto, R. M., & Pollard, D. (2003). Rapid cenozoic glaciation of Joachimski, M. M. et al. (2004). Oxygen isotope evolution of biogenic
antarctica induced by declining atmospheric CO2. Nature, 412, calcite and apatite during the Middle and Late Devonian. Interna-
245–248. tional Journal of Earth Sciences,93(4): 542–553.
Dera, G. et al. (2011). Climatic ups and downs in a disturbed Jurassic Katz, M. E. et al. (2005). Biological overprint of the geological carbon
world. Geology, 39(3), 215–218. cycle. Marine Geology, 217, 323–338.
Dessert, C., et al. (2001). Erosion of deccan traps determined by river Kump, L. R., et al. (1999). A weathering hypothesis for glaciation at
geochemistry: Impact on the global climate and the 87Sr/86Sr ratio of high atmospheric pCO2 during the late ordovician. Palaeogeogra-
seawater. Earth and Planetary Science Letters, 188(3/4), 459–474. phy, Palaeoclimatology, Palaeoecology, 152, 173–187.
Dickens, G. R. (2003). Rethinking the global carbon cycle with a large, Le Hir, et al. (2011). The climate change caused by the land plant
dynamic and microbially mediated gas hydrate capacitor. Earth and invasion in the Devonian. Earth and Planetary Science Letters, 310
Planetary Science Letters, 213(3–4), 169–183. (3–4), 203–212.
Donnadieu, Y., et al. (2011). A mechanism for brief glacial episodes in Lefebvre, V. et al. (2013). Was the Antarctic glaciation delayed by a
the mesozoic greenhouse. Paleoceanography, 26. https://doi.org/10. high degassing rate during the Early Cenozoic? Earth and
1029/2010pa002100. Planetary Science Letters, 371–372, 203–211.
Donnadieu, Y., et al. (2006). A GEOCLIM simulation of climatic and Lenton, T. M. et al. (2012). First plants cooled the Ordovician. Nature
biogeochemical consequences of Pangea breakup. Geochemistry Geoscience, 5, 86–89.
27 The Phanerozoic Climate 383

Loi, A. et al. (2010). The Late Ordovician glacio-eustatic record from a Royer, D. L., Berner, R. A., & Beerling, D. J. (2001). Phanerozoic
high-latitude storm-dominated shelf succession: The Bou Ingarf atmospheric CO2 change: Evaluating geochemical and paleobio-
section (Anti-Atlas, Southern Morocco). Palaeogeography Palaeo- logical approaches. Earth-Science Reviews, 54, 349–392.
climatology Palaeoecology, 296(3–4), 332–358. Rubinstein, C. V. et al. (2010). Early middle ordovician evidence for
McElwain, J. C., Wade-Murphy, J., Hesselbo, S. P. (2005). Changes in land plants in Argentina (eastern Gondwana). New Phytologist, 188
carbon dioxide during an oceanic anoxic event linked to intrusion (2), 365–369.
into Gondwana coals. Nature, 435(7041), 479–482. Scher, H. D., & Martin, E. E. (2006). Timing and climatic
McInerney, F. A., & Wing, S. L. (2011). The paleocene-eocene thermal consequences of the opening of drake passage. Science, 312,
maximum: A perturbation of carbon cycle, climate, and biosphere 428–431.
with implications for the future. Annual Review of Earth and Scott, A. C., & Glasspool, J. (2006). The diversification of Paleozoic
Planetary Sciences, 39, 489–516. fire systems and fluctuations in atmospheric oxygen concentration.
Mountanez, I. P., & Poulsen, C. J. (2013). The late paleozoic ice age: Proceedings of the National Academy of Science, 103(29), 10861–
An evolving paradigm. Annual Review of Earth and Planetary 10865.
Sciences, 41, 629–656. Shaviv, N. J., & Veizer, J. (2003). Celestial driver of phanerozoic
Nardin, E., et al. (2011). Modeling the early paleozoic long-term climate? GSA Today, 13(7), 4–10.
climatic trend. Geological Society of America Bulletin, 123, 1181– Sluijs, H. et al. (2006). Subtropical arctic ocean temperatures during the
1192. Palaeocene/Eocene thermal maximum. Nature, 441(7093), 610–
Nelsen, M. P. et al. (2016). Delayed fungal evolution did not cause the 613.
Paleozoic peak in coal production. Proceedings of the National Stein, W. E. et al. (2007). Giant cladoxylopsid trees resolve the enigma
Academy of Science, 113(9), 2442–2447. of the Earth’s earliest forest stumps at Gilboa. Nature, 446(7138),
Pagani, M., Zachos, J. C., Freeman, K. H., Tipple, B., & Bohaty, S. 904–907.
(2005). Marked decline in atmospheric carbon dioxide concentra- Thomas, D. J. et al. (1999). New evidence for subtropical warming
tions during the Paleogene. Science, 309, 600–603. during the late Paleocene thermal maximum: stable isotopes from
Page, A. A. et al. (2007). Were transgressive black shales a negative the Deep Sea Drilling Project Site 527, Walvis Ridge. Paleo-
feedback modulating glacioeustasy in the Early Palaeozoic ice- ceanography, 14, 561–570.
house? In Deep-time perspectives on climate change: marrying the Trotter, J. A. et al. (2008). Did cooling oceans trigger Ordovician
signal from computer models and biological proxies. pp. 123–156. biodiversification? Evidence from conodont thermometry. Science,
Pohl, A. et al. (2016). Glacial onset predated Late Ordovician climate 321, 550–554
cooling. Paleoceanography, 31(6), 800–821 Veizer, J., et al. (1999). 87Sr/86Sr, d13C and d18O evolution of
Pucéat, E., et al. (2010). Revised phosphate-water fractionation phanerozoic seawater. Chemical Geology, 161, 59–88.
equation reassessing paleotemperatures derived from biogenic Veizer, J., Goddéris, Y., & François, L. M. (2000). Evidence for
apatite. Earth and Planetary Science Letters, 298, 135–142. decoupling of atmospheric CO2 and global climate during the
Pucéat, E. et al. (2003). Thermal evolution of Cretaceous Tethyan phanerozoic eon. Nature, 408, 698–701.
marine waters inferred from oxygen isotope composition of fish Zachos, J. C. et al. (2003). A transient rise in tropical sea surface
tooth enamels. Paleoceanography, 18,(2), Article Number: 1029. temperature during the Paleocene-Eocene Thermal Maximum.
Rasmussen, C. M. O. et al. (2016). Onset of main Phanerozoic marine Science, 302(5650), 1551–1554.
radiation sparked by emerging Mid Ordovician icehouse. Scientific Zachos, J. C., Dickens, G. R., & et Zeebe, R. E. (2008). An early
Reports, 6, Article Number: 18884. Cenozoic perspective on greenhouse warming and carbon-cycle
Raymo, M. E. (1991). Geochemical evidence supporting T.C. cham- dynamics. Nature, 451(17). https://doi.org/10.1038/nature.
berlin’s theory of glaciation. Geology, 19, 344–347. Zhuravlev, A. Y., & Riding, R., (2001). The ecology of the Cambrian
Royer, D. L. (2006). CO2-forced climate thresholds during the radiation - Introduction. Perspectives in paleobiology and Earth
phanerozoic. Geochimica and Cosmochimica Acta, 70, 5665–5675. history series, pp. 1–7.
 Climate and Astronomical Cycles
28
Didier Paillard

reached it), were only greeted by the whistles of the northern

A Little History winds and the thunder of the crevasses that opened on the sur-
face of this vast ocean of ice.
From the Discovery of Ice Ages to the First (L. Agassiz, Studies on glaciers, 1840)
Climate Theories For geologists of that era, climate changes were primarily
related to changes in topography. Although the continental
The scientific notion that climate could evolve over time first drift was not yet well established, the scientists tried to
appeared along with the birth of paleoclimatology in the explain their field observations by means of vertical move-
nineteenth century when the existence of ice ages was dis- ments of the continents: uplift of the mountains, erosion,
covered. Although the scientists of the time were aware that change in sea level etc. Under the influence of Charles Lyell,
the Earth had undergone many upheavals, notably through these changes came to be seen as slow and progressive. The
the successions of various animal and vegetable fossil spe- assumption of catastrophism faded away to be replaced with
cies, it was only when evidence of previous glacial periods gradualism which presupposes that the modifications of the
emerged that the idea of a changing climate really took Earth’s surface obey the physical laws that now apply but
shape. The traces left behind by the movement of glaciers over immensely long durations. This new point of view was
during the last glacial period: moraines, erratic boulders, opposed to any external (especially heavenly) influence,
glacial striations and features, were clearly identified in which no doubt explains the reluctance of geologists when
many regions of Europe and North America and promoted faced with the first astronomical theories of glaciations:
the idea that the climate, at least in the northern hemisphere,
may have been considerably colder in the past. The But though I am inclined to profit by Croll’s maximum eccen-
tricity for the glacial period, I consider it quite subordinate to
nineteenth-century scientists’ point of view on glaciations
geographical causes or the relative position of land and sea and
was quite different from ours, deeply rooted as it was in a abnormal excess of land in polar regions.
‘catastrophic’ perspective on the evolution of the Earth, (C. Lyell to C. Darwin, 1866)
which was at that time imagined to have been punctuated by
The idea that the climate stems above all from geography
‘deluges’ and other cataclysms, as highlighted in the excerpt
and topography has indeed merit. One need only look at the
below.
etymology of the word climate (jkila = inclination, i.e. the
The appearance of these large sheets of ice must have led to the height of the Sun above the horizon, in other words, the
annihilation of all organic life on the surface of the earth. The latitude of the location) or look for ‘climatology’ in an
soil of Europe, formerly ornamented with tropical vegetation
and inhabited by troops of great elephants, enormous hip-
academic organization chart to be convinced that this idea is
popotamuses and gigantic predators, was suddenly buried under still valid. The physical principles underlying the function-
a vast mantle of ice covering the plains, lakes, seas and plateaus. ing of the climate system were also updated in the nineteenth
The movement of a powerful creation was succeeded by the century, notably by Joseph Fourier, who established the laws
silence of death. The springs dried up, the rivers ceased to flow,
and the rays of the sun, rising on this frozen beach (that’s if they
of heat diffusion, explaining how heat is redistributed by the
surface fluids of the atmosphere and the ocean, and also
discussed the essential role of the greenhouse effect:
D. Paillard (&)
Laboratoire des Sciences du Climat et de l’Environnement,
LSCE/IPSL, CEA-CNRS-UVSQ, Université Paris-Saclay, 91190
Gif-sur-Yvette, France
e-mail: [email protected]

© Springer Nature Switzerland AG 2021 385

G. Ramstein et al. (eds.), Paleoclimatology, Frontiers in Earth Sciences,
https://doi.org/10.1007/978-3-030-24982-3_28
386 D. Paillard

The temperature can be increased by the interposition of the does not change. How then could this have any effect on
atmosphere, because the heat finds less obstacle to penetrate the climate? According to Adhemar, although the astronomical
air, being in the state of light, than it finds it to pass through the
air when converted into dark heat. forcing is effectively anti-symmetric with respect to the
(J. Fourier, 1824) seasons and zero for the annual average, the climate pro-
cesses are probably not.
It is in this context that the two main physical theories are In 1864, James Croll clarified this concept. According to
presented, which are still relevant today, and which make it him, the accumulation of snow occurs chiefly in winter, with
possible to explain the existence of glacial periods: the melting occurring in summer. Croll emphasized the role of the
astronomical theory and the variations in the atmospheric winter accumulation which essentially supports Adhemar’s
concentration of CO2. argument: longer or colder winters favor a greater accumula-
tion of ice allows the initiation of a glaciation. In addition,
aware of the progress that had been made in celestial
From Adhémar to Milankovitch: The Role mechanics notably by Pierre Simon de Laplace and Urbain le
of Insolation Verrier, Croll went on to introduce the effect of variations in
the eccentricity of the Earth’s orbit. These variations modulate
Although the notion that the climate is influenced by the
the intensity of seasonal contrasts. Indeed, in the case of a
stars has undoubtedly been around for a very long time, the circular orbit, the effect of the precession on climate is zero,
first astronomical scientific theory of the ice ages was for-
since there would be neither a perihelion (nearest point) nor an
mulated by Joseph Adhemar in 1842. It was based simply on
aphelion (farthest point). The greater the eccentricity, the
common sense: since ancient times, astronomers have greater the climate effects of the precession. Croll therefore
highlighted the ‘three movements of the Earth’: the diurnal
linked the great glaciations with eccentricity maxima.
cycle, the annual cycle, and the precession of the equinoxes
Hence, he pushed back the estimate of the last glaciation to
(Hipparchus, about 130 before J.-C.). While it is clear that 80,000 years ago, and suggested an even more intense
the annual and diurnal cycles generate temperature varia-
glaciation 240,000 years ago. Although Croll proposed a
tions, the same must be true of the third movement of the
much more solid and evolved astronomical theory, he did not
Earth. As will be explained a little later, the precession of succeed in convincing the scientific community of his time.
equinoxes has the consequence of modifying the position of
However, the interglacial-glacial alternations discovered in
the perihelion (point of the Earth’s orbit closest to the Sun)
some sediments argued in favor of a more or less periodic
in relation to the seasons: today, the Earth is closest to the mechanism. Yet the first dating of elements available from that
Sun around January 4 but this date changes slowly to cover
time, extrapolating the rates of erosion or counting lake varves,
the whole of the year over about 21,000 years. In contrast to
indicated a much more recent glaciation. Croll went on to
today, 10,500 years ago, the Earth was far from the Sun in introduce a third important astronomical parameter for the
January and close to it in July. Adhemar suggested that this
calculation of the variations of the solar energy received in a
mechanism could modify the climate. More specifically, the
given place: the obliquity of the terrestrial axis, that is to say its
winters of the northern hemisphere now occur when the inclination with respect to the plane of the Earth’s orbit.
Earth is close to the Sun and, conversely, those of the
However, obliquity has little effect on winter insolation, and
southern hemisphere when the Earth is far from the Sun.
therefore was of little importance in Croll’s theory.
Adhemar proposed that this explains the absence of a large Milankovitch (1941) formulated the astronomical theory
ice cap in the north, due to milder and shorter winters, and
which still applies today. The main criticism that can be lev-
conversely, the presence of a large Antarctic cap. The situ-
elled at Croll’s theory is that it considered winter to be the most
ation would have been exactly the opposite 10,500 years pertinent season. This objection was already expressed by
ago, which allowed him to explain the periods of great
Joseph Murphy during Croll’s time, as observations of the
glacial expansion that had just been revealed by geologists.
eternal snow and mountain glaciers showed that summer
Adhemar’s theory was criticized for many reasons, some
melting had much more impact on the ice mass balance than
largely unfounded, but it was on the very foundations of his
snow accumulation. However, the precession was often
theory that his detractors, Charles Lyell and Alexander von
advanced to explain the current asymmetry of temperatures
Humbolt, would find compelling arguments. In fact, the
between the northern hemisphere and the southern hemi-
mechanism of precession works in an anti-symmetrical way
sphere, and the presence of Antarctica, since today, the austral
between the poles, but also between the seasons. It can easily
winter is longer and further from the Sun than the northern
be shown that although, for example, less energy is received
winter. Croll’s theory was therefore probably based on this
in winter, this is compensated for by an equivalent excess of
nineteenth century misconception of the current climate.
energy received in summer. If the seasonal contrast varies
Milankovitch resolved this problem by deducing that the
with the precession, the full complement of energy received
current north-south asymmetry was linked to geography and
28 Climate and Astronomical Cycles 387

not to astronomical forcing. The critical season for the close to the most recent estimates of about 3.5 °C, another
evolution of ice caps is therefore summer, which completely prediction likely to become true in the not too distant future.
reverses the reasoning of Adhemar and Croll. The obliquity All these arguments were underscored by the American
of the Earth’s axis, that is to say its inclination with respect geologist Chamberlin, who highlighted the succession of at
to the plane of Earth’s orbit, becomes the most important least five glacial stages in the United States. According to
parameter for glacial-interglacial evolution. The foundations Chamberlin, there was a sort of oscillation between the cli-
of modern astronomical theory were thus laid down. mate and the geochemistry of the Earth: a decrease in CO2
leading to cooling, with the effect of reducing carbon sinks
on Earth by reducing the burial of organic matter as well as
From Tyndall to Arrhénius: The Role of Carbon the erosion of silicates. This would then lead to a gradual
Dioxide increase in atmospheric CO2 until it switches over to the
opposite situation. Chamberlin therefore attempted to for-
The importance of the role of the greenhouse effect was well mulate an ‘internal oscillation’ to explain the succession of
understood by the nineteenth century, notably through the ice ages, without calling on a ‘Deus ex machina’ such as the
work of Joseph Fourier. As early as 1845, Jacques Joseph astronomical forcing.
Ebelmen, a French chemist, first suggested that changes in It is interesting to note that these two opposing theories of
the atmospheric concentration of CO2 might have conse- the glacial periods have existed since the middle of the
quences for the climate (Bard 2004). Indeed, by focusing on nineteenth century and are still largely valid today: it
the chemistry of minerals, he established the bases of carbon remains to be understood how they relate and complement
geochemistry: a source primarily of volcanic origin and each other.
sinks related to the erosion of silicates and the burial of the
organic material. Since all these processes appear to be
disconnected, it seems unlikely that the atmospheric con- Astronomical Parameters and Insolation
centration of CO2 would be constant over geological time. In
1861, John Tyndall further authenticated the theory of the Before proceeding further, it is useful to review the various
greenhouse effect. By measuring the absorption and infrared astronomical parameters that influence the energy received at
emission of the various gases present in the air, he demon- the top of the Earth’s atmosphere, which we call ‘insolation’.
strated that nitrogen or oxygen are essentially transparent to
infrared rays and that the greenhouse effect of our planet is
caused primarily by gases in very small quantities, in large Eccentricity
part by water vapor, but also carbon dioxide, methane,
nitrous oxide and ozone. Tyndall then suggested that all of According to Kepler’s first law, the Earth’s orbit is an
the climate changes discovered by geologists, including ice ellipse. This is characterized by a major parameter, the
ages, could be explained by changes in the levels of atmo- semi-major axis, often denoted a, by a shape or flattening
spheric greenhouse gases. parameter, the eccentricity, often denoted e, and also by
But it was the Swedish chemist Svante Arrhenius (1896) three parameters defining the position of this ellipse in space,
who managed to calculate the effect of carbon dioxide on the two of which define the orbital plane (the inclination i with
climate, in an attempt to explain the ice ages. Based on respect to a reference plane and the longitude of the
geological data on moraine positions during glacial periods, ascending node X defined by the intersection of these two
he estimated a cooling of 4 or 5 °C and calculated that this planes), and another to define the absolute position of the
could be explained by a reduction of about 40% in the perihelion (the longitude p). In fact, as soon as the system is
atmospheric concentration of CO2. The measurement of made up of three material bodies (the Sun with two planets)
pCO2 from this glacial period, carried out on air bubbles or more, the movement is no longer strictly an ellipse, and
from ice cores taken from Antarctica in the years 1980– there is no analytical solution to the problem of celestial
1990, confirmed his calculations: the atmospheric concen- mechanics at N bodies, for N > 2. It is therefore appropriate
tration of CO2 was indeed 30% lower during the ice age. to calculate the perturbations or the approximate numerical
This scientific prediction, nearly a century before it could be resolutions. The notion of terrestrial orbit nevertheless still
confirmed through observation, is a good illustration of the makes sense because the perturbations are secondary. It is
essential role of greenhouse gases in the deployment of the therefore useful to reason in terms of elliptical orbit, which
Quaternary cycles. Arrhenius also considered that future deforms and moves over time.
global warming would be linked to anthropogenic CO2 The perturbations induced by the other planets do not
emissions. He calculated an increase in global temperatures modify the semi-major axis of the ellipse a, only the ter-
of about 5 °C for a doubling of CO2, a figure surprisingly restrial trajectory, i.e. the eccentricity e and the orientation
388 D. Paillard

Earth These tiny variations have almost no effect on the climate.

They are of the same order of magnitude as the observed
v
variations in solar flux S over the 11-year cycle, which
r generate temperature variations of the order of 0.1 °C. On
r’
Fo the other hand, as we will see a little later, it is through the
cu
2 s Focus 1 modulation of the effects of precession that eccentricity plays
Aphelion a Perihelion
c (Sun) an essential role in climate.
It is relevant to note that the solar system is chaotic. This
means that the calculation of the orbital parameters in gen-
eral, and in particular that of eccentricity, is only possible for
b instants not too far from the current period. In fact, errors
increase exponentially with time, and it is not possible to
Fig. 28.1 An ellipse can be defined as the locus of the points whose arrive at an estimation beyond a certain time. For the
sum of distances to the two foci is constant: r + r = 2a. The eccentricity, this timeframe is only around 20–30 million
eccentricity is defined as the ratio e = c/a. The semi-minor axis b is years, which is very short compared to the age of the Earth.
therefore given by the theorem of Pythagoras: b = a(1 − e2)1/2
Beyond that, although variations in eccentricity remain
similar in nature (with identical periodicities of approxi-
parameters of the ellipse (i, X, p). So a will be a constant (at mately 100,000 and 400,000 years), it becomes impossible
least over hundreds of millions of years). Moreover, the to say whether the eccentricity was minimal (close to zero)
orientation of the ellipse in space does not have a direct or maximum (close to 0.06) 600 million years ago. In other
consequence on the solar radiation received by the Earth. words, the phase of the oscillations becomes theoretically
The only orbital parameter which is liable to modify the unreliable over the long term (Laskar et al. 2004).
insolation is therefore the eccentricity e.
The eccentricity e is defined by the ratio between the
distance from the focus to the center of the ellipse c, and the Obliquity
semi-major axis a, as shown in Fig. 28.1. Although today it
is 0.0167 (i.e. a flattening of 1.67%), it has varied between In addition to the parameters of the Earth’s orbit, the position
values of almost-zero and 0.06, with periodicities appearing of the rotational axis of the Earth relative to the orbital and
to be around 100,000 and 400,000 years. ecliptic planes must also be taken into account. This position
Eccentricity is the only parameter capable of modifying is given by two axial parameters; the obliquity, denoted e,
the average annual energy received by the Earth. As the representing the inclination of this axis relative to the
semi-major axis is constant, the average distance between the ecliptic; and the precession of the equinoxes, which indicates
Earth and the Sun depends on its eccentricity. The second law its absolute position compared to the stars. The position of
of Kepler (conservation of angular momentum) is written as: the Earth’s axis is modified by the differential attraction of
pffiffiffiffiffiffiffiffiffiffiffiffiffi the Moon (and, to a lesser extent, the Sun) at the equatorial
2 dv 2pa2 1  e2 bulge. In fact, our planet is slightly flattened, because of the
r ¼
dt T Earth’s rotation, and the gravitational pull of the Moon
towards the Earth is therefore not exactly symmetrical. The
The ‘solar constant’ S0, defined as the average energy
equatorial bulge, at an incline relative to the lunar orbit, is
received by the Earth, is deduced by integrating it into a
subject to attraction forces which create a torque on the
complete orbit:
Earth’s axis and modify its orientation. Contrary to orbital
ZT Z2p parameters, such as eccentricity, which depend only on point
S a2 dv S mechanics, the axial parameters (obliquity and precession)
S0 ¼ dt ¼ S pffiffiffiffiffiffiffiffiffiffiffiffiffi ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffi
T r2 2p 1  e 2 1  e2 depend on the shape of the Earth, which introduces new
0 0
sources of error and uncertainty. The timeframe beyond
The average annual energy received by the Earth, S0, which the calculation of the axial parameters becomes
depends on the energy S (assumed here to be truly constant) impossible is therefore probably shorter than for the eccen-
received at the distance a from the Sun and on the eccen- tricity. The calculations become more complicated further
tricity e. A higher eccentricity leads to a more flattened orbit, back than a few million years if the shape of the Earth
on average closer to the Sun, and thus to greater overall solar changes slightly under the influence of glaciations due to the
energy being received by the Earth. However, the variations enormous volumes of ice accumulating on the continents of
remain very small, since for e = 0.06, the maximum is cal- the northern hemisphere at the glacial maxima. It has been
culated as S0 = 1.0018 S, i.e. an increase of only 0.18%. suggested that this could have consequences for the
28 Climate and Astronomical Cycles 389

calculation of axial parameters further back than only a few Moreover, the phenomenon of the seasons is directly
million years. Similarly, internal convection in the Earth’s linked to the obliquity, and it is all the more marked when
mantle potentially induces changes in the distribution of the obliquity is large. For example, for a circular orbit
masses which are difficult to take into account in these (e = 0), we obtain the following expression of daily inso-
astronomical calculations. The phase of the obliquity, like lation at the poles during solstices:
that of the precession, is therefore subject to caution when
extrapolating calculations for the distant past or future Wsummer ðpoleÞ ¼ Ssine; Wwinter ðpoleÞ ¼ 0:
beyond about ten million years. Moreover, the periodicity of If insolation at the winter solstice remains zero, the
these axial parameters depends on the Earth-Moon distance, summer solstice will vary between Wsummer(pole) = 0.373
but the lunar recession is rather poorly constrained in the S and Wsummer(pole) = 0.415 S, when the Quaternary e goes
distant geological past, and the frequencies of the obliquity from 21.9° to 24.5°, i.e. an increase of around 4%, that is to
and precession also become uncertain. say more than 50 W/m2. This is far from negligible.
The obliquity today is 23° 27′, which defines the latitude In addition, it is essential to notice that, contrary to the
of the polar circles (67° 33′ north and south) and the tropics precession which we detail below, the effect of the obliquity on
(23° 27′ north and south). This value oscillates between the insolation is symmetrical relative to the equator. This is an
extremes of around 21.9° and 24.5°, with a periodicity of important aspect of Milankovitch’s theory: contrary to the
around 41,000 years. It is clear that any change in obliquity theories of Croll and Adhemar, which involve winter insola-
will have consequences for the climate by altering the size of tion (mainly dependent on the precession), Milankovitch’s
the polar and tropical areas. Thus, on the island of Taiwan, theory is based on summer insolation. which strongly depends
in the county of Chiayi (Jia-Yi), for almost a century, there on the obliquity. On a planet with symmetrical topography,
has been a monument marking the Tropic of Cancer. this would imply the presence of ice caps oscillating largely in
However, the current decline in obliquity, at a rate of 0.46 phase in both hemispheres. Of course, the distribution of
arc-seconds per year, means that there has been a displace- continents on Earth is not symmetrical and several other fac-
ment of the tropics of 14.4 m per year, i.e. 4 cm per day and tors will also affect how climates are distributed on Earth.
thus more than 1 km since the monument was first erected.
The Taiwanese have therefore regularly built new monu-
ments to follow the southward movement of the Tropic of Precession of the Equinoxes and Climate
Cancer. Precession
Although the average global incident radiation has not
changed, its geographical distribution depends on the In addition to the alternation of day and night and the phe-
obliquity. To be precise, if one calculates the average annual nomenon of seasons which have been known since the dawn
solar incident radiation, it is found that this depends mainly of time, astronomers have observed since ancient times a
on the obliquity and, to a lesser extent, on the eccentricity as slow drift of the polar axis relative to the celestial sphere.
mentioned above. An increase in obliquity e results in an This discovery is generally attributed to Hipparchus
increase in insolation at high latitudes and a decrease in the (130 B.C.) who estimated the drift at approximately 1° per
tropics. At the poles (north and south) and at the equator, century (that is to say, a periodicity of about 360 centuries).
incident is calculated as follows: This estimate is remarkable since we now know that the
S precession does have a periodicity of 25,765 years (1.397°
WYear ðpoleÞ ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffi sin e; per century). It is also likely that the Egyptians and the
p 1  e2
2S Mesopotamian astronomers were already well aware of this
WYear ðequatorÞ ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffi Eðsin eÞ phenomenon because, centuries earlier, they identified the
p 1  e2
2
position of the celestial pole as well as the constellations of
where E(x) = E(p/2, x) is the secondary complete elliptic the zodiac. changed the orientation of certain temples to
integral. For a eccentricity e of zero, when e goes from 21.9° ‘follow’ this movement. As Nicolas Copernic already noted,
to 24.5°, changes of around 1%, i.e. 18 W/m2 are obtained this is the ‘third movement’ of the Earth, the first two cor-
for WYear (pole) and changes of around 0.4%, i.e. 5 W/m2 responding to the day and year. It was therefore logical for
are obtained for WYear (equator). It should be noted that for Adhémar to focus on the consequences for the climate of this
very large obliquities (for p sin e > 2 E(sin e), in other ‘third movement’.
words, for e > ec = 53,896°), the poles receive a higher Nevertheless, a distinction should be made between the
annual energy average than the equator. This is currently the precession of the equinoxes and the climate precession.
case on Uranus and Pluto. Indeed, the ‘absolute’ position of the axis of the Earth,
390 D. Paillard

Cassiopea
Cepheus

P Precession motion
Cygnus

00
P0

0
+2
0
Ursa
minor -200
0
Lyra
-4
00
0
Draco

Herculus Ursa major

P0
P Precession motion

VIRGO LEO

axis

= 0)
LIBRA CANCRI

Earth

is (t
h ax
Eart
SCORPIO
γ'
GEMINI
Apparent motion of the Sun
(annual cycle) 0)
2 00
S T e (+
lin
Or xes
b it no
al p ui
Eq

n e ( t = 0)
la n
e (e
c l ipt TAURUS
SAGITTARIUS ic )
γ γ0
ori pla
al
Equat
E qu
atoria γ (t = -4000)
CAPRICORN l p la n e ( t ARIES
= +2000)

γ (t = +2000) AQUARIUS PISCES γ (t = -2000 from today)

γ0 (t = 0, today or 2000 AD)

Fig. 28.2 The precession of the equinoxes corresponds to the equinox line cc′ (intersection of the orbital plane and the equatorial
rotational motion of the Earth’s axis (T). The pole thus moves relative plane). The current point c0 is in the Pisces constellation and is drifting
to the stars (from P0 today, pointing towards the polar star, to P in the towards the constellation of Aquarius
future in the constellation of Cepheus). The same applies to the

relative to the stars, has, in itself, no effect on the climate. Pisces) certainly has close symbolic links, via astrology,
However, when this axis is oriented differently, so too is the with the history of religions (the pre-Hebraic bull, the ram or
equatorial plane of the Earth. The equinox line is defined as the lamb for the Jews, and the fish for the Christians). The
the intersection of the equatorial plane and the ecliptic (or- vernal point, and therefore the position of the seasons on the
bital) plane. The precession of the equinoxes (from which Earth’s orbit, moves with the precession of the equinoxes.
the origin of the name) corresponds to a drift of the equi- Knowing that the orbit is elliptical makes it easy to under-
noxes relative to the constellations, as shown in Fig. 28.2. stand that the seasons will be situated at different distances
Thus, while the Sun is now in the constellation of Pisces from the Sun, depending on whether the vernal point is
on the day of the spring equinox (thus defining what closer or further from the perihelion or the aphelion. The
astronomers call the vernal point c), it was in Aries of the effect of the precession on climate is therefore measured by
time of the Greeks and in Taurus at the time of the Egyp- the relative position of the vernal point and the perihelion.
tians. The symbol used (c) since antiquity comes from the The latter also moves, as mentioned in the paragraph con-
zodiacal symbol of the ram. This succession (Taurus, Aries, cerning eccentricity (this is the orbital parameter p)
28 Climate and Astronomical Cycles 391

according to a movement called ‘precession of the perihe- λ

lion’. The combination of the precession of the equinoxes March equinox
and the precession of the perihelion thus makes it possible to Earth
define the relative position of the seasons and the principal
axes of the ellipse. The climate precession, denoted by ῶ
v
(‘curvilinear pi’), is defined as the angle between the vernal r
point and the perihelion. If the vernal point carries out a December solstice
complete cycle in approximately 25,700 years, the perihe-
lion does the same in about 112,000 years. As these two
a c Perihelion
movements occur in opposite directions, we can deduce an
average periodicity of 21,000 years for climate precession June
(1/25.7 + 1/112 * 1/21). solstice
However, there is a small additional complication. When
ϖ
the orbit is circular (e = 0), there is no longer a perihelion.
The angle ῶ is then not defined. Moreover, it is clear that the
γ September equinox

effect of changes in precession ῶ on the climate will be Fig. 28.3 Definitions of the longitude k, of the climate precession ῶ,
greater as the eccentricity increases, since the distance and the anomaly v, with respect to the seasons, the perihelion and the
between the Earth and the Sun will be greater between its vernal point c
maximum a(1 + e) (aphelion) and its minimum a(1 − e)
(perihelion).
This effect will be zero when e = 0. For all these reasons, summer solstice, or k = 270° at the winter solstice), as shown
it is appropriate to introduce the ‘climate precession in Fig. 28.3. It is then assumed that this longitude k is ‘fixed’
parameter’ e sin ῶ, which cancels out when - is not defined during the day, as the astronomical parameters are. The only
(for e = 0) and which increases with e. In fact, it is mathe- movement that is taken into account and which is averaged is
matically useful to replace the pair of parameters (e, ῶ), therefore the rotation of the Earth on itself in a day.
defined only if e is not zero, with the pair (e cos ῶ, e sin ῶ), By formulating:
which is always well defined, in other words, a
s ¼ Maxð0; 1  sin2 /  sin2 dÞ ¼ Maxð0 ; 1  sin2 /
polar-Cartesian coordinate change. The effect of the pre-
cession is thus modulated by the eccentricity, as can be seen  sin2 esin2 kÞ
in the following insolation formula. This results in a dupli- p ¼ sin / sin d ¼ sin / sin e sin k
cation of frequencies (more precisely, a multiplication, since
e has itself multiple periodicities). If e varies with a single The following expression of daily insolation can be used:
periodicity of 100 000 years, as for example the function |e0 if s ¼ p ¼ 0; WD ¼ 0
cos(t/200)|, and ῶ has a cycle of 21 000 years, we can pffiffiffiffiffiffiffiffiffiffiffiffi 
deduce:  otherwise s þ p2 6¼ 0 ;
0   1
~ ¼ je0 cosðt=200Þjsinðt=21Þ p pffiffi
e sin x  2 p arccos pffiffiffiffiffiffiffiffi2ffi þ s
1  e cosðk  x~Þ B B
sþp C
C
WD ¼ S 2 @ A
hence the periodicities of 19,000 and 23,000 years (1/ 1e p
21 + 1/200 * 1/19 and 1/21 − 1/200 * 1/23), which have
been detected in oceanic paleoclimate records and form a
strong argument in favor of Milankovitch’s theory. In this formulation, three factors can be identified:

1. The solar constant (S);

Calculations of Insolation, Calendar Problems 2. A term for the Earth-Sun distance that depends on the time
of year (k), climate precession (ῶ) and eccentricity (e);
Knowing the three astronomical parameters e, e, ῶ, it is easy 3. A geometric term that depends only on the time of year
(with the application of some trigonometry) to determine the (k), the obliquity (e) and the latitude of the location (/).
radiation received by the Earth, or insolation, for each loca-
tion (latitude /) and for each season. It is common practice to In particular, the geometrical term remains unchanged if
use the daily insolation, by giving an orbital position with / is changed to −/ and k is changed to k + p simultane-
respect to the spring equinox (i.e. the moment in the year) ously (if the hemisphere and half-year are changed). Simi-
identified by a longitude k (for example, k = 90° at the larly, the distance term remains unchanged if ῶ is changed to
392 D. Paillard

ῶ + x and k is changed to k + x simultaneously (if the results in a significant lag with the astronomical seasons, up
precession and the moment of the year are shifted by the to about two weeks, especially in September, the month
same quantity). We note also that the integral over the year furthest from the reference point. The alternative is to rely on
(between k = 0 and 2p) of the insolation does not depend on the real astronomical seasons which do not have the same
the precession ῶ. number of days, leading to diagnoses more complicated to
In these formulas, it is important to note that the longitude implement (Joussaume and Braconnot 1997).
k is the angle that represents the position of the Earth in its
seasonal cycle, but that k is not quite proportional to the time
that elapses during the year. In fact, when the Earth is close Which Astronomical Forcing Should Be Applied
to the perihelion, its velocity is greater and k changes more to the Climate?
rapidly than when the Earth is at the aphelion.
To be precise, the equation of the ellipse (in polar coor- In addition to the definition of the seasons, another critical
dinates, centered on the Sun) is: issue is the clarification of the concept of ‘summer’ insola-
tion, which is used as a forcing term for Milankovitch’s
að1  e2 Þ theory of the evolution of ice cover in the northern hemi-
r ¼
1 þ e cos v sphere. Should the value of this insolation be taken on a
given day (for example, the summer solstice)? Or should an
where v is the position relative to the perihelion
average for the whole season be taken? Or possibly over
(v = k − ῶ + p).
 pffiffiffiffiffiffiffiffi another orbital interval [k1, k2]? Or over a time interval? Or
2pa2 1e2
The second Kepler equation r 2 dv
dt ¼ T can be should every day of all the seasons be taken and applied to
written: an explicit coupled climate-ice cap physical model? While
pffiffiffiffiffiffiffiffiffiffiffiffiffi this latter solution is the most relevant one, it is, in practice,
2
 
2 2Z dv 2pa2 1  e2 Z difficult to implement and it is still useful to understand the
a 1e ¼ ¼ dt
1 þ ðe cos vÞ2 T bases behind the theory by formulating simpler versions.
Milankovitch calculated an average of the isolation for
This fits into: half the year, centered on the June solstice which he called
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ‘calorific insolation’. The usual practice since the 1970s is to
2p E 1e v
E  e sin E ¼ t with tan ¼ tan choose a given day, often the summer solstice. The typical
T 2 1þe 2 forcing used is thus the daily insolation at 65°N on the June
which is called the ‘Kepler equation’, where the new angle solstice. Nevertheless, insolations averaged over durations
E is called the eccentric anomaly. With these equations, the greater than one day may be useful.
time t2 − t1 necessary to pass from orbital position k1 to k2 is For example, it was suggested (Huybers 2006) that a much
deduced. more relevant astronomical forcing would be an integral of
The duration of the seasons thus changes with the climate insolation above a critical threshold, since this forcing will
precession ῶ. This poses a problem to define the calendar. In melt or not, the ice or snow cover, and corresponding to
particular, our Gregorian calendar is partly adjusted for temperatures above or below the zero degree Celsius thresh-
shorter winters and longer summers (CDD (Cooling Degree old. An insolation integral above a threshold is a reasonably
Days): 90.25 days, MAM: 92 days, JJA: 92 days, SON: good fit with what glaciologists use as a climate forcing, i.e. a
91 days), partially in line with the true duration of the sea- temperature integral called Positive Degree Days (PDD).
sons (winter: 89.0 days, spring: 92.8 days, summer: In practice, the longer the integration period chosen, the
93.6 days, autumn: 89.8 days). While it is important for more important the role of obliquity in the result. If one
paleoclimate data to be based on the astronomical calendar integrates over the whole year, precession is eliminated,
(and therefore on seasons defined by the solstices and leaving only the obliquity. In any case, it is essential to be
equinoxes), models need above all a temporal seasonal axis aware of the non-uniform motion of the Earth in its orbit.
and have to take into account variable durations for the Indeed, the integration of insolation WD must be done
seasons. Most often, these are defined by a fixed time according to the time variable, even if the result concerns an
interval (either a quarter of a year or based on the current orbital interval [k1, k2]. The calculation leads to elliptic
schedule) using the March equinox as a reference point. This integrals that can then easily be evaluated numerically.
28 Climate and Astronomical Cycles 393

The Successes and Difficulties is astronomy that drives the Quaternary climate cycles.
of Milankovitch’s Theory Nevertheless, this paper also highlights the main problem.
The 100,000 year cycle is the dominant cycle, but according
From Hypothesis to Evidence to the theory, it should hardly appear. It seems that there is a
link between the dominant cycle of 100,000 years and
Milankovitch’s theory was not accepted by the majority of eccentricity, but the causes are unknown. In other words, this
geologists for a long time. In fact, stratigraphic studies by ‘historical’ paper demonstrates both that the astronomical
Penk and Brückner had made it possible to define only four theory is necessary, but that, alone, it does not explain the
successive glacial episodes in the Alps (Günz-Mindel-- observations.
Riss-Würm), and not a succession of regular events. It was
not until the middle of the twentieth century that thanks to
studies on marine sediments that the number of glacial A Quasi-linear System for Precession
cycles identified increased considerably (see Chap. 20, vol- and Obliquity
ume 1). In particular, in the 1950s, C. Emiliani carried out
the first isotopic measurements on marine carbonates, Although the astronomical periodicities are indeed present in
showing more than a dozen glacial-interglacial successions paleoclimate records, a simple relation connecting the two
with a clear cyclicity, which put an end to the traditional has yet to be verified. Various analytical techniques have
denomination (Günz-Mindel-Riss-Würm). From then on, been used to this end. It is thus possible to assess the con-
astronomical theory became accepted. sistency, in other words, the correlation in the spectral
Emiliani thus defined the ‘isotopic stages’ which are still domain (or frequency domain) between the astronomical
used today to designate glacial and interglacial periods, odd forcing and the paleoclimatic, geochemical and micropale-
numbers for interglacial periods and even numbers for gla- ontological data. The results are quite significant for the
cial periods. It is interesting to note that Stage 3 appears to periodicities of 23,000 and 41,000 years, i.e. the variations
be an exception, since it is now unanimously considered to of obliquity and precession. There is therefore a close link
be part of the last glacial period (which includes Stages 4, 3 between astronomical changes and climate for these two
and 2). This apparent inconsistency stems from the simple frequencies, which can be interpreted in terms of a
fact that astronomical theory predicts a dominant periodicity ‘quasi-linear’ model. Another way to be sure of this is to
associated with variations in obliquity, i.e. cycles of observe the amplitude modulation of the astronomical
41,000 years. forcing and that of the climate for these periodicities. For
The very rare chronological information available from example, for obliquity, we note that the greater the amplitude
this time suggests a stage 3 around 30–50 ka BP, preceded of the variation in the obliquity, the greater the climate
by many other older cycles of greater amplitude, but with response around the corresponding periodicities (i.e.
uncertain dating. It was therefore logical to begin the num- 41,000 years), as illustrated in Fig. 28.4.
bering of past interglacials from stage 3, in accordance with This close relationship between astronomical forcing and
the idea of a dominant cyclicity linked to the obliquity. It ‘climate’ thus seems sufficiently well established to be used
was not until the 1960s and 1970s, thanks to the Pa–Th
datings, that the main cyclicity appeared to be around
100,000 years, an observation which seemed strange in the
δ18O filtered (‰)

context of astronomical theory (Broecker and van Donk 0.4

1970). This famous ‘problem of the 100,000 years’ is still a 0.2

0
major obstacle, as will be explained later. -0.2
Advances in dating techniques using radioisotopes have -0.4
25
resulted in a gradual refinement of the chronology, in par-
Obliquity (°)

ticular through the use of coral reefs and magnetic reversals 24

that can be identified in both marine sediments and volcanic 23

flows. This enabled a more precise time frame to be pro-
22
posed and the paleoclimate periodicities to be precisely
0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000
defined. The paper by Hays et al. (1976) identified cycles of
Age (ka BP)
23,000 years, 41,000 years, and 100,000 years, which cor-
respond well to astronomical frequencies (Berger 1978). Fig. 28.4 Isotope recording (ODP 659, Tiedemann et al. 1994) filtered
This demonstrated unambiguously the astronomical imprint around 41 ka (above) and variations in obliquity (Laskar et al. 2004)
on the climate and the value of Milankovitch’s theory that it (below). It is clear that when the obliquity variations increase, so too do
the associated climate variations
394 D. Paillard

to date past recordings. Indeed, a persistent difficulty in -3

paleoclimate reconstructions is the establishment of a reli- -2

Ice volume (V)

able chronology. As such, absolute dating obtained from the -1
radioactive decay of certain isotopes (14C, 40K/39Ar, 0
40
Ar/39Ar, U/Th) is often rare and difficult to obtain. 1
Moreover, they are imprecise. For example, a 1% error in the 2
measured age translates to an error of 10,000 years for 3
0 100 200 300 400 500 600 700 800 900
samples of one million years. The error only grows as one
Age (ka BP)
goes back in time. It is therefore extremely tempting to use
Insolation (21st june 65°N) W/m2
astronomical theory to identify the cycles measured in a
420 440 460 480 500 520 540 560 580 600
paleoclimate recording by referring to the cycles of the 0.3
astronomical forcing. This link seems particularly relevant 0.2
for precession or obliquity, at least for the Quaternary. In the

Ice volume variations (dV/dt)

0.1 dV/dt = -0.0041 inso + 2.0257
case where correct identification of all the cycles is possible, 0
which turns out to be quite frequently, this methodology has -0.1
the enormous advantage of having an associated error which -0.2
does not increase disproportionately with time. This is -0.3
because if one does not ‘skip’ a cycle, then only the phase -0.4
-0.5
relationship between the forcing and climate is cast into
-0.6
doubt, which limits the error to a few thousand years, even if
-0.7
we go back tens of millions of years into the geological past.
-0.8
In this case, isotopic stratigraphy shows how extraordinarily
efficient it is, and the establishment of an astronomical Fig. 28.5 On top: SPECMAP isotopic recording (Imbrie et al. 1984)
chronology for a substantial part of the Earth’s history is (normalized) interpreted like a record of ice cap volume. Bottom, this
volume of ice is derived according to the time (taking the difference
currently underway. However, the chaotic nature of celestial between two successive points), then the derivative is represented as a
mechanics mentioned above imposes limits on the calcula- function of summer insolation at 65°N. Good correlation between the two,
tion of certain astronomical parameters. Conversely, it is for the majority of points, is interpreted as a proof of Milankovitch’s
possible that we may be able to place geological constraints theory. The points which deviate substantially from this correlation are
marked by symbols which are also shown in the top figure. It is these
on the evolution of the parameters of the solar system (Pälike specific points that are the terminations
et al. 2004).

The Difficulty of the 100,000-Year Cycles Quaternary these have only existed for about a million years
and so we only have about ten of these cycles. As a result,
What is true for the 23,000 and 41,000 year cycles is not statistics have struggled to attribute a specific periodicity to
true for the major climate cycles that occur more or less them. It seems that the periodicity of 100,000 years is
every 100,000 years. The first two are highly asymmetric a merely an average between cycles with each having signif-
marked by a much shorter deglaciation phase than the icantly different durations (see Paillard 2001, Table 1).
100,000-year cycle, and as was already noted in the 1970s, Some authors even suggest that these so-called ‘100,000-
the latter did not fit very well into the framework of asro- year cycles’ are really a double or triple obliquity period (i.e.
nomical theory (Broecker and van Donk 1970). These 2  41 = 82 ka, or 3  41 = 123 ka) (Huybers and Wun-
deglaciations are consequently called ‘terminations’. This is sch 2005).
not only a visual impression, and it is possible to define these To put it simply, as noted above, variations of eccentricity
mathematically, by observing that they all correspond sys- only have a negligible role on the energy received by the
tematically to an accelerated decrease in the volume of the Earth, so, it is necessary in any case to imagine relatively
ice caps, as illustrated in Fig. 28.5. As such, the terminations complex processes to achieve a climate response in this
are therefore, from the start, outside the scope of Milanko- frequency band where the forcing is almost non-existent.
vitch’s theory. The simplest way of doing this is to assume the existence
Moreover, we can no longer observe any real link (in of thresholds in the climate system. For example, the system
consistency or amplitude modulation) between the variations would not function in exactly the same way during the large
of eccentricity and the major climate cycles. The very notion terminations as during the rest of the cycle. This strategy can
of the 100,000-year cycle is problematic because during the be assessed by very simple conceptual models.
28 Climate and Astronomical Cycles 395

A Few Simple Models accumulation episodes is introduced, with a different k co-

efficient in the two cases, k = kF and kA depending on
Milankovitch’s accomplishment was to calculate the orbital whether i > i0 (melting) or not. In addition, V is imposed as a
parameters of the Earth and the associated insolation positive value. Although this model fails to closely repro-
occurring over hundreds of thousands of years. Neither a duce the observations, it nevertheless has some remarkable
model of evolution of the ice caps nor a climate model was characteristics. In particular, it correctly predicts the position
available at this time. He simply linked the minimums in the of the large terminations, where many other more sophisti-
astronomical forcing with episodes of glaciation. He cated models fail, see Fig. 28.6. We will return to this point.
explained that due to a very high probably inertia in the caps, This model is quite unstable and small changes in the
there would be a gap of several thousand years between the parameters (i0 or kA/kF) lead to significantly different results.
astronomical forcing and the evolution of the caps, even to This is easily understood, since the volume of ice is, ulti-
the point of ‘smoothing’ the astronomical forcing if the caps mately only the integral of the insolation. Small changes in
do not have the time to react. Indeed, the volume of the caps the threshold or the coefficient quickly lead to very different
is not directly related to the insolation, but rather it is the volumes.
variation in the volume of ice which is a function of inso- A much more robust model was formulated by Imbrie
lation. In other words, an extremely simple model of evo- and Imbrie in 1980 (see Paillard 2001):
lution of the caps consists of integrating the astronomical
forcing over time. As the important physical object is the dV=dt ¼ ði  VÞ=s
cap, it is essential to look at the dynamics of the cap, not where, in a way, the threshold i0 has been replaced by the
only the forcing. volume of ice V. This time, the insolation i is taken as nor-
Surprisingly, no scientist seems to have focused on this malized (by subtracting its average value and dividing it by
task, following either in the footsteps of Adhémar or Croll, the standard deviation) and so it is possible for the volume of
or those of Milankovitch. It was therefore a journalist ice to also be negative. As before, the coefficient, or time
(Calder 1974) who would publish a scientific article constant, s, will have two different values depending on
describing, for the first time, a model of ice cap evolution in whether the cap melts (-i-V < 0) or grows (-i-V > 0). This
the Quaternary, more than a century after the ice ages were model reproduces quite well the evolution of the cycles
observed. This model was both very simple and informative. linked to precession and obliquity, but fails to reproduce the
It is based on a simple integration of the forcing [see Paillard 100,000-year cycles. In particular, it produces a strong
(2001, 2010, 2015)]: 400,000 year cycle, clearly present in the eccentricity, but not
dV=dt ¼ kði  i0 Þ: in the paleoclimate data on the volume of the ice caps, see
Fig. 28.7. This model thus illustrates the ‘Stage 11 problem’.
where V is the volume of the cap and i(t) is the astronomical About 430,000 years ago, the eccentricity was very low and
forcing, i.e. summer insolation at the high latitudes of the therefore the variations in the precession parameter e sin ῶ
northern hemisphere. Above a fixed isolation value i0, the were minimal. This was reflected in very little variation in the
volume of ice V decreases proportionally to i-i0 and, below Imbrie model. Conversely, the paleoclimate data show that
this value, V increases. This model works relatively well this period corresponds to a major transition between a very
provided that an asymmetry between the melting and intense glacial stage (stage 12) and a very marked interglacial

Magnetic reversal

2 4 6 8 10 12 14 16 18 20 22
δ18O (‰)

-2.0

-1.0
(arbitrary units)

2 4 6 8 10 12 14 16 18 20
ice volume

0 100 200 300 400 500 600 700 800

Thousands of years before present

Fig. 28.6 Comparison between the Calder model (bottom) and the the Brunhes-Matuyama magnetic reversal, now fixed at around 772 ka
V28-238 core (top) (see Paillard 2015). The glacial-interglacial (good match between cycles indicated by the dotted lines). In hindsight,
transitions are predicted for the correct dates by the model, while the the prediction made by the Calder model was quite remarkable
isotopic data in 1974 were displaced in time, with a too-early date for
396 D. Paillard

Summer insolation at 65°N

Calder’s model

Imbrie’s model

Paillard’s model

LR04

-200 0 200 400 600 800 1000

Time (thousands of years before present)
Fig. 28.7 Comparison between the various simple models discussed, 1998 and marine isotopic data LR04 (Lisiecki and Raymo 2005). Note
over the last million years and the next 200,000 years. From top to that in Paillard’s threshold model, there are two possible solutions for
bottom: daily summer insolation (65°N, June solstice) (Laskar et al. the future climate cycles, depending on whether the threshold for entry
2004), results from the Calder model; Imbrie model; Paillard model into glaciation has already been crossed or not (see Paillard 2001)

(stage 11). How can small variations in insolation lead to the dV=dt ¼ ðVR  V Þ=sR  F=sF
greatest of transitions? A natural solution is to view the great This time, the volume of ice is ‘relaxed’ towards different
cycles at 100,000 years, not as a linear oscillation around an VR values: the ‘climate mode’ R is changed as a function of
equilibrium point whose amplitude is inevitably linked to the certain threshold overruns on the astronomical forcing i and
amplitude of the forcing, but rather as a relaxation oscillation on the volume of ice V. In particular, an essential point
between two different climate modes between which the emerging from the study of this model is that in order to
system can switch as soon as certain thresholds are crossed. predict deglaciations at the right position, they must be
This is what is proposed in the Paillard model (1998): linked to the glacial maxima: the switch between
28 Climate and Astronomical Cycles 397

glacial-interglacial regimes needs to be triggered by a Although it is relatively easy to apply Milankovitch’s theory
threshold overrun of the ice cap volume. This relationship is to most of the past cycles, considerable difficulties arise for
what allows Calder’s model to finally work ‘well’, because the deglaciations as is highlighted in Fig. 28.5. It is therefore
Calder is more successful at predicting glacial maxima than for these specific moments, when the astronomical theory
transitions. In other words, for still unknown physical rea- alone is insufficient, that other mechanisms need to be
sons, deglaciations are facilitated by the occurrence of a explored.
glacial maximum, which triggers a relaxation oscillation, In other words, glacial-interglacial changes are not lim-
propelling the system into the opposite state. ited to changes in the expansion of the ice caps that could
Although this astronomical forcing is at the root of the subsequently influence the rest of the climate system. They
climate variations in the Quaternary, the understanding of are, on the contrary, a combination of changes for the caps,
the climate mechanisms in play is limited. In Milankovich’s but also for the biogeochemical cycles and the climate as a
theory there is a very simple hypothesis which, in a way, whole. Milankovitch’s theory only accounts for part of this
equates the concept of ‘climate’ with the expansion of the ice reality, the other part most likely involves the carbon cycle
caps in the northern hemisphere. Milankovitch’s theory, coupled with climate variations.
strictly speaking, is not a climate theory, but rather, an ice Unfortunately, our understanding of the carbon cycle dur-
cap theory, a point emphasized by the conceptual models ing the Last Glacial Maximum is very patchy. To make a first
above. The action of insolation on the ice caps of the approximation, it is reasonable to consider the ocean + at-
northern hemisphere accounts for certain Quaternary phe- mosphere + terrestrial biosphere system as isolated, i.e. with
nomena, but does not explain all the observations, in par- no significant exchanges of geological carbon (via volcanoes
ticular the existence of the 100,000-year cycles, punctuated or rivers). The problem is therefore to reduce the atmospheric
by exceptional deglaciations (or terminations). This theory reservoir by about 200 GtC (billion tons of carbon) while
therefore needs to be further completed to make it a true increasing the others by the same amount. However, the ter-
climate theory. restrial biosphere was considerably reduced during the glacial
period (between 300 GtC and 700 GtC), making the problem
all the more difficult, since all this atmospheric and biospheric
Recent Advances carbon needs to be trapped in the ocean. Many hypotheses
have been put forward to try to explain this low level of pCO2
The Vital Role of Atmospheric CO2 during the Last Glacial Maximum, but no consensus has yet
emerged. A complex combination of multiple factors (physi-
Shortly after the discovery of the role of astronomical peri- cal and biogeochemical) is one possibility which would
odicities in the climate (Hays et al. 1976), analysis of air explain a glacial-interglacial difference of 100 ppm, but the
bubbles in Antarctic ice cores demonstrated that the last high level of similarity between the climate recordings around
glacial period was also characterized by a significantly lower the Antarctic and the pCO2 records argue for a relatively
atmospheric concentration of CO2, in line with the predic- simple mechanism which would link the Southern Ocean and
tions of Arrhenius. Since the work on the Vostok ice cores its climate with the atmospheric concentration of CO2.
(Petit et al. 1999) and Dôme C ice cores (Monnin et al.
2001), it is now well established that the glacial-interglacial
cycles also correspond to cycles in atmospheric greenhouse Towards a Consolidation of Astronomical
gas composition, and in particular of CO2, which varies and Geochemical Theories?
between about 280 ppm (cm3/m3 of air) during the inter-
glacial period and 180 ppm during the glacial period. These Nevertheless, very recent progress makes a forthcoming
measurements make it possible to demonstrate that the two solution possible. The conceptual models mentioned above
traditional theories, astronomical and geochemical, are not suggest viewing the ‘glacial’ and ‘interglacial’ states as
mutually exclusive, but that both are necessary. This was distinct states able to account for a relaxation oscillation
largely confirmed by numerous numerical experiments in the between two (or more) different modes of operation. In this
simulation of the glacial climate: in order to explain the context, it is interesting to mention the hypothesis of a gla-
paleoclimate observations, it is essential to take into account cial ocean with very cold and above all salty bottom waters.
the 30% decrease in partial CO2 pressure. In addition, during This hypothesis is largely supported by measurements of
the terminations and in particular during the last deglacia- interstitial fluids in marine sediment cores (Adkins et al.
tion, it is well established that the concentration of CO2 2002), which directly estimate the salinity of the ocean floor
increased several thousand years before the rise of the sea in the past. The glacial ocean was therefore likely to have
level associated with the melting of the ice caps, or, in other been profoundly different from the current ocean, with
words, the actual deglaciation, as illustrated in Fig. 28.8. strong stratification between the upper and lower halves of
398 D. Paillard

530

Insolation (W/m2)
520 insolation
510 (65°N, 21st june)
500
490 0

480

MWP 1A
-20

(meters below current level)

470
-40

Sea level
-60

280 Sea level

Atmospheric CO2 (ppm)

-80
CO2 (from corals)
260 Dome C -100

240 -120

220

200 -400

-410

δD (‰)
Antarctic -420
temperature
(Dome C) -430

-440

5 10 15 20 25
Age (kyr BP)

Fig. 28.8 The last deglaciation. From top to bottom: daily insolation present day), while the sea level remains close to its ice age level
(65°N summer solstice) (Laskar et al. 2004); sea level (Bard et al. (about −100 m), atmospheric CO2 has already increased by about
1996); atmospheric CO2 and temperature at Dome C in Antarctica 60 ppm above its ice age value, more than half of the full transition
(Monnin et al. 2001). At about 15 kyr BP (15,000 years before the

the water column. These highly saline bottom waters are In this model, terminations are explicitly induced by an
likely to form around Antarctica, due to the salt deposits that increase in atmospheric CO2, itself caused by the previous
occur during the formation of sea ice. Indeed, thanks to the glacial maximum, which destabilizes the stratification of the
continental shelf, these very salty waters are able to flow deep ocean. Since this model is based on a bi-modal system,
along the topography (Ohshima et al. 2013) and reach the it correctly reproduces certain characteristics already present
abyss of the Southern Ocean. The consequences for the in even simpler models (Paillard 1998), such as the possi-
carbon cycle of this ocean configuration are considerable. bility of switching between dominant periodicities, from
With a mechanism of this type, it is possible to store a lot of 23,000-year cycles before the Quaternary glaciations became
carbon at the bottom of the ocean and to explain the low established, to 41,000-year cycles 3 million years ago, and
level of atmospheric CO2. This mechanism was recently then to 100,000-year cycles in the last million years.
confirmed in a relatively simple model coupling climate and
carbon cycle (Bouttes et al. 2011). This makes it possible to
formulate a scenario that accounts for the glacial-interglacial Pre-quaternary Astronomical Cycles
cycles, both in terms of changes in the expanse of the ice
sheets, but also in terms of atmospheric CO2 and hence Periodic variations of insolation have existed throughout the
global climate (Paillard and Parrenin 2004), as shown in history of our planet. Although their effects on the climate
Fig. 28.9. have been particularly marked for about a million years
28 Climate and Astronomical Cycles 399

Age (kyrBP)
0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000

LR04 data

Model : ice volume

Model: atmospheric CO2

100 ka
41 ka model spectrum
23 ka
19 ka
100 ka
41 ka LR04 spectrum
23 ka
19 ka
0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000
Age (kyrBP)

Fig. 28.9 Results of the Paillard and Parrenin model (2004) which about 3 million years ago, and then the 100,000-year cycles for the last
includes a mechanism coupling the evolution of ice caps and changes million years. From top to bottom: ice volume data (Lisiecki and
in the carbon cycle, forced by summer insolation at 65°N. By adding a Raymo 2005), the model results in terms of ice caps and CO2, and then
slow drift to the critical threshold parameter, it is possible to account the decomposition into periodicities (spectrum) for the model and the
for periodicity switches with the emergence of the 41,000-year cycles data

through the glacial-interglacial alternations, it is normal to layers may be from very different origins, and they are
expect possibly more subtle types of changes in climate sometimes rather poorly understood. There are examples of
when the Earth was less ice-covered or even not at all. this in more or less all the epochs of the history of the Earth.
Climate variations of astronomical origin may well involve A prime example concerns deposits of organic matter in
other components of the Earth system besides the ice caps, the Mediterranean Sea, which occur regularly in the form of
such as monsoons, biological productivity, or other aspects clearly identifiable layers of black silt called sapropeles.
of our planet. These astronomical cycles are also often These layers rich in organic matter are explained either by an
incorrectly called ‘Milankovitch cycles’, although, for the increase in biological production at the surface or by a
most part, they have nothing to do with changes in the change in the circulation of the deep waters of the
expansion of the ice caps, which were, more often than not, Mediterranean, which would have been poorly oxygenated
non-existent throughout the history of the Earth. Milanko- during these events, just as the Black Sea is today. In fact,
vitch’s theory is a theory of the evolution of the ice caps. It is these sapropel events correspond to rainfall episodes in
not a climate theory and therefore does not apply to all of the Saharan Africa, or even over the Mediterranean basin as a
changes in climate due to astronomical causes. Theoreti- whole, which brought significantly large supplies of fresh-
cally, there is no reason to select summer insolation in the water via the Nile and the rains and disrupted the formation
northern hemisphere as a dominant forcing parameter for of deep waters and therefore the oxygenation levels of the
components of the system other than the ice caps in the Mediterranean. Whatever the mechanism, these sedimentary
northern hemisphere. According to measured indicators levels appear, with a few exceptions, to be governed by
(sedimentology, isotopes of oxygen or carbon, color, etc.), precessional variations since the Miocene, about 14 million
depending on the sites and the geological periods under years ago, until the last event designated S1, at the beginning
consideration, the cycles line up with different astronomical of our interglacial period, about 7000 years ago. As shown
periodicities corresponding to variations in precession, in Fig. 28.10, this cyclicity is sufficiently well-marked to be
obliquity or eccentricity. These alternating sedimentary used not only as a dating method but also to calibrate the
400 D. Paillard

Messâdit Sorbas / Nijar basin summer 65°N Ages 40Ar/39Ar

(Maroc) (Espagne) composite La2004(1,1) analytic error ± 2σ
max min
BGC VU
+ + + + + + + + + +
Mes-19

ME35 UA34
? UA33 6.0
ME32 UA32
ME31
Mes-18 UA31
ME28
+ + + + + + + + + +
UA30
Mes-17

C3An.1n
ME26
+ + + + + + + + + +
UA29
+ + + + + + + + + +

Mes-16 UA28 6.1

ME25
Mes-15 UA26
UA25
ME24 Mes-14
+ + + + + + + + + +

+ + + + + + + + + +
UA24
ME22 Mes-13
UA22
6.2
ME20
Mes-12 UA21
ME19
UA20
ME18 UA19
UA18
6.3
ME17
UA17
ME16
UA16 Mes-14
ME15 Mes-11 UA15
ME14
+ + + + + + + + + +

+ + + + + + + + + +
Mes-10
UA14
6.4
ME13 Mes-9
UA13 Mes-12
UA12
ME12 UA11

UA10
6.5

C3An.2n
ME11 UA9 Mes-11
Mes-8 UA8
UA7
Mes-9
ME10
ME9 UA6
6.6
ME8 + + + + + + + + + +
Mes-7 UA5 Mes-8

Mes-6 UA4
ME7 + + + + + + + + + +

UA3
6.7
ME6 UA2
ME5 + + + + + + + + + +
Mes-5 UA1 + + + + + + + + + + A1.3
ME4 LA20
LA19
6.8 Mes-4
Mes-4 LA18 + + + + + + + + + + A1.2
LA17
LA16
LA15
ME3 6.9 Mes-1
ME2 Mes-3 LA14
4m + + + + + + + + + +
?
ME1 LA13
Mes-2
+ + + + + + + + + +
?
Mes-1 LA12
0 LA11 7.0
LA10

Lithology LA9 Age (in Ma)

LA8
Diatomite/Laminite LA7
7.1
LA6
Dark blue clay
C3Bn

LA5
Marls LA4
LA3
Evaporites LA2
LA1 7.2
A1.1
Halimeda-algae packstone
+ + + + + + + + + +

Sapropel
Tephras

Fig. 28.10 According to Kuiper et al. 2008. Astronomical calibration calibration is then used to better constrain the 40Ar/39Ar method
of the Messinian (between 7.2 and 5.3 Ma) where the alternating (measured here in tephras), since the disintegration constant of 40K is
marl/sapropel series can be correlated with the astronomical forcing (in only known with to an error of about 3 or 4% ((5.463 ± 0.214)  10
−10
agreement with many other stratigraphic markers). This astronomical an−1)

40
Ar/39Ar radiometric method (Kuiper et al. 2008). As geological recordings concerns the alternating marl-
mentioned above, this way of establishing age scales, or limestone series. As early as the end of the nineteenth cen-
cyclostratigraphy, makes it possible to achieve a level of tury, Gilbert suggested that these sedimentary successions
precision far superior to the usual radiometric methods, the from the Cretaceous, which he studied in the limestone
uncertainties of which increase as we go back in time. formations of the Green River in Colorado, were probably
The chronology is thus often the first point of interest caused by astronomical changes. Extrapolating from the
when identifying astronomical cycles in old sedimentary limited outcrops that he had at his disposal, he assumed that
series. The most famous example of cyclicity found in they were linked to the precession cycles, which led him to
28 Climate and Astronomical Cycles 401

125 ky
405 ky 25.2 ky
95 ky
50
21.4 ky Late Triassic
45
Lockatong formations
Relative power 40 and Lower Passiac
35
30.0 ky
30

25

20

15

10

0
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0,2
cycles / foot
Fig. 28.11 Spectrum of the Triassic lake levels in New Jersey (Olsen and Kent 1996)

0.10 400 ky 100 ky Late Cretaceous

40 ky band

1684 ky 20 ky Early Jurassic width Selbuchra, Crimea

0.08 Belemnites marls 20 ky
Puissance relative

140 ky
(arbitrary linear scale)

99 %
0.06
98 %
Spectral power

95 %
0.04 90 %

0.02

0.00
0.00 0.02 0.04 0.06 0.08 0.1
cycles / 1000 years 0 25 50
Frequency (cycles / Ma)
Fig. 28.12 Spectrum of the percentage of carbonate, Jurassic, Eng-
land (Weedon et al. 1999) Fig. 28.13 Spectra of sedimentary reflectance, Cretaceous, Crimea
(Gale et al. 1999) with various chronological hypotheses

suggest that the Upper Cretaceous lasted between 20 and 40

million years. This is in line with recent estimates (34 mil- Carboniferous, alternating marine sediment and coalbed lay-
lion years). Numerous studies have since confirmed the ers, called cyclothems, probably correspond to changes in sea
existence of these Triassic, Jurassic and Cretaceous cycles, level in the delta regions, where abundant vegetation during
as illustrated in the following Figs. 28.11, 28.12 and 28.13. low sea levels (regressions) was followed by sedimentary
In particular, it can be seen that although the precession is deposits during high sea levels (transgressions). This is ulti-
often dominant, the cycles linked to the obliquity or to the mately to be expected, as the Carboniferous, like the Quater-
eccentricity also play very important roles. nary, corresponds to a ‘glacial’ period with consequential ice
The influence of astronomical cycles has also been caps which are likely to fluctuate in line with astronomical
demonstrated in the Paleozoic, although there are fewer changes. Similarly in the Devonian, sedimentary cyclicities
studies on this subject. This is due to the difficulty in finding are observed and these are usually interpreted in terms of the
chronological benchmarks that are sufficiently precise and evolution of the large ice caps present at that time. It is also
reliable to be able to unambiguously attribute the cyclicality possible to find much older cycles, such as the Archean, for
found to astronomical parameters. For example, during the example, more than 2 billion years ago (Hofmann et al. 2004).
402 D. Paillard

The further back in time we go, the more important it is to δ13 C(‰) 405 ka ε (°) δ18 O(‰)
take possible changes in astronomical periodicities into 1.5 1.0 0.5 0 22 24 3.0 2.5 2.0 1.5 1.0
account. In particular, the movements of the Earth’s axis
(precession and obliquity) will be strongly affected by the 80

Miocène Oligocène
22 55
Earth-Moon distance which increases over time due to tidal 90
dissipation. Thus, the periodicity of the precession of the 58
Mi-1
100
equinoxes, between 25,700 years ago and today, was
24
noticeably faster before. The same applies to the cycles of 61 110
climate precession and obliquity, which are directly related 120
64
to it, and which are today 19, 23 and 41 ka. Considering the

Depth (m)
26 130

Age (Ma)
current rate of lunar recession, we obtain respective peri- 67 140
odicities of 16, 18.7 and 29 ka for 500 million years ago. 150
These values are too quick and are inconsistent with geo- 28 70 160
logical observations, which underscores the need for a 170
slower lunar recession in the past, due to the current isostatic 73 180
rebound but also to changes in sea level, in the topography 30 190
76
of the ocean floors, and even ocean stratification. Generally 200
speaking, even more than finding a given periodicity, the 79 210

Oligocene Eocene
aim is to find a coherence between several periodicities 32 220
which corresponds well to the astronomical forcing. For 82 230
Oi-1
example, the presence of three periodicities in a 1:5:20 ratio 240
34 85
is often interpreted as the precession (*20 ka) and the
eccentricity (*100 ka and *400 ka) periodicities, even
though the frequencies have varied in the past, and even if Fig. 28.14 According to Pälike et al. 2006. Stable 13C and 18O
the chronological information does not allow the recorded benthic isotopes of the ODP 1218 site, obliquity (e) with its amplitude
periodicities to be determined with confidence. modulation, and filtering of the eccentricity and the isotopic signals at
If one had to choose a particularly stable periodicity in the 405 ka. The 405 ka periodicity is clearly visible in the isotopic signals,
particularly 13C, which makes it possible to define an astronomical
past, the 405 ka cycle associated with variations in eccentricity chronology (numbering the 405 ka cycles from the current one). The
would probably be the best choice (Laskar et al. 2004). As minimum amplitude in variation of the obliquity very often correspond
stated in paragraph IIa, the solar system is chaotic and it is to maxima for the 18O
impossible to calculate precisely its evolution further back
than a few tens of millions of years. However, if the pertur-
bations associated with the internal planets (from Mercury to These 405 ka eccentricity cycles probably play an important
Mars) become quickly unpredictable, the outer planets (in role in the evolution of climate in the Quaternary. Although
particular, Jupiter and Saturn) have much more regular changes in ice cap volume are largely dominated by the 100 ka
long-term movements. This produces high stability for the cycle, this is not necessarily the case for other indicators. In
periodicity at 405 ka, which can therefore be used as a particular, we find a clear signature in this frequency band in
chronological reference over several hundred million years. the 13C of benthic foraminifera with cycles between 400 and
According to Laskar et al. (2004), the spread of solutions over 500 ka (Wang et al. 2004). These changes in the global carbon
the last 250 million years for this periodicity is less than one cycle are probably the cause of the different climate phases of
cycle (<400 ka). This therefore, creates the opportunity to the last million years: the beginning of the glaciations
construct a very precise absolute chronology not only for the (Pliocene-Pleistocene transition around 2.6 Ma), as well as the
Cenozoic (since 65 Ma) but also for the whole of the Mesozoic transition between the 41 ka cycles and the 100 ka cycles
(between 65 Ma and 250 Ma). These cycles can be system- (Mid-Pleistocene transition around 0.7 Ma) corresponding to
atically numbered from the present time to the distant past, thus the amplitude modulation of this 400 ka eccentricity cycle
offering a new way of stratigraphic tracking on the geological (Paillard 2017).
scale. For example, in Fig. 28.14, the clear presence of these Although cycles were probably present throughout the
cycles is observed during the Oligocene period. Moreover, the Earth’s history, it is important to avoid systematically
Eocene-Oligocene (Oi-1 event) or Oligocene-Miocene (Mi-1 ascribing an astronomical origin to all periodicity observed
event) transitions correspond to particularly ‘cold’ climate in sedimentary records. Indeed, it is quite conceivable that
periods according to the oxygen isotopes. These extreme certain components of the Earth system could generate more
values can be related to the minima of the astronomical forcing or less periodic internal oscillations at frequencies that have
linked to the obliquity (minima of the amplitude modulation). little to do with celestial dynamics. Thus, in the Quaternary,
28 Climate and Astronomical Cycles 403

there is a so-called ‘sub-Milankovitch’ variability corre- References

sponding to the Dansgaard-Oeschger and Heinrich events. It
is likely that cyclicities of this type have also existed at other Adkins, J., McIntyre, K., & Schrag, D. (2002). The salinity, temper-
times in Earth’s history. For example, the 470 m thick Tri- ature and d18O of the glacial deep Ocean. Science, 298, 1769–1773.
assic strata, observed in Latemar in the Italian Dolomites, Arrhenius, S. (1896). On the influence of carbonic acid in the air upon
the temperature of the ground. Philosophical Magazine and Journal
have a well-marked periodic structure showing about 600
of Science, 41, 237–276.
cycles. There is controversy between those who support the Bard, E. (2004). Greenhouse effect and ice ages: Historical perspective.
hypotheses of astronomical cycles, who see precession C. R. Geoscience, 336, 603–638.
cycles and therefore a total recording time of around 10 Bard, E., Hamelin, B., Arnold, M., Montaggioni, L., Cabioch, G., Faure,
G., et al. (1996). Deglacial sea-level record from tahiti corals and the
million years, and the proponents of much quicker envi-
timing of global meltwater discharge. Nature, 382, 241–244.
ronmental variations, using radiometric dating to estimate a Berger, A. (1978). Long-term variations of daily insolation and
total duration of around a million years only, hence an order quaternary climatic change. Journal of the Atmospheric Sciences,
of magnitude faster. Although the debate is not yet settled, it 35, 2362–2367.
Bouttes, N., Paillard, D., Roche, D. M., Brovkin, V., & Bopp, L.
highlights that it is not enough to identify the cycles in
(2011). Last glacial maximum CO2 and d13C successfully recon-
geological records to systematically find an astronomical ciled. Geophysical Reseach Letters, 38, 1–5.
signature. The dynamics of the Earth system are indeed Broecker, W., & van Donk, J. (1970). insolation changes, ice volumes
likely to reveal many more surprises. and the O18 record in deep-sea cores. Reviews of Geophysics and
Space Physics, 8, 169–197.
Calder, N. (1974). Arithmetic of ice ages. Nature, 252, 216–218.
Gale, A. S., Young, J. R., Shackleton, N. J., Crowhurst, S. J., & Wray,
Conclusions D. S. (1999). Orbital tuning of cenomanian marly chalk successions:
Towards a milankovitch time-scale for the late cretaceous. Philo-
The dynamics of the glacial-interglacial cycles of the Qua- sophical Transactions of the Royal Society A, 357, 1815–1829.
Hays, J., Imbrie, J., & Shackleton, N. J. (1976). Variations in the earth’s
ternary are far from fully understood and this is even more orbit: Pacemakers of the ice ages. Science, 194, 1121–1132.
true for the pre-Quaternary astronomical cycles. Moreover, Hofmann, A., Dirks, P. H. G. M., & Jelsma, H. A. (2004).
the climate during glacial periods can sometimes change Shallowing-upward carbonate cycles in the belingwe greenstone
very abruptly during the Dansgaard-Oeschger or Heinrich belt, zimbabwe: A record of archean sea-level oscillations. Journal
of Sedimentary Research, 74, 64–81.
events, for reasons unrelated to astronomical forcing or to Huybers, P. (2006). Early pleistocene glacial cycles and the integrated
atmospheric CO2 variations. It is noteworthy that the last summer insolation forcing. Science, 313, 508–511.
deglaciation was punctuated by such rapid events. It is Huybers, P., & Wunsch, C. (2005). Obliquity pacing of the late
therefore quite possible that they also play a decisive role in pleistocene glacial terminations. Nature, 434, 491–494.
Imbrie, J., Hays, J., Martinson, D., McIntyre, A., Mix, A., Morley, J. J.,
the dynamics of large cycles, and even more so if one Pisias, N., Prell, W., Shackleton, N., Berger, A., Kukla, G., &
considers that they are based on tipping points that allow a Saltzman, B. (1984). The orbital theory of pleistocene climate:
shift from a glacial state to an interglacial state. Support from a revised chronology of the marine d18O record. In A.
The conceptual models presented here are far from suffi- dans Berger (Ed.), Milankovitch and climate (pp. 269–305),
Dordrecht: Kluwer Academic Publishers (Nato ASI Ser. C).
cient to account for the physical and biogeochemical inter- Joussaume, S., Braconnot, P. (1997). Sensitivity of paleoclimate
actions involved in these climate changes. The most simulation results to season definitions. Journal of Geophysical
sophisticated climate models that are used to simulate the Research D, 102, 1943–1956.
twenty-first century, or ‘general circulation models’, are Kuiper, K., Deino, A., Hilgen, F. J., Krijgsman, W., Renne, P. R., &
Wijbrans, J. R. (2008). Synchronizing rock clocks of earth history.
unfortunately unable to simulate these changes because the Science, 320, 500–504.
time scales involved are far too long. In addition, the ice cap Laskar, J., Robutel, P., Joutel, F., Gastineau, M., Correia, A. C. M., &
changes and the dynamics of the main biogeochemical cycles Levrard, B. (2004). A long-term numerical solution for the
must also be taken into account, which these models are not insolation quantities of the earth. Astronomy & Astrophysics, 428,
261–285.
yet able to do correctly. For all these reasons, simpler models Lisiecki, L. E., & Raymo, M. E. (2005). A pliocene-pleistocene stack of
of the ‘Earth system’ (or models of intermediate complexity) 57 globally distributed benthic d18O records. Paleoceanography,
are used to address this type of question. Recent advances, 20, PA1003. https://doi.org/10.1029/2004pa001071.
both in terms of paleoclimate reconstructions but also in Milankovitch, M. (1941). Kanon der Erdbestrahlung und seine
Andwendung auf das Eiszeiten-problem. (p. 633).
terms of modeling (Bouttes et al. 2011), provide hope that a Monnin, E., Indermühle, A., Dällenbach, A., Flückiger, J., Stauffer, B.,
new theory of the glacial-interglacial cycles of the Quater- Stocker, T., et al. (2001). Atmospheric CO2 concentrations over the
nary that takes into account the various aspects of terrestrial last glacial termination. Science, 291, 112–114.
environmental changes (ice caps, carbon, vegetation, climate) Ohshima, et al. (2013). Antarctic bottom water production by intense
sea-ice formation in the Cape Darnley polynya. Nature Geoscience,
can be developed in the years to come. An overview of this 6, 235–240.
type would provide a much better understanding of the great
climate changes experienced so far by homo sapiens.
404 D. Paillard

Olsen, P. E., & Kent, D. V. (1996). Milankovitch climate forcing in the Pälike, H., Norris, R. D., Herrle, J. O., Wilson, P. A., Coxall, H. K.,
tropics of pangaea during the late triassic. Palaeogeography Lear, C. H., Shackleton, N. J., Tripati, A. K., & Wade, B. S. (2006).
Palaeoclimatology Paleoecology, 122, 1–26. The Heartbeat of the oligocene climate system. Science, 314, 1894–
Paillard, D. (1998). The timing of pleistocene glaciations from a simple 1898.
multiple-state climate model. Nature, 391, 378–381. Petit, J.-R., Jouzel, J., Raynaud, D., Barkov, N., Barnola, J.-M., Basile,
Paillard, D. (2001). Glacial cycles: Toward a new paradigm. Reviews of I., et al. (1999). Climate and atmospheric history of the past 420,000
Geophysics, 39, 325–346. years from the Vostok ice core, Antarctica. Nature, 399, 429–436.
Paillard, D., & Parrenin, F. (2004). The antarctic ice-sheet and the Tiedemann, R., Sarnthein, M., & Shackleton, N. (1994). Astronomic
triggering of deglaciations. Earth Planet Science Letters, 227, timescale for the pliocene atlantic d18O and dust records of ocean
263–271. drilling program site 659. Paleoceanography, 9, 619–638.
Paillard, D. (2010). Climate and the orbital parameters of the earth. C. Wang, P., Tian, J., Cheng, X., Liu, C., & Xu, J. (2004). Major
R. Geoscience, 342, 273–285. pleistocene stages in a carbon perspective: The South China sea
Paillard, D. (2015). Quaternary glaciations: from observations to record and its global comparison. Paleoceanography, 19, PA4005.
theories. Quaternary Science Reviews, 107, 11–24. https://doi.org/10.1029/2003pa000991.
Paillard, D. (2017). The plio-pleistocene climatic evolution as a Weedon, G. P., Jenkyns, H. C., Coe, A. L., & Hesselbo, S. P. (1999).
consequence of orbital forcing on the carbon cycle. Climate of the Astronomical calibration of the jurassic time-scale from
Past, 13, 1259–1267. cyclostratigraphy in British mudrock formations. Philosophical
Pälike, H., Laskar, J., & Shackleton, N. (2004). Geologic constraints on Transactions of the Royal Society A, 357, 1787–1813.
the chaotic diffusion of the solar system. Geology, 32, 929.
 Rapid Climate Variability: Description
and Mechanisms 29
Masa Kageyama, Didier M. Roche, Nathalie Combourieu Nebout,
and Jorge Alvarez-Solas

The previous chapters focused on climate variations stem-

Rapid Climate Changes During Glacial
ming from factors external to the climate system: tectonics,
Periods: Heinrich and Dansgaard-Oeschger
orogeny, variations in insolation. Yet the study of glacial
Events
records and marine sediments over the past three decades has
revealed major changes within the climate system over much
The Discovery
shorter time scales than previously envisaged. These abrupt
reorganisations of the climate system, which cannot be
Abrupt climate changes were first discovered in the context
explained by forcings external to the system, are true climate
of the last glacial period (see the review by Hemming 2004).
‘surprises’. They have been the subject of many studies, both
In 1977, Ruddiman showed that during Marine Isotopic
to describe the expression of these events on a global scale
Stages 4, 3 and 2, large quantities of coarse detrital material
and to model these events and their impacts. This work is
transported by icebergs detaching from the Northern
still going on to better characterize and understand this type
Hemisphere ice sheets were deposited at the mid-latitudes of
of climate variability, called millennial variability, in con-
the North Atlantic between 40 and 65° N in a band now
trast to the time scales associated with the Milankovitch
known as the Ruddiman Belt, while the main deposition area
forcings (see Chap. 7), or abrupt (rapid) variability, because
for Stage 5 is closer to Greenland and Newfoundland. The
the transitions between climate states take place on even
concentration of detrital material measured in marine sedi-
shorter time scales, ranging from ten to one hundred years.
ments is thus linked with the size and expansion of conti-
This chapter presents a synthesis of current knowledge on
nental ice sheets over the Milankovitch timescale. In 1988,
rapid climate variability.
H. Heinrich showed that six major events of coarse detrital
material deposition occurred in the Ruddiman band during
the last glacial period at intervals of about 10,000 years, and
M. Kageyama (&)
D. M. Roche therefore at time scales shorter than those of Milankovitch.
Laboratoire des Sciences du Climat et de l’Environnement,
These six events are often accompanied by a major change
LSCE/IPSL, CEA-CNRS-UVSQ, Université Paris-Saclay, 91191
Gif-sur-Yvette, France in the composition of assemblages of planktonic for-
e-mail: [email protected] aminifera, with a predominance of the left-coiling
N. C. Nebout Neogloboquadrina pachyderma polar species, an indicator
UMR 7194 CNRS/UPVD/MNHN, HNHP-Histoire Naturelle de of a particularly cold environment. Heinrich initially based
l’Homme Préhistorique, Département Homme et Environnement, his interpretation of these results on the fact that the period
Muséum National d’Histoire naturelle, Paris, France
of 10,000 years is about half a precession cycle. He specu-
N. C. Nebout lated that the coarse detrital material was transported either
Institut de Paléontologie Humaine, 1 rue René Panhard, 75013
by icebergs (in orbital conditions favoring a ‘cold’ period) or
Paris, France
by the melting of the ice sheets (‘warm’ situation). This
J. Alvarez-Solas
hypothesis did not stand up to more precise dating and
Departamento de Física de la Tierra y Astrofísica, Facultad de
Ciencias Físicas, Universidad Complutense de Madrid, 28040 analyses of marine records in the North Atlantic but it shows
Madrid, Spain that, at the time of this discovery, the variations in orbital
J. Alvarez-Solas insolation were considered the main contributors to the
Instituto de Geociencias, Consejo Superior de Investigaciones evolution of the climate system. At that time, sudden
Científicas-Universidad Complutense de Madrid, 28040 Madrid, changes to the system were not envisaged. The discovery
Spain

© Springer Nature Switzerland AG 2021 405

G. Ramstein et al. (eds.), Paleoclimatology, Frontiers in Earth Sciences,
https://doi.org/10.1007/978-3-030-24982-3_29
406 M. Kageyama et al.

however initiated much research on these major deposits of end of a Heinrich event, by the first Dansgaard-Oeschger
coarse detrital material, which were named Heinrich events event, which has a large amplitude. This is followed by a
(HE) by Broecker et al. in 1992. Heinrich events are defined few other Dansgaard-Oeschger cycles, of diminishing
by these authors in terms of both their characteristic low amplitude. The last cycle ends with a massive discharge of
concentration of planktonic foraminifera and high concen- icebergs from the Laurentide sheet, in other words, by a
tration of detrital material. They further demonstrate that Heinrich event. Each cycle lasts approximately 7000–
these events are related to armadas of icebergs that broke off 10,000 years.
from the ice sheets rather than to the melting of these ice Since these discoveries in the early 1990s, millennial
sheets. The layers of detrital material may be several meters variability has become a subject of intense research.
thick in the Labrador Sea. This thickness is less, although Numerous records were analyzed with as fine a resolution as
still significant, close to the European coasts. Subsequent possible and they showed that variability at shorter time
studies of the composition and properties of this scales than those of Milankovitch was not limited to the
ice-rafted detrital material show that its origin is mostly from North Atlantic and adjacent regions. The challenge is then to
the Canadian shield and has therefore been transported by be able to synchronize these different records to better
icebergs which detached from the North American ice sheet. characterize this type of variability, as well as to better
During the same period, the analysis of the d18O isotopic understand the connections between climate signals recorded
signal from ice cores taken from the summit of the Green- all over the globe. In the following sections, we focus on the
land ice cap (Dansgaard et al. 1993) shows the contrast signature of the Heinrich and D/O events, on their impact on
between interglacial periods, and particularly between the climate and on the efforts to understand these climate
remarkably stable Holocene, and the last glacial period, instabilities through modeling.
characterized by high amplitude oscillations. This amplitude
can be as much as half of the glacial-interglacial difference in
Greenland. The warming events were named The Regional Impacts of the Heinrich
‘Dansgaard-Oeschger events’ (D/O), because these two and Dansgaard-Oeschger Events: North Atlantic
authors had already detected rapid d18O fluctuations in the and Adjacent Regions
ice in a core from Camp Century. At the time, the record
appeared dubious due to large simultaneous variations in the Atlantic Ocean
CO2 content in the air bubbles trapped in the ice. We now
know that the latter are artefacts caused by the presence of The Heinrich Events
carbonate dust attacked by the sulfuric acid present in the
ice. However, the rapid fluctuations of d18O are indeed The melting of massive armadas of icebergs profoundly
significant. modified the ocean surface conditions in the North Atlantic.
D/O warming appears to have occurred particularly The changes are recorded in ocean sediment cores, particu-
quickly, over a few decades, solely when ice sheets devel- larly in d18O signals from planktonic foraminifera. Indeed,
oped on the continents of the northern hemisphere. The excursions towards lighter d18O are measured in the calcite
relatively warm period following the D/O event is called the of these foraminifera, indicating either less saline surface
‘interstadial period’ and is characterized by gradual cooling. waters or warmer temperatures. The fauna of fossil for-
It ends with a rapid return to the coldest levels recorded, aminifera in sediments reveal very cold conditions during
known as ‘stadial periods’. This return completes a D/O Heinrich events (see, for example, Hemming 2004; Cortijo
cycle, which lasts a total of about 1500 years. There are et al. 2005). The melting of icebergs is responsible for a
therefore more D/O events than Heinrich events. huge inflow of freshwater, desalinating the surface of the
The existence of abrupt variations in both the surface ocean and introducing a highly negative d18O signal, typical
climate in Greenland and in the ocean conditions over time of ice contained in the ice sheets.
scales far shorter than the Milankovich cycles contributed to Other paleoceanographic records reveal significant reor-
the emerging idea that the climate system could have mul- ganizations during the Heinrich events. In particular, the
tiple equilibrium points and that it could be abruptly reor- study of d13C in sediment cores proves that the bottom
ganized as it transfers from one equilibrium to another. This waters of the Atlantic Ocean were less well ventilated (Elliot
idea was reinforced by the correlations between marine and 2002), demonstrating a reorganization of ocean circulation.
glacial records, first demonstrated by Bond et al. (1993). These studies were corroborated by analyses of the magnetic
These authors suggest a perspective of glacial variability that properties of sediment cores collected in the North Atlantic
integrates the Heinrich and Dansgaard-Oeschger events. The (see Kissel 2005 for a compilation), which show that the
glacial millennial variability is organized in cycles, later melting of icebergs was accompanied by a slowing of deep
called “Bond cycles” (Fig. 29.1). Each cycle begins at the currents and of the thermohaline circulation.
29 Rapid Climate Variability: Description and Mechanisms 407

Fig. 29.1 Bond cycles (from Bond et al. 1993). The top two curves Greenland. These curves show how events recorded in marine cores
represent the abundance of N. Pachyderma (s.) in sediment records and those recorded in ice cores correspond with each other. The fourth
from the DSDP-609 and VEMA23-51 cores from the North Atlantic at curve is a schematic representation of the Dansgaard-Oeschger and
50 and 54° N, respectively. The Heinrich events and the Younger Heinrich events organized into cycles now called ‘Bond cycles’. The
Dryas are indicated by arrows under the second curve. The third curve figures over the age axis indicate the depth in the GRIP core
shows the d18O record in an ice core from the GRIP project in

The isotopic study (Sr/Nd) of the grains brought by the iceberg armada events and other records sensitive to tem-
icebergs to the North Atlantic during the Heinrich events perature, precipitation and other climatic factors have been
made it possible to determine where these icebergs came highlighted. This has led, in some publications, to a shortcut
from. Grousset et al. (1993) first demonstrated that within being made between the iceberg armada event and its sup-
the Heinrich levels, although most of the detrital material posed climate impact. It should be noted that, strictly
brought to the Ruddiman belt came from the Laurentide ice speaking, a Heinrich event can only be defined by the pres-
sheet, there were some detrital elements from European and ence of coarse detrital elements in sedimentary cores taken
Icelandic ice sheets. Subsequent, more detailed studies of the from the Ruddiman band. This implies in particular that the
Heinrich events complemented this panorama by showing presence of detrital elements in an ocean core from
that during each Heinrich event a level of detrital material the Fennoscandian ice sheet margin does not necessarily
from the European ice sheets preceded the level coming indicate a Heinrich event, but could be due to the melting of
from the North American ice sheet. This almost systematic icebergs from European ice sheets. In addition, it is not
phase relationship led these authors to question the possi- generally possible to define Heinrich events from marine
bility of a causal relationship: could the icebergs from the records outside the Ruddiman band and, even less so, from
European ice sheet be triggering the discharge of icebergs continental records. As a result, the term ‘cold Heinrich
from the North American ice sheet? Another interpretation event’ is often inaccurately used because it applies only to the
of these observations is to imagine that the European ice Ruddiman band, and is simplistic, because the drop in tem-
sheets were smaller and oscillated faster than the larger perature is only one aspect of the climate changes associated
Laurentide ice sheet, causing them to precede it during the with a Heinrich event. One exception is pollen records from
Heinrich events, with no obvious causality between the two. marine cores containing detrital material. These provide
The Heinrich events continue to attract a lot of attention simultaneous records from the adjacent continent (pollen)
from researchers, and many connections between these and from the iceberg armadas originating from the North
408 M. Kageyama et al.

American ice sheet. They can therefore safely be used to during this period, rapid climate changes, whether D/O
reconstruct continental climate changes associated with oscillations or Heinrich events, had repercussions on
Heinrich events (Sánchez-Goñi et al. 2002; Combourieu- vegetation.
Nebout et al. 2002). In Western Europe, D/O events resulted in periodic changes
in forest cover (Fig. 29.2), with warm events corresponding to
Dansgaard-Oeschger Events the expansion of oak forests, also implying wet conditions (e.g.
Sánchez-Goñi et al. 2002; Combourieu-Nebout et al. 2002).
When the cores are studied at century-scale resolution, On the other side of the Atlantic, in Florida, over the same
records of calcite d18O from foraminifera from the ocean periods, the oak is associated with large quantities of herba-
surface indicate that abrupt changes also occurred during ceous plants (Ambrosia, Poaceae), indicating drier phases
Dansgaard-Oeschger events. The variations recorded by (Grimm et al. 2006). During the Heinrich events, the vegeta-
marine indicators are very similar to those obtained from tion cover of western Europe became steppe-like, indicating a
Greenland ice data. It is therefore tempting to directly link an cold and very dry climate, especially in the Mediterranean
observed variation in one set of records to one observed in region (Fig. 29.2). These cold conditions prevailing over
the other and to assume that they are responses to a common Europe are associated with high d13C values recorded in sev-
cause. While this is certainly partially true (the d18O of polar eral stalagmites in the south of France. They show up in a
ice is above all a response to temperature and the d18O of the slowing down, or even a curtailment of growth as the cold
foraminiferal calcite also contains a temperature signal), conditions prevent the infiltration necessary for the formation
especially in the case of marine records close to the ice of stalagmites (Genty et al. 2005).
sheets, it should be borne in mind that it is paleoclimate At the same time, vegetation in Florida was characterized
indicators and not climate variables, such as temperature, by an abundance of pine trees and the regression of herba-
that are being measured. Directly correlating every small ceous plants and oak trees which is interpreted as a conse-
variation in the two records is therefore hasty and inordinate. quence of a warmer and more humid climate (Grimm et al.
Despite these difficulties, marine records reveal that during 2006). The strong contrast between the responses by climate
Dansgaard-Oeschger events, the surface ocean and the deep and vegetation to abrupt climate events on the two sides are
ocean undergo changes of great amplitude in the surface examined with the cases of Europe and Florida. In Florida,
temperature and/or the d18O of the sea water (see Rasmussen this response may seem to be counter-intuitive, since the
et al. 1996; Shackleton et al. 2000). The reconstruction of climate becomes warmer and more humid during the Hein-
ocean surface temperatures from independent indicators rich events, although they are responsible for a major cool-
shows that the temperature signals in Greenland cores are ing in the North Atlantic. However, as records obtained for
matched in the neighboring North Atlantic. Florida have a relatively poor temporal resolution, it is
Associated with these changes in temperature and possible that this contrasting response with the European one
hydrological conditions are signals of relatively small is due to problems of synchronization in the reconstructions.
amplitude in oceanic d13C records (Elliot 2002). Since d13C Another possibility is that the climates of Europe and Florida
is an indicator of the ventilation of ocean water bodies, it do not have the same sensitivity to freshwater incursions into
would appear that at least for the North Atlantic and the the North Atlantic and the associated cooling. Europe,
Arctic, transitions between stadial and interstadial periods indeed, has a regime of prevailing westerly winds and is
are not associated with large anomalies in ventilation, and therefore under the direct influence of the cooling of the
therefore with any drastic change in thermohaline circulation North Atlantic, unlike Florida on the other side of the
in the Atlantic. This observation shows the different mech- Atlantic. It is thus crucial to be able to explain the com-
anisms operating during the Dansgaard-Oeschger events and plexity of these different signals. Modeling, as will be shown
Heinrich events, the latter having very marked anomalies in later in this chapter, can provide a coherent framework for
ocean d13C records. reconstructions of climate changes in regions geographically
distant from each other.
Neighbouring Continents
The evolution of continental paleo-conditions is recorded in
numerous environments, such as in caves (e.g. in concretions Millennial-Scale Variability in Other Regions
or speleothems), lakes (lake sediments), peat bogs, loess or of the World
marine cores. For continental regions adjacent to the Rud-
diman belt, these records show a clear correlation between After the discovery of abrupt climate changes in Greenland,
the evolution of the ocean and continent during periods of the North Atlantic and adjacent regions, rapid changes in the
strong glacial variability. Pollen data, for example, show that characteristics of the climate system were discovered in
29 Rapid Climate Variability: Description and Mechanisms 409

GREENLAND
MEDITERRANEAN AREA: ODP Site 976
GISP 2

)
(%
m ina
st

.c.

ns ium
er dr

ïd B
) ore

llo PD
al
yd ua

%
kie in
(% te f

es
e(
) ert

ch oq

bu )
pi d

G. O (‰
te ato
(% des
ra

pa lob
pe

ct

8
og
i-

δ 1
m
δ 18O (‰) PDB

te
m

Ne
Te

Bi
Se
- 45 -40 -35 0 20 40 60 0 20 40 60 0 10 0 10 20 3 2 1 0
0 0

10
YD
1 1 1 5
H1

20
Age (103 calendar yr)

2 2

Depth (mcd)
2
H2

3 3 3 10
4 4 4
30
H3
5
6 5 5
6 6
7
7 7
8 h
40 9 8 8
10 H4 15
11 10 10
11 11
12 12 12
H5

50 20

Fig. 29.2 Comparison of the d18O curves of the GISP2 ice core from on the GISP2 curve, with the corresponding numbers on the
Greenland with the paleoenvironmental reconstructions obtained for the Mediterranean curves. The Heinrich events and the Younger Dryas
ODP 976 (Alboran Sea) site. The curves are presented as a function of are indicated by the gray bands. Adapted from Combourieu-Nebout
calendar age (on the left) and as a function of depth in the ODP 976 et al. (2002)
core, in m (on the right). D/O events are indicated by the numbers 1–12

other parts of the world, in some cases far from the North type are recorded. Since this review, new records have been
Atlantic. Abrupt changes in the dD of the ice at Vostok in produced, confirming the global nature of these abrupt cli-
Antarctica were reported as early as 1994 by the team of M. mate variations, and suggesting the involvement of a
Bender, who established a correlation with the Greenland mechanism operating on a global scale.
records, through the oxygen isotopes in the air preserved in Rapid climate variations of great amplitude have thus
the ice. In 1995, abrupt changes in ventilation were observed been found at many points on the globe. The most difficult
in the Santa Barbara Basin off California. In 1998, abrupt part of interpreting these records is to synchronize them. The
changes in biological productivity in the Arabian Sea, ideal would be to have an absolute dating for each record, on
strongly related to monsoon intensity, were shown over time the same scale of the abrupt events, i.e., on a decadal scale.
scales similar to Dansgaard-Oeschger cycles. In 2000, sig- For most paleoclimate indicators, this is impossible to
nificant variations in the hydrological cycle in the low lati- obtain. A dating method such as that based on 14C does not
tudes of the North Atlantic (Cariaco basin) were shown to offer sufficient precision over the glacial period to be able to
correspond with events observed in the North Atlantic. link abrupt events recorded in two different places, in other
Voelker et al. (2002) list 183 sites where variations of this words, to establish the phase differences between the various
410 M. Kageyama et al.

records. In addition, knowing that the residence time of 14C explain the observations or even to establish the connection
in the ocean (reservoir age variations) varies during abrupt between the inflow of freshwater to the North Atlantic and
events makes absolute dating based on this single variable the climate consequences. Each study focuses on one aspect
even more difficult. However, these variations are interesting of the rapid climate variability. For the time being, no study
because they give us information on changes in ocean has tried to include all of the factors of rapid climate vari-
ventilation and ocean circulation (Chap. 21). ability in a single model, which would require a represen-
However, it is possible to synchronize the different tation of the ice sheets, climate, ocean, chemical composition
records with each other using parameters known to be the of the atmosphere, conditions of land surfaces, etc. consti-
same or to vary in a similar manner in different locations on tuting an almost complete model of the Earth system, which
the planet. This allows us to study the chronology of events does not currently exist. The following sections present
recorded in different places and to propose hypotheses about some of the suggested mechanisms behind Heinrich events
the mechanisms behind glacial abrupt climate variability. and how they impact on global climate as well as on the
Specific events, such as volcanic eruptions, whose falling Dansgaard-Oeschger cycles.
ash appears in ice, marine and lake sediments, can help to
synchronize records (Rasmussen et al. 2014). Other param- Heinrich Events
eters can be used depending on the type of record.
One example is the use of variations in the atmospheric Ice-Sheet Instabilities
concentration of methane to place Greenland and Antarctic
ice records on the same time scale. Methane is well mixed in One of the first explanations of the cycles of spectacular
the atmosphere and so variations in its concentration are the armadas of icebergs discharged from the North American ice
same everywhere on the globe. Members of the EPICA sheet was proposed by MacAyeal in 1993. He constructed a
community (2006) used this method to compare the model of the part of the Laurentide ice sheet positioned on
expression of abrupt events at the two poles. Comparisons what is today the Hudson Bay, that is, on a sediment-
show an anti-correlation between isotopic signals in covered bedrock. When the base of the ice sheet reaches its
Greenland and Antarctica. In addition, the strength of the melting point, causing liquid water to be present there, the
warming in Antarctica is related to the length of the cooling ice slides much more easily than it would if the base of the
in Greenland. ice sheet and the sediments were frozen or if it was on a
Another example is the use of magnetic properties to syn- bedrock not covered with sediment. The ice basal layer can
chronize marine records (see Chap. 7). At the scale of the warm up through the input of geothermal energy when the
ocean basin (e.g., the North Atlantic, Elliot 2002), variations in ice sheet is thick enough, because the temperature at the base
rock magnetism can be used to put marine records on a com- of an ice column is dependent on the pressure exerted by this
mon timescale. At the interhemispheric scale, paleomagnetic column and because of the ice insulating its base from the
intensity can be used, especially for episodes of sudden and cold conditions at its top. Based on these properties,
intense variation, such as around the Laschamp event. MacAyeal proposes the following cycle: (1) the ice sheet
grows due to the accumulation of snow on the surface (as-
sumed to be constant over time in this model) and the base of
Mechanisms the ice is frozen; (2) once the ice sheet is thick enough, its
base melts and a layer of liquid water forms at the
The climate reconstructions described above demonstrate the base-sediment-ice interface; (3) the ice sheet then quickly
existence of rapid climate variability of great amplitude that slides towards the ocean and its elevation decreases; (4) the
involve all the components of the climate system. Under- elevation of the ice sheet decreases sufficiently so that the
standing all aspects of this type of variability is still a basal layer freezes again, thus slowing down the movement
challenge for the scientific community. An approach to test of the ice, and returning to step 1. MacAyeal calls this model
our understanding of these phenomena is to build models ‘binge/purge’. In this model, the time between two “purge”
that include processes that are thought to contribute in a events (i.e. between two episodes of iceberg break-up) is
critical way to rapid variability. Well-informed choices from determined by the characteristic time necessaryfor the ice
the possible forcings and processes can highlight the sheet to ticken and hence depends on the configuration of
importance of each one for a particular type of event or for a the ice sheet (size, distance to margins, characteristics of the
level of variability (see Chap. 25, volume 2). Models have underlying surface) and on its surface mass balance. It
thus been constructed to explain the iceberg armadas (the should be stressed that this mechanism works even for a
Heinrich events themselves, strictly speaking) or to estimate constant climate. MacAyeal demonstrates an inherent
the inflow of freshwater to the North Atlantic that could oscillation in the ice sheets and shows that for the Laurentide
29 Rapid Climate Variability: Description and Mechanisms 411

(2013) postulated that such an ice-shelf break-up could be

triggered by warmer subsurface waters. By using the hybrid
ice-sheet/ice-shelf thermomechanical model GRISLI (Ritz
et al. 2001), they simulated the effects that such an ice-shelf
removal has on the inner part of the Laurentide ice sheet.
Under this theory, not only the floating parts of the ice sheet
contribute to the generation of the iceberg armadas but also
the subsequent acceleration of the ice streams terminating in
the ocean. For this mechanism to work, the ice shelves need
to be destabilized from beneath and thus it assumes the
existence of an external (to the ice sheet) forcing of Heinrich
events. At the same time, Marcott et al. (2011) showed
through the analysis of Mg/Ca in benthic foraminifera that
the North Atlantic Ocean indeed experienced marked
warmings of its subsurface waters during Heinrich events.
Later on, in 2015, in a study entitled “Icebergs not the trigger
Fig. 29.3 Modeling of the instabilities of the atmosphere-ocean-North for North Atlantic cold events”, Barker et al. analysed sev-
American ice sheet system in the context of a glaciation (Calov 2002).
The climate model used is the CLIMBER-2 model, the model for the eral marine sediment records. They demonstrated that the
North American cap is the SICOPOLIS model in which the possibility melting of icebergs into the North Atlantic could not be the
of the ice cap sliding when the base is not frozen is incorporated. The cause of the observed surface cooling because the icebergs
model simulates a series of instabilities of the North American simply arrive too late to these sites. They concluded that
cap. a Ice volume of the ice cap. b Altitude of the ice cap above Hudson
Bay (85.5° W, 61.5° N). c Altitude of the ice cap near the mouth of although the freshwater from icebergs could provide a pos-
Hudson Bay (67° W, 60.75° N). The periods indicated in bold indicate itive feedback for lengthening stadial conditions, it does not
periods when the ice cap releases armadas of icebergs, in other words, trigger northern stadial events. Recently, Bassis et al. (2017)
Heinrich events simulated by the model further expanded on the idea of the Heinrich events being
triggered by the ocean by simulating the response of the
ice sheet the length of the cycle is about 7000 years, which Laurentide grounding line and the associated ice-streams
is of the same order of magnitude as the observed period. acceleration.
One criticism of this explanation is precisely that the ice Therefore, recent advances both in modelling and paleo-
sheets can oscillate independently, whereas a certain con- record analyses suggest that Heinrich events are a conse-
sistency between the break-up events in North America and quence of oceanic circulation changes rather than their primary
Scandinavia has been observed (Grousset et al. 1993). cause. In this theory, a shift into a cold surface stadial condition
However, the MacAyeal mechanism was reproduced in a in the Northern Hemisphere caused by a weakening of the
more complex model (Calov 2002), which included the Atlantic oceanic circulation would be accompanied by a
ocean, atmosphere, vegetation and ice sheets of the northern warming of the subsurface waters of the Labrador Sea, facil-
hemisphere, the characteristics of the bedrock at the base of itating the occurrence of a Heinrich event. An important fact
the Laurentide ice sheet being imposed and spatially variable here is that Heinrich events appear at the middle of the stadial
(Fig. 29.3). This latter study shows that although instabilities phases. Subsequently, the massive presence of icebergs in the
of the Laurentide ice sheet exist under constant forcing, the North Atlantic will amplify the decrease of the oceanic cir-
rest of the climate system can play a synchronization role for culation intensity or even halt it, therefore enhancing stadial
the events. conditions. This interpretation of the chain of events involved
Hulbe et al. (2004), with MacAyeal as co-author, pro- in Heinrich events remains fully compatible with the observed
posed an alternative to the “binge-purge” mechanism. Their excursions towards the lighter d18O observed in planktonic
explanation relies on the speculated presence of a massive foraminifera. Nonetheless, this theory does not explicitly
ice shelf on the Labrador Sea. A break-up of this floating explain yet why Heinrich events do not appear for every sta-
part of the Laurentide ice sheet would indeed be very effi- dial. The likeliness of this theory does not remove any value to
cient source of the icebergs defining the Heinrich events in the modeling exercises based on mimicking Heinrich events
the North Atlantic. A drawback of this theory, however, by means of freshwater flux injections into the North Atlantic.
consists in the difficulty of identifying the ultimate reason of These have been very helpful in understanding the mecha-
the ice-shelf break-up during cold surface conditions such as nisms by which a weakened or suppressed Atlantic circulation
those observed during Heinrich events. Alvarez-Solas et al. can have impacts on the rest of the climate system.
412 M. Kageyama et al.

Evaluation of the Freshwater Influx Associated with a to provide a new estimate of the volume of icebergs dis-
Heinrich Event: Example of a Joint Model-Data charged and the duration of these events: for the Heinrich 4
Approach event (about 45 ka BP) the most likely duration is
300 ± 100 years and the volume of icebergs discharged is
The thickness and location of the detritic levels found in the equivalent to about 3 m of sea level. These results were later
North Atlantic open the question of the volume of icebergs confirmed by a method based on a sediment/iceberg model
released in this region, especially as the records of sea level (Roberts et al. 2014).
change during the deglaciation do not show the Heinrich 1
event having any impact. This would lead us to assume that Interpretation of the Isotopic Signal Measured in the
the volume of icebergs is low (a maximum of 3–5 m of sea North Atlantic
level). However, estimates achieved using other methods
(Roche et al. 2004) have quite different results. Yet, in order We have seen that the d18O records from Greenland cores
to reproduce the evolution of these events in climate models, and from ocean sediment cores appear to have very similar
we must know the influx of freshwater represented by the signals during periods of rapid climate variability. These
corresponding armada of icebergs. It is therefore necessary similarities should not lead to the conclusion that both sig-
to determine not only the volume of icebergs released into nals have the same climatic cause. Indeed, d18O records
the ocean, but also the timeframe over which they were from foraminiferal calcite are complex, sensitive to both
released. The classic 14C dating method presents a major temperature and hydrological changes. Hydrological chan-
problem related to the reorganization of the ocean. Indeed, ges have a particularly strong impact. To separate out the
the postulated shutdown of the thermohaline circulation is influences of the different causes of d18O variations in cal-
causing a change in the distribution of carbon within the cite, one solution is to simulate this indicator within a cli-
ocean reservoir, the largest in the Earth system. Basically, mate model to analyze the importance of the various
the stopping of the thermohaline circulation would make the processes at work. Roche and Paillard (2005) performed this
surface waters appear younger, and the bottom waters appear type of simulation for the series of Dansgaard-Oeschger
older during the event, and it would make the surface waters events surrounding the Heinrich 4 event (about 40 ka BP).
appear older, and the bottom waters younger after the event, The result is shown in Fig. 29.4. The first outcome is the
making it difficult to evaluate the duration of the event. accuracy of the model in reproducing the variations mea-
As the results from conventional methods assessing the sured in the ocean record. The advantage of using a climate
duration and volume of icebergs emitted during these events model is that not only the d18O of calcite can be simulated
have a very high degree of uncertainty, a more detailed but also temperature and salinity, prognostic variables for the
estimation using different methods is necessary. An climate model. The ‘temperature’ and ‘hydrological cycle’
approach combining model and data was undertaken by components can then be extracted from the d18O signal of
Roche et al. (2004), based on the simulation of d18O of water the calcite. The surprising result from Fig. 29.4 is that, in the
in the ocean. In fact, there is a large number of d18O records model, the strongest changes in both temperature and d18O
of foraminiferal calcite from marine sedimentary cores from of the water over the entire simulated period are shown to be
the North Atlantic, which constrain tightly the evolution of during the Heinrich 4 event. In the d18O record in calcite, the
this indicator during the Heinrich 1 and 4 events. The basis greatest variation is associated with the Dansgaard-Oeschger
of this new evaluation method is to consider that the geo- events. The reason for this difference is that since the vari-
graphical distribution of the d18O anomaly recorded in ations in temperature and in the d18O of water have, from the
planktonic foraminifera provides information about the point of view of the calcite d18O, opposing signatures, their
duration and volume of icebergs emitted. If the thermohaline effects partially cancel each other out in the d18O of the
circulation slows down, the slight anomaly in d18O created calcite, hiding the strongest signal in the marine sedimentary
by the melting of the armadas of icebergs will tend to remain data. This example shows how difficult it is to interpret
longer at the surface (and vice versa). The maximum isotopic paleoclimate indicators in terms of climate and how
recorded anomaly should also be related to the maximum of an integrated data-model approach can provide a better
the iceberg flux. To assess this relationship, Roche et al. understanding of climate dynamics in the past.
(2004) performed a large set of simulations by varying the
duration and influx of additional freshwater for all of the Transmission of the Signal to the Continents
feasible values (determined from the data). Then, the simu-
lated d18O anomaly was compared with the anomaly mea- The pollen records extracted from the marine cores provide
sured in the marine sediment cores. The simulations that best information on the evolution of the vegetation contempora-
represent the distribution described by the data were selected neous with the evolution of the oceanic conditions. The two
29 Rapid Climate Variability: Description and Mechanisms 413

0.2 Atlantic, Mediterranean and the west coast of Europe

Fresh water flux

0.15 (a) (Sánchez-Goñi et al. 2002; Combourieu-Nebout et al. 2002,

(Sverdrup)

0.1 cf. Figure 29.2). The question is then to understand these

relationships. The climate in a region such as Western
0.05
Europe depends not only on the ocean surface conditions in
0
the North-East Atlantic and the Mediterranean, but also on
1 much more distant marine and continental conditions which
(b) impact on atmospheric and oceanic circulation.

(per mil)
1.5

δ18Oc
To continue with the example of the impact a Heinrich
2 event has on the climate and on the vegetation of Western
2.5 Europe, we have seen that the signature of this event in terms
of ocean surface temperature is firstly, a cold anomaly of
12 several degrees Celsius at the mid-latitudes of the North
Temperature

10 Atlantic. We know, moreover, that at these latitudes the

(°C)

atmospheric circulation is dominated by westerly winds.

8 (c) A simple transmission mechanism of the oceanic signal
6 would be its advection, i.e. its transport, by the prevailing
atmospheric circulation, towards the ‘leeward’ continents, in
(d) 0 (per mil) this case, Western Europe. However, it is difficult to estimate
how far the signal can travel and the atmospheric circulation
δ18Ow

0.4 itself can be modified by the changes in sea surface tem-

peratures. The initial response may also be reinforced or
0.8
negated by positive or negative feedbacks, such as, for
3 example, snow cover or clouds (Chap. 1). Another question
(e)
is: what can be said a priori about changes in precipitation
(per mil)

3.5
δ18Oc

associated with a cold event in the North Atlantic? Two

4 processes need to be taken into account. A first constraint is
that a colder atmosphere contains less water vapor
4.5 (Clausius-Clapeyron relation) which makes it less conducive
10000 8000 6000 4000 2000 0
Run time (years) to the formation of precipitation. Second, in the
mid-latitudes, over the oceans and the western side of the
Fig. 29.4 Analysis of glacial variability (D/O and Heinrich events) in continents, most precipitation comes from weather systems
the North Atlantic Ocean, in the intermediate complexity model that form over the oceans and pushed towards the western
CLIMBER2-ISO, an atmosphere-ocean-vegetation coupled model
including calculations of 18O concentration, which allows direct coasts of the continents by the prevailing winds. These
comparison of the model results with measurements from marine perturbations are the result of atmospheric instabilities rela-
sediments (foraminiferal calcite). Adapted from Roche and Paillard ted to the meridional temperature gradient. A shift of the
(2005). a Scenario of freshwater influx in Sverdrup (106 m3/s). All areas of higher gradient is important because it causes a
model outputs are given at 41.25° N in the Atlantic for easy
comparison with the MD95-2042 core. b d18O of the calcite in the
displacement of the precipitation zones. A negative anomaly
ocean model’s surface layer, i.e. the first 50 m (expressed in parts per of ocean surface temperatures at the mid-latitudes of the
thousand versus SMOW). c Evolution of the temperature at the surface North Atlantic implies a shift of the high meridional gradient
simulated by the model (in °C). d Results of the simulation in d18O of zone of ocean surface temperatures towards the south. This
the water (in parts per thousand versus SMOW). e) d18O of simulated
calcite at a depth of 3000 m. The results for the temperature and the should favor a southward migration of the prevailing wind
d18O of the water are not included for this depth because their belt and the precipitation associated with weather systems. If
variations are in the same direction and each contribute half of the d18O these weather systems are stronger, this can offset the direct
signal of the calcite influence of temperature on precipitation. This shows how
important it is to have information from multiple locations in
types of reconstructions, marine and continental, are syn- order to interpret a given record. For example, in order to
chronized because they come from the same sediment core. interpret reconstructions of precipitations in Western Eur-
Thus, by looking for pollen associations that appeared at the ope, it is important to know the temperatures in the region,
same time as the levels of detritus, the vegetation present which, in general, is relatively easy to infer from the same
during Heinrich events can be reconstructed. More gener- core, but it is also important to know the meridional tem-
ally, it is possible to study the relationship between ocean perature gradient over the North Atlantic and this requires
conditions and vegetation, as has been done for the Eastern having data from several cores located far from, but yet
414 M. Kageyama et al.

synchronized with, the original core. This shows how Kageyama et al. (2005) also show the impact of this
important it is to synchronize the different cores. This can be change in climate on vegetation, as simulated by the
complex. Given the application of the comparison of results dynamic ORCHIDEE vegetation model. The climate chan-
from different cores, it is important to understand how this ges simulated by the climate model for a cold event in the
synchronization is done. North Atlantic, as weak as the impacts may seem on the land
At this stage of interpretation, climate models (Chap. 25) masses, result in a significant decrease in vegetation cover,
can also be useful to better understand the possible links of both trees and herbaceous plants. This result suggests that
between changes in climate recorded at different locations. during glaciations vegetation in Europe and the Mediter-
We now continue our analysis of the impact of a Heinrich ranean is, as indicated by pollen records, extremely sensitive
event on climate and vegetation in Western Europe. In a first to changes in climate.
series of numerical experiments we hypothesized that for the It may be noted that there is an area on the Mediterranean
climate of Western Europe, the main forcing linked to a side of the Iberian peninsula where precipitation increases in
Heinrich event is a cooling by 4 °C of the North Atlantic at the sensitivity experiment on colder ocean surface temper-
the mid-latitudes. We used the LMDZ atmospheric general atures at the mid-latitudes of the North Atlantic. This
circulation model (developed at the Laboratoire de anomaly, in opposition to the existing reconstructions for
Météorologie Dynamique—Dynamic Meteorology Labora- this zone (Combourieu-Nebout et al. 2002; Kageyama et al.
tory, Paris) to estimate the impact of such an anomaly in the 2005), shows the limitations of this experiment. The
context of glaciation, in this case, the Last Glacial Maxi- hypothesis that the factor responsible for the changes in
mum. We therefore carried out two experiments by forcing climate in western Europe was the differences in surface
the atmospheric model with glacial boundary conditions (see temperatures of the North Atlantic Ocean at mid-latitudes is
http://pmip1.lsce.ipsl.fr): ice sheets as reconstructed for the not sufficient to explain all the climate anomalies recon-
LGM, concentrations of greenhouse gases as measured for structed around the Mediterranean region. A first explana-
this period from ice cores, orbital parametres as they were tion could be that the differences in surface temperature of
21,000 years ago. In the first experiment, we used the sur- the Mediterranean Sea were not taken into account. A sec-
face temperatures of the oceans as reconstructed by the ond explanation could simply be that the model is not able to
CLIMAP project (Chap. 21). The only difference in the simulate the climate correctly for this region. A third
second experiment is the surface temperatures of the oceans explanation could be an error in the interpretation of the
in the North Atlantic. In this experiment, we decrease these records. In a case like this, it is very instructive for both
temperatures by 4 °C between 40 and 50 °N. The second modelers and palynologists to compare their data. It is also
experiment is a sensitivity experiment to the ocean surface for this reason that the results of the models should be
temperatures of the North Atlantic at the mid-latitudes. analyzed not only from the point of view of surface climate
Kageyama et al. (2005) show the results of this experi- variables but also from the point of view of the circulation
ment for climate in France and the Iberian Peninsula. Fig- and physics of the atmosphere, was this makes it possible to
ure 29.5 summarizes their results. The model does not identify the mechanisms responsible for the simulated
simulate a propagation of the cooling imposed in the North changes in climate and to improve the model and future
Atlantic very far inland over the European continent. The experiments.
place on the European continent where this cooling is most These experiments allow the influence of the various
important in terms of temperature of the coldest month is in mechanisms suggested above to be quantified: propagation
the northwest of the Iberian Peninsula. It brings about an of the ocean temperature anomaly to the adjacent continents
increase of only 1 °C in this region, which is low compared by the mean circulation, the shift of the prevailing winds
to the 4 °C imposed in the North Atlantic. On the other following the migration of the zones with a strong meridian
hand, the precipitation anomaly simulated by the model in temperature gradient, the impact of atmospheric instabilities
response to the cooling imposed in the North Atlantic is and average temperature on precipitation. In this way, a
much greater: it reaches −200 mm/year (a shortfall of model can help interpret reconstructions and, in particular,
200 mm/year, a drop of about 30%) over the Iberian can confirm or rebut scenarios established using several
Peninsula. Examining the results of the model for the North records located far away from each other.
Atlantic and Europe, we see that the band of strong westerly It is important to be aware of the limitations of experiments
winds is shifted southward, contributing to the decrease of of this type: regions where the anomaly is imposed, lack of
precipitation over Europe. The slight precipitation increase vegetation or ocean feedback, assumptions in the model
over northwestern Africa can also be partly attributed to this design and experimentation. These aspects can be evaluated
atmospheric circulation change. by additional sensitivity experiments (see Chap. 25).
29 Rapid Climate Variability: Description and Mechanisms 415

Fig. 29.5 Simulated climate variables for the LGM and the Heinrich 1 event (H1). Left column: LGM results: annual temperatures (°C), annual
precipitation (mm/year), surface zonal wind (m/s). Right column: H1—LGM anomalies (simulations described in Kageyama et al. 2005)

Dansgaard-Oeschger Cycles The component of the climate system most likely to be

The Dansgaard-Oeschger cycles modify the climate on a the cause of the Dansgaard-Oeschger events is undoubtedly
global scale (see Voelker (2002) for a review). Therefore, the ocean, and more particularly ocean circulation. It has two
the underlying mechanism must either have a global origin main characteristics of interest here: it reacts on a global
as well, or else must modify a major element of the climate scale, on time scales compatible with the Dansgaard-
system, which has a global impact. We will now consider Oeschger cycles, and it can respond to a relatively local
three possible mechanisms that fulfil these requirements. forcing. Studies such as Ganopolski and Rahmstorf (2001)
One of the simplest ways to obtain a global response is a have shown that a semi-periodic freshwater forcing can,
variation in a forcing external to the climate system. It could be under certain conditions, reproduce oscillations that resem-
a mechanism operating on long time scales (Milankovitch ble the cycles observed in the paleoclimate data. However,
forcing) or on short ones (volcanic changes, for example). It this does not provide any indication of the origin of this
has therefore been proposed that variations in solar forcing freshwater forcing and therefore gives no new explanation as
may be responsible for Dansgaard-Oeschger events (Braun to the mechanism at work.
et al. 2005). However, the relationship between the small To compensate for this lack of a proven periodic climate
variations in the solar cycle at the relevant highlighted times forcing, some authors propose mechanisms with oscillations
and the observed climate variations remain questionable. internal to the climate system. One of the most interesting
416 M. Kageyama et al.

explanations today relies on an internal oscillation of oceanic The Observations

circulation under given boundary conditions (Paillard 2004),
without involving any external forcing. The big advantage of The 8200-year event has been recorded in numerous climate
this hypothesis is that it accounts both for the frequency of archives, including the Greenland polar ice cap, continental
cycles, compatible with what we know of the ocean, and for archives (lakes, vegetation, speleothems) and the ocean.
their global character. A coupled ocean-atmosphere internal In Greenland cores, this event shows up as a decrease in
oscillation has more recently been proposed as the reason for d18O of about two permil over a period of about 100 years.
the occurence of Dansgaard-Oeschger events as well (Peltier This decrease is interpreted as a drop in local temperature in
and Vettoretti 2014). This study has the advantage of Greenland of 6 ± 2 °C (see Thomas et al. 2007). It was
showing such an internal oscillation of the climate system in estimated at 7.4 °C using an independent method based on
a more realistic General Circulation Model. oxygen isotopes (d15N) (Kobashi et al. 2007). This was a
Nevertheless, opposing theories coexist nowadays when fairly dramatic event, at least over Greenland. This oscilla-
trying to explain the triggering mechanisms of Dansgaard- tion is found in all the cores of adequate resolution obtained
Oeschger events. Banderas et al. (2015) also relate their from the Greenland ice sheet: GRIP, GISP2, NGRIP, Dye-3.
occurence to an internal climatic oscillation, but in their This is therefore an event with a regional impact, at least, on
case, it is controlled by the effects of Southern-Ocean wind the scale of Greenland.
stress and CO2 on the oceanic circulation. Similarly, Dima The cold period equating to the 8200-year event in
et al. (2018) proposed that the meridional overturning cir- Greenland is also recorded in other paleoclimate indicators.
culation changes are ultimately controlled by the Southern Although this event was not recognized as such in polar ice
Ocean with links to gradual global climate changes. Besides, records, it was well identified in records from Norwegian
it has recently been proposed that Dansgaard-Oeschger lakes and in the extended ice sheets reconstructed from the
events owe their existence to the interactions between the end moraines. Indeed, the cold and dry period ‘Finse1’ was
North Atlantic ice shelves, sea ice and oceanic circulation identified (Dahl and Nesje 1994), and then connected to the
(Boers et al. 2018). 8200-year event in the ice cores. Simultaneously, this event
To conclude, there is no consensus today on the cause of was associated with a cold spell reconstructed from varia-
the Dansgaard-Oeschger events, despite an increasing tions in planktonic foraminiferal fauna in a marine core from
understanding of the effects that these events have had on the south of Norway. This record proves that it is indeed a
global climate. significant climate event on the scale of the North Atlantic,
but also that the ocean was affected almost simultaneously.
This interpretation is reinforced by other more distant
Sudden Events During Interglacial Periods records of the event. One of the first studies to show this was
the study of the d18O of the ostracod shells from a Bavarian
The Discovery lake (Ammersee, Germany), which recorded changes in the
d18O of the precipitations associated with cooling (Chap. 15,
Although the climate oscillation called ‘the 8200-year event’ von Grafenstein 1998).
left traces in the Greenland cores, the first publications In the oceans, records marking the 8200-year event are
presenting the isotopic records from the ice in Greenland rare, as identification requires a temporal resolution of much
from GRIP cores (Dansgaard et al. 1993) described the cli- less than a century. Of these records, those from the northern
mate of the Holocene as stable, as the weak oscillation seas show a decrease of about 3 °C in ocean surface tem-
occurring around 8200 years before today was considered to perature. This study proves that this event was the largest
be background noise. Even the discovery of a second record during the entire Holocene in this region. Finally, these
with exactly the same oscillation did not allow the correct authors show that the oceanic change affected not only the
recognition of this climate event. It was not until four years surface but also the deep ocean. These last changes have
later that the notion of ‘the 8200-year event’ was introduced been confirmed recently in a core from southern Greenland,
in an article by a team led by R. Alley. Efforts to place this which shows changes in the formation of deep water bodies
climate evolution event on a hemispheric scale have indi- in the North Atlantic.
cated that this event had in fact been identified since 1973 On the continent, the 8200-year event is expressed as a
from other records. cooling especially evident in Europe. Studies of the changes

1
Named after the Finsevatn, the Norwegian Hardanger Lake where the
record originated.
29 Rapid Climate Variability: Description and Mechanisms 417

in the level of European lakes indicate that during this event than is accepted today. Even more importantly, it revealed
the climate was wetter between 50 and 43° N, while north disparities in the seasonal climate response, with a greater
and south of this zone, it was drier. Pollen records rarely anomaly in summer than in winter over northern Europe and
show this event, perhaps because these climate variations a north-south bipolarity, with cooling in the north and
were not significant enough to modify the vegetation in a warming at 60° S. Subsequent studies confirmed that a
major way, or because it did not affect the growing period of freshwater influx, even one lasting only one year, was suf-
the main plants characteristic of these paleoenvironments, as ficient to obtain a climate response consistent with the one
was shown in Northern Europe and North America. obtained from the analysis of paleoclimate indicators. The
climate impact was not immediate, as can be seen in
Fig. 29.6. For example, in Greenland, the maximum point
The Mechanisms was reached after about thirty years.
Other authors have also highlighted the role of climate
The scale of changes observed as well as the multiplicity of ‘noise’ in the duration of the ocean’s response to the
their impacts (precipitation, temperature, oceanic circula- freshwater forcing. Indeed, they show that by adding noise
tion), very quickly pointed to the Atlantic thermohaline (with an average value of zero) to the freshwater forcing, the
circulation as a major actor in this abrupt change in climate. duration of the response by the model can be changed from a
The experience of Heinrich events (see above) suggested few decades to two or three centuries. This highlights the
that here too, a sudden influx of fresh water to the ocean importance of the initial climate state to the response to a
could have altered the thermohaline circulation and the cli- given forcing, as the response can have a more or less global
mate around the North Atlantic. However, 8200 years ago, effect as was the case for the 8200-year event, or have a local
only a small part of the Laurentide ice sheet could have been impact on a decadal scale. The 8200-year event was also
involved, with the disappearance at this time of the dome reproduced in a coupled general circulation model incorpo-
covering what is currently the Hudson Bay. Traces of rating water isotopes (Legrande et al. 2006), demonstrating
paleo-shorelines of lakes also indicate the disappearance of that the results were consistent not only with temperature
two large pro-glacial lakes (the Ojibway and Agassiz lakes) estimates but also with the paleoclimate indicators them-
around this period, which could have contributed an influx selves, directly simulated in the model.
of fresh water to the ocean. However, the first 14C datings These authors were able to correctly predict the variations
did not enable a precise chronology of these events. More in d18O associated with the 8200-year event from the calcite
precise dating, as well as an evaluation of the volume of marine sediments before the publication of the first data
released, make it possible to pinpoint the massive draining of showing the 8200-year event in marine sediment cores.
Lake Agassiz as a cause of this cold event (Clarke et al. It is important to note that although the 8200-year event is
2004). These data indicate a draining of about 100,000 km3 considered an abrupt event during the interglacial period,
of water in about a year, thus causing a massive influx to the this classification is not quite accurate. In fact, it is a final
ocean. manifestation of the glacial climate through the draining of a
periglacial lake, the result of a melting ice sheet . To date, no
abrupt temperature changes involving glacial ice sheet
Modeling and the Global and Hemispheric behavior have been observed during the Holocene.
Consequences

Nevertheless, can the consequences for the climate of this Outlook

extremely abrupt but yet short-lived influx be replicated in
coupled climate models? Although the impact of this huge Global Connections
influx to the ocean was quickly identified, the first simulation
of the 8200-year event using boundary conditions consistent The work of the last thirty years has made it possible to
with the time period was achieved in the early 2000s demonstrate a variability in the climate system on millennial
(Renssen et al. 2001). These authors were able to show that time scales, characterized by transitions occurring, in some
by forcing an ocean general circulation model coupled with cases, over only a few decades. This variability was first
a simplified atmospheric model using an idealized water flux observed in North Atlantic marine sediment cores for
of 0.75 Sv for twenty years, they produced a slowing of the Heinrich events, and in Greenland ice cores for Dansgaard-
thermohaline circulation in the Atlantic for about three Oeschger events. Gradually, climate records have been
centuries, with a temperature drop of 1–5 °C on the conti- analyzed at increasingly fine temporal resolutions and vari-
nents bordering the North Atlantic. This result fits well with ations with temporal characteristics similar to those in the
the data, although the freshwater forcing is relatively larger North Atlantic and Greenland have been discovered in areas
418 M. Kageyama et al.

Time (years)
-200 0 200 400

Fresh water -24

pulse
Ensemble results for
simulated temperature
at Greenland summit
-26

Temperature (ºC)
-28

1
-30

0
-32
δ18O anomaly (‰)

Decadal average of
cooling from nitrogen
-1 isotopes from GISP2:
~3.3 +/- 1.1 ºC

δ18O anomaly of
-2 combined
NGRIP, GRIP
and DYE-3 ice
cores

-3

8500 8400 8300 8200 8100 8000

Time (yearsB2K)

Fig. 29.6 Simulated surface air temperatures for the summit of to the model results, by aligning the first decrease in d18O with the first
Greenland for a set of ten simulations (gray). The average is drawn decrease in simulated temperature. The temperature axis is selected so
in black. These results are compared with the reconstructed d18O profile that the decrease of 3.3 °C reconstructed from the nitrogen isotopes in
from measurements from Dye-3, GRIP and NGRIP ice cores. The same the GISP2 core corresponds to the minimum anomaly in d18O.
scale is used for the time axes, the d18O curve is positioned in relation Adapted from Wiersma et al. (2011)

very far from the original regions, as far away as the South advances made in methods of relative synchronization (such
Pacific and Antarctica, and including China and the northern as methane concentration in ice cores or paleo-magnetic
Indian Ocean, influenced by monsoons. The interpretation of properties in marine cores) for the scientific community
these data is based on the assumption that the time scales seeking to characterize rapid variability.
associated with each record can be accurately synchronized. A comprehensive description on a global scale is very
This crucial synchronization is still a challenge despite important for our understanding of the climate and of the
29 Rapid Climate Variability: Description and Mechanisms 419

variations that may have been experienced contemporane- large longer term variations. In particular, these events are
ously in parts of the globe very distant from each other. more numerous and of greater amplitude in glacial periods
Finding the explanation for these connections is also a than in interglacial periods. This suggests a different
challenge faced by modellers, because attempting to model expression of millennial variability depending on the size of
these events means testing our knowledge of climate as it is the ice sheets in the northern hemisphere. The
interpreted in the models which are also used to predict high-resolution records of the most recent climate transitions
future climates. In terms of modeling, the first experiments (the deglaciation between LGM and the Holocene as well as
testing the sensitivity of the ocean-atmosphere system to the entry into the last glaciation) may provide information on
freshwater discharges in the North Atlantic in a glacial the conditions favoring events of larger amplitude. However,
context have been carried out in the years 2000. these transitions are punctuated by abrupt climate events.
These experiments, if they are to be carried out with a This raises the question of the role that these events might
resolution sufficiently fine to enable comparison between play in the transition itself, knowing that a rapid event can
model results and reconstructions, require a calculation time have, in the case of Dansgaard-Oeschger events in the
at the forefront of the capability of the most powerful Greenland cores, an amplitude equivalent to half the differ-
modern computers. For example, Kageyama et al. (2009), ence between glacial and interglacial states.
using an ocean-atmosphere general circulation model, show Let us consider the case of the last deglaciation
that the link between the cooling in the North Atlantic and (Fig. 29.7). A few thousand years after the LGM came the
the decrease in the Indian monsoon can be explained by a Heinrich H1 event. Climate conditions returned to an almost
decrease in the summer temperature gradient in the upper glacial level. This event was followed by a warm phase, with
half of the troposphere between the Indian Ocean and the the transition between these two events showing up as abrupt
Tibetan plateau. However, these results are obtained for in many records, particularly around the North Atlantic. This
conditions corresponding to the LGM and not to Marine is the Bølling-Allerød phase, whose climate is almost
Isotopic Stage 3, and further experiments will be required to interglacial. However, this period was followed by the cold
better understand the signal transmission mechanisms period of the Younger Dryas, which is sometimes considered
between the North Atlantic and Asia. Furthermore, although to be the most recent Heinrich event (H0), which is wrong
these experiments reproduce certain climate events con- because there is no corresponding layer of detritic elements
temporaneous with the Heinrich events (southward migra- in the Ruddiman belt. It is after this last cold phase that the
tion of the Intertropical Convergence Zone, reduction of the climate of the current interglacial, the Holocene, became
Indian monsoon, cooling and drying in Western Europe), definitively established (apart from the 8200-year event).
they do not manage to reproduce others, such as variations in This shows that the last deglaciation was not a
the Southeast Asian monsoon. The modeling and under- smooth transition. On the contrary, it is a series of abrupt
standing of the millennial variability within the climate events, as if the climate system ‘hesitated’ between two
system thus remains in many respects a challenge for equilibria, one glacial and the other interglacial. The role of
modellers. abrupt events during this climate transition is therefore
We have, until now, attempted to describe climate vari- important, but they still need to be understood and modeled.
ability on the millennial scale mainly in terms of the Entry into the last glaciation at the end of the Eemian is
“physical” Earth system, that is, including the ocean, the also characterized by the appearance of abrupt events,
atmosphere and the cryosphere. However, the records con- *110 ka before today, in a context where glacial ice sheets
tain more information than this simple description. Take, for had already developed over Canada. One might think that
example, the case of methane concentrations. We have these cold events would promote entry into glaciation, but
shown that they are useful to synchronize the records from this assumption ignores the fact that cooler air at high lati-
Greenland and Antarctica but this important greenhouse gas tudes also contains less water and is therefore less able to
emitted mainly from the wetlands in the tropics and high supply the water needed to build the ice sheets up at a sig-
latitudes shows significant variations. The role of biogeo- nificant rate. These compensating factors need to be asses-
chemical phenomena in millennial climate variability also sed. Here again, the influence of abrupt events on the
needs to be better understood and modeled. evolution of the ice sheets has yet to be assessed, in com-
parison with other mechanisms and feedbacks, such as the
slower changes in state of the ocean, cryosphere, vegetation,
Abrupt Event Interactions—Large Climate atmospheric concentrations of greenhouse gases, as well as
Transitions external changes such as changes in insolation. Models of
the Earth system can lead to a better understanding of the
Although abrupt events are the result of internal variability reconstructed signals by conducting sensitivity experiments
in the climate system, they are nonetheless sensitive to its for each of these factors.
420 M. Kageyama et al.

Proceedings of the National Academy of Sciences, 115(47),

E11005–E11014.
Bond, G., et al. (1993). Correlations between climate records from
North Atlantic sediments and greenland ice. Nature, 365, 143–147.
Braun, H., Christl, M., Rahmstorf, S., Ganopolski, A., Mangini, A.,
Kubatzki, C., et al. (2005). Possible solar origin of the 1470-year
glacial climate cycle demonstrated in a coupled model. Nature, 438
(7065), 208.
Broecker, W., et al. (1992). Origin of the northern Atlantic’s Heinrich
Greenland events. Climate Dynamics, 6, 265–273.
Calov, R., et al. (2002). Large-scale instabilities of the Laurentide Ice
Simulated T sheet simulated in a fully coupled climate-system model. Geophys-
ical Research Letters, 29, 2216. https://doi.org/10.1029/
Sea level 2002gl016078.
Clarke, G. K. C., et al. (2004). Paleohydraulics of the last outburst flood
from glacial lake agassiz and the 8200 BP cold event. Quaternary
Science Reviews, 23, 389–407. https://doi.org/10.1016/j.quascirev.
2003.06.004.
Combourieu-Nebout, N., et al. (2002). Enhanced aridity and atmo-
spheric high-pressure stability over the western mediterranean
during the North Atlantic cold events of the past 50 k.y. Geology,
AntarcƟca 30, 863–866.
Cortijo, E., et al. (2005). Heinrich events: Hydrological impact.
Comptes Rendus Geoscience, 337, 897–907. https://doi.org/10.
1016/j.crte.2005.04.011.
Years before present Dahl, S. O., & Nesje, A. (1994). Holocene glacier fluctuations at
hardangerjøkulen, Central-Southern Norway: A high-resolution
Fig. 29.7 An overview of the last deglaciation. From top to bottom: composite chronology from lacustrine and terrestrial deposits. The
atmospheric CO2 as measured in air bubbles from Antarctic ice cores; Holocene, 4, 269–277.
d18O measured in Greenland ice (GRIP members 1993) as a first-order Dansgaard, W., et al. (1993). Evidence for general instability of past
indicator of local temperature, superimposed with the simulated average climate from a 250-kyr ice-core record. Nature, 364, 218–220.
annual temperature in Greenland from the iLOVECLIM model; the Dima, M., Lohman, G., & Knorr, G. (2018). North Atlantic versus
scaling between model and data implies that the sudden increase of global control on dansgaard‐oeschger events. Geophysical Research
d18O at the HE1/B/A limit is 10 °C. The periods corresponding to the Letters.
Heinrich 1 event (HE1), Bolling-Allerod (B/A) and the Younger Dryas Elliot, M., et al. (2002). Changes in North Atlantic deep-water
(Y/D) are indicated below. It may be noted that there is no abrupt event formation associated with the dansgaard-oeschger temperature
in the temperature simulated by the climate model. Sea-level (Peltier oscillations (60–10 ka), Quaternary Science Reviews, 21, 1153–
and Fairbanks 2006) (black triangles) in meters relative to the current 1165.
sea level, as reconstructed from corals with the abrupt event “Meltwater EPICA community members. (2006). One-to-one coupling of glacial
Pulse-1A” (MWP-1A) indicated, representing a rise of about 20 m in climate variability in Greenland and Antarctica. Nature, 444, 195–
*300 years. Isotopic abundance of deuterium in Antarctic ice (Peltier 198.
and Fairbanks 2006) as a first-order indicator of local temperature with Ganopolski, A., & Rahmstorf, S. (2001). Rapid changes of glacial
the Antarctic Cold Reversal (ACR) indicated, a cold spell interrupting climate simulated in a coupled climate model. Nature, 409, 153–
the warming of the deglaciation 158. https://doi.org/10.1038/35051500.
Genty, D., et al. (2005). Rapid climatic changes of the last 90 kyr
recorded on the European Continent. Comptes Rendus Geoscience,
337, 970–982.
References Greenland Ice-core Project (GRIP) Members. (1993). Climate instabil-
ity during the last interglacial period recorded in the GRIP ice core.
Alvarez-Solas, J., Robinson, A., Montoya, M., & Ritz, C. (2013). Nature, 364, 203–207. https://doi.org/10.1038/364203a0.
Iceberg discharges of the last glacial period driven by oceanic Grimm, E. C., et al. (2006). Evidence for warm wet heinrich events in
circulation changes. Proceedings of the National Academy of Florida. Quaternary Science Reviews, 25, 2197–2211. https://doi.
Sciences, 110(41), 16350–16354. org/10.1016/j.quascirev.2006.04.008.
Banderas, R., Alvarez-Solas, J., Robinson, A., & Montoya, M. (2015). Grousset, F. E., et al. (1993). Patterns of ice-rafted detritus in the
An interhemispheric mechanism for glacial abrupt climate change. Glacial North Atlantic. Paleoceanography, 8, 175–192.
Climate Dynamics, 44(9–10), 2897–2908. Heinrich, H. (1988). Origin and consequences of cyclic ice rafting in
Barker, S., Chen, J., Gong, X., Jonkers, L., Knorr, G., & Thornalley, D. the Northeast Atlantic ocean during the past 130000 years.
(2015). Icebergs not the trigger for North Atlantic cold events. Quaternary Research, 29, 142–152.
Nature, 520(7547), 333. Hemming, S. R. (2004). Heinrich events: Massive late pleistocene
Bassis, J. N., Petersen, S. V., & Mac Cathles, L. (2017). Heinrich detritus layers of the North Atlantic and their global climate imprint.
events triggered by ocean forcing and modulated by isostatic Review of Geophysics, 42, RG1005.
adjustment. Nature, 542(7641), 332. Hulbe, C. L., MacAyeal, D. R., Denton, G. H., Kleman, J., & Lowell,
Boers, N., Ghil, M., & Rousseau, D. D. (2018). Ocean circulation, ice T. V. (2004). Catastrophic ice shelf breakup as the source of
shelf, and sea ice interactions explain Dansgaard-Oeschger cycles. Heinrich event icebergs. Paleoceanography, 19(1).
29 Rapid Climate Variability: Description and Mechanisms 421

Kageyama, M., et al. (2005). Le Dernier Maximum glaciaire et INTIMATE event stratigraphy. Quaternary Science Reviews, 106,
l’événement de Heinrich 1 en termes de climat et de végétation 14–28. https://doi.org/10.1016/j.quascirev.2014.09.007.
autour de la mer d’Alboran: une comparaison préliminaire entre Renssen, H., et al. (2001). The 8.2 Kyr BP event simulated by a global
modèles et données. Compte Rendus Geoscience, 337, 983–992. atmosphere-sea-ice-ocean model. Geophysical Research Letters, 28,
Kageyama, M., et al. (2009). Glacial climate sensitivity to different 1567–1570. https://doi.org/10.1029/2000gl012602.
States of the Atlantic meridional overturning circulation: Results Ritz, C., Rommelaere, V., & Dumas, C. (2001). Modeling the evolution
from the IPSL model. Climate of the Past, 5, 551–570. of Antarctic ice sheet over the last 420,000 years: Implications for
Kissel, C. (2005). Magnetic signature of rapid climatic variations in altitude changes in the Vostok region. Journal of Geophysical
Glacial North Atlantic, a review. Comptes Rendus Geoscience, 337, Research: Atmospheres, 106(D23), 31943–31964.
908–918. https://doi.org/10.1016/j.crte.2005.04.009. Roberts, W. H. G., et al. (2014). A new constraint on the size of
Kobashi, T., Severinghaus, J. P., Brook, E. J., Barnola, J. M., & Heinrich Events from an iceberg/sediment model. Earth and
Grachev, A. M. (2007). Precise timing and characterization of Planetary Science Letters, 386, 1–9.
abrupt climate change 8200 years ago from air trapped in polar ice. Roche, D. M., & Paillard, D. (2005). Modelling the oxygen-18 and
Quaternary Science Reviews, 26(9–10), 1212–1222. rapid glacial climatic events: A data-model comparison. Comptes
Legrande, A., et al. (2006). Consistent simulations of multiple proxy Rendus Geoscience, 337, 928–934. https://doi.org/10.1016/j.crte.
responses to an abrupt climate change event. Proceedings of the 2005.03.019.
National Academy of Sciences of the United States of America, 103, Roche, D. M., et al. (2004). Constraints on the duration and freshwater
837–842. https://doi.org/10.1073/pnas.0510095103. release of Heinrich event 4 through isotope modelling. Nature, 432,
MacAyeal, D. R. (1993). Binge/purge oscillations of the Laurentide ice 379–382. https://doi.org/10.1038/nature03059.
sheet as a cause of the North Atlantic’s heinrich events. Paleo- Ruddiman, W. (1977). Late quaternary deposition of ice-rafted sand in
ceanography, 8, 775–784. the subpolar North Atlantic (lat 40° to 65° N), GSA Bulletin, 1813–
Marcott, S. A., Clark, P. U., Padman, L., Klinkhammer, G. P., Springer, 1827.
S. R., Liu, Z., et al. (2011). Ice-shelf collapse from subsurface Sánchez-Goñi, M. F., et al. (2002). Synchroneity between marine and
warming as a trigger for Heinrich events. Proceedings of the terrestrial responses to millennial scale climatic variability during
National Academy of Sciences, 108(33), 13415–13419. the last glacial period in the Mediterranean region. Climate
Paillard, D. (2004). Modelling rapid events within the climate system. Dynamics, 19, 95–105. https://doi.org/10.1007/s00382-001-0212-x.
Comptes Rendus Geoscience, 336, 733–740. https://doi.org/10. Shackleton, N. J., et al. (2000). Phase relationships between
1016/j.crte.2003.12.019. millennial-scale events 64,000–24,000. Paleoceanography, 15(6),
Peltier, W. R., & Fairbanks, R. G. (2006). Global glacial ice volume 565–569.
and Last Glacial Maximum duration from an extended Barbados sea Thomas, E. R., et al. (2007). The 8.2 Ka event from greenland ice
level record. Quaternary Science Reviews, 25, 3322–3337. cores. Quaternary Science Reviews, 26, 70–81. https://doi.org/10.
Peltier, W. R., & Vettoretti, G. (2014). Dansgaard-Oeschger oscilla- 1016/j.quascirev.2006.07.017.
tions predicted in a comprehensive model of glacial climate: A Voelker, A. H. L., et al. (2002). Global distribution of centennial-scale
“kicked” salt oscillator in the Atlantic. Geophysical Research records for marine isotope stage (MIS) 3: A database, Quaternary
Letters, 41(20), 7306–7313. Science Reviews, 21, 1185–1212.
Rasmussen, T. L., et al. (1996). Rapid changes in surface and deep von Grafenstein, U., et al. (1998). A Mid-european decadal
water conditions at the faeroe margin during the last 58,000 years. isotope-climate record from 15,500 to 5000 years BP, Science,
Paleoceanography, 11, 757–771. https://doi.org/10.1029/ 284, 1654–1657.
96PA02618. Wiersma, A. P., et al. (2011). Fingerprinting the 8.2 Ka event climate
Rasmussen, S. O., et al. (2014). A stratigraphic framework for abrupt response in a coupled climate model. Journal of Quaternary
climatic changes during the Last Glacial period based on three Science, 26, 118–127. https://doi.org/10.1002/jqs.1439.
synchronized Greenland ice-core records: refining and extending the
 An Introduction to the Holocene
and Anthropic Disturbance 30
Pascale Braconnot and Pascal Yiou

(Fig. 30.1), and 5 W/m2 higher about 3000 years ago. Also,
The Major Trends of the Holocene
the amplitude of the seasonal cycle of insolation at the
beginning of the Holocene was greater in the northern
The Different Radiative Disturbances
hemisphere but less in the southern hemisphere. This vari-
ation in amplitude is not symmetrical on either side of the
The Holocene started about 10,000 years ago at the end of
equator, with greater variations in the northern hemisphere
the last glaciation. The last thousand years of this period is
and in the tropics (Fig. 30.1). The precession also alters the
marked by the growing impact of human activity (changes in
length of the seasons. According to Kepler’s laws, the boreal
land use and atmospheric composition). At first sight, the
summer, defined as the time between the spring and autumn
numerous data show us variations that are less spectacular
equinoxes, lasted 172 days 9500 years and 176 days
than the great upheavals engendered by deglaciation. Nev-
6000 years ago with 180 days currently. In the northern
ertheless, the general natural trend, driven by changes in
hemisphere, summer insolation was therefore more intense
solar radiation at the top of the atmosphere, is characterized
over a shorter period.
by radical changes in the monsoon and the El Niño phe-
Greenhouse gases, volcanism and the solar constant are
nomenon in the tropics. In the mid-latitudes of the northern
other factors that have influenced the climate of the Holo-
hemisphere, the changes are seen in the characteristics of the
cene through their impact on the radiative balance
main modes of variability. Several abrupt events also
(Fig. 30.2). These are also the dominant factors of the last
punctuate the unfolding of the process.
2000 years, ever since insolation has reached approximate
Slow variations in solar radiation at the top of the
current values. Indeed, the combined effect of these green-
atmosphere, caused by variations in orbital parameters, are
house gases has resulted in a reduction in the radiative
the driving force behind the evolution of the main climate
balance of the planet of around 0.5 W/m2 from the begin-
characteristics on this 10,000-year scale. Variation in the
ning of the Holocene to the beginning of the industrial era.
obliquity, from 24.23° at the beginning of the Holocene to
This slight perturbation of the radiative balance is due to a
23.44° at the present time, has brought about an increase in
7 ppm increase in atmospheric carbon content at the
the average annual solar radiation at high latitudes of
beginning of the Holocene, followed by a decrease of
1.5 W/m2 over this period. During the same period, solar
20 ppm up to the beginning of the industrial era. Methane
radiation in the low latitudes dropped by 1.10 W/m2. Added
levels decreased from 730 ppb at the beginning of the
to this effect of obliquity is the precession of the equinoxes.
Holocene to 580 ppb in the Middle Holocene (6000 years
At the beginning of the Holocene, the summer solstice was
ago), and gradually returned to early Holocene levels in the
located at the perihelion of the ecliptic. It is now at the
pre-industrial era. Levels of atmospheric N2O follow the
aphelion. Thus, the solar radiation received at the top of the
variations of CO2 and have varied between 2 and 10 ppb. In
atmosphere (insolation) in June was 48 W/m2 higher at
more recent times, the evolution of greenhouse gases is
60° N at the beginning of the Holocene than it is today
dominated by anthropic emissions. The combined effects of
human activity correspond to an increase of 1.6 W/m2 in
the radiative balance. This estimate takes into account the
P. Braconnot (&)
P. Yiou
Laboratoire des Sciences du Climat et de l’Environnement
dominant effect of the increase in greenhouse gases and the
LSCE/IPSL, CEA-CNRS-UVSQ, Université Paris-Saclay, negative contribution of aerosols for the twentieth century
91191 Gif-sur-Yvette, France (Solomon et al. 2007; IPCC 2013).
e-mail: [email protected]

© Springer Nature Switzerland AG 2021 423

G. Ramstein et al. (eds.), Paleoclimatology, Frontiers in Earth Sciences,
https://doi.org/10.1007/978-3-030-24982-3_30
424 P. Braconnot and P. Yiou

There are many proofs of volcanic activity during the

Holocene, but there is no quantified series for the overall
period. Data has become more accurate for the last century,
although the eruption timing, geographic location and
quantities of aerosols emitted are not known with precision.
Thus, depending on the sources of information, the ampli-
Insolation

tude and the date of the Holocene eruptions vary greatly.

The Evolution of Temperature in the Various

Records

The recorded variations in temperatures during the Holocene

years
reflect the response of the climate system to the various
Seasonal amplitude and month of the maximum radiative disturbances listed above, as well as other various
feedbacks, such as water vapor and melting snow and ice
caps. The climate optimum at the beginning of the Holocene
did not occur simultaneously at all latitudes (Fig. 30.3), as
insolation depends on latitude and the season and feedbacks
produce a delay between the forcing and its maximum
summer temperatures 10,000–8000 years ago. Reconstruc-
tions for the mid-latitudes of the northern hemisphere are
characterized by slow decreases in sea-surface temperature
(SST) during the Holocene. In North America, the hottest
period was between 7000 and 5000 years ago. Pollen data
and macrofossil remains (see Volume 1, Chap. 10 for
methodologies) indicate that temperate forests were found
years
further north than they are today and that glaciers had
Fig. 30.1 a Difference in insolation at 60°N between the Holocene retreated. The early warming at high southern latitudes at the
and the current period. b Change in the amplitude of the seasonal cycle beginning of the Holocene cannot be explained simply as a
during the Holocene, calculated as the difference between the month of
response to local insolation conditions, and appears to be a
maximum insolation and the month of minimum insolation (isolines,
W/m2), and months of maximum difference (shaded, number of the manifestation of a large-scale reorganization of atmospheric
month) and oceanic heat transport. Yet most tropical regions show a
gradual warming during the Holocene, reflecting an increase
Estimates of the forcings induced by changes in the solar in insolation in these regions.
constant and volcanism are more open to controversy The growing abundance of chronicles relating the impacts
because they are more difficult to measure directly. Esti- of climate extremes (due to the expansion of humanity)
mates from satellite data suggest that the differences between provides a description of the intra-annual climate variability
periods of solar activity and inactivity are due to fluctuations over the last 2000 years. This can be superimposed on the
of about 0.08% in solar irradiance (1 965 W/m2) over the secular fluctuations mentioned above. Figure 30.4 summa-
last twenty years. For older periods, cosmogenic isotopes, rizes the current state of knowledge of temperature changes
the number of sunspots and observations of the aurora since the year 700 AD. Taking uncertainties into account,
borealis are all indicators of solar activity (see Chap. 1, the temperature of the northern hemisphere is fairly stable
Volume 1). The amplitude of solar variations over the last between the year 700 and the end of the nineteenth century,
millennium has recently been revised downwards and cur- and begins to increase continuously from the twentieth
rent best estimates suggest an increase of between 0.05 and century onwards (Fig. 30.4c). The shape of this curve
1.2% in solar irradiance between the Maunder minimum resembles a hockey stick, hence its iconic name. Excluding
(between 1650 and 1720) and the current period, which the twentieth century, for which there is an abundance of
corresponds to a radiative disturbance of 0.1 to 0.3 W/m2. meteorological recordings, the value of the curve is that it
Very intense volcanic eruptions introduce sulfate aerosols shows a relatively warm phase between the year 1000 and
into the stratosphere, which reflect back solar radiation and 1300 (often called the ‘Medieval Warm Period’) and a rel-
thus contribute to a cooling of the climate in the year fol- atively cold phase between 1350 and 1850 (called the ‘Little
lowing the eruption. Ice Age’). The causes of the inter-annual variations in
30 An Introduction to the Holocene and Anthropic Disturbance 425

0
−1
−2
External forcings (W/m2)

−3
−4

1
0
−1
Volcanic activity
Solar activity (low estimate)
2

Solar activity (high estimate)

GHG + SO2
1
0

1000 1100 1200 1300 1400 1500 1600 1700 1800 1900 2000

Year AD
2
Fig. 30.2 Evolution of the estimates of radiative forcing (W/m ) since reconstructions); bottom panel: greenhouse gases (GHG) and sulfate
the year 1000 AD. Top panel: volcanic activity; middle panel: solar (SO2) aerosols (Jansen 2007)
activity (with low and high estimates from paleoclimate

Fig. 30.3 Epochs (in thousands

of years) and intensities of the
maxima of temperature
differences relative to the
pre-industrial era, by latitude. The
colors indicate the level of
variance from pre-industrial
levels (according to Jansen 2007)

temperature over the last two millennia are still not fully more modest effect on the radiative balance but one that is
understood, especially since climate reconstructions are much more durable, as the phases of solar activity
sometimes in disagreement for certain cold periods. (Fig. 30.2) last several decades. In particular, the solar
Large-scale volcanic eruptions cool the atmosphere by one irradiance minimum, the ‘Maunder Minimum’ occurring in
to two degrees, but these effects don’t last longer than two the middle of the seventeenth century corresponds to a
years (Fig. 30.2). On the other hand, solar activity has a marked cooling in the northern hemisphere.
426 P. Braconnot and P. Yiou

The End of the Humid Period in Africa

About 4500 years ago, dry tropical Africa experienced

variations in precipitation of much greater amplitude than it
has for the last century. Lake levels fluctuated by several
tens of meters and the surface area of Lake Chad reduced
from 350,000 to 5000 km2. Several ‘climate crises’, occur-
ring over just a few years, punctuated this period and led to
the complete drying up of lakes and the cessation of surface
runoff (Gasse 2000). Wind-carried sediment recorded in a
marine core off Mauritania suggests that a sharp change in
atmospheric circulation occurred off the coast of dry tropical
Fig. 30.4 Ensemble of reconstructions of the temperature anomalies Africa between 5500 and 4000 years BP (deMenocal 2000).
of the northern hemisphere since the year 700 (shaded areas) (Jansen This period therefore corresponds with the end of the Afri-
2007) and average measured temperature anomalies (solid line).
Anomalies of temperature are defined as differences with respect to a can Wet Period and the establishment of the current climate
modern climatology (here 1960–1990). The shading accounts for the conditions. The resulting aridification of the Sahara condi-
uncertainties of the reconstructions tioned the settlement of this area, where the activities of the
populations were mainly farming and breeding of animals
(Kropelin and Kuper 2006). However, rapid fluctuations do
not seem to have impacted on the different regions of the
Important Events of the Holocene Sahara-Sahel complex simultaneously, and signs of rapid
events range are found from 4500 years to 2000 years BP
Several events punctuate the history of the Holocene. Here depending on the latitude considered and the specific local
we will focus on the end of the humid period in Africa, about conditions (Fig. 30.5). For example, the Dahomey dry
4500 years ago, on the Medieval Warm Period, on the Little interval was established about 4000 years ago in the middle
Ice Age and on the first lasting traces of human activity. The of an area occupied by the Guineo-Congolese forest. The
cooling observed about 8 200 years before now is discussed forest, fed by underground galleries, disappeared between
in Chap. 15 of volume 1 and not here because it is consid- 4500 and 3500 years ago in northern Nigeria but only
ered to be linked to the end of deglaciation. 2000 years ago in the western Sahel. There are still many

21° North 3° South

Sahara Sahel Sudanian zone Guineo-Congolian zone
Radiocarbon chronology (B.P.)

Desert Grassland Sahelian wooded grassland Degraded forest, shrubland Forest

Fig. 30.5 Evolution of vegetation based on pollen data from different main transitions are not contemporaneous in the different locations
sites in Africa covering the transect between the Sahara at 21° N and (Courtesy of Anne-Marie Lézine)
the Guineo-Congolese zone in the south. This figure shows that the
30 An Introduction to the Holocene and Anthropic Disturbance 427

grey areas concerning this period. Nevertheless, as insolation The strongest cooling was undoubtedly in northern Europe,
seasonality varied gradually, with reduced insolation during with lakes being systematically frozen over in winter.
boreal summer, over the course of the Holocene, the rapid There are also signs of cooling in Equatorial America and
establishment of these dry conditions are the result of non- New Zealand (Bradley 1999). This relatively cold period (on
linear interactions between climates and ecosystems. average, 1 °C cooler than currently in France) was respon-
sible for the loss of crops and spikes in wheat prices, and
often for insurrections, as well as an increase in the number
Medieval Warm Period of witches burned alive in public squares (Le Roy Ladurie
1967). After 1850, the glaciers began to retreat and the
Through the study of historical archives, the British meteo- temperature to rise steadily, marking the end of the Little Ice
rologist Lamb reconstructed in the 1960s, the severity of the Age.
winters and the humidity of the summers since the year 1000 The Little Ice Age is marked by a few particularly cold
(Le Roy Ladurie 1967). He concluded that Europe experi- decades in Europe between 1650 and 1720. A decrease in
enced an escalation of dry summers during the period 1080– the number of sunspots was observed at this time. These
1200, unparalleled since then. He named this period the decades are called the Maunder Minimum, after the Amer-
Medieval Warm Period. It is important to note that there are ican astronomer Edward W. Maunder (1851–1928), who
no direct temperature measurements for this time, since the studied the relationship between sunspots and sun activity. It
thermometer did not yet exist. More recent studies (Jones should be noted that it was the German astronomer
and Mann 2004) suggest that Lamb’s conclusions cannot be Friederich W. G. Spörer (1822–1895) who first noticed the
generalized to the entire planet. decrease in the number of sunspots between 1650 and 1720.
From more extensive datasets, several studies have con- History bestowed the Spörer Minimum between 1420 and
firmed that many regions of the northern hemisphere expe- 1570 on him. The variation in temperature caused by the
rienced warmer conditions during the eleventh and twelfth direct influence of solar activity on the radiative balance at
centuries. However, several regions do not present signs of the Earth’s surface (0.5 W/m2) is less than 0.08 °C for the
having experienced this optimum, or else did at a different northern hemisphere (Jansen 2007) and is not sufficient to
time in history. For example, it appears that even though the explain a cooling of between 0.2 and 0.5 °C in the northern
summers were warmer at that time, winters were often very hemisphere.
harsh in Western Europe until 1170. These apparent con- The Little Ice Age was also marked by intense volcanic
tradictions, probably due to a small number of observations, eruptions that affected temperatures in a visible but not very
may be completed by new climate reconstruction programs durable manner (Fig. 30.4). Eruptions having a global effect
in the future. Meanwhile, it should not be ruled out that this generally took place in the tropics (Tambora, Krakatoa,
Medieval Warm Period was purely a regional phenomenon. Agung, Pinatubo etc.), and injected enough dust into the
It is also not a direct analogue to the current warming, as it stratosphere to be homogeneously distributed globally
was not associated with an increase in the atmospheric (Robock 2000). Extratropical volcanoes (e.g., Laki in Ice-
content of greenhouse gases. land in 1783, St Helens in the United States in 1980) have
had significant effects regionally in the northern hemisphere
due to dust transport in the troposphere. However, their
The Little Ice Age impact on a global scale was minor, because the volcanic
dust they emitted did not reach the stratosphere.
The Medieval Warm Period was interrupted at the beginning
of the fifteenth century by the arrival of the Little Ice Age,
which took hold until 1850. Again, there is no precise date The Anthropocene Era
for the beginning of this period: the Scandinavians felt a
cooling at the end of the nineteenth century, whereas in The Anthropocene is a recently devised term that refers to
France, the degradation of the weather was felt around 1430 the period when the climate and the environment became
(Bradley 1999; Le Roy Ladurie 1967). During this Little Ice influenced by human activity. It is accepted that this period
Age in France there was a growth of alpine glaciers, very started at the beginning of the industrial era (in the middle of
harsh winters and a succession of terrible summers. For the nineteenth century), which also coincides with the first
example, the Bossons glacier, near Chamonix, descended functioning weather service networks. It is possible, how-
1000 m lower than presently. Though possibly influenced by ever, to argue that man began to impact on the climate from
the artistic trends of the time, the study of old engravings and the middle of the Holocene. Land-clearing, the start of
paintings shows a striking difference with the present period. farming and slash-and-burn cultivation have emitted enough
428 P. Braconnot and P. Yiou

greenhouse gases into the atmosphere to be detectable in the Indirect Indicators

glacial and sedimentary archives (Ruddiman 2007). The climate has an influence on the evolution of the matu-
ration of certain fruits. This is especially the case for grapes
(vitis vinifera). In France, the harvesting dates for grapes have
Climate Reconstructions for the Holocene been recorded in parish or municipal records for several
centuries. These records provide valuable indicators of tem-
The Different Archives perature variations, as explained in Chap. 17 of Volume 1.
Similarly, there are series of harvest dates for other crops
The reconstruction of climate conditions during the Holo- in France dating back to the Middle Ages, series of the
cene poses several major challenges for the scientific com- bloom dates of fruit trees (e.g. the Kyoto series in Japan,
munity. The low amplitude of the variations in mean dating back to 800 AD). These historical phenological
temperature, compared to the amplitude observed during the indicators are responses to climate variations and provide
glacial ages, makes it difficult to use the usual indicators to information on the temperature conditions during the months
identify a signal. Moreover, the need to take sub-annual time preceding the harvest. For example, making the reasonable
scales into account to describe the significant events means assumption that wheat is harvested when the grain is mature
pushing the interpretation of many climate indicators to the (or shortly afterwards), an understanding of the vegetative
limit, or else finding new ones compared to the ones tradi- cycle of wheat and a knowledge of the date of harvest
tionally used in paleoclimatology. (which usually takes place at the beginning of summer)
The typology of paleoclimate indicators for the Holocene gives an idea of the spring temperatures for the region. An
allows them to be divided into two main categories: natural important point for the climatologist is to identify the species
archives and societal archives. The characteristics of these of wheat grown at that time so as to make an accurate
archival types are summarized in Table 30.1. estimate of the phenological cycle.
This section does not discuss the natural archives treated Societal archives are comprised mainly of chronicles of
in volume 1 (ice cores, marine cores, corals, lake sediments extreme climate or meteorological events (droughts, heat
etc.). The focus here is on societal archives, because their waves, storms, intense rain etc.) which had destructive
study requires an expertise not called on previously: that of impacts on agriculture or buildings. These records tend to
historians. contain such records as rogations, i.e. religious processions
carried out to ask for divine help when drought or persistent
Societal Archives rain threaten crops. The invoked saints are usually thanked
One big difference between the study of climate during the when the requested effect occurs. These registers provide
Quaternary and during recent centuries is that detailed and indications of the length of these climate episodes (Brazdil
quantitative records made by direct witnesses of the climate 2005; Garnier 2009).
events exist for recent times. Societal archives have been In farming accounts in newspapers and in journals kept
mined mainly by historians, most notably by Emmanuel Le by local scholars, we find chronicles describing more or less
Roy Ladurie, who pioneered the study of climate through precisely the influence of certain climate events on society:
history on a European scale (Le Roy Ladurie 1967). freezing days, ice jams on rivers, storms etc. This climate
The study of climate through history (or of historical information is obtained through accounts of the damage
climatology) relies on the study of old archives describing caused and the costs incurred.
chronicles, empirical measures, events (Brazdil 2005). The One of the reasons for the name ‘Little Ice Age’ is
most important aspect of this study is not necessarily the linked to the progress of the Alpine glaciers into the val-
production of databases that can be used by climatologists, leys that contained them from the fifteenth century on.
but the verification of sources and the demonstration of their Using dated indicators, such as moraines displaced as the
relevance to our understanding of climate. glaciers advanced, it is possible to have an idea of how the

Table 30.1 Types of climate Natural archives Societal archives

archives used to reconstruct the
climate of the Holocene Direct climate data None Narrative descriptions, instrumented
measurements
Indirect organic data Tree rings, pollens etc. Vegetation growth
Indirect non-organic Ice cores, sediments, glaciers, Precipitations, ice floes, frosts etc.
data corals
30 An Introduction to the Holocene and Anthropic Disturbance 429

volume of these glaciers evolved over the centuries. To get calculation, was also an influential politician of the Second
a better idea, climatologists also use graphical representa- Empire. Like his astronomer predecessors of the Observa-
tions by painters, and later by photographers. It is thus toire de Paris, Cassini, Maraldi and de la Hire, he was also
clear that after a spread to maximum size at the beginning interested in meteorology and, as a politician, in the strategic
of the nineteenth century, the majority of the Alpine gla- advantage that could be derived from weather forecasting.
ciers retreated by several kilometers over a century and a Following the disaster of Sevastopol in 1854, when the
half, with an acceleration of this retreat at the end of the Allied fleet was destroyed by a storm, Le Verrier claimed
twentieth century. that it was possible to predict this event through a network of
ad hoc meteorological observations. This marked the birth of
Direct Indicators centralized meteorological networks via the telegraph, which
Historians classify direct observations of climate and mete- would for a long time be under the control of the army in
orology since the year 1000 into several types of historical most countries in the world.
phases. These phases qualify the type, abundance and
quality of the information.
Before 1300, these were isolated accounts of extreme Statistical Methods for Climate Reconstruction
anomalies and natural disasters (Brazdil 2005; Le Roy
Ladurie 1967). These accounts describe in particular the Methods for obtaining the temperature curves from Fig. 4
ravages caused by natural disasters: destruction of crops, are based on statistical regressions between several cate-
buildings, floods, increased mortality etc. A detailed study of gories of climate indicators. The general concept behind
them, by cross-referencing sources and checking them, reconstructing a hemispheric temperature is to use a set of
makes it possible to track the chronology of extreme events. indicators (thickness of tree rings, isotopic concentrations,
From 1300 to 1500, more or less continuous descriptions pollen concentrations, harvest dates etc.) evenly distributed
of the character of summers and winters (and to a lesser over the hemisphere. This is called a ‘multi-proxy’ approach
extent those of spring and autumn) become available, con- because it uses a mix of several types of climate indicators
taining indications of everyday conditions. (referred to as proxies from now on).
From 1500 to 1800, more or less regular descriptions of The strategy of these reconstructions (Jones and Mann
monthly or daily conditions start to become available. These 2004) is generally in three steps and requires a temperature
descriptions can be corroborated by records of processions dataset with sufficiently good coverage of the globe, hemi-
or rogations organized by the local parishes to end droughts sphere or well-defined region (e.g., The North Atlantic or the
or avert events that might endanger crops (Le Roy Ladurie Equatorial Pacific).
1967). The first step is to determine a small number of general
Between 1680 and 1860, the very first instrumental statistical characteristics (space or time) for recent observa-
measurements appeared. The barometer was invented by tions. The best known temperature reconstructions (Mann
Torricelli and the thermometer by Galileo. The first attempts et al. 1998) use statistical techniques of decomposition into
to establish international meteorological networks were principal components (von Storch and Zwiers 2001), how-
made during this time. In France, the first network of sys- ever there are alternatives, depending on the distribution of
tematic observations dates from the reign of Louis XVI. In the proxies and their properties. Thus, the temperature field
1776, Félix Vicq d’Azyr, secretary of the Academy of T, which depends on time t and space x, can be written in the
Medicine, asked the doctors of the kingdom to record the form:
temperature of the air three times a day, as well as to write a
summary of the diseases treated during the month. This X
5
T ðx; tÞ  ak ðtÞEk ð xÞ:
initiative was founded on the idea that variations in climate t¼1
could have an impact on the health of the population. This
work was maintained for a few decades, and many doctors in In this equation, the Ek(t) are the spatial modes of the
France contributed to the exercise, scrupulously noting variability of T, and the ak(t) are the associated time coeffi-
temperatures and diseases. Unfortunately, this directive was cients (von Storch and Zwiers 2001), generally called prin-
discontinued, and it is not possible to have continuous data. cipal components (denoted PC). In this equation, we only
In addition, the physicians of the day obviously had no keep 5 modes.
knowledge of meteorology or instrumentation, and not all of The climate indicators (proxies) are then compared with
their measurements were reliable. the evolution of the statistical characteristics of observations
From 1860 onwards, meteorology was developed within over a calibration period, where proxies and observations are
the framework of national and international networks. available. This calibration period may cover all or part of the
Urbain Le Verrier, who discovered the planet Neptune by twentieth century. There is a technical debate about the
430 P. Braconnot and P. Yiou

selection of this learning period (Jansen 2007) and the periods. It should be noted in particular that the latter part of
number of statistical components to be used, but the result in the twentieth century emerges significantly from the error bars
the end is not very sensitive to this selection. During this of the temperature variations for the preceding millennium.
period, a linear regression is performed between the proxies
and the statistical characteristics observed. This regression
attaches weights to proxies in order to maximize their cor- Climate Simulations
relation with temperatures. This analysis also makes it pos-
sible to eliminate the proxies which have a poor correlation Climate simulations provide an understanding of how dif-
with the temperature signal during the learning period. ferent forcings affect the climate and quantify the main
A ‘verification’ period prior to the learning period for feedbacks. Moreover, comparison between the model results
which instrumental temperature data are available is then and the data makes it possible to determine whether the
used. This verification period identifies the errors incurred in models are capable of representing a climate different from
the regression obtained for the calibration period. For this the current one. The models used for Holocene simulations
verification period, it is also possible to determine the error cover the spectrum of models presented in Chap. 4 of this
linked to the omission of proxies in the estimation of the volume which describes the different hypotheses and the
temperature. This calculation is essential because it is protocols to run such simulations. In the case of the Holo-
obvious that the further back in time we go, the fewer cli- cene, it is mainly the characteristics of the seasonal cycle that
mate series are available and the more uncertain the recon- have been analyzed, as changes in insolation, driven by
struction of temperatures becomes. precession, strongly modulate seasonality but have little
Finally, temperatures can be reconstructed for the last impact on the annual average of the different climate vari-
millennium. It should be noted that this reconstruction has a ables. For the last 2000 years, the emphasis is on under-
spatial aspect. It is important to bear in mind that this type of standing the forcing associated with fluctuations in the solar
method relies on basic assumptions about the temporal sta- constant and in volcanism, and the identification of associ-
bility of the climate modes identified in the temperature data ated feedbacks. This is a major step towards gaining per-
during the calibration period and the temporal stability of the spective on recent centuries strongly disrupted by human
relationship between proxies and temperatures. These two activity. Increased understanding of interannual to centennial
types of stability can be quantified over the verification variability is necessary in order to detect climate change and,
period, but it is impossible to exclude the possibility of where appropriate, to attribute it to human activities.
changes over a longer period. Another problem with this
type of reconstruction comes from the statistical regression
between proxies and main components over the calibration Holocene Simulations
period. Since a regression is generally imperfect, it inevi-
tably leads to an underestimation of the variance when this is Major Trends
used to reconstruct the climate, which can lead to a poor There are very few simulations that cover the entire Holo-
estimation of long-term climate variations. It is possible to cene. Those that exist were carried out using models of
partially solve this problem by using proxies that represent intermediate complexity (see Chap 4), because it is impos-
different time scales, and are thus sensitive to scales varying sible to simulate a period of 10,000 years within an
from inter-annual to centennial (Jansen 2007). acceptable time frame using general circulation models,
A major advance in the quality of climate reconstructions which need an average of one month to achieve 100 years of
lies in the improved understanding of the mechanism linking simulation on supercomputers.
the ‘proxy’ to the climate variation, which helps the statis- The main objective of simulations of the whole Holocene
tical steps described above to be guided by knowledge of the period is to reproduce the major climate trends caused by
physics. This research topic is particularly active at the variations in orbital parameters and by the concentration of
moment. greenhouse gases. They generally do not take into consid-
Several research teams have proposed temperature eration the full set of forcings such as volcanism or the
reconstructions for the last millennium, based on different evolution of the solar constant, factors which are not well
proxy datasets (Jansen 2007). These reconstructions often known for the whole period. The applied models of inter-
have common foundations (often, tree ring data) but the mediate complexity take account of atmospheric and ocean
spatial distribution of the proxies used varies considerably circulation in a simplified way, as well as sea ice and veg-
from one reconstruction to another. etation (Crucifix et al. 2002; Renssen et al. 2005). They
The evolution of the error bars of these reconstructions nevertheless have different levels of complexity.
shown in Fig. 30.4 shows the discrepancies between the The first model considers the ocean and the atmosphere in
estimates of temperature changes, especially during cold terms of latitudinal sectors, while the second takes a
30 An Introduction to the Holocene and Anthropic Disturbance 431

three-dimensional ocean model coupled with a simplified involved and to assess the ability of climate models used for
atmospheric model. The simulated annual temperature climate projections for the next century to represent a dif-
changes are very low, in accordance with the low amplitude ferent climate than the current one (Cane et al. 2006; Bra-
of the change in insolation. Between 60° N and 70° N, these connot et al. 2012).
studies show cooling during the Holocene in response to the
reduction in summer insolation relatively in line with Feedbacks from Snow, Vegetation and from Sea
observations (Fig. 30.3). The increase in greenhouse gas Ice in High Latitudes
concentrations since the Middle Holocene partly offsets the Feedbacks from ocean, vegetation and snow and ice cover
reduction in temperature by about 0.5 °C. The beginning of are the main factors that have conditioned the evolution of
the Holocene is also a time when the ocean was still dis- temperature in the different regions of the northern hemi-
turbed by the recent deglaciation and when there were larger sphere during the Holocene. When forest is replaced by
ice caps than are currently present in North America and grass, the reflective power of the combination of snow and
Greenland. Simulations (Renssen et al. 2005) indicate that vegetation is more marked. This results in a positive loop of
the main effect of the Fennoscandian cap at the beginning of increasing snow, cooling and withdrawal of the forest. This
the Holocene was to delay the climate optimum in eastern mechanism in particular describes entry into the last
Canada and Greenland (Fig. 30.3). These studies also sug- glaciation (de Noblet 1996). The ocean response leads to a
gest an increase in variability during the Holocene, mainly phase shift in the seasonal cycle due to its high thermal
around the northern seas, resulting from feedbacks between inertia and evaporation capacity, which limits surface heat-
atmospheric circulation, ocean and sea ice. ing. Thus, spring and autumn are more affected than summer
The simulations show no major changes in thermohaline and winter by oceanic feedbacks. These effects combine
circulation of the ocean during the Holocene, except perhaps with the feedback from sea ice and snow to reduce seasonal
a slight increase between 9000 and 8000 years before the contrasts at high latitudes and to amplify the cooling during
present. It also appears that the initial conditions and the the Holocene (Wohlfahrt et al. 2004).
long-term variations of the intermediate and deep ocean have The feedback from snow and sea ice is manifested in a
had little impact on the characteristics of the climate during modification of the surface albedo. It is therefore possible to
the Holocene. Thus, it can be taken that the different periods quantify it in terms of energy (W/m2), using a simple
are more or less in equilibrium with the forcings. Hence, approach to climate sensitivity analysis (Taylor 2007). This
equilibrium simulations for a particular period of the Holo- allows us to define this effect as:
cene can be used to correctly identify the main characteris-
tics of the climate. This result is important because the Albedo effect ¼ ðSWncspal  SWncsOk ÞSWf
ignorance of oceanic conditions at the global scale at the
where SWf = (1 − aOk)DSWi represents the radiative forcing
beginning of the Holocene is a source of uncertainty for
related to the change in insolation and SWncs represents the
numerical simulations. Nevertheless, the equilibrium is not
net solar radiation (SW: short wave) in a clear sky (ncs) for
necessarily verified over the last 4000 years because of a
the paleoclimatological (pal) simulation and for the control
200-year lag between the forcing and the climate response of
simulation (Ok). The radiative forcing takes into account the
the reorganization of the boreal forest in the northern
model characteristics via the albedo of the control simula-
hemisphere (Crucifix et al. 2002).
tion. Indeed, although the same insolation disturbance SWi is
applied to different models, the difference in net incident
Reference Periods and Analysis of Feedbacks solar radiation depends on the selected model as a function
Simulations from intermediate complexity models describe
of the surface characteristics, the representation of clouds
the major trends and main feedbacks that have shaped the
etc. which characterize the planetary albedo in the model
Holocene in the mid and high latitudes of the northern
(Fig. 30.6).
hemisphere but do not allow for a detailed analysis of
As an example, Fig. 30.6 shows that between February
feedbacks or changes in variability. Some periods, such as
and April, the estimated radiative forcing for two different
the beginning of the Holocene (9500 years ago) and the
models ranges from −12 to −4 W/m2 between 20° N and
Middle Holocene (6000 years ago), have been the subject of
50° N. The cooling associated with this insolation deficit
numerous modeling studies and benefit from international
allows the snow to remain longer on land.
coordination (Joussaume and Taylor 1995; Joussaume et al.
This snow introduces a local feedback of −4 to
1999; Kutzbach 1988; Braconnot et al. 2007a).
−16 W/m2 on the continent which reinforces the insolation
In particular, there is an effort to compile data at the
deficit and the cooling. Both simulations produce the same
global scale for the Holocene (Prentice and Webb 1998), and
type of feedback, but with different amplitudes and geo-
this helps, along with simulations, to improve our under-
graphical localization. On the other hand, in summer, the
standing of the mechanisms and the different feedbacks
432 P. Braconnot and P. Yiou

surface albedo retroaction (model1) surface albedo retroaction (model2)

Fig. 30.6 Radiative forcing due to the change in incident radiation at for the months of February to April from two climate models. Adapted
the top of the atmosphere during the Mid Holocene (W/m2) and from Braconnot et al. (2007b)
feedback resulting from surface changes (mainly change in snow cover)

lower coverage of sea ice and snow induces a heating sur- Holocene. The first contributory factor is the amplification of
plus of 5 W/m2 above 30° N. The water vapor feedback the seasonal insolation cycle in the northern hemisphere. The
adds to this effect, contributing an additional 2.5 W/m2 to monsoon is the result of differential heating between the
summer warming in these regions. These different feedbacks hemispheres and between the continents and the oceans.
amplify the direct effect of insolation. The way in which they Higher summer insolation increases the contrast in temper-
are represented in the models is the source of the differences ature between the land and the ocean. This contrast alone
between simulations and of the uncertainty on their does not explain all the monsoon characteristics. The orog-
amplitude. raphy and the warming at high altitudes of the Tibetan
plateau constitute a source of energy which contributes to the
establishment of this planetary thermal current. As the
Monsoons and Ocean and Vegetation Feedbacks monsoon becomes established, the release of latent heat at
the time of condensation is an additional source of heat
The amplification of monsoon regimes in the northern enhancing the convergence of moisture in areas of high
hemisphere is the main climate characteristic in tropical convective activity.
regions during the first half of the Holocene. The wet belt For the climate of 6000 years ago, Fig. 30.7 shows the
extending from Africa to India and South Asia recorded changes in temperature, precipitation and large scale circu-
strong changes in precipitation regimes, in line with the lation linked to the reinforcement of the monsoon in Africa
northward extension of the intertropical convergence zone. and India in response to increased insolation of the northern
These fluctuations are recorded in both continental and hemisphere, as well as the uncertainties arising from differ-
ocean data. Several factors explain the stronger monsoons ences between the models. Continental warming in
between the beginning of the Holocene and the Mid July-August reaches about 2.5–3 °C. This heating is
30 An Introduction to the Holocene and Anthropic Disturbance 433

temperature (°C) and pressure (hPa) regions of the different tropical ocean basins (Zhao et al.
2005). Coupled models capture the main features of the SST
evolution for the current climate. It should be noted, how-
ever, that differences between the simulations range from
0.5 °C to more than 2 °C. The amplitude and phase of the
seasonal cycle are relatively well reproduced in most of the
models. The results show the greatest amount of disparity for
the Indian Ocean as models do not correctly capture the
semi-annual SST signal.
Despite the differences for the current climate, the
response of the ocean to the change in insolation is very
consistent from one model to another. All of the tropical
rainfall (mm/d) regions are marked by a cooling in the first part of the year, a
direct response to the change of insolation. Warming occurs
one to two months after the change in insolation hits its
peak, depending on the region. The shift in phase between
the regions comes from the discrepancy in seasonal insola-
tion between the northern hemisphere and the southern
hemisphere and from localized changes in the thermal inertia
of the ocean. Although the main changes are relatively
similar between the models, dispersion between results can
be up to a factor of 2. The phase differences between the
models are greater in winter than in summer because the
change in insolation is weaker and lasts longer. The tropical
ocean is relatively cold in late spring, when the monsoon
Fig. 30.7 a Temperature change (shaded, °C) and surface pressure starts, and accentuates the contrast between land and ocean,
(hPa) between the Mid Holocene 6000 years ago and the current period
based on the average of the set of simulations from the international
promoting moisture advection on the continent. In the
PMIP project. b As for a, but for ocean surface temperatures (isolines, ° Atlantic Ocean, a dipole occurs on both sides of 5° N, with
C) and precipitation (gray). In figures a and b, the hatching indicates the warmer temperatures in the north and colder in the south
regions with the most variable results from one model to another than is currently the case. It strengthens the low pressure
zone between 10 and 20° N, the convergence of humidity in
this region and the monsoon influx in West Africa. This
associated with a deepening of the thermal depression which differential between the two hemispheres originates in the
is reinforced where warming is at a maximum. Large-scale heating differential linked to the insolation on either side of
circulation leads to increased convergence of ocean winds 5° N. The heating of the surface ocean is reinforced to the
towards the interior of the continents. Large-scale moisture north of 5° N by a decrease in evaporation due to a weak-
convergence follows this circulation in the lower layers of ening of the trade winds, part of which converges towards
the atmosphere, explaining the increase in precipitation in the African continent instead of crossing the Atlantic. In
Africa and northern India. These main features are repro- addition, enhancing the monsoon influx initiates an Ekman1
duced by all of the models. However, there is great disparity transport in the surface layers of the ocean, contributing to a
in the magnitude of the simulated changes in many regions delay in the warming of the region south of 5° N (Zhao et al.
such as East Africa, Arabia and northern India. 2005). The system relaxes in autumn when insolation on the
Several factors amplify or offset the atmosphere’s equator is sufficient to smooth the temperature gradient. In
response to insolation. These have been described in several India and South-East Asia, warming of the western part of
overview articles (e.g. Braconnot et al. 2012) of which the the Indian Ocean and of the ‘warm pool’ promotes the
main points are summarized here. The method used to study convergence of surface circulation to these warm waters to
the ocean and vegetation feedbacks consists mainly of the detriment of the continent. The Indian monsoon seems
removing the feedbacks one by one.
1
The Role of the Ocean The current generated by the surface wind is rotated to the right of the
The ocean plays a major role in modulating the amplitude wind in the northern hemisphere under the effect of the Coriolis force
and reduces with depth under the effect of friction in the form of a spiral
and phase of the seasonal cycle. Figure 30.8 summarizes the (Ekman’s spiral). The transport generated (Ekman transport) over the
simulated SST (Sea Surface Temperature) changes for key entire Ekman layer (about 100 m) is perpendicular to the wind.
434 P. Braconnot and P. Yiou

Month Month

Month Month

Month Month

Month Month
Fig. 30.8 Surface temperature simulated by different models of the 0° W; 10° S–0° S), IND (55° E–75° E; 5° N–15° N) et WPL (110° E–
PMIP project for the current period (left) and the difference between the 160° E; 0° N–20° N). Each curve and acronym represents a different
Middle Holocene and the current period (right) and four key regions of model
the tropical ocean, NAT (60° W–20° W; 10° N–20° N), SAT (30° W–
30 An Introduction to the Holocene and Anthropic Disturbance 435

mm/d between two iterations. New versions of land surface models

include the carbon cycle and vegetation dynamics (Krinner
Africa (12°N-23°N ; 17°W-30°E)
2005), which allows for interactive representation of vege-
tation changes, taking into account both slow changes in
precipitation
vegetation (replacement of one vegetation type by another)
evaporation
and the interactions of these vegetation changes with the
climate. There are still many uncertainties in how these
couplings are represented, which results in marked differ-
ences in the amplitude of how vegetation reinforces pre-
OAV cipitation in monsoon regions (Braconnot et al. 2007b).
Nevertheless, some of the broad outlines appear robust.
When vegetation is interactive, the role of local recycling
of water becomes more important. Figure 30.9 shows the
OA
contributions from advection and local recycling (evapora-
tion) in the Sahelian region, estimated from coupled
Month ocean-atmosphere-vegetation simulations. In this case, the
vegetation was coupled asynchronously to the atmosphere
Fig. 30.9 Seasonal evolution of the change in evaporation (mm/d) and model. Simulations with interactive vegetation indicate that
in precipitation (mm/d) between the Middle Holocene and the current the desert in the Sahelian region was replaced by steppe. The
time as simulated by the IPSL-CM1 model coupled (OAV) and
uncoupled (OA) asynchronously with a vegetation model (from
albedo was thus decreased in the region, reinforcing the local
Braconnot et al. 1999) warming over land at the beginning of the monsoon season,
and favoring the advection of humidity and precipitation.
During the monsoon season, vegetation is more efficient than
less amplified in the case of an interactive ocean than in bare soil at recycling water, and this also increases precipi-
older simulations, carried out with atmospheric models tation. Finally, at the end of the monsoon season, vegetation
forced by the current SST. absorbs the soil water, which contributes to a lengthening of
the rainy season (Texier et al. 2000). All of these effects
The Role of Vegetation explain why simulations incorporating interactive vegetation
Another factor that affects the characteristics of the monsoon are generally in better agreement with the data for the region
is changing vegetation. Vegetation modifies the surface of West Africa stretching from 15 to 20° N.
albedo and the ways in which soil water is used. Depending However, there are still great disparities between the
on the type of vegetation, water losses through interception, results of the different models. These come from small-scale
evaporation from bare soil and transpiration are distributed phenomena (clouds, boundary layer, turbulent flows etc.)
differently. All climate models incorporate a more or less which need to be better represented in climate models.
complex land surface model to determine the exchanges of
heat and water between the surface and the atmosphere. The Vegetation and the End of the Wet Season
types of vegetation and the resulting surface characteristics in Africa
(albedo, roughness, resistance to evaporation) are prescribed The role of vegetation and the changing variabilities are two
from current data. Simulations of the Middle Holocene tend factors that have also been suggested to explain the abrupt
to underestimate the hydrological changes in the Sahelian end of the wet period in Africa. Early studies using a model
region (Joussaume et al. 1999). Vegetation characteristics of intermediate complexity (Claussen 1999) showed sudden
are generally set in line with current ones although pollen aridification around 5000 years BP in Africa, accompanied
data indicate changes in vegetation (Jolly 1998). Change in by a rapid reduction in precipitation, in line with the data
vegetation, not taken into account in these simulations, is a (deMenocal 2000). Rapid vegetation feedback is the source
likely candidate to explain this underestimate. of this abrupt variation, as vegetation can produce multiple
The first coupled atmosphere-vegetation simulations equilibria in which green Sahara conditions and desert
involved an iterative coupling between climate and vegeta- conditions can coexist.
tion. For this, a vegetation model was used to simulate a In the early part of the Holocene, vegetation gradually
vegetation in equilibrium with the climate produced by the decreased in response to the reduced summer insolation and
atmospheric model and then was reintroduced as a boundary the associated reduction in precipitation. When the system
condition to the climate model (Claussen and Gayler 1997; reached a threshold where ‘green Sahara’ conditions could
de Noblet 1996). The process is reiterated until climate and no longer be maintained, the coupled system abruptly pro-
vegetation are in equilibrium, i.e. with no major difference duced a desert state accompanied by a sudden drop in
436 P. Braconnot and P. Yiou

precipitation. This sudden change in state is therefore asso- reactions with stratospheric ozone (Shindell 2001) which
ciated with a strong positive feedback from the vegetation. can produce a warming. This option is more complicated
Liu (2007) have another interpretation based on coupled to implement because most of the models used for this
ocean-atmosphere simulations in which the rapid disap- exercise do not include stratospheric chemistry, which is
pearance of vegetation is not accompanied by sudden arid- very demanding on computing time.
ification implying no strong positive feedback induced by • Volcanic eruptions, particularly the most powerful, which
vegetation. The mechanism in this case is linked to the emit sulfurous SO2 and dust into the stratosphere. The
internal variability of precipitations. This variability can be solid particles are quickly pulverized but the sulfurous
considered to be a stochastic forcing that generates slow gas turns into sulfuric acid, which is only washed away
stochastic variability of soil moisture. This variability after several months. Large volcanic eruptions (e.g.
interacts with the nonlinearities of the plant system to gen- Pinatubo 1991) have the effect of cooling the planet for
erate a rapid decrease in vegetation. It is therefore the non- two to three years (Robock 2000). Volcanism is generally
linear response of the vegetation to the strong internal accounted for rather crudely in climate models (Jansen
variability that is responsible for the sudden change of state 2007) and is often done by decreasing the solar constant
of the vegetation when a bioclimate threshold is reached. All by a few W/m2 for two years. This representation,
these studies provide insights into the interactions between although it saves calculation time, cannot account for the
long-term insolation variations, climate and vegetation. dispersion of the volcanic particles across latitude in the
Nevertheless, we do not yet know the exact reasons and the stratosphere and therefore for its impact on atmospheric
dominant mechanisms in these African regions. Progress is circulation.
essential to better represent the semi-arid zones and to obtain • Greenhouse gases added to those already present in the
hydrological and ecological data at high temporal resolution. atmosphere. This additional greenhouse effect is notice-
able from the twentieth century onwards. These are
almost exclusively emissions related to human activity.
The Recent Climate (Recent Centuries) Sulfate aerosols are also linked to human activity and
have a cooling effect on the atmosphere locally.
Simulating the climate variations of the last millennium is a
major scientific challenge for the beginning of the Estimates of the radiative effects of these forcings during
twenty-first century. The first problem is a technological the last millennium are shown in Fig. 30.2.
one: these simulations require several months of intensive The details of the atmospheric circulation responses vary
calculations on the most powerful computers available. This widely from one model to another, particularly for GCMs.
explains why the number of available simulations have Simpler models such as EMICs, on the other hand, show
gradually increased from one IPCC report to the other (IPCC similar behaviors because the atmospheric responses have
2007, 2013). The second problem is how to define an initial few degrees of freedom, which constrains these models to
state for the ocean and the atmosphere. This problem is behave in the same way.
avoided by using the output from an ocean model output
using pre-industrial conditions and for which no drift is
observed. The idea behind this approximation is that the Characteristics of Climate Variability
climate will adjust sufficiently quickly to forcing conditions
so that the initial conditions of the beginning of the last Extratropical Circulation
millennium are not critical. The third problem is purely
physical and is the definition of the climate forcings over The North Atlantic Oscillation (NAO) is the climate regime
time. On this time scale, the main forcings are: of the atmospheric circulation dominating the North Atlantic
region. This regime was first detected in temperature
• Changes in solar activity. The role of solar activity is not anomaly structures around the North Atlantic basin. A tem-
yet clear. Fluctuations in the energy received in the tro- perature anomaly, the term used by meteorologists, is the
posphere are theoretically very low. Fluctuations in solar difference between the observed temperature and the average
activity also affect the chemistry of the stratosphere. The temperature. Sir Gilbert Walker, the British meteorologist,
modulation of reactions with ozone, in particular, can noticed opposing temperature anomalies between Labrador
cause heating of the stratosphere and its expansion, thus and North Africa on the one hand and the eastern United
favoring atmospheric circulation patterns in the tropo- States and Western Europe on the other. Thus, when it is
sphere (Shindell 2001). Variations in solar activity in warmer than usual in the American East (a positive tem-
numerical climate simulations can have a purely radiative perature anomaly), it is generally colder than usual in the
effect. More sophisticated models include chemical south of France (a negative anomaly).
30 An Introduction to the Holocene and Anthropic Disturbance 437

This temperature structure is related to fluctuations in the pressure differences, for which the contemporary period can
atmospheric circulation in the extra-tropics of the northern be used to verify. Of course, such reconstructions can yield
hemisphere. Theoretically, this circulation blows from west ambiguities and may have diverged before the nineteenth
to east between 30° N and 70° N, and is diverted to the century, depending on the indices used, because the rela-
north by the Coriolis force (due to the rotation of the Earth). tionship between the pressure gradient (the NAO index) and
The North Atlantic has a characteristic zone of low pressure variables such as temperature or precipitation may have
(depression) around Iceland and a zone of high pressure changed over time.
(anticyclone) around the Azores. It is the modulation of the Looking at the daily weather charts, it can be seen that on
respective influences of these two phenomena that is at the some days in winter an anticyclone settles over Scandinavia,
origin of the NAO. causing high pressures that prevents the western winds from
At the level of the North Atlantic basin, the atmospheric reaching France. This kind of weather situation, often refer-
flow (of density q) is governed by an equilibrium between red to as a ‘Scandinavian blocking’, can last several weeks
the pressure forces (P) and the Coriolis force (f perpendicular resulting in generally dry weather. Another meteorological
to the movement). It is the geostrophic equilibrium which is situation that can occur, this time with an abnormally high
valid on a large spatial scale and over a few weeks: pressure zone in the middle of the Atlantic is the ‘North
Atlantic ridge’. The air crossing the Atlantic is diverted to the
1~
V¼ k  grad P: north of this zone and passes over Greenland, before
qf returning to Europe, causing a cold snap. We can thus count
In this equation, the velocity field V is related to the several types of weather with longer or shorter lifetimes. The
most common types of weather patterns are the two phases of
pressure gradient between low and high latitudes and ~ k is a
the NAO, the Scandinavian blocking and the North Atlantic
unit vector on the vertical. A strong gradient not only deflects
ridge. Figure 30.10 shows the four weather patterns that
the flow but also accelerates it. The pressure gradient term
dominate the atmospheric circulation in winter around the
can be approximated by the pressure difference between two
North Atlantic. These four situations are obtained through
selected locations, for example the Azores and Iceland. This
statistical calculations on pressure data for the last fifty years.
pressure difference is called the North Atlantic Oscillation
The isolines indicate the general direction of the wind.
Index (NAO). A reason for the choice of these locations is
As part of the heat is transported by atmospheric flow,
that pressure data is available for them since 1825.
each of these weather regimes has a different influence on
So when the pressure gradient between the Azores and
temperature in Europe. In the case of winter in France, cold
Iceland is pronounced (high NAO index), circulation
episodes are more linked to patterns such as the North
becomes more active and flows towards northern Europe. In
Atlantic ridge or the negative phase of the NAO and mild
this way, moisture is transported to Scandinavia, and during
spells are more related to the positive phase regimes of the
this time, southern Europe is mild and dry. On the other
NAO or the Scandinavian blocking. A change in the distri-
hand, if this gradient is low, the atmospheric flow is towards
bution of these patterns can affect the transport of heat, and
Southern Europe, causing it to experience wet and cold
thus have an impact on the surface temperature. The effect of
climate conditions.
this change in atmospheric circulation is comparable to the
During the 1990s, the NAO index was consistently very
effect of external forcings (greenhouse gases, volcanism and
positive, resulting in a rise in temperatures in Europe (while
solar activity) on the variability of temperatures. During the
cold records were being broken in Canada). This consistent
exceptionally hot summer of 2003, these high temperatures
index has been interpreted by some as a link with global
were shown to be partly due to the persistence of an anti-
warming. However, a similar situation with a very strong
cyclone over Europe, which was in turn related to a pre-
NAO index was already encountered at the beginning of the
cipitation anomaly over the Sahel, which dried up the
twentieth century and is probably not exceptional. More-
atmospheric column. It is therefore essential to consider
over, the index has returned to negative values at the
changes in the distribution of atmospheric circulation
beginning of the 21st century, while the temperature of
regimes in order to interpret climate change.
Western Europe continued to increase.
This representation facilitates the interpretation of rela-
Thanks to the NAO signature on temperature and pre-
tionships between meteorology (fluctuations on very short
cipitations, it has been possible to use paleoclimate data (tree
time scales, such as weather regimes) and climatology
rings or glacial drilling) to extend this ‘instrumental’ NAO
(centennial climate fluctuations). This concept was applied
index to the last millennium. This application makes it
to a very different climate, such as that of the Last Glacial
possible to determine an index based on atmospheric pres-
Maximum 21,000 years before the present. Using numerical
sure without having access to pressure itself. It is based on
simulations of the climate of this period, Kageyama et al.
large-scale relationships between climate variables and
(1999) found that the notion of weather regimes was not
438 P. Braconnot and P. Yiou

NAO− Blocking

70

70
4 6

6
8
60

60
2 −2

4
−4

8
10

2
0
−8

50

50
40

40
−10

30 −6

30
−2

0
0
0
−80 −60 −40 −20 0 20 40 −80 −60 −40 −20 0 20 40

Atlantic Ridge NAO+

70

70
−6
−6
−10
2 −2
−12
60

60
8

−8
50

50
4
6

−4
10 −4
0
−2
40

40

2
0
4
30

30
0

−80 −60 −40 −20 0 20 40 −80 −60 −40 −20 0 20 40

Fig. 30.10 Four common weather regimes around the North Atlantic. pressure zones. From top left clockwise, the weather regimes are: the
These regimes are calculated from sea-level pressure data from 1948 to negative phase of the North Atlantic Oscillation (NAO−), Scandina-
2009. General wind direction follows the isobars (iso-pressure lines) vian Blocking (Blocking), the positive phase of the North Atlantic
from west to east, turning in a clockwise direction around the high Oscillation (NAO+) and the North Atlantic Ridge (Ridge)

applicable at that time, and that the types of stationary cir- These oceanic structures are strongly linked to the
culation observed today were not present then. atmosphere and to the southern oscillation and modulate the
intensity of the trade winds, large-scale tropical circulation
and precipitation. El Niño occurs every three to seven years
The Equatorial Pacific (ENSO) and alternates with La Niña periods, which, on the contrary
introduces cold conditions in the eastern equatorial Pacific.
El Niño events are a major manifestation of the coupled The impact of El Niño events is not limited to equatorial
ocean-atmosphere variability of tropical regions originating regions and spreads to mid-latitudes via an atmospheric
in the Pacific Ocean (Philander 1990). The normal condi- wave train. Nevertheless, the intensity and frequency of
tions of the Pacific Ocean are characterized by the ‘warm these events, as well as the links between ENSO (El
pool’ (using the language of climatologists) in the western Niño-Southern Oscillation) variability and the average cli-
part of the basin and by the development of an upwelling in mate state, are still poorly understood.
the eastern part, brought about by the divergence of the The Holocene is a period of particular interest to under-
southeast trade wind and, closer to the coast, by the meridian stand these different aspects. Variations in the frequency and
component of the northerly wind along the coast of South intensity of El Niño events for this period were recorded in
America, which favors the development of a coastal corals, tropical glaciers and lake sediments. The different
upwelling. studies are in agreement that El Niño events were less
The El Niño years are characterized by abnormally warm intense during most of the Holocene and that the recent
waters between the date line and the South American coast, period is unusual in this context (Cobb et al. 2013; Carre
and are accompanied by a weakening of the temperature et al. 2014; Emile-Geay et al. 2016). Although all modeling
gradient across the Pacific basin (Fig. 30.11). studies agree on this reduction in activity in the Mid
30 An Introduction to the Holocene and Anthropic Disturbance 439

and, strengthens the easterly trade winds and hence favors

the development of equatorial upwelling and the temperature
gradient. This feedback is called the Bjerknes mechanism
(1969).
Although simulations from general circulation models
also show wind intensification and a reduction in El Niño
events, they do not show a coupling between winds and the
temperature gradient across the basin at an annual scale.
Rather, the results of these simulations indicate a link with
large-scale changes in circulation. In particular, the intensi-
fication of the Asian monsoon promotes the intensification
of the winds in the Pacific and their convergence towards the
southeast of the Asian continent. These winds counteract the
development of El Niño events (Zheng et al. 2008). Also the
seasonal changes in insolation prevents the development of
the upwelling Kelvin wave associated with El Ninõ (Luan
et al. 2012).
This result is very different from that obtained for other
climates, such as the climate of the future or the climate of
the Last Glacial Maximum. It shows the strong constraint
exerted by the changes in large-scale dynamics, which are
dominant over the changes in local heat fluxes in tropical
regions. For the Mid Holocene, this decrease in intensity of
El Niño events is also accompanied by a reduction in the El
Niño impact on other regions of the world, such as on the
Sahelian region (Zhao et al. 2007). However, it is not yet
Fig. 30.11 a Interannual variability of simulated sea surface temper- possible to estimate whether these changes in teleconnection
atures for the current period. b Changes in sea surface temperature have played a role in the environmental modifications in
variability for the Middle Holocene. c Changes in simulated variability
by eight climate models for the Eastern Pacific. Figures (a) and Africa, nor to test this result against available data.
(b) represent the median of the results of the various models shown in Fluctuations in El Niño events have also been reported for
figure (c) the last millennium, particularly in corals in the Pacific
(Bradley 1999). The results indicate that the ENSO had
Holocene, they have come to slightly different conclusions characteristics in the mid-seventeenth century similar to
about the origin of this evolution. All of the studies, how- those of the current period, with events that may have been
ever, attribute the main source of variation to solar radiation as intense as the El Niño of 1997–1998. In the period from
and a change in the seasonal cycle. the twelfth to the fourteenth centuries, however, events
The first simulations were carried out with a model of appear to have been less intense. However, it is not possible
intermediate complexity dedicated to the study of this phe- to make the connection between the average conditions and
nomenon (Clement et al. 2000). The area covered by the the characteristics of the El Niño events. Other authors have
model was limited to the tropical Pacific Ocean. The dif- found statistical links between the intensity of El Niño
ferent equations are treated with regard to their difference events and volcanic eruptions (Adams et al. 2003). ENSO
from the mean state. These simulations do not therefore variability is reduced in the years following a volcanic
explicitly include changes in the seasonal cycle brought eruption, and this is confirmed by simulations conducted by
about by changes in insolation during the Holocene. Nev- Mann et al. (2005), but had not been identified in previous
ertheless, they have helped to improve our understanding of studies. These relatively contradictory results come from the
the response by the tropical ocean to changes in surface difficulty in representing all of the facets of El Niño events in
fluxes. The basic mechanism implies a behavioral difference current models. Certain systematic biases in the models
between the western part of the basin, where warm waters interfere with the representation of interannual variability,
accumulate, and the eastern part, where equatorial upwelling for example, the dual ‘intertropical convergence zone’
develops. During the northern polar summer, the increase in (ITCZ) which is overly pronounced in the eastern Pacific in
insolation warms the western part of the basin more than the coupled ocean-atmosphere simulations, or the fact that the
east. It increases the temperature gradient across the basin structure of the equatorial upwelling is too confined along
440 P. Braconnot and P. Yiou

the equator and extends too far into the western part of the Let’s assume that temperature can be represented by a
basin [Chap. 8 of Solomon et al. (2007)]. Gaussian shape, a debatable assumption but one which
simplifies the subject. If an event is defined as extreme when
it is more than two standard deviations from the mean
Climate Extremes Relative to the Average (which theoretically happens *2.5 times out of 100) then, at
constant standard deviation, it is obvious that the probability
We have seen how it is possible to reconstruct average cli- of reaching a given higher temperature increases as the
mate conditions from environmental indicators or simulate average temperature increases.
them with climate models of varying levels of complexity. This hypothesis of Gaussian representation doesn’t hold
By average we mean both in terms of space (hemispheri- for daily precipitation, for which the standard deviation is
cally, globally or regionally) and time (seasonal, annual and ill-defined. One of the crucial parameters to describe pre-
decadal averages). Thus, most indicators show that the cipitation is the one that describes the tail of the distribution,
average temperature of the planet rose by *0.74 °C in the that is, the way rare values occur. A Gaussian possesses
course of the last century. It is obvious that behind this what is referred to as a ‘light’ tail, so that if big events are
average variation there are significant regional and temporal possible, they are exponentially rare.
disparities: not all the places on the planet heat up at the However, the occurrence of rare major precipitations, for
same rate (high latitudes seem to be more affected than low most locations on the planet, decreases much more slowly
ones), and gradual warming does not eliminate the possi- than in a Gaussian graph. The theory of extreme values
bility of local cold spells that can last several days. It is even makes it possible to statistically describe these phenomena
statistically possible for local cold records to be beaten in a (Coles 2001). Armed with this statistical theory, it is possible
climate that is warming globally. to model the different ways in which average precipitation
If we examine precipitation patterns, we can observe a and extreme precipitation behave which do not necessarily
trend towards aridification in southern Europe over the last respond in the same way to climate change. This statistical
few decades. This is reflected in a lower average seasonal characterization of climate variability makes it possible to
water balance than during the first part of the twentieth predict the evolution of the probabilities of intense events, as
century. On the other hand, observations show that extreme well as the uncertainty associated with this prediction.
precipitation events (such as the flooding episodes that occur
during the autumn in southern France or the wet winters in
the UK) may become more frequent and more intense. Outstanding Questions at the Beginning
Typical extreme climate events can be classified into of the Twenty-First Century
several categories:
Weather or Climate
• Heatwaves and cold snaps, i.e. when the temperature
exceeds certain thresholds for a certain duration. The Variations in the climate of the Holocene up to the present
definition of a heatwave differs depending on the loca- time have been punctuated by very short-lived meteorolog-
tion: a heatwave in Paris is a typical summer in Seville! ical events that have had a variety of consequences, but were
• Intense precipitation and droughts. On a one-month scale, often disastrous for societies. Since the end of the twentieth
it is important to distinguish between long episodes of century, there appears to have been an upsurge in extreme
‘average’ precipitation and those that last only a few events on the planet (the storms of 1999, record monsoons,
hours. They are responses to distinctly different meteo- deadly heat waves etc.). Assuming that this increase is not
rological conditions. due to improved communication of information on the pla-
• Storms, cyclones and other hurricanes. net, the first question to be answered is whether this state of
events has been ‘observed’ in the recent past (such as over
These examples illustrate the use of statistics and probabil- the last millennium). To answer this question, it is essential
ities to describe climate phenomena (von Storch and Zwiers to gather as much information as possible on extreme events
2001). These are particularly relevant when describing the (of all types) and put them into perspective. The gathering of
interactions between normal climatological magnitudes this information is the subject of many national and inter-
(mean temperatures and precipitations) and the occurrence of national programs on historical research and the climate
extreme or rare events. A typical paradigm for describing reconstruction of extreme events.
temperatures is to represent them as an average and a stan- A parallel approach is the numerical simulation of climates
dard deviation. These are the statistical parameters necessary at a resolution sufficiently high to enable the extremes to be
to carry out reconstructions of average temperatures over adequately represented (over the current period). The most
recent millennia. sophisticated models can be used to simulate the reported
30 An Introduction to the Holocene and Anthropic Disturbance 441

weather over recent centuries, and thus to examine how One of the teams has developed the weather@home system
meteorological statistics (regimes, extremes etc.) respond to (Massey et al. 2015), which was devised to calculate tens of
forcings over the long-term. The confrontation of models with thousands of high resolution climate simulations. Those
observations, within the context of extremes, becomes a cru- large ensembles are performed with present day greenhouse
cial issue, involving complex statistical theories. These gas concentration and natural variability, and with natural
mathematical tools allow for the regionalization of the climate, variability only. They allow estimating the probability den-
with the aim of forecasting climate variables on a very small sity function that a key climate variable (temperature, pre-
scale (e.g. precipitation at the scale of a city or a field) based on cipitation) exceeds a large threshold, with and without
understanding of large-scale variables (such as atmospheric climate change. Other statistical methods have been devised
circulation on the scale of the North Atlantic basin). since then to attribute extreme events to climate change
(Stott et al. 2016; Jézéquel et al. 2018).

Detection and Attribution of Climate Change

References
Understanding the role of human activity in climate change
has become a major issue in recent years. The identification Adams, J., et al. (2003). Proxy Evidence for an El Niño-like Response
of this role is not limited to superimposing demographic, to Volcanic Forcing. Nature, 426(6964), 274–278.
greenhouse gas and temperature curves on each other and Bjerknes, J. (1969). Atmospheric teleconnections from equatorial
pacific. Monthly Weather Review, 97(3), 163–172.
highlighting their simultaneous growth. The difficulty in Braconnot, P., Harrison, S. P., Kageyama, M., Bartlein, P. J.,
assessing this role is that there is no ‘control’ version of the Masson-Delmotte, V., Abe-Ouchi, A., et al. (2012). Evaluation of
Earth on which the climate with no human activity can be climate models using palaeoclimatic data. Nature Climate Change,
evaluated. 2, 417–424. https://doi.org/10.1038/nclimate1456.
Braconnot, P., Joussaume, S., Marti, O., & de Noblet, N. (1999).
To attribute climate change to any one factor, it is neces- Synergistic feedbacks from ocean and vegetation on the African
sary to evaluate the role of all other possible factors and to monsoon response to mid-Holocene insolation. Geophysical
show that the characteristics of the change are compatible Reseach Letters, 26, 2481–2484.
with this factor alone. This endeavor is usually done by Braconnot, P., et al. (2007a). Results of PMIP2 coupled simulations of
the mid-holocene and last glacial maximum—Part 1: Experiments
researchers through climate model simulations, taking into and large-scale features. Climate of the Past, 3(2), 261–277.
account all of the physical factors minus one, and then Braconnot, P., et al. (2007b). Results of PMIP2 coupled simulations of
comparing the simulations obtained with the observations. the mid-holocene and last glacial maximum—Part 2: Feedbacks
The key is to then identify the crucial factor(s) that best allows with emphasis on the location of the ITCZ and mid- and high
latitudes heat budget. Climate of the Past, 3(2), 279–296.
simulations to resemble observations. The difficulty with this Bradley, R. S. (1999). Paleoclimatology: Reconstructing climates of
evaluation is defining the statistical criteria for the compar- the quaternary, 2nd edn. San Diego: Harcourt/Academic, 613p.
isons and to obtain the necessary confidence intervals. Brazdil, R., et al. (2005). Historical climatology in Europe—The state
In order to make progress in this direction, improved of the art. Climatic Change, 70(3), 363–430.
Cane, M., et al. (2006). Progress in paleoclimate modeling. Journal of
models able to take the many interactions in the climate Climate, 19(20), 5031–5057.
system into account are essential. This requires considering Carre, M., Sachs, J. P., Purca, S., Schauer, A. J., Braconnot, P., Falcon,
not only physical interactions involving energy and water R. A., et al. (2014). Holocene history of ENSO variance and
cycles, but also the interactions between climate and the asymmetry in the eastern tropical Pacific. Science, 345, 1045–1048.
Claussen, M., et al. (1999). Simulation of an Abrupt change in saharan
biogeochemical cycles. In particular, land use, deforestation,
vegetation in the mid-holocene. Geophysical Research Letters, 26
different types of aerosols (natural and anthropic) are often (14), 2037–2040.
excluded from climate simulations. The distinction between Claussen, M., & Gayler, V. (1997). The greening of the sahara during
meteorology and changes in average climate also necessi- the mid-holocene: Results of an interactive atmosphere-biome
model. Global Ecology and Biogeography Letters, 6, 369–377.
tates improvement in the resolution of the models used,
Clement, A. C., et al. (2000). Suppression of El Nino during the
which goes hand in hand with better representation of mid-holocene by changes in the earth’s orbit. Paleoceanography,
small-scale processes. All of these elements combined with 15(6), 731–737.
high-resolution data make it possible to better understand the Cobb, K. M., Westphal, N., Sayani, H. R., Watson, J. T., Di Lorenzo,
E., Cheng, H., et al. (2013). Highly variable El Nino-southern
risks associated with the profound changes in our environ-
oscillation throughout the holocene. Science, 339, 67–70.
ment linked to the ongoing global warming. Coles, S. (2001). An introduction to statistical modeling of extreme
The attribution of extreme events is a rather recent field of values (p. 208). London, New York: Springer.
research, that poses many scientific challenges (NAS 2016). Crucifix, M., et al. (2002). Climate evolution during the holocene: A
study with an earth system model of intermediate complexity.
Since 2012, the Bulletin of the American Meteorological
Climate Dynamics, 19(1), 43–60.
Society (BAMS) publishes a summary of the analyses of de Noblet, N., et al. (1996). Possible role of atmosphere-biosphere
past extreme events and their attribution to climate change. interactions in triggering the last glaciation. Nature, 23(22), 3191–3194.
442 P. Braconnot and P. Yiou

deMenocal, P., et al. (2000). Coherent high- and low-latitude climate Ecosystem in the Holocene, Quaternary Science Reviews, 26(13–
variability during the holocene warm period, Science, 288(5474), 14), 1818–1837.
2198–2202. Luan, Y., Braconnot, P., Yu, Y., Zheng, W., & Marti, O. (2012). Early
Emile-Geay, J., Cobb, K. M., Carre, M., Braconnot, P., Leloup, J., and mid-Holocene climate in the tropical Pacific: Seasonal cycle
Zhou, Y., Harrison, S. P., Correge, T., McGregor, H. V., Collins, and interannual variability induced by insolation changes. Climate
M., Driscoll, R., Elliot, M., Schneider, B., & Tudhope, A. (2016). of the Past, 8, 1093–1108.
Links between tropical Pacific seasonal, interannual and orbital Mann, M., Bradley, R., & Hughes, M. (1998). Global-scale temperature
variability during the Holocene. Nature Geoscience, 9, 168-+. patterns and climate forcing over the past six centuries. Nature, 392,
Garnier, E. (2009). Les Dérangements du temps: 500 ans de chaud et 779–787.
de froid en Europe. Paris: Plon. Mann, M., et al. (2005). Volcanic and solar forcing of the tropical pacific
Gasse, F. (2000). Hydrological changes in the African tropics since the over the past 1000 years. Journal of Climate, 18(3), 447–456.
last glacial maximum. Quaternary Science Reviews, 19(1–5), 189– Massey, N., Jones, R., Otto, F. E. L., Aina, T., Wilson, S., Murphy,
211. J. M., et al. (2015). Weather@home—development and validation
IPCC. (2007). Climate change 2007: The physical science basis. of a very large ensemble modelling system for probabilistic event
contribution of working group I to the fourth assessment report of attribution. Quarterly Journal of the Royal Meteorological Society,
the intergovernmental panel on climate change. In: S. Solomon, D. 141(690), 1528–1545. https://doi.org/10.1002/qj.2455.
Qin, M. Manning, Z. Chen, M. Marquis, K.B. Averyt, M. Tignor & National Academies of Sciences Engineering and Medicine. (2016). éd.
H.L. Miller (Eds.), Cambridge, United Kingdom and New York, Attribution of extreme weather events in the context of climate
NY, USA: Cambridge University Press, 996pp. change. Washington, DC: The National Academies Press. https://
IPCC. (2013). Climate change 2013: The physical science basis. doi.org/10.17226/21852.
contribution of working group I to the fifth assessment report of the Philander, S. G. H. (Ed.). (1990). El Niño, La Niña, and the Southern
intergovernmental panel on climate change. In: T. F. Stocker, D. Qin, Oscillation (p. 312). San Diego: Academic Press.
G.-K. Plattner, M. Tignor, S. K. Allen, J. Boschung, A. Nauels, Y. Xia, Prentice, I. C., & Webb, T. (1998). BIOME 6000: Reconstructing
V. Bex, & P. M. Midgley (Eds.), Cambridge, United Kingdom and Global Mid-Holocene Vegetation Patterns from Palaeoecological
New York, NY, USA: Cambridge University Press, 1535pp. Records, Journal of Biogeography, 25(6), 997–1005.
Jansen, E., et al. (2007). Palaeoclimate, dans Solomon, In S. et al., Renssen, H., et al. (2005). Simulating the holocene climate evolution at
(Ed.), Climate change 2007: The physical science basis. contribu- northern high latitudes using a coupled atmosphere-sea
tion of working group I to the fourth assessment report of the ice-ocean-vegetation model. Climate Dynamics, 24(1), 23–43.
intergovernmental panel on climate change, Cambridge: Cam- Robock, A. (2000). Volcanic eruptions and climate. Reviews of
bridge University Press. Geophysics, 38(2), 191–219.
Jézéquel, A., Dépoues, V., Guillemot, H., Trolliet, M., Vanderlinden, Ruddiman, W. (2007). The early anthropogenic hypothesis: challenges
J.-P., & Yiou, P. (2018). Behind the veil of extreme event and responses, Reviews Of Geophysics, 45(3).
attribution, Climatic Change. https://doi-org.insu.bib.cnrs.fr/10. Shindell, D., et al. (2001). Solar forcing of regional climate change
1007/s10584-018-2252-9. during the maunder minimum, Science, 294(5549), 2149–2152.
Jolly, D., et al. (1998). Biome reconstruction from pollen and plant Solomon, S., et al. (Eds.). (2007). Climate change 2007: The physical
macrofossil data for Africa and the Arabian Peninsula at 0 and 6000 science basis: Contribution of working group I to the fourth
Years. Journal of Biogeography, 25(6), 1007–1027. assessment report of the intergovernmental panel on climate
Jones, P., & Mann, M. (2004). Climate over past millennia, Reviews of change (p. 996). Cambridge, New York: Cambridge University
Geophysics, 42(2). Press.
Joussaume, S. & Taylor, K. E. (1995). Status of the paleoclimate Stott, P. A., Christidis, N., Otto, F. E. Sun, Y., Vanderlinden, J.-P., van
modeling intercomparison project, dans. In Proceedings of the first Oldenborgh, G. J., et al. (2016). Attribution of extreme weather and
international AMIP scientific conference, WCRP-92, Monterey, climate-related events, Wiley Interdisciplinary Reviews, Climate
USA, pp. 425–430. Change, 7(1), 23–41, https://doi.org/10.1002/wcc.380.
Joussaume, S., et al. (1999). Monsoon changes for 6000 years ago: Taylor, K. E., et al. (2007). Estimating shortwave radiative forcing and
Results of 18 simulations from the paleoclimate modeling inter- response in climate models. Journal of Climate, 20(11), 2530–2543.
comparison project (PMIP). Geophysical Reseach Letters, 26(7), Texier, D., et al. (2000). Sensitivity of the African and Asian monsoons
859–862. to mid-holocene insolation and data-inferred surface changes.
Kageyama, M., et al. (1999). Weather regimes in past climate Journal of Climate, 13(1), 164–181.
atmospheric general circulation model simulations. Climate Dynam- von Storch, H., & Zwiers, F. W. (2001). Statistical Analysis in Climate
ics, 15(10), 773–793. Research. Cambridge: Cambridge University Press.
Krinner, G., et al. (2005). A dynamic global vegetation model for Wohlfahrt, J., et al. (2004). Synergistic feedbacks between ocean and
studies of the coupled atmosphere-biosphere system, Global vegetation on mid- and high-latitude climates during the
Biogeochemical Cycles, 19(1). mid-holocene. Climate Dynamics, 22(2–3), 223–238.
Kropelin, S., & Kuper, R. (2006). Climate-controlled holocene Zhao, Y., et al. (2005). A multi-model analysis of the role of the ocean
occupation in the Sahara: Motor of Africa’s evolution. Science, on the African and Indian monsoon during the mid-holocene.
313(5788), 803–807. Climate Dynamics, 25(7–8), 777–800.
Kutzbach, J. E. (1988). Climatic changes of the last 18,000 years— Zhao, Y., et al. (2007). Simulated changes in the relationship between
observations and model simulations, Science, 241(4869), 1043–1052. tropical ocean temperatures and the western african monsoon during
Le Roy Ladurie, E. (1967). Histoire du climat depuis l’an mil (p. 381). the mid-holocene. Climate Dynamics, 28(5), 533–551.
Paris: Flammarion. Zheng, W., et al. (2008). ENSO at 6 Ka and 21 Ka from
Liu, Z., et al. (2007). Simulating the Transient Evolution and Abrupt ocean-atmosphere coupled model simulations. Climate Dynamics,
Change of Northern Africa Atmosphere-Ocean-Terrestrial 30(7–8), 745–762.
 From the Climates of the Past
to the Climates of the Future 31
Sylvie Charbit, Nathaelle Bouttes, Aurélien Quiquet, Laurent Bopp,
Gilles Ramstein, Jean-Louis Dufresne, and Julien Cattiaux

This chapter connects everything we have learned about past that the energy balance at the surface of our planet is
climates (both from the analysis of natural archives and from dominated by incoming solar radiation which is the primary
numerical simulations), and future climate projections. The source of energy, and infrared emission exchanges which
models used to explore the future are similar to those used for control energy losses. He concluded that any change in
past climates, except that the results are based on emission surface conditions could lead to a change in climate and
scenarios of greenhouse gas emissions whereas past climate argued that the development and progress of human societies
simulations may be compared to reconstructions from the can notably change the state of the ground surface over vast
natural archives described in volume 1. The climate of the regions, as well as the distribution of waters and the great
past 1000 years can be seen as the ‘background noise’ of the movements of the air and that such effects have the ability to
recent natural evolution of the climate system. On this basis, cause the mean degree of heat to vary over the course of
the current climate change can be analyzed. Over the longer several centuries. Joseph Fourier also identified the trapping
term, analysis of ice cores allows us to trace back the history of infrared radiation by the gases in the atmosphere (Fourier
of atmospheric CO2 concentration over the last 1824).
800,000 years, and shows that the anthropogenic disturbance This was the beginning of the greenhouse effect theory.
(producing an atmospheric CO2 concentration currently in Starting from this seminal work, numerous studies were
excess of 410 ppm) is completely outside the documented conducted throughout the nineteenth and twentieth cen-
glacial-interglacial variations over the last million years, turies. Within a scientific context where the understanding of
ranging from 180 to 280 ppm. With this awareness you are the glacial-interglacial cycles was the matter of a very hot
well equipped to now explore the climate of the future. debate, the Swedish chemist Svante Arrhenius was the first
one to quantify the effect of the atmospheric carbon dioxide
concentration on the average surface temperature of the
Climate Observations in Recent Decades: The Earth and to suggest that substantial variations of the
First Signs of Warming atmospheric CO2 levels could explain the glacial advances
and retreats. However, it is only since the late 1970s that we
The principles of the physical laws governing the tempera- can calculate precisely the radiation exchanges using radia-
ture on the Earth’s surface were formulated at the beginning tive transfer codes and spectral databases to break down
of the nineteenth century by Joseph Fourier, who established energy by wavelength.
Long time refuted, or at least underestimated, global
S. Charbit (&)
N. Bouttes
A. Quiquet
L. Bopp
G. Ramstein warming has now become an incontrovertible reality: in
Laboratoire des Sciences du Climat et de l’Environnement,
LSCE/IPSL, CEA-CNRS-UVSQ, Université Paris-Saclay, 91190 2001, the scientific review by the Intergovernmental Panel
Gif-sur-Yvette, France on Climate Change (IPCC) concluded that there was a
e-mail: [email protected] growing body of evidence confirming global warming as
J.-L. Dufresne well as other changes in the climate system. Based on the
Laboratoire de Météorologie Dynamique/IPSL, Université Pierre results of numerical experiments with coupled
et Marie Curie, BP 99, 4 Place Jussieu, Cedex 05, 75252 Paris, atmosphere-ocean general circulation models (AOGCMs),
France
the Fifth IPCC Assessment Report, published in September
J. Cattiaux 2013 (IPCC 2013), established with a probability of more
Centre National de Recherches Météorologiques, Université de
Toulouse, CNRS, Météo-France, 42 Avenue Gaspard Coriolis, than 95%, that the global warming observed in recent dec-
31057 Toulouse Cedex, France ades is due to anthropogenic activities. This situation has no

© Springer Nature Switzerland AG 2021 443

G. Ramstein et al. (eds.), Paleoclimatology, Frontiers in Earth Sciences,
https://doi.org/10.1007/978-3-030-24982-3_31
444 S. Charbit et al.

direct equivalent in the past climates of the Earth. The last periods. For periods prior to 1950, analysis of air bubbles in
time we find a climate with the current level of atmospheric the ice is the only reliable way to track the chemical com-
CO2 is several million years ago. The climate at this period position of the atmosphere. In recent times, the first direct
corresponds to a warm Earth, with a reduced cryosphere (no measurements (i.e. in situ) were obtained in 1958 at the
Greenland) and a smaller Antarctic ice sheet yielding a much Mauna Loa site in Hawaii (Keeling et al. 1995). These mea-
higher sea level (15–30 m higher as reported by Haywood surements revealed for the first time that not only was CO2
et al. 2011, 2016). increasing in the atmosphere, but that this increase was
modulated in line with seasonal variations due to photosyn-
thesis of the terrestrial biosphere. This first measurement
Evolution of Greenhouse Gases campaign was then supplemented by campaigns covering
other sites in the northern and southern hemispheres. Cur-
We have seen in previous chapters that the greenhouse effect rently, a wide range of direct and indirect measurements
is above all a natural phenomenon, without which the sur- confirm that atmospheric CO2 levels have increased since the
face temperature of the Earth would be about −18 °C, beginning of the preindustrial era, rising from 275–285 ppm
making life, as we know it today, impossible. The main between the years 1000 and 1750, to about 380 ppm in 2005
greenhouse gases naturally present are water vapor (H2O), (Fig. 31.1) and to more than 410 ppm in 2018, that is a dif-
carbon dioxide (CO2) emitted by volcanic eruptions and ference of more than 100 ppm compared to the pre-industrial
forest fires, methane (CH4) produced by wetlands and vari- period. About 30% of the current atmospheric CO2 has been
ous fermentation processes, ozone (O3), and the nitrous emitted by anthropogenic sources. In addition, the amount of
oxide (N2O) emitted by soils. Industrialization has led annual anthropogenic emissions has been continuously
societies to discharge massive amounts of these gases increasing throughout the industrial era. In 1990, 2000, 2010
through the combustion of fossil fuels (oil, gas, coal), and 2017, global CO2 emissions from human activities (fossil
deforestation, agriculture, intensive livestock breeding and fuel combustion, cement production, land-use change)
fertilizer production. An inhabitant of an industrialized reached 22, 25, 32 and 37 Gt CO2/yr respectively. The
country releases on average ten tons of carbon per year (or increase in annual CO2 emissions accelerated from 1.1% per
CO2 equivalent) compared with only two tons per year for year in the 1990s to 3.3% per year in the 2000s. This growth
an inhabitant of most emerging countries. However, these rate of atmospheric CO2 is ten times faster than the highest
estimates mask large disparities from one country to another, rates recorded in ice cores, and is mainly due to the rapid
with differences ranging from 0.12 ton of CO2 per year and growth of developing countries and a drop in the efficiency of
per inhabitant for Ethiopia to 49.3 tons of CO2 per year and fossil fuel use in the global economy. Between 65% and 80%
per inhabitant for Qatar (from Global Carbon Atlas 2017). In of CO2 released in the atmosphere is trapped and/or dissolved
addition, human activities produce fluorinated gases (CFCs, in the ocean and the terrestrial biosphere in 20–200 years,
HFCs, PFCs, SF6) used especially in refrigeration and air depending on the various estimations. The rest is removed by
conditioning systems, as well as in aerosol cans. In total, slower processes, including chemical weathering and rock
more than forty of these gases have been identified by the formation that take several thousands of years, indicating that
IPCC. The greenhouse effect produced by human activities the effect of anthropogenic CO2 will persist for hundreds to
is called additional greenhouse effect. The contribution of thousands of years into the future (Archer et al. 2009).
each gas to the additional greenhouse effect can be estimated Atmospheric methane is the third most important green-
by taking into account the increase in their concentration and house gas after H2O and CO2 in terms of atmospheric
their ‘radiative efficiency’. Between 1750 and 2011, the concentration. Averaged over 100 years, the radiative effi-
variation in their concentration increased the greenhouse ciency of methane is estimated to be 28–36 times greater
effect by about 2.83 W/m2, with main contributions of 64% than that of CO2 but its lifetime in the atmosphere (i.e. the
for CO2, 17% for CH4, 12% for O3 and 6% for N2O time it takes for a CH4 molecule to be removed from the
(Fig. 31.1). Since 1750, in other words, since the beginning atmosphere by chemical reaction) is much less (* 9–
of the industrial era, this anthropogenic phenomenon has 12 years) than that of CO2. Ice core records indicate that
produced an energy imbalance of the Earth and has caused a CH4 levels in the atmosphere also show variations from
warming of the lower layers of the atmosphere. The addi- about 350 ppb (during glacial periods) to 700 ppb (during
tional radiative forcing corresponds to approximately 1% of interglacial periods). In 2011, the level of methane in the
the total radiation received. atmosphere, established from a network of measurements
Analysis of air bubbles trapped in the ice of the Antarctic covering both hemispheres, was at 1803 ppb, a level never
ice sheet revealed that over the last 800,000 years, CO2 levels attained, at least throughout the last 800,000 years. Since the
have changed by no more than 100 ppm, going from pre-industrial era, CH4 has increased by approximately
180 ppm during glacial periods to 280 ppm in interglacial 250%. Although the growth rate of methane was over 1%
31 From the Climates of the Past to the Climates … 445

Fig. 31.1 Atmospheric CO2 (top), CH4 (middle), and N2O (bottom) and firn air (color symbols) and from direct atmospheric measurements
concentrations history since the beginning of the industrial era (right) (blue lines, measurements from the Cape Grim observatory) (Source
and from years 0 to 1750 (left), determined from air trapped in ice cores IPCC 2013)

per year during the 1970s and early 1980s, the 1990s and The greenhouse gas with the fourth contribution to
early 2000s saw this rate stabilize, resulting in relatively radiative forcing is nitrous oxide (N2O). Its level has steadily
stable concentrations. However, atmospheric CH4 concen- increased from 270 ppb in 1750 to 323 ppb in 2011. The
trations started growing again from 2007, although the true main natural emissions of this gas come from soil microbial
cause of this renewed increase is still unclear. CH4 is emitted activity and ocean processes. As for anthropogenic emis-
by many agricultural activities (ruminant farming, rice cul- sions, they come mainly from the use of nitrogen fertilizers
tivation), by industrial activities (biomass combustion, the in agriculture, fossil fuel combustion and chemical industry.
oil and gas industry), as well as by natural processes (wet- Halocarbons (or halogenated hydrocarbons), responsible
lands, permafrost, peat bogs). There are no available data on in particular for the destruction of stratospheric ozone,
annual CH4 emissions from industrial activities as these are generate a lower radiative forcing (about 0.35 W/m2 in
difficult to quantify. When the climate warms up, CH4 2011) than the three main greenhouse gases (CO2, CH4 and
emissions from natural processes can increase. This has been N2O), whose total contribution is 2.30 W/m2. The emissions
observed with permafrost thawing in Sweden, but no of these gases are almost exclusively anthropogenic. To
large-scale evidence is available to clearly relate this process combat the destruction of stratospheric ozone, the Montreal
to the recent increase in methane. If the observed increase is Protocol regulated the production of halocarbons containing
caused by the response of natural reservoirs to global chlorine (chlorofluorocarbons or CFCs) and bromine. Sub-
warming, this could last for several decades, even centuries, stitute products adopted to replace CFCs, for example in
and thus reinforce the enhanced greenhouse effect (positive refrigeration processes, do not affect the ozone layer but
feedback). remain powerful greenhouse gases. Perfluorinated
446 S. Charbit et al.

Fig. 31.2 Main components of

radiative forcing involved in
climate change coming from both
natural processes and human
activities. The values correspond
to the estimated difference
between 2011 and 1750.
Anthropogenic contributions are
responsible for most of the
radiative forcing. Positive
forcings cause global warming,
while negative forcings lead to
cooling. The black lines
associated with each box show
the associated uncertainties
(Source IPCC 2013)

hydrocarbons (PFCs, such as CF4 and C2F6) and sulfur Evolution of Surface Temperatures
hexafluoride have extremely long residence times in the
atmosphere and are excellent absorbers of infrared radiation. Instrumental observations documented for the past 150 years
Thus, even though these compounds are released in small show an overall increase in temperature on the Earth’s sur-
quantities, their impact on the greenhouse effect and the face (Fig. 31.3). According to the synthesis presented in the
climate is far from negligible. Fifth IPCC Assessment Report (IPCC 2013), it has increased
Although ozone is also a greenhouse gas, it is not emitted by an average of 0.89 ± 0.20 °C over the last century (1901–
directly, but is formed from photochemical reactions 2012), that is an increase of about 0.08 ± 0.02 °C per dec-
involving other precursor gases of natural and anthropogenic ade. This multi-decadal signal (warming trend) emerges as
origin. Its impact on the radiative budget depends on the significant from the noise of the internal climate variabil-
altitude at which the changes in its concentration occur, as ity – a global warming is thus detected. Furthermore, dedi-
these vary spatially. Moreover, once formed, its residence cated studies have shown that the observed warming cannot
time in the atmosphere is very short, unlike the greenhouse be explained solely by natural forcings (solar and volcanic
gases mentioned previously. For this reason, it is difficult to activities), and that anthropogenic forcings necessarily con-
establish precisely its role in the radiative budget. tribute: global warming is thus attributed to both natural and
In addition to the production of greenhouse gases, human anthropogenic causes. In particular, “it is extremely likely
activities also produce aerosols. These can have a direct that human activities caused more than half of the observed
impact on radiative forcing by absorbing or reflecting solar increase in global average surface temperature from 1951 to
and infrared radiation. Some of them contribute negatively 2010” (IPCC 2013). However, this increase has not been
to radiative forcing, others positively. Finally, aerosols can steady over time, due to decadal fluctuations in both the
have an indirect effect by modifying the reflective properties internal climate variability and the external natural and
of clouds. Taking the totality of aerosols into account, the anthropogenic forcings. Indeed, the observations highlight
overall contribution is negative and therefore partially two periods of accelerated warming: one from 1910 to about
compensates the effect of greenhouse gases. In 2011, the 1940, and the other, even more important, since 1970, while
estimated forcing of anthropogenic aerosols is about temperatures were relatively stable between 1940 and 1970
−0.9 W/m2, albeit with large uncertainties (−1.9 to (Fig. 31.3). Trends computed over short time periods are
−0.1 W/m2, IPCC 2013). Overall, taking into account both highly uncertain. For instance, the warming rate over the
greenhouse gases and aerosols, anthropogenic activities are 15-year period 1998–2012 is about 0.05 °C per decade,
responsible for a positive radiative forcing since the begin- which is weaker than the 1901–2012 warming trend men-
ning of the industrial era, estimated at 2.3 W/m2 in 2011 tioned above (0.08 °C per decade). However, the associated
(IPCC 2013). The main components involved in climate confidence interval is large (± 0.10 °C per decade), so that
change are summarized in Fig. 31.2. there is no inconsistency between both values. More
31 From the Climates of the Past to the Climates … 447

more difficult to obtain than at the surface, but available data

unanimously reveal a warming troposphere and a cooling
stratosphere since the mid-twentieth century. This provides
further evidence for the anthropogenic nature of the recent
climate change. However, there has been a recent contro-
versy about the amplitude of these changes. The IPCC’s
Fourth Assessment Report published in 2007 (IPCC 2007)
indicated persistent uncertainty on temperature trends in the
middle atmosphere since 1979. Indeed, most of the available
data from radiosondes and satellite measurements indicated
a lower warming in the tropical high troposphere (between
10 and 15 km altitude) than that recorded at the surface,
Fig. 31.3 Global annual average land-surface air temperature (LSAT)
anomalies relative to the 1961–1990 climatology from the latest whereas all of the climate models projected an amplified
versions of four different data sets (Berkeley, CRUTEM, GHCN and warming in this zone of the atmosphere, especially in the
GISS) (Source IPCC 2013) tropics. This apparent difference between data and numerical
generally, global warming is a long-term process that is model outputs has been widely cited to emphasize the
superimposed on the internal climate variability, and it inconsistency between observational data and models. Some
remains possible to have occasional short-term cooling trends have used this example to call into question the impact of
in a warming world. Importantly, the decade of the 2000s human activities on the climate, even going so far as to deny
appears to be the warmest compared to the entire period the existence of the current warming. But more in-depth
covered by instrumental data. The decade of the 2010s is analyzes have revealed that the observations on which this
likely to break this record, since the years 2014–2018 have controversy was based were unreliable, particularly because
been the five hottest years observed at global scale. the measurements did not take into account interannual cli-
The values mentioned above are global mean annual mate variability. Since then, researchers have re-analyzed
values. There are, in fact, large disparities from one region to these measurements using more rigorous techniques (Allen
another, as well as large seasonal disparities. In general, and Sherwood 2008). New estimates of these observations
warming is amplified in high latitude regions in summer, show greater warming than previously reported, and this
partly due to the albedo effect induced by the decrease in new larger set of estimates now falls within the model
snow cover and/or in the sea-ice extent. In addition, warm- trends, thus removing any concerns expressed in the 2007
ing over the continents is much faster than over the oceans Fourth IPCC Assessment Report.
because of their weaker thermal inertia but also due to
changes in evaporation. Not all regions exhibit a statistically
significant warming over the observational period (e.g. Precipitations and Water Balance
South Greenland and North-West Atlantic); but no region
exhibits a statistically significant cooling. Climate change While the increase in mean surface temperatures is one of
detection is a signal-to-noise ratio problem: the higher the the most obvious manifestations of ongoing climate change,
temperature variability, the harder it is to detect long-term spatial and temporal changes in the hydrological cycle (e.g.
warming. In other words, internal variability can hide cli- precipitation, evaporation, runoff) appear to be just as
mate change over small spatial domains and/or short time important but are much more difficult to simulate. The term
periods; but this does not call into doubt the existence of a precipitation refers to all the meteoric waters that fall to
long-term global warming trend, nor the fact that it is attri- Earth in the form of liquid (rain, mist, showers) or solid
butable to the anthropogenic fingerprint. (snow, hail, sleet) water. The formation of precipitation
requires the condensation of water vapor around what are
called condensation nuclei, which allow the water molecules
Evolution of Temperature in the Troposphere to aggregate together. This phenomenon is called coales-
cence. Condensation only occurs when the amount of water
In response to an enhanced greenhouse effect (e.g. due to vapor per unit of volume exceeds a threshold value, called
anthropogenic activities), the physical theory of radiative the saturation value. This is an increasing function of tem-
transfer indicates that not only the surface, but the entire perature. Warmer air can therefore hold more water vapor
lower atmospheric layer, namely the troposphere (altitude before condensation occurs: this physical law is called the
0–10 km), should warm. Conversely, the atmospheric layer Clausius-Clapeyron relationship. For the Earth’s atmo-
above, namely the stratosphere (altitude 10–50 km), should sphere, the additional humidity that can be held by warmer
cool. Observations of the vertical temperature profile are air is estimated at 7% per °C. Once the water vapor
448 S. Charbit et al.

condensates, the different precipitation episodes can be cat- trend towards drought still persists, it has become less pro-
egorized according to their intensity, duration, frequency, nounced since the early 1990s.
and by type (stratiform, like those caused by depressions in There are numerous uncertainties surrounding the deter-
the mid-latitudes, or convective, like heavy rains and tropical mination of the hydrological cycle variables. This is due to a
cyclones in the intertropical convergence zone). These lack of data for some regions (for example, Canada,
characteristics depend to a large extent on local temperature Greenland and Antarctica, some desert regions such as the
and weather conditions (wind speed and direction, pressure, Sahara, the Tibetan plateau and over the oceans), and to the
humidity, evaporation). It is thus clear that a change in any fact that accurate measurements have only been available for
of these parameters will affect the hydrological cycle as a a very short time. In addition, it is very difficult to measure
whole. precipitation rates and to accurately quantify their changes at
Throughout the twentieth century, for example, annual the global and regional scales.
precipitation increased on the eastern side of the South and In situ measurements are affected by atmospheric condi-
North American continents. In contrast, a significant rainfall tions (e.g. the effect of strong winds especially on snowfall).
deficit has been observed in south and west Africa, as well as Spatial observations, on the other hand, provide only
in the Sahel. In northwestern India, an increase of about 20% instantaneous measurements and are affected by the uncer-
was observed for the period 1901–2005, despite a sharp tainties associated with the algorithms used to convert
decrease between 1979 and 2005. However, changes in radiometric measurements into precipitation rates. Because
precipitation are hard to measure using the existing records, of these difficulties, climatologists explore the coherence
and there is only medium confidence that observed trends between the whole set of complementary variables associ-
are due to the anthropogenic influence. ated with the hydrological cycle. One way to represent
Observed trends in relative humidity (i.e. air precipitation changes over the past century or over decades
humidity/saturation humidity) suggest that these have is to calculate the Palmer Drought Severity Index. This index
remained constant through the tropospheric column down to is a measure of drought, in other words, the accumulated
the surface. However, if the amount of water vapor at satu- surface soil moisture deficit compared to average local
ration increases and the relative humidity remains constant, conditions. It is based on recent rainfall and atmospheric
then this means that the absolute humidity (and thus the humidity (determined from temperatures). In general, ana-
amount of water vapor) has increased in the atmosphere. lyzes of the last century suggest a trend towards drying for
Observations indicate that tropospheric water vapor has much of Africa, southern Eurasia, Canada and Alaska. Other
increased by about 3.5% over the past 40 years, which is direct or indirect measurements (e.g. from river flow esti-
consistent with the observed temperature change of 0.5 °C mations or ocean salinity measurements) show that during
over the same time period. Climate models confirm these the twentieth century precipitation has generally increased
empirical observations: a warmer climate leads to increased on land surfaces between 30°N and 85°N, but significant
moisture content in the atmosphere and more intense pre- reductions have been observed over the last thirty or forty
cipitation events (although total precipitation over a full year years between 10°S and 30°N.
is reduced) and therefore flooding is more likely to occur. At present, the main challenge is to determine the inter-
Thus, in winter, it is observed that for most of the extrat- annual variations and trends in precipitation changes over
ropical land surface areas of the northern hemisphere, the the oceans. Global averages are often unrepresentative and
greatest precipitation is linked to higher temperatures. Con- mask large regional disparities. However, particularly pro-
versely, in areas with low rainfall, such as the Mediterranean nounced droughts in the last 30 years, as well as heavy
basin, rising temperatures are associated with a higher risk of rainfall events in many regions clearly illustrate an intensi-
drought. This general intensification of the hydrological cycle fication of the hydrological cycle.
can be summarized by: wet gets wetter, dry gets drier.
Added to these complex phenomena is the variability in
atmospheric circulation. In Sect. “Climate Variability”, the Extreme Weather Events
modes of atmospheric variability will be examined in greater
detail. However, we already know that fluctuations in Extreme weather events result from exceptional fluctuations
atmospheric circulation over the North Atlantic brought of a climate variable, and are generally associated with
heavy rainfall in the 1990s to northern Europe, and, in significant impacts on society and the environment. As
contrast, led to a drying-up of the Mediterranean basin. In illustrated by the frequent exposure by the media of the most
addition, the severe Sahelian drought over more than impressive events, the statistical evolution of climate
20 years (1970–1990) was linked to changes in both atmo- extreme characteristics (frequency, amplitude) is of major
spheric circulation and surface ocean temperatures in all concern for current climate change, especially in the devel-
three of the Pacific, Indian and Atlantic basins. Although this opment of adaptation strategies. The term ‘weather extreme’
31 From the Climates of the Past to the Climates … 449

covers a wide spectrum of events at spatio-temporal scales Precipitation Extremes, Tropical Cyclones
ranging from intense local precipitation lasting a few hours and Extra-Tropical Storms
to exceptional hot years for the whole planet. In addition, Changes in the frequency and intensity of droughts and
because they are by definition rare, extreme events offer few floods, in response to global warming, are also of major
case studies, and the detection of possible man-induced concern, as our industrialized societies become increasingly
trends requires analysis over long periods of time and/or vulnerable to rainfall extremes. Although the study of
large spatial domains. The study of extreme events is thus droughts is similar to that of heat waves, since they both
often restricted by the availability of datasets of sufficient have impacts over large areas, the analysis of extremes of
spatial and temporal resolution. intense precipitation is more difficult because it requires data
with finer spatio-temporal resolution.
Temperature Extremes As illustrated by the record floods in summer 2002 in
Extreme temperature events typically affect large areas Europe, intense precipitation events have been on the
(usually several thousand kilometers), and are often increase since 1950 in the mid-latitudes of the northern
accompanied by extremes in other climate variables (e.g. hemisphere. Even in the Mediterranean Basin, where rainfall
drought during a hot summer, snowstorms during a cold is decreasing on average, the episodes of heavy rainfall are
winter). Their impacts on ecosystems and human activities more intense. A recent example is given by the intense pre-
are thus particularly important. cipitation that occurred in South of France in autumn 2018.
One of the most striking recent examples is the European This upward trend is noticeable on a global scale, although
heat wave in summer 2003 which, with a temperature 2.5 °C the increases are more moderate than those observed for
higher than the seasonal mean, exceeded by three standard temperature. Changes in precipitation extremes are consistent
deviations the distribution of summer temperatures in Eur- with the Clausius-Clapeyron relationship: warmer air can
ope. It had dramatic socio-economic and environmental hold more humidity, so there is more water to be mobilized
impacts: high mortality, loss of energy production, acute by condensation when rainfall events occur.
urban pollution, fires, accelerated melting of glaciers etc. Paradoxically, the extent of regions affected by drought is
The return period of this particular event was estimated at also increasing, illustrating the fact that climate change not
250 years at the time it occurred (2003); it would have taken only affects the mean of statistical distributions, but also its
4 times longer (1000 years) without human intervention variability. Africa, southern Eurasia and North America are,
(Stott et al. 2004). according to the Fifth IPCC Assessment Report, the regions
This illustrates an expected feature of climate change: hot most affected by these droughts in recent times, as illustrated
extremes become more frequent in a warmer world, and by the dry conditions persisting from 2014 to 2018 over the
conversely, cold extremes become less common, but remain west of the United States (California). The increase in the
still possible. For instance, in the U.S., the ratio of the record frequency and/or intensity of both intense rainfall episodes
of high maximum daily temperatures to the record of low and droughts is consistent with the overall intensification of
minimum temperatures—which should be 1 to 1 in a sta- the hydrological cycle in a warmer world (wet gets wetter,
tionary climate—is currently about 2 to 1 and is likely to dry gets drier).
reach about 50 to 1 by the end of the twenty-first century Lastly, climate change may also affect meteorological
(Meehl et al. 2009). systems like tropical cyclones or extra-tropical storms. In a
At the global scale the geographical distribution of warmer world, the former are expected to occur less fre-
changes in the frequency and amplitude of temperature quently in general, due to less frequent atmospheric condi-
extremes is consistent with the distribution of average tions favorable to the cyclogenesis (reduced temperature
warming (IPCC 2013). This finding is nonetheless nuanced difference between the surface and the high-troposphere).
by the fact that changes in temperature distributions are often However, once triggered, future cyclones should get more
more complex than a uniform shift towards warmer values: energy from a warmer ocean: the intensity of the strongest
spreading, tightening, and/or asymmetry of values may also tropical cyclones is therefore expected to increase. Unfortu-
show up in the statistical distribution. For instance, in Eur- nately, so far, trends are difficult to detect from the past due to
ope and Central U.S., the variability of summer temperatures the lack of homogeneous data. For extratropical storms,
is expected to increase in a warmer climate, due to the forecasting is even more difficult: projections performed with
increase in evapotranspiration which leads to drier soils climate models do not unanimously agree on their frequency
(Douville et al. 2016). This would result in a widening of the and/or intensity. The only robust future signal seems to be a
temperature distribution, and further amplify the increase in poleward shift of the storm tracks, but again, available
the frequency of hot extremes in these regions. observations are insufficient to capture any past trend.
450 S. Charbit et al.

Evolution of the Cryosphere the role it plays in thermal insulation. In winter, snow cov-
ered grounds cool much less quickly than bare grounds,
The exact estimate of the influence of human activities on hence the importance of snow depth for plant and animal
climate is still limited because it is critically dependent on life. Finally, melting snow in spring and summer requires a
our ability to distinguish the signal related to this additional high latent heat of fusion, so that snow cover represents a
radiative forcing from the natural variability of the climate. significant heat loss for the atmosphere during the melting
However, there is a growing number of tangible factors season. As a result, seasonal snow produces thermal inertia
indicating that man has had a perceptible influence on the within the climate system, as it involves significant energy
climate. In particular, the global warming observed over the exchanges, with little or no change in temperature.
past century has been accompanied by a rise in sea level In the northern hemisphere, snow cover varies seasonally
(Clark et al. 2016), largely attributed to the thermal expan- with a maximum in winter and a minimum in summer, but
sion of the oceans, but also to significant changes in the with large inter-annual variations. Since the end of the
cryospheric components of the climate system over the nineteenth century, daily records of snowfall and snow depth
whole planet. The cryosphere represents all the water in have been kept by many countries. Nevertheless, these
solid form on Earth and contains more than 70% of the measurements were only fully developed after 1950, and in
Earth’s freshwater reservoir. It is an excellent indicator of particular from 1966 onwards, with the arrival of satellites.
climate change. It includes ice sheets, floating ice-shelves, All of these data series reveal that snow cover has
mountain glaciers, snow and sea ice, but also the water in decreased in spring and summer since the 1920s, with an
rivers and lakes that freezes in winter, as well as the per- even more striking decrease since the end of the 1970s.
mafrost, that is to say the permanently frozen ground, cov- According to the Fifth IPCC Assessment report, this
ered by an ‘active’ soil layer which melts each summer and decrease in March-April snow cover extent ranges
whose thickness is variable. from −0.8% per decade over the 1922–2012 period
Each of these components interacts in various ways with to −2.2% per decade between years 1979 and 2012. For the
the other components of the climate system over a wide fall and winter seasons, the signal is less clear: some data
range of time scales, from seasonal (snow, permafrost, riv- sets suggest positive trends as a result of increased snowfall
ers, lakes, sea ice) to a hundred thousand years for the in a warming climate, while others suggest negative trends
glacial-interglacial cycles. While the cryosphere is particu- similar to what is observed in spring and summer. Never-
larly important in the polar regions, there are also many theless, there is a consensus that the mean annual snow
glaciers in the low and mid-latitude regions, which provide cover has decreased with a shift from February to January of
an overview of the relationship between climate change and the maximum extent, an earlier onset of melting (*5.3 days
changes in the cryosphere. since winter 1972–1973) and thus a reduction of snow cover
duration. This drop in snow cover is mainly observed in the
Snow Cover northern hemisphere. In the southern hemisphere, very few
Estimating the extent of snow cover and the physical prop- data exist outside of Antarctica, and these are often of much
erties of snow is of paramount importance for both hydro- lower quality than in the northern hemisphere.
logical applications, such as modeling or predicting runoff
due to snowmelt, and for the understanding of local or Evolution of Sea Ice
regional weather patterns. Typical snow parameters, derived Sea ice is frozen seawater. When freezing occurs, salt is
from radar data, include the extent of snow cover, the water expelled from the ice crystals, thus raising the density of the
equivalent of the snow layer, and the state of the snow (wet surface ocean waters. The formation of sea ice can therefore
or dry). have a direct impact on the intensity of the thermohaline
The extent of snow cover has a direct influence on the circulation. Sea ice is a highly reflective surface with a high
energy balance on the Earth’s surface, but also on the soil albedo of about 0.8. Conversely, when sea ice melts, the
water content. Fresh snow reflects between 80 and 90% of ice-free ocean surface absorbs about 90% of the radiation
incident solar radiation. The warming trend decreases the due to large albedo changes, causing the ocean to warm up,
snow cover, which in turn decreases the fraction of solar followed by a further increase in surface temperature. This
energy reflected back to space, and increases the absorption phenomenon is a positive feedback between temperature and
of incoming radiation, thereby increasing warming, which in albedo. Thus, sea ice regulates heat exchanges between the
turn accelerates the snow melting. This amplifying mecha- atmosphere and the polar ocean. It isolates the relatively
nism is known as the ‘temperature-albedo’ effect. Thus, the ‘warm’ ocean waters from the much colder atmosphere,
surface temperature is strongly dependent on the presence or except when there are winter ‘leads’ occurring as a result of
absence of snow. Another important aspect of snow cover is sea-ice break-up. These leads allow the exchange of heat and
31 From the Climates of the Past to the Climates … 451

water vapor between the atmosphere and ocean which can

affect local cloud cover and precipitation rate.
The extent (i.e. area of the sea covered by sea ice), and
the thickness of sea ice are the two indicators of sea ice
conditions. Typically, the average sea-ice extent ranges from
14 to 16  106 km2 at the end of winter (7 to 9  106 km2
at the end of summer) for the Arctic and from 17 to
20  106 km2 (3 to 4  106 km2 at the end of summer) for
Antarctica. The first systematic monitoring of sea ice con-
ditions began in 1972 with the first satellite observations. For
older periods, only a few scattered data are available. Most
of the analysis of the variability and trends in sea ice cover
relates to the post-1978 period. The results from different
types of satellites give very consistent results, and all high-
light an asymmetry between the Arctic and the Austral
Ocean, with a clear decreasing trend for the Arctic and a
slight growth for Antarctica, although the latter trend is not
statistically significant.
Before the satellite era, sea-ice data are relatively sparse
and inconsistent. Nevertheless, although these data indicate
significant regional variability, all confirm an increase in
sea-ice cover during the nineteenth century and the first half
of the twentieth century, and a sharp decline in sea ice from
1970 onwards. According to the assessment by the National
Snow and Ice Data Center, sea ice coverage in September
has decreased at a rate of *12.8% per decade from 1979 to Fig. 31.4 Evolution of Arctic sea-ice extent from 1870 to 2011.
2018 relative to the 1981–2010 average (Fig. 31.4). One a Annual sea-ice extent and b seasonal ice extent using averages of
striking result was the record reached in 2007 when the total mid-month values derived from in situ measurements and remote
sensing observations In (b), data from the different seasons are shown
area of sea ice fell to 4.1  106 km2 in mid-September. In in different colors to illustrate variation between seasons (blue:
2012, this record was broken with a minimum sea ice cov- January–February–March; green: April-May-June; red:
erage of 3.4  106 km2 (i.e. *20% below the previous July-August-September; orange: October–November–December). The
record) making 2012 the lowest minimum since the begin- black lines in a and b correspond to data coming from the Scanning
Multichannel Microwave Radiometer and passive microwave data from
ning of satellite era. Although measurements did not reach the Special Sensor Microwave Imager (Source IPCC 2013)
new records between 2012 and 2018, the recorded
September minima for sea ice extent for the 2015–2018
period are all below 5.0  106 km2. The persistence of these recent thinner ice remains in the region and is more prone to
minima shows that the decrease in Arctic sea ice is accel- disappear during the melt season (Rigor and Wallace 2004).
erating. In the context of global warming, this observation is Between 1989 and 1995 the Arctic Oscillation entered a very
of growing concern for scientists, especially since the phe- positive mode, thereby reducing the extent of sea ice. Since
nomenon is amplified by the temperature-albedo feedback. the mid-1990s, only a few years have seen a positive or a
Moreover, in the past, a year marked by a very small sea-ice neutral mode. Meanwhile, the thickness of Arctic sea ice
extent was generally followed the year after by a return to continues to drop due to increasing temperatures. In March
normal conditions. Observations made in recent years sug- 1985, the fraction of first-year ice was *50% against 70% or
gest that a progressive disappearance of the oldest ice could more in 2015, and during the 1985–2015 period, the
occur if the decrease in sea ice continues at the same rate as multi-year ice (4 years and older) dropped dramatically from
in the last two decades. 20% to only 3% (Tschudi et al. 2016).
The decline in sea ice observed in recent years seems to The situation is different in Antarctica mainly because of its
be mainly driven by rising temperatures. However, atmo- geographic location (surrounded by the Southern Ocean and
spheric variability, and in particular the Arctic oscillation, centered on the South Pole). Atmospheric winds and the
may also favor the decreasing trend if the Arctic oscillation oceanic circumpolar currents act as barriers to warmer air and
is in a positive mode. This has the effect of shifting the jet warmer waters coming from the north. Changes in sea-ice
stream from the mid-latitudes to higher latitudes and causes coverage are more tenuous, with an increase of 1.2–1.8% per
the older, thicker sea ice to be pushed out of the Arctic. More decade between 1979 and 2012 (IPCC 2013). However,
452 S. Charbit et al.

changes in the distribution of Antarctic sea ice show major

regional differences. For example, the Weddell and Ross seas
are experiencing an increase in sea ice extent due to large-scale
changes in atmospheric circulation, while in West Antarctica,
the surface covered by sea ice is drastically decreasing, con-
sistent with the observed warming in this region.

Permafrost
Permafrost is defined as soil that remains permanently frozen
for at least two consecutive years. It is topped by a so-called
‘active layer’ that thaws each summer, and whose thickness
can vary from a few centimeters to a few meters, depending
on altitude and latitude. In areas where it has persisted for
several glacial-interglacial cycles, the permafrost can be
several hundred meters thick, and even exceeds 1000 m in
some parts of Siberia and Canada. During the last
glacial-interglacial cycle, there have been large variations in
area and depth of permafrost over North America (Tarasov
and Peltier 2007) or Europe (VandenBerghe 2011). Cur-
rently, permafrost covers 22.8  106 km2 of the northern
hemisphere, or about 24% of the continental areas. Per-
mafrost occurs mainly in polar and circumpolar areas and in
mountain regions at lower latitudes (e.g. Chile, the Alps, the
Himalayas). It can also be found in the seabed of the Arctic
Ocean in the continental shelf areas. When surface condi- Fig. 31.5 Time series of mean annual ground temperatures at depths
tions are not spatially homogeneous (e.g. snow cover, veg- between 10 and 20 m for boreholes located throughout the circumpolar
etation) permafrost can occur in patches. Such permafrost northern permafrost regions (Romanovsky et al. 2010a). Data sources
are from Romanovsky et al. (2010b) and Christiansen et al. (2010).
areas are called ‘discontinuous permafrost zones’. When Measurement depth is 10 m for Russian boreholes, 15 m for Gulkana
temperatures drop at higher latitudes, gaps in permafrost are and Oldman, and 20 m for all other boreholes. Borehole locations are:
less frequent. When surface conditions become homoge- ZS-124, 67.48°N 063.48°E; 85-8A, 61.68°N 121.18°W; Gulkana,
neous, permafrost is referred to as ‘continuous permafrost’. 62.28°N 145.58°W; YA-1, 67.58°N 648°E; Oldman, 66.48°N 150.68°
W; Happy Valley, 69.18°N 148.88°W; Svalbard, 78.28°N 016.58°E;
The presence of permafrost is critically dependent on the Deadhorse, 70.28°N 148.58°W and West Dock, 70.48°N 148.58°W.
soil temperature, which is itself controlled by the surface The rate of change (degrees Celsius per decade) in permafrost
energy balance, and thus, by several climatic factors such as temperature over the period of each site record is: ZS-124:
incoming solar radiation, cloudiness, snow cover, vegetation 0.53 ± 0.07; YA-1: 0.21 ± 0.02; West Dock: 0.64 ± 0.08; Dead-
horse: 0.82 ± 0.07; Happy Valley: 0.34 ± 0.05; Gaibrath Lake:
cover, surface and subsurface hydrology, and carbon 0.35 ± 0.07; Gulkana: 0.15 ± 0.03; Old Man: 0.40 ± 0.04 and
exchanges between the soil and the atmosphere. One of the Svalvard: 0.63 ± 0.09. The trends indicate the very likely range,
key factors affecting permafrost distribution is the insulating 90% (Source IPCC 2013)
effect of snow. When snow is present, ground temperatures
are generally warmer than those which would occur under
smaller snow cover or snow-free conditions. In continuous temperatures to lower. Interception of precipitation by veg-
permafrost areas, snow cover exerts a direct influence on the etation also has an impact on ground temperatures through
active layer thickness. For example, it has been shown that a evaporation and transpiration and the associated turbulent
doubling of snow cover from 25 to 50 cm may increase the heat exchanges between atmosphere and surface layers.
mean annual surface soil temperature by several degrees However, the direct effect of vegetation on ground temper-
(Fig. 31.5). On the other hand, if seasonal snow melting atures is less important than its role on snow cover. Inter-
occurs in late spring or early summer, ground warming is ception of snow in boreal forests reduces snow cover on
delayed. soils and acts to reduce ground temperatures. These exam-
The permafrost thermal state is also influenced by rain- ples show that the formation or degradation of permafrost is
fall. Firstly, ground temperatures can be increased through strongly influenced by climate. It is thus studied as an
the energy flux released by liquid precipitation penetrating indicator of climate change by a global network of
into the soil. Secondly, rain falling on a snow covered sur- researchers (Romanovsky et al. 2010a, b) who rely on
face may alter the snow insulating effect causing temperature measurements taken from boreholes and from
31 From the Climates of the Past to the Climates … 453

satellite tracking. In the framework of the International Polar transformation of nitrogen trapped in soils into nitrous oxide,
Year (2007–2009), a large permafrost- monitoring network which could increase with increasing temperatures and thus,
was developed and ground temperatures (which control the in turn, amplify global warming.
thermal state of the permafrost) have been measured in 575 In any case, even if there are still many uncertainties, the
sites located in arctic regions (North America, Nordic Eur- Fifth IPCC assessment report emphasized on the positive
asian regions and Russia). In most sites belonging to the feedback of permafrost melting on climate warming: “the
network, permafrost temperatures have increased in recent release of CO2 or CH4 to the atmosphere from thawing
decades. The observed rate of change of mean annual permafrost carbon stocks over the twenty-first century is
ground temperatures from mid-1970s to 2010 ranges from assessed to be in the range of 50 to 250 GtC for RCP8.5”
0.15 ± 0.03 to 0.82 ± 0.07 °C per decade, depending on (IPCC 2013). The RCP scenarios are defined in the section
the site (Fig. 31.5). In cold permafrost regions (mostly in “Projecting the Future of the Climate System” of this chapter.
continuous permafrost zones), the mean annual ground
temperatures have increased by up to 2 °C, compared to less Glaciers
than 1 °C in warm, forested permafrost areas (discontinuous A glacier is a mass of ice formed by the successive accu-
permafrost). This can be explained by the fact that the mulation of layers of snow year after year. Over the years,
amount of ice in warm permafrost is usually larger than for under the pressure of its own weight, the snow hardens and
cold permafrost. Indeed, in case of ice melting, the overall becomes granular (firn), then turns into ice and expels the air
warming trend is counteracted by the latent heat effect. it contained. Under the action of gravity, the ice flows along
Moreover, in forested areas, the snow insulating effect is the slope, thus supplying the lower parts of the glacier.
limited due to the interception of snow by vegetation. A glacier is in constant movement and carries mass from
One of the main consequences of permafrost warming is high altitudes to lower altitudes. In winter, the glacier grows
increased thickness of the active layer, although some per- due to snow accumulation on its surface. During the fol-
mafrost areas exhibit only modest thickening or even a lowing summer, the glacier loses all or part of the mass it
thinning. Indeed, a study based on the analysis of 169 cir- gained during winter. The disappearance of ice through
cumpolar and mid-latitude sites revealed that only 43.2% of surface processes is called ablation. The difference between
them have experienced an increase of the active layer accumulation and ablation determines the surface mass
thickness since the 1990s (Luo et al. 2016). However, there balance of the glacier. This brings about a change in ice
is great spatio-temporal variability from one site to the other volume. A positive balance overall causes the glacier to
ranging from a few tenths of cm/yr to more than 10 cm/yr. grow while a negative balance leads to a loss of ice volume
Thickening of the active layer is a matter of great concern which can be accompanied by a retreat of the glacial front.
since it may have large consequences on the stability of the The mass balance of temperate glaciers in the mid-latitudes
surface due to the melting of shallow ground ice. Potential is mainly dependent on winter precipitation, summer tem-
impacts include thaw settlement, soil creeps, slope failures perature and summer snowfalls (temporally reducing the melt
and ponding of surface water. All these features can cause due to the increased albedo). In contrast, the glaciers in low
severe damages to infrastructures, such as roads, dams or latitudes, where ablation occurs throughout the year and
structural building foundations but also to vegetation. In multiple accumulation seasons exist, are strongly influenced
forested areas, thaw modifies the hydrological conditions by variations in the atmospheric moisture content which
and can lead, for example, to the destruction of tree roots, affects incoming solar radiation, atmospheric long-wave
causing drastic changes in the ecosystems. When permafrost emission, albedo, precipitation and sublimation. In mon-
thaws and the active layer thickens, more organic matter is soon areas, such as the Himalayas, accumulation and ablation
likely to be decomposed by bacteria that produce either occur mainly in summer. Glaciers at high altitudes and in
methane or carbon. In both cases, the bacterial action polar regions can experience accumulation in any season.
enhances greenhouse gas emissions and thus promotes glo- The retreat of mountain glaciers is one of the most visible
bal warming. However, the magnitude of this thaw-related examples of climate change. A compilation of observations
feedback is a great unknown. The total amount of carbon made by the World Glacier Monitoring Service on 19
stored in the permafrost has been estimated at 1672 Gt, of regions around the world, gathering more than 40,000
which 277 Gt is found in peat bogs. This is about twice the observations, shows that the retreat is an almost global
amount of carbon in the atmosphere. phenomenon, despite some intermittent readvances related to
We still do not know for sure if the increase in atmo- dynamical instabilities (e.g. Island, Svalbard) or to specific
spheric methane concentration observed in recent years, after climatic conditions (i.e. increased winter snowfall) observed
about ten years of relative stability, is due to the warming of on a few individual glaciers in Scandinavia or New Zealand
the high northern latitudes. Another amplification reaction in the 1990s. However, the periods of glacier front advances
observed at high northern latitudes involves the microbial are short compared to the overall ice retreat. By compiling
454 S. Charbit et al.

the data obtained on 169 glaciers since 1700, Oerlemans freshwater supplies. Antarctica is composed of two effec-
(2005) shows that the retreat of glacier fronts began in the tively distinct ice sheets in the east and west, separated by
nineteenth century, and accelerated strongly from 1850 the Transantarctic Mountains. Its ice volume is close to
onwards, with a continuation throughout the twentieth cen- 27  106 km2 and its surface, almost 98% covered by ice, is
tury and early decades of the twenty-first century. about 14  106 km2. A large part of the western ice sheet
Regional analyses have shown that, until around 2000, lies below sea level. The West Antarctic ice sheet extends
the average mass balance cumulated over all European locally over the sea to form floating ice shelves, mainly in
glaciers was close to zero, with significant mass losses for the embayments of the coast, as in the Weddell and Ross
Alpine glaciers being compensated for by advances of gla- Seas. In contrast, East Antarctica, which is larger, rests lar-
ciers in western Norway stemming from a sharp increase in gely on bedrock. It forms a plateau with an area exceeding
precipitation in response to a positive phase of the North 10  106 km2 covered by a large ice layer of more than
Atlantic Oscillation. From the year 2000 onwards, the 4000 m thick in the center.
Norwegian glaciers began to retreat in response to a decrease The evolution of the part of an ice sheet grounded on the
in precipitation. Over the period 2003–2009, the most neg- bedrock depends on its surface mass balance and its flow due
ative mass balances occurred for glaciers in the northwestern to the deformation of the ice itself. When the temperature is
United States and southwestern Canada, Central Europe, high enough, the ice melts at the surface. As for glaciers, the
Southern Andes and low latitude areas. In the Alps, glaciers surface mass balance is determined by the difference
have been retreating since the mid-nineteenth century. In between accumulation and ablation. In addition, under the
Switzerland, they currently cover only 60% of the area they effect of its own weight, the ice flows by plastic deformation
occupied in 1850. Due to the heat wave, an exceptionally along the line of steeper slope, as well as by sliding on the
high loss of mass occurred in 2003, corresponding to a bedrock when the local temperature is close to the melting
reduction of 2500 kg m−2 yr−1 for the nine glaciers studied. point: this is called basal sliding. As ice is an insulating
This value exceeds the previous record 1600 kg m−2 yr−1 in material, a temperature gradient is established between the
1996 and is four times higher than the average measured colder surface and the warmer base. Furthermore, by
between 1980 and 2001 (600 kg m−2 yr−1). In Africa, the changing the ice viscosity, the temperature also affects the
glacier area at the top of Kilimanjaro is now only 20% of flow velocities from the surface to the base of the ice sheet.
what it was at the beginning of the twentieth century. In Thus, the processes involved in the ice deformation are not
Patagonia, the ‘icefields’ have lost between 3 and 13 km3/yr the same at the surface and at the base of the ice sheet.
of ice since the 1970s. The retreat of the glaciers in Nepal In the case of the Antarctic ice sheet, the surface tem-
and Himalayas seems to have accelerated over the past perature generally remains low enough so that ablation is
twenty years, and in Tibet, the number of glaciers in retreat negligible. The ice then drains into the ocean or feeds the
has recently multiplied. floating ice shelves through ice streams which are charac-
The Fifth IPCC Assessment Report estimated that the terized by a rapid outflow (i.e. low basal friction). The
contribution of glaciers and small ice caps to sea level rise sources of these ice streams are found far upstream, and their
over the 1993–2010 period was about 0.76 mm/yr (IPCC contribution to the evacuation of grounded ice from the
2013). Since then, new estimates based on new data indicate center of the ice sheet towards the edges is estimated at
a higher contribution, demonstrating that mass loss from nearly 90%. The flow regime through the ice shelves is very
glaciers and small ice caps has accelerated significantly since different from that of grounded ice. In fact, whereas groun-
the early 1990s, and currently contributes between 1.05 and ded ice is characterized by a shear regime in the vertical
1.12 mm/yr to the rise of the global sea level. plane, the predominant constraints on the ice shelves are the
horizontal shear and the pressure forces exerted by the sea.
Polar Ice Sheets The destabilization of these glacial platforms is due to the
Polar ice sheets are huge masses of ice formed, like glaciers, increase in sea level, but also to the basal melting beneath
by continual accumulation of snow in excess of ablation, the platforms. This melting is therefore related to the ocean
which gradually turns into ice under the effect of com- temperatures under the ice shelves and to the energy released
paction. As for glaciers, this transformation occurs in a by ocean currents. The dislocation of these glacial plateaus
transition zone about one hundred meters thick called firn. leads to the formation of icebergs (i.e. ice calving). This is
Currently, the ice sheets are located at high latitudes, one exactly what happened with the dislocation of the Larsen B
near the North Pole, Greenland, the other centered on the ice shelf in 2002 which resulted in a surface loss of about
South Pole, Antarctica. The area of the Greenland ice sheet 3250 km2. This could also occur in the coming years with
is about 1.8  106 km2. In the center, the ice thickness is the break-up of the Larsen C ice shelf in July 2017
greater than 3000 m. The Greenland ice volume (6000 km2 of surface loss) and the continuous acceleration
(* 3.0  106 km2) represents about 10% of the worldwide of the Twaithes glacier reported by satellite observations
31 From the Climates of the Past to the Climates … 455

since 1992. Moreover, as the ice shelves exert a buttressing example, remote sensing techniques may underestimate
effect for the upstream grounded ice, their disintegration can the extent of the ablation area and the increased rainfall
cause the destabilization of a large part of the ice sheet. over some parts of the ice sheet during rainfall events.
There are different methods to measure the amount of ice
stored in an ice sheet: Past published estimates of Greenland and Antarctic ice
sheets obtained with these methods have often diverged.
(a) The measurement of inflow (snow accumulation) and This was due to the way in which the different sources of
outflow (ablation, discharges to the ocean). Snow uncertainty were estimated and to the fact that the different
accumulation is often estimated from annual layers in measurements did not cover the same regions and the same
ice cores and interpolated between different drilling time periods. Within the IMBIE framework (Ice-sheet Mass
sites. The use of high resolution atmospheric models is Balance Intercomparison Exercise), scientists made a huge
also becoming more common. Discharges of ice to the effort to combine, over common survey periods and common
ocean are estimated from seismic or radar measure- regions, various observations from satellite geodetic tech-
ments of the ice thickness and from measurements of niques (altimetry, interferometry, radio-echo sounding and
ice flow velocity. Ablation is usually determined from gravimetry) with simulated surface mass balance estimates
ice models forced with atmospheric reanalyses, clima- inferred from regional atmospheric models. This allowed for
tology or outputs from global climate models calibrated a reconciliation of the apparent disparities between the dif-
with surface observations. The loss of mass beneath the ferent methods and for a consistent picture of ice-sheet mass
ice shelves remains very difficult to quantify. In general, balance to be created (Shepherd et al. 2012).
inflows and outflows cannot be estimated with a margin For Greenland, this compilation effort confirmed with a
error of less than 5%, which implies uncertainties of 40 high confidence level that the ice sheet is continuously losing
and 140 Gt/yr on the estimate of the mass balance of mass and that this process now affects all sectors of the ice
Greenland and Antarctica respectively. sheet (Fig. 31.6a–c). However, after a record mass loss in
(b) Remote-sensing techniques. These include altimetry summer 2012, Greenland has seen a slight decrease in the
measurements from radio-echo sounding, interferome- short-term mass loss trend. The mass loss is partitioned
try and gravimetry. Interferometry provides information between surface melting and dynamic ice discharges. Shep-
on ice flow velocities. The altimetric measurements herd et al. (2012) estimate that the ice mass loss is
provide information on the spatial and temporal varia- about −142 ± 49 Gt/yr over the IMBIE time period (1992–
tions of the topography, which makes it possible to 2011) with an acceleration of the mass loss rate as illustrated
trace the ice volume variations, after correcting for the by the comparison between the estimations made for 1992–
altitude of the bedrock and for variations in the thick- 2000 (–51 ± 65 Gt/yr) and 2005–2010 (–263 ± 30 Gt/yr).
ness and density of the firn. These measurements are Using gravimetry observations, Velicogna et al. (2014) pro-
strongly dependent on the nature of the terrain (flat, vide a more recent estimate of the Greenland ice sheet mass
sloping or hilly surfaces) and the snow surface condi- loss for the 2003–2013 decade of 280 ± 58 Gt/yr with an
tions (density, viscosity, etc.) and this may, in some acceleration of the loss rate bringing it to 25.4 ± 58 Gt/yr2.
cases, make interpretation difficult. Finally, satellite The case of the Antarctic ice sheet (Fig. 31.6b) is a bit
measurements of the gravity field provide, for the first different since recent observations have shown that mass loss
time, estimates of the ice mass changes. However, there was mainly driven by dynamic ice discharge resulting from
are still significant uncertainties, especially regarding enhanced ice flow of marine-terminating glaciers. The main
the cause of the change in mass, as this could come region experiencing mass loss is the West Antarctic ice sheet
from the ice sheet itself, the air column, the evolution of (WAIS), especially in the Amundsen/Bellingshausen Sea
the subglacial bedrock or even from masses near the ice sectors (e.g. Thwaites and Pine Island glaciers) and, to a
sheet (i.e. ocean mass, masses of water or snow con- lesser extent, in the Antarctic Peninsula. According to a
tained on nearby continents). These different effects are recent update of the IMBIE estimates (IMBIE team, 2018),
evaluated and then corrected, but a significant uncer- the mass loss from the Amundsen and Bellingshausen Sea
tainty remains, mainly on how to correct for the altitude sectors increased from 53 ± 29 Gt/yr to 159 ± 26 Gt/yr
of the underlying Antarctic bedrock. Other sources of over the 1992–2017 period, and from 7 ± 13 Gt/yr to
uncertainty relating to satellite measurements come 33 ± 16 Gt/yr in the Antarctic Peninsula. It has long been
from the fact that they do not provide complete cov- considered that the East Antarctic ice sheet (EAIS) was
erage of the ice sheets. Moreover, in Greenland, for gaining mass due to enhanced precipitation, despite no firm
456 S. Charbit et al.

consensus being established (Velicogna and Wahr 2006; shelves is enhanced. Eventually, this may lead to the dislo-
Ramillien et al. 2006). However, recent estimates suggest cation of ice shelves and to the removal of the buttressing
that some sectors, such as the Wilkes Land region, are losing effect mentioned above. This causes an inland retreat of the
mass. As a result, the rate of change in ice-sheet mass is grounding line (i.e. the limit beyond which ice starts to
estimated to be +11 ± 58 Gt/yr in 1992 (mass gain) and float), and subsequently, an acceleration of the upstream
−28 ± 30 Gt/yr (mass loss) in 2017 (IMBIE team, 2018). grounded ice. Theoretically, this process is only valid for
Using a different technique, Rignot et al. (2019) estimate an ice-shelves confined within their embayment. It is respon-
even larger mass loss from EAIS with a strongly reduced sible for more than half of the ice mass loss at the margins of
uncertainty. the Antarctic ice sheet. For unconfined ice shelves, another
Overall, taking the Antarctic and Greenland ice sheets process, known as the Marine Ice Sheet Instability, may also
together, it appears that mass losses have accelerated in apply. For a marine-based ice sheet, such as the WAIS,
recent years. This trend is correlated with an increase in bedrock is often more depressed in the center of the ice sheet
surface ablation due to increasing temperatures, but also with than it is at the margins. As ice flux increases with ice
an acceleration of ice flow and subsequent dynamic ice thickness, the position of the grounding line becomes highly
discharges. unstable in areas of reverse bed slopes, and any change in ice
Several processes are at the origin of ice mass loss. The thickness in the vicinity of the grounding line creates an
increasing surface ablation, mainly observed in Greenland, is irreversible retreat of the grounding line position. A second
a direct response to increased atmospheric temperatures. hypothesis is the lubrication of the subglacial substratum
However, the ocean warming also plays a key role. As caused by meltwater produced at the surface that percolates
oceanic temperatures rise, basal melting under the ice to the base of the ice sheet. A third process that may be

Fig. 31.6 Temporal evolution of ice loss in Greenland (top) and year for the periods 2003–2012, 2003–2006 and 2006–2012, color
Antarctica (bottom) determined from time variable gravimetry obser- coded red (loss) to blue (gain) (Source IPCC 2013)
vations from the GRACE satellite, shown in centimeters of water per
31 From the Climates of the Past to the Climates … 457

responsible for ice flow acceleration is hydro-fracturing time scale being considered, these types of variations have
(DeConto and Pollard 2016). This mechanism is related to different origins. Over geological time scales, changes in the
water coming from ice melting at the surface of the ice shelf shape of ocean basins and in the continent/ocean distribution
that may percolate inside the ice-shelves. Crevasses can form are the main factors affecting sea level. Over
and widen when the water pressure is high enough, thereby glacial-interglacial cycles, sea-level variations are mainly
favoring iceberg calving. Ultimately, this process may lead related to changes in continental ice volume and to the iso-
to ice-shelf collapse. It may also favor the marine ice sheet static adjustment due to the vertical movements of the
instability. Indeed, once the ice shelves have collapsed, ice Earth’s crust in response to changes in the mass of land ice.
cliffs become unstable and fall down if their height is greater Following the last deglaciation initiated about 21,000 years
than * 90 m. However, this process remains poorly con- ago, sea level rose by * 120 m and then stabilized around
strained and is still a matter of debate. 6000 years ago. Geological data indicate that sea level has
A special attention is given to the evolution of ice sheets. not changed by more than 30 cm from that time until the end
Indeed, ice sheets strongly interact with the other components of the nineteenth century.
of the climate system. These interactions can lead to a highly On time scales ranging from a few years to a few decades,
non-linear climate response. This means that the effects of the variations in the mean sea level are the result of two factors,
radiative disturbance (caused by variations in insolation or in mainly related to climate change:
the amount of greenhouse gas in the atmosphere) can be
amplified or mitigated by these feedback processes, as (a) Variation of the ocean volume caused by changes in sea
illustrated in Chap. 28 for abrupt events. Many studies on these temperature. As temperature increases, the volume of
interactions have been published over the past two decades. water expands. This process is called thermal
They would deserve a full chapter. Here, we only give some expansion;
quick examples for time scales ranging from a few decades to (b) Variations in the bodies of oceanic water resulting
a few centuries. Ice sheet melting is first of all accompanied mainly from exchanges with continental reservoirs,
by possible changes in albedo and therefore in the surface such as rivers, lakes and inland seas, snowpack, ground
energy balance. This effect is almost instantaneous. In turn, a water, but also mountain glaciers and polar ice sheets.
change in the energy balance can lead to changes in the mass
balance of the ice sheets. Another consequence of the melting Other factors such as ocean circulation or atmospheric
and/or mechanical destabilization of the ice sheets, widely pressure can bring about local variations, without altering the
discussed in the literature, concerns the freshwater flux mean global sea level. In addition, some human interven-
released in the ocean. Locally, this release leads to a decrease tions have the effect of modifying regional hydrology by
in ocean surface temperatures, a change in sea ice cover and a modifying the runoff of freshwater released to the ocean, and
reduction of ocean density in the vicinity of the ice sheets. thus the sea level. This occurs in the case of dams, irrigation,
Density changes also cause a disruption of large-scale ocean urbanization, water extraction from aquifers and deforesta-
circulation by altering deep-water convection. For example, tion. Some of these processes have the effect of increasing
meltwater from Greenland has the potential to weaken the runoff (urbanization, deforestation); others, such as dams and
Atlantic Meridional Overturning Circulation. These changes irrigation, contribute to the sequestration of freshwater on the
can have effects in regions far from the polar zones. For continents. Current estimates of the net land water storage
example, the recent study by Defrance (2017) showed that for are based on observations and models, and vary between
a substantial melting of Greenland under a climate change −0.33 and 0.23 mm/yr over the period 2002–2014/15
driven by the RCP8.5 scenario, Greenland meltwater could (WCRP Global Sea Level Budget Group 2018).
cause a massive alteration of the monsoon regime with a Measurements of current sea-level variations are based on
drastic decrease in West African rainfall, and subsequently two different techniques: tide gauges, which began to be
with a significant reduction in cultivable areas. Therefore, in installed in the nineteenth century, and altimetry data from
addition to the direct and obvious consequences on sea level, satellite observations since 1992. The two main limitations
the future of polar ice sheets is of primary importance for the of tide gauges are their inhomogeneous spatial and temporal
future of human societies, all the more so if they are already coverage, and the inclusion of vertical land motion in their
economically fragile. record, which needs to be corrected for to obtain relative
sea-level changes. To limit the uncertainties related to these
movements, which are difficult to quantify in current models,
Sea Level Changes only a few tens of geologically stable sites, mainly located
along the coasts of North America and Europe, are taken into
Sea level variations are the result of changes in both the account to inform us of the sea level evolution in the course
volume of the oceans and ocean basins, as well as changes in of the twentieth century. Based on a compilation of the most
the mass of water contained in the oceans. Depending on the recent estimates, the sea-level rise as indicated by tide gauge
458 S. Charbit et al.

data is estimated at 1.7 ± 0.2 mm/yr between 1901 and Data on ocean temperature changes allow us to quantify the
2010 for a total sea level rise of 0.19 ± 0.02 m (IPCC contribution from thermal expansion. Recent advances in
2013). temperature measurements have greatly improved our
Since 1992 and the launch of the TOPEX/Poseidon knowledge. Since 2000, the Argo project has deployed thou-
satellite, altimetric data have provided a new way of esti- sands of free-drifting profiling floats measuring temperature
mating sea-level variations, based on the time required for (and salinity) in the ocean at depths between 0 and 2000 m,
the round trip of the radar wave and on the satellite altitude providing a continuous record of heat penetrating in the ocean.
defined above a standard reference surface. These mea- In addition, ship-based data were collected during the World
surements are thus independent on the vertical land motion, Ocean Circulation Experiment, providing the means to esti-
in spite of a small correction (around 0.3 mm/year) to mate the deeper ocean temperature change (Johnson et al.
account for the change in the reference level at the ocean 2007; Purkey and Johnson 2010; Kouketsu et al. 2011). These
bottom due to the post-glacial rebound (Peltier et al., 2015). data show that, globally, the ocean has warmed significantly
This results in an increase of about 3.2 ± 0.4 mm/yr over the last 50 years, and in particular, over the past two
between 1993 and 2012 (IPCC 2013, Fig. 31.7), now ree- decades. By vertically integrating the temperature data along
valuated at 3.1 ± 0.4 mm/yr between 1993 and 2017 the water column at each oceanic point, sea-level changes due
(WCRP Global Sea Level Budget Group 2018). Satellite to oceanic thermal changes can be estimated over the past
altimetry data now cover a time period of 25 years, long 50 years. Based on data gathered over a depth of 700 m, the
enough to identify an acceleration of the sea level rise esti- contribution of thermal expansion to sea level rise was esti-
mated at around 0.084 ± 0.025 mm/yr2 (Nerem et al. 2018). mated at 0.60 ± 0.2 mm/yr for the period 1971–2010
These altimetry data show significant regional disparity, (0.8 ± 0.3 mm/yr if the deep ocean contribution is included).
with some regions showing a sea-level rise well above the For the more recent period 1993-2017, this contribution is
global mean and other regions showing a decrease in sea 1.3 ± 0.4 mm/yr (Table 10.1). In recent decades, this has
level. This regional variability partially explains the differ- been the dominant contribution to global sea level rise.
ences between tide gauge data and altimetry data, and is due The other factors contributing to the rise in sea level come
to several factors including density variations, ocean circu- from changes in the oceanic water mass (Table 31.1). One of
lation, atmospheric pressure, and variations within the solid the most important contributions is from mountain glaciers
Earth or the geoid. For example, the Scandinavian shield (excluding Greenland and Antarctica) which cause a global
continues to rise at a faster rate than the mean and, so sea level increase of 0.62 ± 0.37 mm/year for the period
paradoxically, a decrease in local sea level is recorded there. 1971–2010 and 0.65 ± 0.15 mm/year for the period 1993–
Tide gauges and altimetry data provide ways of quantifying 2017. Both polar ice sheets also contribute to the recent rise
the global mean sea level rise and its different components. in global sea level with 0.48 ± 0.10 mm/yr for Greenland

Fig. 31.7 a Yearly average global mean sea level (GMSL) recon- National Oceanic and Atmospheric Administration (NOAA), Goddard
structed from tide gauges by three different approaches: orange from Space Flight Centre (GSFC), Archiving, Validation and Interpretation
Church and White (2011), blue from Jevrejeva et al. (2008), green from of Satellite Oceanographic (AVISO), Commonwealth Scientific and
Ray and Douglas (2011), b Changes in the mean global sea level from Industrial Research Organisation (CSIRO) with the mean of the five
altimetry data sets from five groups (University of Colorado (CU), shown as a bright blue line (Source IPCC 2013)
31 From the Climates of the Past to the Climates … 459

Table 31.1 Sea level rise from Components Sea level Sea level rise (mm/yr) Sea level rise (mm/yr)
different sources, adapted from rise (mm/yr) 1993–2017 2005–2017
IPCC (2013), Chapter 13, with 1993–2010 (WCRP Global Sea Level (WCRP Global Sea Level
additional data from WCRP (IPCC, Budget Group, 2018) Budget Group, 2018)
Global Sea Level Budget Group 2013)
(2018). The percentages are
relative to the sum of 1. Thermal expansion 1.1 ± 0.2 1.3 ± 0.4 (48%) 1.3 ± 0.4 (44%)
contributions 2. Glaciers (excluding 0.76 ± 0.37 0.65 ± 0.15 (24%) 0.74 ± 0.1 (25%)
Greenland and Antarctica)
3. Greenland ice sheet 0.33 ± 0.08 0.48 ± 0.10 (18%) 0.76 ± 0.1 (26%)
4. Antarctic ice-sheet 0.27 ± 0.11 0.25 ± 0.10 (9%) 0.42 ± 0.1 (14%)
5. Land water storage 0.38 ± 0.11 / −0.27 ± 0.15 (−9%)
Total of contributions (1 + 2 2.8 ± 0.5 2.7 ± 0.23 2.95 ± 0.21
+3 + 4 + 5)
Observed global mean sea 3.2 ± 0.4 3.07 ± 0.37 3.5 ± 0.2
level rise
Thermal 3.6 ± 0.4
expansion + GRACE-based
ocean mass

and 0.25 ± 0.10 mm/yr for Antarctica for 1993–2017. The only the temperatures change, and that this affects only the
ice-sheet contribution to sea-level rise has been increasing radiation emission law, without modifying any radiative
from 27% of the total contributions for 1993–2017 to 40% property of the atmosphere or the surface. For example, with
for 2005–2017. a doubling of atmospheric CO2 concentration, a temperature
The sum of all contributions, including land water stor- increase of 1.2 ± 0.1 °C is obtained. However, these sim-
age, amounts to 2.8 ± 0.5 mm/yr for the period 1993–2010, plified assumptions are not reliable, because when the tem-
which is lower than the observed global mean sea level rise perature changes, all the other climate variables (e.g.
of 3.2 ± 0.4 mm/yr (Table 31.1). Similarly, for the most humidity, wind, clouds, rain, snow cover) also change. These
recent period 2005-2017, the sum of contributions is changes can in turn modify the energy balance of the surface
2.95 ± 0.21 mm/yr, lower than 3.5 ± 0.2 mm/yr which is and the atmosphere and thus have an additional effect on
the observed global mean sea level rise. This discrepancy is temperatures. These are called feedback processes. They are
due to uncertainties in the estimation of the different com- said to be positive when they amplify the initial disturbances,
ponents of ocean mass contributions (glaciers, ice sheets and and are said to be negative when the opposite is true,
land water storage). Instead, if the ocean mass contribution is when they work towards the stabilization of the system.
taken from gravity measurements using GRACE (Gravity The first studies which took these feedbacks into account
Recovery And Climate Experiment), the sum of thermal were carried out using radiative-convective models, with
expansion and GRACE-based ocean mass contributions is only one vertical dimension. For example, Manabe and
3.6 ± 0.4 mm/yr, which is in the error bar of the observed Wetherald (1967) showed with their model, that the surface
global mean sea level rise for this period (3.5 ± 0.2 mm/yr). warming due to a doubling of the atmospheric CO2 con-
centration was 1.3 °C when the absolute humidity of the
atmosphere remained constant, but reached 2.4 °C in case of
Climate Modeling and Recent Changes constant relative humidity. Numerous other studies have
confirmed the crucial importance of feedback mechanisms
Simple Radiative Climate Models and Their on the magnitude of global warming: they can amplify twice
Limitations to four times the temperature variation compared to the sit-
uation where no feedback is taken into account. These
To estimate the changes in the mean Earth’s temperature in studies have also shown that the magnitude of these feed-
response to different radiative forcings (solar irradiance, backs is strongly dependent on complex physical processes
greenhouse gases, energy re-emitted by the surface etc.), a (less understood than radiative transfer), such as turbulence,
first approach is to use purely radiative models. With these convection, cloud formation and precipitation (Ramanathan
models, one can easily and accurately calculate the temper- and Coakley 1978). These processes, and in particular, the
ature changes with some simplifications: it is assumed that atmospheric circulation which determines how energy and
460 S. Charbit et al.

water vapor are redistributed within the atmosphere, cannot In parallel, ocean general circulation models were
be represented in a useful way in the radiative-convective developed to simulate heat transport and to study the role of
models. Thus, even rough estimates of the changes in the the ocean in the Earth’s energy balance. Progressively, they
global mean temperature require the atmospheric dynamics included sea-ice models and, from the 1990s onwards, they
to be taken into consideration and, for more precise calcu- were coupled with atmospheric models to simulate the
lations, three-dimensional models representing the general atmosphere-ocean-sea ice interactions. These early models
circulation of the global atmosphere of the Earth are did not simulate the heat and water fluxes at the air-sea
needed. interface, resulting in strong biases in the simulated oceanic
surface temperatures. To fix these shortcomings, the fluxes at
the air-sea interface were corrected in an ad hoc way, before
General Circulation Models: Progress these corrections were gradually removed after the end of the
and Limitations 1990s, thanks to continuous model improvement.
These coupled atmosphere-ocean models gradually
The Evolution of Climate Models became the basic tool for studying both past and future
A general circulation climate model is a simplified repre- climate variations. For example, in preparation for the
sentation of the climate system, but including as best as Fourth IPCC Assessment Report (IPCC 2007), some twenty
possible most processes that influence the climate. It is based of these coupled models performed a whole series of climate
on a preliminary physical analysis, in order to reduce the change simulations, and only six of them were based on flux
number of processes to be incorporated, as well as on correction at the air-sea interface. These models can simulate
appropriate mathematical and numerical formulations. a natural climate variability that can then be compared to
Numerical modeling of the climate began in the 1970s and observations at different time scales: a few days, a few years
has since greatly progressed, thanks to the steady increase in (interannual variability, the best known of which is El Niño),
the processing power of computers. The general philosophy or a few tens, or even several hundred years.
behind this development, established by Charney and his In a schematic way, climate models simulate the energy
collaborators in the 1950s, was to understand the problem on and water cycles. Progressively, representations of chemical
a global scale, even if this meant making very general reactions in the atmosphere, biogeochemical cycles and the
approximations initially, with the aim of gradually improv- transport of species were introduced into modeling so as to
ing the model by identifying its drawbacks and its limita- study new aspects of climate variations: the effect of aero-
tions. The first models only described the atmosphere and sols, coupling between climate change and the chemical
the continental surfaces. In order to reduce model com- composition of the atmosphere, and between climate change
plexity, the oceanic surface temperature was imposed: even and the carbon and methane cycles. This required advances
if the energy balance of the ocean surface is very different in our understanding of each of the components of the
from the observed measurement, the surface temperature is system: atmosphere, ocean, vegetation, continental surface,
maintained at its prescribed value. However, to investigate sea ice. Numerical climate models progressively incorporate,
the variations in climate accounting for oceanic temperatures in a coherent way, a wide range of physical processes
is required. governing the climate variations and the interactions
The first studies of the impact of a CO2 doubling using between the different climate components. On the other
this type of model were carried out in the 1970s at the GFDL hand, the inclusion of a growing number of physical pro-
(Geophysical Fluid Dynamics Laboratory, Princeton, USA) cesses has made the models more complex and more difficult
with a representation of the ocean with no circulation and to develop and evaluate.
zero heat capacity, so that equilibrium with the atmosphere As ice sheets interact with atmosphere, ocean and vege-
could be achieved quickly. In this model, there was no tation, the next key challenge for the climate modeling
diurnal or annual variation in insolation, and ad hoc cor- community is to incorporate ice-sheet models in general
rections were applied to heat fluxes at the air-sea interface to circulation climate models. This is a necessary step to obtain
keep the ocean surface temperature close to observations. a comprehensive representation of the climate system for
The use of this type of model became widespread during the past, present and future time periods.
1980s, with a gradual increase in sophistication and realism.
For example, both annual and diurnal insolation variations What Are the Uncertainties Inherent in Climate
were incorporated, modeling of cloud formation processes Models?
began, etc. At the same time, new satellites were being used The climate is characterized by a very wide range of both
to estimate global cloud cover and radiative fluxes at the top spatial (from the micrometer to several thousand kilometers)
of the atmosphere, which contributed to the improvement and temporal (from the second to several thousand years or
and evaluation of atmospheric models. more) scales. The processes at these different scales interact
31 From the Climates of the Past to the Climates … 461

with one another, and in principle, it is never possible to However, this does not explain the role that clouds could
know which scales should be considered and how to rep- play in global warming. Depending on how their properties
resent the neglected scales in a simplified way. A typical change, clouds may attenuate or, on the contrary, amplify,
example is the formation of clouds and precipitation. Let’s global warming. The physical formation mechanisms
take the example of convective clouds (of the cumulonimbus involve so many processes and spatial scales (from a
type), whose core is a rising column of moist air in which the micrometer to a thousand kilometers), and their radiative
water vapor condenses as it rises. This ascending column properties depend on so many factors that it is impossible to
mixes with the surrounding drier air, and this mixing conclude on the basis of theory, simple reasoning or analysis
depends on many factors (for example, the intensity of the of available observations, how they will evolve in the future.
upward thermal current and wind shear). In order to take
these mixtures into account, it is first necessary to know
precisely the vertical profile of the atmospheric variables in Simulation of the Current Climate and Recent
the vicinity of the column. The turbulent exchanges between Changes
the column and its environment must also be calculated, as
well as the coupling between these turbulent exchanges and Analysis of simulated climate combined to the comparison
the formation or dissipation of rain drops, hail or snowflakes. of model results with observations is an important step to
This requires a modeling on a very small scale (a few establish the reliability of climate models. The aim is to
hundred nanometers to a few meters), which is not possible evaluate not only the mean climate, but also the climate
with global models. Therefore, a simplified model must be variability at different time scales (from a few days to a few
developed based only on large scale variables which repro- decades) as well as recent climate changes. Unless otherwise
duce the effect of unresolved small scale processes. This type indicated, this section presents the simulated climate char-
of modeling, called parameterization, is based on important acteristics from the twenty AOGCMs that contributed to the
simplifications that nevertheless require a thorough physical preparation of the Fifth IPCC Assessment Report (IPCC
analysis and an in-depth understanding of the phenomena. 2013).
The aim is to obtain a simplified model that is not only as
accurate as possible, but also justified and well understood. Mean Climate
There are many parameterizations in a climate model The difference in mean insolation is what causes the tem-
which can be directly related with the atmospheric circula- perature difference between the equator and the poles. This
tion (for example gravity waves and orographic effects), the difference is the main driver of atmospheric and oceanic
calculation of radiative exchanges, deep convection, or circulations, which act to reduce the equator-pole tempera-
boundary layer phenomena. Many of these parameteriza- ture gradient. It is also influenced by the presence of clouds,
tions play a key role in the water cycle, the formation of reflective surfaces (snow, glaciers, sea ice), large mountain
clouds and their radiative properties, precipitation, heat and ranges and the topography of the ocean. Latitudinal tem-
water fluxes on the surface of continents or oceans, among perature variations are thus a key criterion for evaluation of
others. All these phenomena have an impact on the simu- climate models. All the models simulate this strong
lation of the current climate and as there are strong inter- equator-pole gradient: the simulated temperature is 25 °C at
actions between them, it is often very difficult to identify the the equator, −20 °C at the North Pole and −40 °C at the
precise role of each of the parameterizations on the simu- South Pole, which agrees with observations. However, the
lation of these phenomena, and in particular to understand models also show significant biases over Antarctica,
why some of them are poorly simulated. Greenland, and over large mountain ranges in general, such
Parameterizations also play a very important role in the as the Himalayas. These are due to an approximate repre-
climate response to different forcings, and in the simulation sentation of the topography (due to the limited spatial res-
of past and future climate changes. One example is the olution of the models) and a poor representation of turbulent
simulation of clouds. Clouds exert two opposing effects on exchanges under conditions of strong thermal stability. Over
the terrestrial radiative balance: on the one hand, they reflect the oceans, there is a warm bias on the eastern coasts caused
part of the solar radiation, and on the other, they absorb by a poor representation of stratus clouds observed in these
infrared radiation and thus contribute to the greenhouse regions.
effect. The relative importance of these two effects depends The annual variation in solar radiation is the strongest
on many factors, in particular, cloud altitude. Over the past energy ‘disruption’, apart from the diurnal cycle, to which
twenty years, we have learned that, on average, the first the surface of the Earth is subjected. The observed seasonal
effect outweighs the second, and therefore that clouds have a temperature cycle is generally well reproduced by the
cooling effect on the climate, especially low-lying clouds, models: it is higher at high latitudes than at low ones (30 °C
because they have little impact on infrared radiation. vs. 5 °C, essentially reflecting the seasonal amplitude of
462 S. Charbit et al.

incoming solar radiation), and is higher over continental Climate Variability

areas than above the oceans, mainly because of the lower The mean climate state gives a very incomplete picture of
thermal inertia of the continents. the climate. Indeed, climate varies continuously over a very
The formation of precipitation involves numerous pro- wide range of space and time scales with fluctuations having
cesses, some of them on a small scale, and remains one of the hourly, daily, interannual, decadal or even longer time
phenomena that models have the largest difficulties in cor- scales.
rectly simulating. Around the equator, the area of maximum One way to characterize these fluctuations is to select
precipitation corresponding to the intertropical convergence those having a large-scale spatial structure (typically, the
zone, is well simulated by the models (Fig. 31.8). In the Pacific scale of an ocean basin or a continent), whether fixed or
Ocean, observations show that the maximum is at 10°N, spreading, and to characterize its temporal evolution (am-
indicating that this convergence zone is located mainly in the plitude, phase). To describe these fluctuations, we refer to
northern hemisphere. The models, on the contrary, generally what we call modes of variability. Some of which are well
have two maxima located on either side of the equator, the known, such as the El Niño-Southern Oscillation (ENSO) in
southern maximum extending almost to the coasts of Peru, the tropical Pacific or the North Atlantic Oscillation (NAO),
although observations indicate that precipitation is almost which dominates the weather and climate fluctuations over
absent in this region. In observations, the maximum rainfall Europe.
ranges from northern New Guinea to southern South America. ENSO is the dominant mode of tropical variability at the
On the continents, one of the biggest flaws of the models is that interannual to decadal time scale. The warm phase, or the El
the intensity of rainfall over the Amazon is too low. Niño event, is characterized by a warming of the eastern
The extent and characteristics of sea ice are relatively tropical Pacific (along the equatorial cold water tongue) and
well simulated by current models, both in terms of mean an easterly displacement of the zone of maximum precipi-
value and seasonal cycle. However, in the Arctic basin, few tation, usually located above Indonesia. The cold phase, or
models are able to successfully replicate the distribution of La Niña, is characterized by negative sea surface tempera-
sea ice thickness, which should be less than 1 m north of ture anomalies (i.e. temperatures below the climatological
Siberia, to more than 5 m north of Greenland and of the mean) and can be interpreted as an enhancement of the
Canadian archipelago. This difference in ice thickness is seasonal climatological cycle. This mode of variability is a
mainly due to winds, which move ice from the Siberian coupled ocean-atmosphere oscillation that affects atmo-
coasts to Canadian shores and Greenland. These winds spheric circulation throughout the tropical belt and ocean
allowed FRAM, the F. Nansen’s boat, to cross the Arctic circulation throughout the Indo-Pacific basin. Its periodicity
Ocean at the end of the nineteenth century although it was is between three and seven years, and the characteristics of
trapped in the ice. The spatial distribution of ice thickness is ENSO events observed in the twentieth century can vary
generally poorly simulated by the models due mainly to considerably from one event to another. All current climate
poorly simulated surface winds. models simulate a mode of tropical variability with general

Fig. 31.8 Annual-mean precipitation rate (mm/day) for the 1980– between multi-model mean and precipitation analyses from the Global
2005 period. a Multi-model-mean constructed with one realization of Precipitation Climatology Project (Adler et al. 2003) (Source IPCC
all available AOGCMs used in the CMIP5 (Coupled Models 2013)
Intercomparison Project, Phase 5) historical experiment. b Difference
31 From the Climates of the Past to the Climates … 463

characteristics resembling those of ENSO. This was not the radiative forcing of about 2.5 W/m2. The uncertainty in this
case with the previous generation of climate models. Nev- forcing is estimated to be quite low at less than ±10%. Since
ertheless, the spatial structure of these events is not generally fossil fuels contain sulfur, CO2 emissions are accompanied
well simulated, such as the asymmetry between the El Niño by SO2 emissions which lead to the formation of sulfate
and La Niña episodes. In terms of the recurrence of these aerosols. In 2000, these aerosols produced a radiative forcing
events, the periodicity simulated by the models is usually too of about − 1 W/m2, but, depending on how it is estimated,
short and too regular. In general, the strong diversity of this value varies from − 0.5 to −2 W/m2. In addition, other
observed spatial and temporal characteristics of ENSO is aerosols, such as soot or aerosols from biomass fires, may
often poorly simulated. Research works are being done to also play an important role, but their effects are even less
identify the impact of the different atmospheric and oceanic well known. Thus, about a third of the positive radiative
processes on the characteristics of ENSO, and on the reasons forcing (albeit with a high degree of uncertainty) from the
for model errors. increase in greenhouse gases is masked by the negative
The Madden and Julian—MJO—oscillation (Madden and radiative forcing from aerosols. In addition to anthropogenic
Julian 1994) is the main mode of intra-seasonal variability in forcings, there are natural forcings. At the century time scale,
the tropical region, with a periodicity of between 30 and aerosols are mainly injected into the stratosphere by varia-
90 days. Unlike ENSO, whose spatial structure is stationary, tions in the intensity of incoming solar radiation, and by
this oscillation is characterized by a wave that propagates strong volcanic eruptions where they can remain for several
from west to east, the intensity of the convection alternating months or even years. These aerosols reflect solar radiation,
between reinforced and reduced. An MJO- type signal is creating a negative forcing. At the end of the twentieth
present in the results of most models, but several essential century, strong volcanic eruptions were more frequent than
characteristics of this mode of variability (amplitude, phase, at the beginning, resulting in an enhanced radiative forcing.
propagation) are not realistic. The roles of the different Therefore, regardless of the climate response, there is
processes and their interactions in the characteristics of this already an inherent uncertainty in the radiative forcing of
mode of coupled atmosphere-ocean variability have not been about ± 50% (IPCC 2013).
well identified and several hypotheses have been proposed. When only natural forcings are considered, simulated
In the extratropical regions, an important mode of vari- warming is not in line with observations, especially since the
ability is the North Atlantic Oscillation (NAO). It is a 1980s (Fig. 31.9). However when all forcings are taken into
pressure oscillation between the temperate and the subpolar account (natural + anthropogenic), all models are able to
latitudes, often defined as the difference in normalized correctly simulate the increase in the mean Earth temperature
pressure between the Icelandic low and the Azores anticy- over the past 150 years (Fig. 31.9). Models simulate a
clone. It is associated with changes in prevailing westerly greater increase in temperature over the continents than over
winds throughout the North Atlantic Basin, and affects the the ocean and a geographic distribution of warming in
climate of Europe and its surroundings. For example, the agreement with observations (IPCC 2013). Thus, tempera-
positive phases of NAO (NAO+) are associated with a ture changes over the past 150 years make it possible to
northward shift of low pressure systems with mild and rainy verify that the climate response simulated by the models is
winters and droughts in southern Europe. Current models coherent with the observed temperature variations, although
correctly simulate the spatial properties of the NAO, but are there is too much uncertainty in the forcings to be able to
not as good at simulating its temporal properties. In partic- constrain precisely the climate sensitivity of the models.
ular, the current trend of the NAO (an increase in positive The aerosols have two opposing effects: one warming the
phases) is underestimated by the models. The reasons for surface, the other cooling it. The first one, called the direct
this underestimation are diverse: poor representations of i/the effect, is the diffusion of incoming solar radiation, where part
interactions between the stratosphere and the troposphere, of the solar radiation is returned to space. The second, the
ii/the exchanges between stationary waves and transient indirect effect, is the modification of the optical properties of
activity (storms) and iii/the exchanges with the surface of the clouds: the presence of a large number of aerosols increases
ocean. the number of condensation nuclei. For the same amount of
liquid water, more numerous drops forming the clouds tend
Recent Evolution of the Climate to have a smaller radius and thus to diffuse more solar
Over the last 150 years, the evolution of the global surface radiation. Aerosols are also likely to modify the formation of
temperature is documented with a large set of observations. rain, and therefore the liquid water content of clouds. The
Simulating this evolution is therefore one way to test climate complexity of the radiative properties of aerosols in the
models. Between 1850 and 2000 the steady increase in the atmosphere make it more difficult to model the impact of
concentration of greenhouse gases has led to an increase in cloud physics.
464 S. Charbit et al.

In response to the radiative forcing DQ, the different

climate models simulate a temperature change at equilibrium
DT which is different from one model to another. These
differences are difficult to analyze and interpret directly
because of the high level of model complexity. A standard
method to identify the origin of these differences is the
feedback analysis (see, for example, Dufresne and Saint-Lu
2016). From the temperature change at equilibrium DT, a
‘feedback parameter’ k is defined:

k ¼ DQ=DT:
By writing this equation in the form k  DT +DQ = 0,
we see that k  DT represents the variation in the average
flux at the top of the atmosphere necessary to compensate for
the radiative forcing DQ. We can therefore write k = −dF/
dT, with F being the net radiative flux at the top of the
atmosphere, counted as positive when it is descending. This
derivative can be decomposed, in the first order, as a sum of
partial derivatives:

k ¼ R@F=@X  @X=@T
The sum over X is the sum of all the X variables affecting
Fig. 31.9 Temporal evolution of the global surface air temperature of the radiative balance at the top of the atmosphere and that are
the Earth: observations (black line) model simulations taking only modified when the surface temperature changes. These are
natural forcings into account (top), and taking both natural and mainly the three-dimensional fields of temperature, water
anthropogenic forcings into account (bottom). The red and blue lines
vapor and clouds, and the two-dimensional fields of surface
represent the multi-model average of the CMIP3 and CMIP5 (Coupled
Models Intercomparison Project, Phases 3 and 5) models respectively albedo. The change in the temperature field is generally
(Source IPCC 2013) broken down into two terms, one corresponding to a uniform
temperature change, the other to the non-uniform part of the
temperature change. Finally, the parameter k can be
Projecting the Future of the Climate System decomposed as follows:

k ¼ kP þ kL þ kc þ kw þ ka
Climate Response to a Doubling of CO2: Forcings
and Feedbacks The terms of the right-hand side are the respective feedback
parameters: Planck kP (uniform temperature change), the
Perturbations that modify the energy balance of the climate temperature gradient kL (non-uniform part of the temperature
system are quantified in terms of energy flux at the top of the change), clouds kc, water vapor kw and surface albedo ka.
atmosphere. Quantifying the forcing due to a variation in These parameters are often calculated as follows using the
average insolation is immediate. For a change in the partial radiative perturbation method. For a given climate
greenhouse gas concentration, a radiative model is used to model, two simulations are carried out, a reference one and a
calculate how these changes affect fluxes at the top of the perturbed one. The fluxes at the top of the atmosphere are
atmosphere. Since the late 1980s, it is now possible to per- calculated off-line from the outputs of the reference simu-
form these calculations using radiative transfer codes and lation using a radiative code. They are then recalculated by
spectral databases, provided that all other atmospheric (e.g. replacing the values of some variables (temperature,
clouds, aerosols) and surface (e.g. snow cover) characteris- humidity, clouds, surface albedo) from the reference simu-
tics are assumed to be fixed. For a doubling of the atmo- lation with the corresponding values from the perturbed
spheric CO2 concentration, the global mean annual radiative simulation. The difference between these two fluxes gives
forcing at the top of the atmosphere is 3.7 ± 0.2 W/m2. As the sensitivity of the fluxes at the top of the atmosphere to a
we focus on slow climate variations, the calculation of the perturbation of each of the variables.
radiative forcing takes into account the rapid adjustment of The particular Planck parameter kP corresponds to the
the stratospheric temperature. response of the flux at the top of the atmosphere to a uniform
31 From the Climates of the Past to the Climates … 465

temperature change at the surface and within the whole water vapor. If there is little variation in the relative
atmosphere. Its value is approximately − 3.2 W/m2 K. There humidity, this results in an increase in the concentration of
is very little change in this value from one model to another, water vapor in the atmosphere, and therefore in the green-
and the sign convention used corresponds to a decrease in house effect, thus constituting a very powerful amplification
the Earth’s energy balance when the surface temperature mechanism of the warming: 1.7 °C on average for the
increases. In response to a radiative forcing, DQ, the surface models considered here.
temperature response can be calculated if the Planck For thermodynamic reasons (variation of the adiabatic
parameter kP is the only non-zero feedback parameter: temperature gradient as a function of humidity), it is also
expected that, in the case of a humid tropical atmosphere,
TP ¼ DQ=kP : variations in water vapor will be accompanied by greater
This so-called Planck response causes an increase of 1.2 ° warming at high altitudes than close to the surface (except at
C for a forcing of 3.7 W/m2, resulting from a doubling of the high latitudes). This increases the emission of infrared
CO2 concentration. It is the response of an idealized system radiation from the upper atmosphere and is the only negative
in which only the atmospheric and surface temperatures can feedback: it decreases warming (− 0.8 °C on average). As
change uniformly and where only the radiation emission law these two feedbacks are very strongly correlated for physical
is affected (see Sect. “Simple Radiative Climate Models and reasons, their combined effect is generally considered to
Their Limitations” of this chapter). It can be said that this is contribute about 1 °C to the increase in mean temperature
the response with no feedback from the climate system. By (Fig. 31.10a).
combining the above equations, the increase in temperature The mechanisms behind the surface albedo feedback are
at equilibrium DT can be written as a function of the Planck also well identified: an increase in temperature increases the
response: melting of snow and ice. This decreases the area covered by
surfaces that reflect solar radiation, and therefore increases
DT ¼ DTP =ð1  gÞ the amount of energy absorbed by the Earth. This effect
contributes about 0.2 °C to the temperature increase
where g is the feedback gain of the system: (Fig. 31.10a).
g ¼ ðkL þ kc þ kw þ ka Þ=kP : Finally, the increase in temperature is likely to impact
cloud cover. As seen before, clouds exert two opposing
If the gain g is positive and less than 1, the feedbacks will effects on the terrestrial radiative balance: on the one hand,
amplify the temperature increase DT, relative to DTP. Con- by reflecting solar radiation, they contribute to a cooling of
versely, if the gain is negative, the DT increase will be the Earth, and on the other, by absorbing infrared radiation,
attenuated. In the preparation of the Fifth IPCC Assessment they contribute to increase the greenhouse effect. The rela-
Report (IPCC 2013), climate change simulations were con- tive importance of these two effects depends on multiple
ducted within the CMIP5 project by around forty climate factors. The average contribution of clouds estimated by the
models. In particular, for a doubling of the CO2 concentra- models considered here is equivalent to a warming of 0.7 °C
tion, the models simulate a global warming of 3 °C on (Fig. 31.10a). However, the dispersion between models is
average (between 2.0 and 4.6 °C depending on the model), high (Fig. 31.10b): while some models predict a relatively
until a new energy equilibrium is found. We have seen that neutral cloud response, most predict a decrease in cloud
in the absence of feedbacks, this warming would be 1.2 °C. cover as temperature increases and an increase in global
This means that climate feedbacks amplify this warming by warming of up to 2 °C.
a factor of 2 to 4 depending on the model. Recent studies indicate that this uncertainty stems mainly
Other developments make it possible to use the feedback from the way different climate models predict the response
parameters to estimate the temperature increase due to the of low clouds (stratus, stratocumulus or small cumulus
Planck response and to the various feedbacks (Dufresne and clouds) to global warming. The way other clouds (including
Bony 2008). They were applied to twelve CMIP3 (Coupled large cumulonimbus storm clouds) respond to climate
Models Intercomparison Project) models, and the results are change is also uncertain. This response contributes only
shown in Fig. 31.10 illustrating both the average contribution poorly to the uncertainty on the magnitude of global
of the models and the inter-model dispersion. These results warming, but contributes strongly to uncertainties in regio-
were confirmed for the CMIP5 models (Vial et al. 2013). nal precipitation changes associated with global warming.
Several of the mechanisms governing the feedback Other feedbacks exist in the climate system, for example,
parameter values, and hence the gain, are now well identified those potentially causing changes in the atmospheric carbon
(Bony 2006). For example, an increase in the temperature of storage capabilities by the ocean and the biosphere. They
the atmosphere increases the saturation vapor pressure of the will be discussed below (Sect. “The CO2 Cycle”).
466 S. Charbit et al.

Fig. 31.10 For a doubling of

atmospheric CO2: a bar on the
left, multi-model mean ± 1
standard deviation of increase in
global temperature (DTe, °C) and,
bar on the right, part of this
increase due to the Planck
response and to the different
feedbacks: combined effects of
water vapor and temperature
gradient (WV + LR), surface
albedo and clouds. b Standard
deviation of the inter-model
difference in temperature increase
attributable to radiative forcing,
the Planck response, and various
feedbacks, normalized by the
standard deviation of the increase
in global temperature. According
to Dufresne and Bony (2008)

Future Scenarios models for sulfate aerosols. The concentrations of each of

these constituents can then be used to calculate the corre-
It is important to keep in mind that the future climate cannot sponding radiative forcing. For example, the evolution of the
be accurately predicted. This can be explained by several different forcings from 1860 to 2100 for the RCP8.5 sce-
reasons including model uncertainties, uncertainties in sce- narios is shown at the bottom of Fig. 31.11.
narios of future greenhouse gas emissions and uncertainties
in natural disturbances, such as volcanism, having a strong
impact on radiative forcing. However, we can try to answer The CO2 Cycle
specific questions: independently of the natural forcings,
how would the climate evolve if greenhouse gas emissions As described in Chap. 23, the carbon cycle is strongly linked
followed this or that emission pathway? To this end, various to the living world. Understanding the roles of the two main
socio-economic scenarios for the evolution of human reservoirs, the ocean and the terrestrial biosphere, is there-
activities have been established. Four representative con- fore crucial to understanding how atmospheric CO2 will
centration pathways (RCP) scenarios have been selected, evolve in the coming decades. The ocean is the largest
labeled by a value that corresponds approximatively to the carbon reservoir (holding 40 times the atmospheric content)
radiative forcing in 2100: RCP2.6, RCP4.5, RCP6.0, and can regulate the atmospheric CO2 concentration by
RCP8.5 (top of Fig. 31.11). On one extreme, CO2 emissions exchanging CO2 with the atmosphere. These exchanges
rapidly stabilize and then decrease down to zero before 2100 depend largely on the vertical carbon gradient in the ocean:
for scenario RCP2.6. On the other extreme, CO2 emissions surface waters are depleted of inorganic carbon and mineral
continue to grow until 2100 for scenario RCP8.5, which is salts; conversely, the deep ocean is enriched in carbon and
for now the most realistic one. CO2 emissions are mainly mineral salts. The existence of this vertical gradient is driven
due to the use of fossil fuels (oil, coal, gas) and to the SO2 mainly by ocean biology, which establishes a biological
emissions from the sulfur contained in these fuels. Mainly pump and transfers carbon from the surface to the bottom.
for health reasons, fuels are increasingly purified of sulfur Any modification of this pump may ultimately influence the
before use, resulting in a slower increase in CO2 emissions atmospheric reservoir.
(or a faster reduction) than for CO2 in almost all scenarios. What about anthropogenic disturbances and their impacts
For the last IPCC exercise (IPCC 2013), the concentrations on the carbon cycle? These impacts are manifold: a direct
of the different gases were calculated from their emissions by chemical consequence of the increase in atmospheric CO2 is
biogeochemical cycle models and by chemical-transport the acidification of the oceans; a thermal consequence of the
31 From the Climates of the Past to the Climates … 467

biological pump, this will depend on how life functions and

adapts in a warming world, which is not easy to predict.

Climate Projections for 2100: What

the Models Say: Main Climate Characteristics

In this section, the results are derived, unless otherwise

indicated, from simulations performed by the forty or so
coupled atmosphere-ocean general circulation models (or
climate models) that contributed to the CMIP5 exercise. The
values given below are the average of the models ± 1
standard deviation.

The Amplitude of the Warming

In response to both anthropogenic (greenhouse gases and

aerosols) and natural (volcanoes and solar intensity) forc-
ings, the models simulate an average global increase in air
temperature near the surface of about 0.8 °C between the
beginning and the end of the twentieth century, consistent
with observed measurements. Sensitivity studies, with dif-
ferent forcings, have shown that this warming is mainly due
to anthropogenic forcings. Today, the climate system is out
of balance; if the concentrations of greenhouse gases and
aerosols were maintained at their 2000 values, the climate
would continue to warm by around 0.4 °C through the
twenty-first century. However, the simulated increase in
temperature depends above all on the emission scenarios.
Indeed over twenty models, when the concentration of
Fig. 31.11 Top: Time evolution of the total radiative forcing (in
W/m2) due to human activities for the three SRES emission scenarios greenhouse gases is prescribed, there is on average a varia-
used in IPCC-AR4 (IPCC 2007) and the four RCP scenarios used in tion of a little less than 1 °C for the RCP2.6 scenario and of
IPCC-AR5 (IPCC 2013). Bottom: Temporal evolution of the radiative 4 °C for the RCP8.5 scenario (Fig. 31.12). The dispersion of
forcing (in W/m2) due to the different chemical agents for the scenario
the simulated warming by the different models at the end of
RCP8.5 (Szopa et al. 2013)
the century is ± 0.5 °C, the main reasons for which are
outlined in Sect. “Climate Response to a Doubling of CO2:
warming of the water is a decrease in the capacity of the Forcings and Feedbacks” of this chapter.
ocean to absorb CO2, and a biological effect is a modification
in the distribution of species due to environmental changes
(salinity, temperature) which can cause transformations in the Geographical Distribution of Temperature
trophic chain and therefore in the biological pump. In terms Changes
of CO2 sinks, it is now known that for two CO2 molecules
emitted, only one remains in the atmospheric reservoir while The geographical distribution of the temperature increase is
the other is stored either in the terrestrial biosphere or in the roughly similar for the different scenarios (although the
ocean. This ratio is the result of many processes involved in amplitudes are very different). Figure 31.13 illustrates the
the regulation of the carbon cycle; therefore, due to the RCP2.6 and RCP8.5 scenarios. It shows results that are now
anthropogenic perturbation it is expected to vary. The well established: the temperature increase is higher on land
capacity of continents to store carbon can change either than on the oceans, and is particularly strong in the high
gradually or abruptly. For example, permafrost thawing in latitudes of the northern hemisphere.
Siberia would release the equivalent of an additional In the tropics, the greater increase in temperature over land
100 ppm of CO2 at least into the atmosphere. As for the than over ocean is partly explained by changes in
468 S. Charbit et al.

and where the ocean temperature remains homogeneous to a

considerable depth. In order for the temperature of the ocean
surface to increase, a large amount of water must be heated
up.
The larger increase in temperature in the high latitudes of
the northern hemisphere is partly due to the surface albedo
feedback. The increase in temperature is accompanied by a
decrease in snow cover and sea ice extent in spring and
summer. This reduces the reflection of solar radiation at the
surface, increases the amount of radiation absorbed and
Fig. 31.12 Multi-model time series of air surface temperature
tends to amplify the initial increase in temperature. In
anomaly (°C) (compared to the 1986–2005 average), for the twentieth
and twenty-first centuries under two different scenarios with very regions where the sea ice thins, or even disappears, air
contrasting emission scenarios: RCP2.6 (blue) and RCP8.5 (red). The temperature rises sharply as the temperature of the ocean
mean and associated uncertainties averaged over 2081–2100 are given surface is higher than that of sea ice. Finally, a change in
for all RCP scenarios as colored vertical bars. The line represents the
average of the models and the light-colored regions, the spread
atmospheric circulation (and in particular the increase in
water vapor transported to the high latitudes) is another
reason for the strong increase in temperature in these
evaporation. Over the oceans, the amount of water available regions. In the south and east of Greenland, it is noted that
for evaporation is not limited, while on the continents it the temperature of the air near the surface increases only
depends on the amount of water available in the soil, and slightly. This trend is more or less marked depending on the
thus, on the amount of precipitation. Due to latent heat model, with some even simulating a slight local cooling. The
exchanges, evaporation cools the surface and this cooling reason for this is a change in the ocean circulation, and
effect is larger over the oceans than over continental areas. especially in the thermohaline circulation. In these areas, the
For example, with the IPSL model, evaporative cooling is density of sea water decreases at the surface because of
found to reduce the surface radiative balance by 9.8 W/m2 increasing temperatures and/or precipitation. Surface waters
over the ocean against only 0.2 W/m2 over the continents. are no longer dense enough to sink, reducing oceanic con-
Other processes, such as changing cloud cover or changing vection and the associated North Atlantic drift. The effect of
atmospheric circulation, also play a role in the this density reduction on temperature depends on the model,
ocean-continent warming differential. both in terms of amplitude and geographical extension. It
In the mid and high latitude regions, the smaller increase modulates global warming locally and influences global
in ocean surface temperature is partly due to its thermal warming slightly, but nevertheless this warming remains
inertia. This is particularly true in the southern hemisphere, significant on all continents of the northern hemisphere,
where very strong winds cause great mixing of the ocean, especially in Europe.

Fig. 31.13 Multi-model average of the difference in air surface temperature (°C), between the end of the twenty-first century (2081–2100) and the
end of the twentieth century (1986–2005), for the RCP2.6 (left) and RCP8.5 (right) scenarios (Source IPCC 2013)
31 From the Climates of the Past to the Climates … 469

Changes in Precipitation It can be seen that the models consistently simulate a

year-round increase in precipitation at high latitudes, and a
Although there is a large spread between climate models, on winter increase in the mid-latitudes (Fig. 31.14). Similarly,
average, they project an increase in precipitation over the they consistently simulate a decrease in precipitation in
twenty-first century alongside increased temperatures (IPCC subtropical regions. In Europe, models simulate an increase
2013). Projected precipitation changes exceed 0.05 mm/day in precipitation in the north and, on the contrary, a drying up
and 0.15 mm/day with the RCP2.6 and RCP8.5 scenarios around the Mediterranean Basin. On the other hand, changes
respectively. These changes are far from being spatially in precipitation in the equatorial and tropical regions are not
homogeneous and exhibit a strong seasonal variability consistent from one model to the other, especially on the
(Fig. 31.14). However, if we consider the zonal means, there continents such as South America, West and Central Africa,
is a tendency towards increased precipitation everywhere, India and South-East Asia. In these areas, some models
except in the subtropical regions (around 30°N and 30°S) simulate a decrease in precipitation, while others simulate an
and in the Mediterranean basin where they decrease. The increase. These differences between models are particularly
general increase in precipitation is due to the increase in pronounced in the monsoon regions. In general, changes in
atmospheric water vapor content, while the decrease simu- precipitation on the continents remain very uncertain, even
lated in the subtropical regions is related to a change in in terms of annual mean. This is due to major uncertainties in
atmospheric circulation. the representation of the different processes. Currently, there

Fig. 31.14 Multi-model average

of the precipitation difference (%)
between the end of the
twenty-first century (average
between 2081 and 2100) and the
end of the twentieth century
(average between 1980 and
1999), for the RCP8.5 scenario
for December–January–February
(top) and (bottom) June–July–
August. Hatching indicates
regions where the multi-model
mean change is less than one
standard deviation of internal
variability. Stippling indicates
regions where the multi-model
mean change is greater than two
standard deviations of internal
variability and where at least 90%
of models agree on the sign of
change
470 S. Charbit et al.

Evolution of Sea Ice

The analysis of CMIP3 and CMIP5 models clearly high-

lights a sea-ice decline in the course of the twenty-first
century, the dominant factor being the rising summer tem-
peratures rather than the winter ones which are projected to
remain negative. Though most of the CMIP5 models project
a nearly ice-free Arctic Ocean by 2100 with a sea ice extent
of less than 1  106 km2 (for at least 5 consecutive years) at
the end of summer in the RCP8.5 scenario, some show large
changes occurring before 2050. However, these results are
strongly dependent on the ability of models to reproduce the
climatological mean state of the Arctic sea ice cover over
1979–2012. Based on the CMIP5 multi-model ensemble,
projections of average reductions in Arctic sea ice extent for
2081–2100 compared to 1986–2005 range from 8%
(RCP2.6) to 34% (RCP8.5) in February and from 43%
(RCP2.6) to 94% (RCP8.5) in September. The evolution of
sea ice around the Antarctic is more uncertain. The CMIP5
Fig. 31.15 Compilation of paleo sea level data, tide gauge multi-model mean projects a decrease in sea ice extent
data, altimeter data and central estimates and likely ranges for
projections of global mean sea level rise for RCP2.6 (blue) and ranging from 16% for RCP2.6 to 67% for RCP8.5 in
RCP8.5 (red) scenarios, all relative to pre-industrial values (Source February and from 8% for RCP2.6 to 30% for RCP8.5 in
IPCC 2013) September for 2081–2100 compared to 1986–2005. There
is, however, low confidence in those values because of the
wide inter-model spread and the inability of almost all of the
is no clear way of identifying which of the results are the available models to reproduce the mean annual cycle, the
most reliable (Fig. 31.15). interannual variability and the overall increase of the
Antarctic sea ice coverage observed during the satellite era.

Changes in Storm Patterns

Evolution of Continental Ice
Under climate change, the characteristics of lows in the
mid-latitudes (particularly those reaching the coast of Brit- Retreat of the glaciers
tany) are likely to change for two reasons: the first is a
change in the equator-pole temperature gradient, which Glaciers react quickly to climatic effects and are thus good
tends to decrease near the surface but to increase with alti- indicators of climate change. For more than a century, and
tude. The second reason is an increase in the total amount of especially since the 1980s, the retreat of glaciers has been a
water vapor in the atmosphere, and in turn, in the amount of phenomenon occurring almost everywhere on the globe.
water vapor that can be condensed and thus release latent This phenomenon is likely to increase in the twenty-first
heat. In climate change simulations, we observe a poleward century with the rising temperatures, and could even lead to
shift of low-pressure, a reduction in the total number of the disappearance of certain glaciers in the coming decades.
depressions, but an increase in the number of deeper and Projections of the future evolution of glaciers depend, on the
hence stronger troughs. For example, according to the one hand, on climate scenarios but also, on their sensitivity
diagnoses made by Lambert and Fyfe (2006), the models to global warming.
simulate, on average, in 2100 and for a moderate green- Glaciers located in dry and cold regions (such as the
house gas scenario, a decrease in the total number of Canadian Arctic) have low sensitivity to global warming and
depressions of around 10% in the southern hemisphere and a are more prone to resist. On the contrary, glaciers located in
little less in the northern hemisphere. They also simulate an coastal regions are subject to oceanic influence and have a
increase of 20% of the number of intense depressions in the greater sensitivity to global warming. This is the case of
northern hemisphere and of 40% in the southern Norwegian glaciers, for example. In recent decades, these
hemisphere. have tended to grow due to increased precipitation. But in
31 From the Climates of the Past to the Climates … 471

recent years, this trend seems to have been reversed. Alpine The future of ice sheets
glaciers are in an intermediate situation. We will not review
exhaustively the melting of glaciers, but we chose to illus- There are different approaches to determine how polar ice
trate this behavior through the study of two alpine glaciers. sheets will evolve in the future. The first one is to determine
At the Environment Geosciences Institute of Grenoble, their surface mass balance which is directly dependent on the
two studies were carried out on alpine glaciers, one on the climate. To simulate as precisely as possible the mass bal-
Saint-Sorlin, the other at the dome of Goûter, in the ance of the ice sheets, high-resolution climate models are
Mont-Blanc massif. necessary to properly represent the topography of the ice
The first study simulated the future evolution of the mass sheets, especially the steep slopes at the margins. These
balance of the Saint-Sorlin, located in the Grandes Rousses areas correspond to the ablation zones and are also the
massif whose highest point is at 3400 m altitude. The results locations where most of the precipitation falls. However,
show that even with an optimistic greenhouse gas emission AOGCMs usually tend to overestimate precipitation and
scenario (+1.8 °C in 2100), the equilibrium line (i.e. the underestimate ablation. In addition, snowpack models
lowest limit of eternal snow) is at a higher altitude than the implemented in general circulation models are often too
highest point of the glacier. This means that over a full year, simplistic and do not account for key processes taking place
the glacier does not accumulate snow (or ice) and is therefore in the snowpack. Among the missing processes are the
in a state of chronic ablation. In order to simulate the dynamic refreezing of surface melt water that percolates at depth, the
response of the glacier, these results were then used as inputs transport of snow by winds, the snow metamorphism and the
for a two-dimensional ice flow model. Despite a moderate transformation of snow into ice. These processes are crucial
global warming, the model suggests a complete disappear- for the estimates of surface mass balance. For example,
ance of the glacier around 2070. The dynamic response of the water refreezing modulates surface runoff and snow meta-
glaciers is complex because it depends not only on their morphism strongly modifies the albedo and thus the surface
specific morphology, which is different from one glacier to energy balance.
another, but also on many physical processes, which are Regional atmospheric models generally offer a better
sometimes difficult to simulate, and which determine their representation of the surface energy balance than AOGCMs
temporal response, i.e. on what timescale the glacier will because of their finer resolution and the inclusion of more
grow or, on the contrary, disappear. Nevertheless, this study sophisticated snow models. However, as they are forced at
suggests that alpine glaciers similar in size and located at an their lateral boundaries by outputs from global climate
altitude close to that of the Saint-Sorlin, could undergo the models, they may suffer from the global model deficiencies.
same type of evolution in the twenty-first century under the As a result, the uncertainties associated with the GCM-based
influence of moderate global warming. forcing represent about half of the uncertainty associated
The second study is based on borehole temperature with changes in surface mass balance inferred from regional
measurements (140 m deep) located at the Dome du Goûter climate models (RCM). All RCM-based studies project an
at 4250 m altitude. These measurements showed a warming increase in precipitation over large parts of Greenland and
of 1 to 1.5 °C over the first 60 m of ice, between 1994 and Antarctic ice sheets in response to global warming, but there
2005. A physical modeling of the process of heat diffusion in is great uncertainty as to the magnitude of this increase.
the ice made it possible to show that the observed warming Projections suggest that over Greenland, ablation will largely
results not only from atmospheric warming, but also from exceed the increase in snowfall. Conversely, in East
the heat produced by the refreezing of surface melt water Antarctica, precipitation is expected to exceed ablation
that percolates at depth. throughout the twenty-first century, but the key question is
Moreover, simulations carried out for different global whether the mass gain will be offset by dynamical ice losses
warming scenarios confirm the twenty-first century trend and from the West Antarctic ice sheet.
show that, regardless of the scenario, the Alpine glaciers, The surface mass balance is not the only important
currently classified as ‘cold’ and located between 3500 and parameter in the determination of the evolution of polar ice
4250 m altitude with an internal temperature ranging from 0 sheets. As explained earlier, their dynamic response must
to −11 °C, could become ‘temperate’, with an internal also be considered. This can be achieved through the use of
temperature close to the melting point. three-dimensional ice-sheet models which include most of
Other modeling studies using different greenhouse gas the processes responsible for the ice-sheet evolution as a
scenarios lead to similar and equally disturbing conclusions. function of climate forcing (temperatures and precipitation
A study conducted in the Montana region in the Glacier or surface mass balance). For a more thorough description of
National Park shows that, with a doubling of CO2, some ice-sheet models, we ask the reader to refer to Chap. 24 of
glaciers would disappear by 2030, despite an increase of 5– this volume. However, several problems may arise when
10% in average precipitation in the mid and high latitudes. using such models. As previously mentioned, the surface
472 S. Charbit et al.

mass balance can be computed by a climate model (RCM or expected to refine numerical simulations and thus to improve
AOGCM), but frequently, ablation is still determined from the relevance of future forecasting models. Finally, part of the
an empirical formulation that relates the number of uncertainty in the future ice-sheet responses comes from
positive-degree-days (integral of positive temperatures) to uncertainty in the climate forcing itself. This is due to our lack
the snow and ice melting rates. The main drawbacks of this of knowledge in the future socio-economic pathways and to
approach is that it does not account for albedo changes the biases of the climate models. Moreover, climate-ice sheet
and it has been validated on only a few Greenland sites for interactions are still very rarely taken into account in the
the present-day period. The mass balance derived from such models. Accounting for these interactions may have the
methods is generally more uncertain than that obtained using potential to strongly modify not only the simulated climate
regional models. Another difficulty is linked to the difference and ice-sheet responses but also sea-level projections (Bron-
of resolution between climate and ice-sheet models. As selaer et al. 2018; Golledge et al. 2019).
temperature, precipitation and surface mass balance are
highly dependent on topography, the development of
appropriate downscaling techniques is required to account Sea Level Change
for the effect of altitude.
In addition, researchers still face many challenges in rep- We have seen, in the first part of this chapter, that at the scale
resenting some processes related to rapid dynamics, such as of a few decades, the main causes of sea-level variations are
those occurring at the base of the ice sheet or at the ice-ocean due, on the one hand, to the thermal expansion of the ocean
interface, where small disturbances seem capable of triggering (in the case of a warming climate), and on the other hand, to
strong instabilities, which can propagate up to a thousand changes in the mass of water in the ocean due to exchanges
kilometers upstream, and lead to a destabilization of the entire with continental reservoirs. Based on the results from 21
ice sheet. However, understanding the impact of small-scale AOGCMs, the sea-level projections reported in the
processes on the large scale is still in its infancy. Over the last Fifth IPCC report (IPCC 2013), all show an increase in
two decades, observations have shown an increase in the flow sea-level by the end of the twenty-first century (2081-2100)
of outlet glaciers, suggesting that these could respond much compared to the 1986–2005 period, with an average global
more quickly than previously expected to variations in mean sea-level rise ranging from 0.28 to 0.61 m for RCP2.6
atmospheric and oceanic conditions, making possible a sig- and from 0.52 to 0.98 m for RCP8.5. For the RCP8.5 sce-
nificant retreat of the ice sheet in the more or less distant nario, the rate of sea-level rise increases throughout the
future. The physical laws governing ice flow in these glaciers twenty-first century, going from 3.2 ± 0.4 mm/yr (for the
still remain poorly known. For example, although the marine 1993–2010 period) to 11.2 mm/yr (2081–2100), whereas, in
ice-sheet instability is increasingly well represented in new the RCP2.6 scenario, the rate increases up to 4.4 mm/yr
generation ice-sheet models, there is no consensus on around 2050 and slightly declines in the second half of the
hydro-fracturing yet. Large uncertainties also exist on the century.
physical laws governing iceberg calving and basal melting According to the Fifth IPCC report (IPCC 2013), thermal
under the ice shelves. A last, but not least, problem is related to expansion remains the main contributor (30 to 55%) to
the initial state of the ice sheets. Indeed, starting future global mean sea-level rise over the twenty-first century with
short-term (a few centuries) simulations from a realistic median rates of 0.14 ± 0.04 m (RCP2.6) and 0.27 ± 0.04
present-day state of the ice sheet is of primary importance to m (RCP8.5) in 2081–2100. These contributions can be
avoid as much as possible spurious biases in the results. estimated from changes in ocean heat uptake increasing
Getting an accurate initial state is challenging however roughly linearly with global mean surface air temperatures
because of the scarcity of observations on vertical velocity and simulated by the AOGCMs. It should be noted that AOGCM
temperature profiles, and also on basal conditions (e.g. frozen simulations do not include volcanic forcing which may
or thawed bed areas, bedrock topography). Several approa- reduce the projected contribution of thermal expansion.
ches have been developed ranging from long free-evolving The AOGCM simulations forced by the different RCP
simulations over one or several glacial-interglacial cycles to scenarios highlight a spatial variability of sea level
more formal optimization approaches often based on inver- (Fig. 31.16). As they do not take into account the freshwater
sion techniques, with various target criteria (either surface fluxes coming from ice sheets, this regional variability,
velocities or topography in agreement with observations, or previously observed in satellite measurements over the 1993–
internal properties accounting for the past history of ice 2010 period, is linked to a variability in the atmosphere-
sheets). Each of these techniques presents some advantages ocean-sea-ice system. More specifically, regional variability
and drawbacks, but their efficiency should improve in the can be due not only to variations in temperature and salinity
future as the number of observations grows. Recent remote (i.e. precipitation/evaporation ratio) and therefore density, but
sensing observations made over Greenland and Antarctica are also to variations in ocean and atmospheric circulations
31 From the Climates of the Past to the Climates … 473

0.3 grounding line migration, which plays a crucial role in

60°N
ice-sheet evolution as mentioned earlier. However, signifi-
0.2
30°N cant progress has been made in the development of ice-sheet
0.1 models, such as the Full Stokes models (e.g. Gillet-Chaulet
0° et al. 2012). These models do not rely on simplifying
0.1
assumptions and have an adaptive mesh that may be less
30°S
0.0 than 1 km at the ice-sheet margins, thereby providing a more
60°S accurate representation of ice-sheet dynamics. However,
0.0
90°E 180° 90°W 0° these models are highly expensive in terms of computational
(m)
time and have not yet been applied to the whole Antarctic ice
Fig. 31.16 Ensemble mean projection of the time-averaged dynamic sheet. A second limiting factor is related to the fact that some
and steric sea-level changes for the period 2081–2100 relative to the processes remain poorly constrained due to the scarcity of
reference period 1986–2005, computed from 21 CMIP5 climate models
observations (e.g. basal conditions or hydro-fracturing), or
(in meters), using the RCP4.5 experiment. The figure includes the
globally-averaged steric sea level increase of 0.18 ± 0.05 m (Source poorly understood (e.g. basal melting under the ice shelves
IPCC 2013) or iceberg calving). Finally, as mentioned in the previous
section, there is a crucial need to develop Earth System
(variations in wind direction and intensity). From one model to Models which account for the feedbacks between climate
another, the spatial distributions of sea level variations are not and ice sheets so as to refine sea-level projections.
identical. However, some common features emerge in the It is important to note that since the publication of the last
projections for the end of the twenty-first century that are IPCC report (IPCC 2013), many projections coming from
identified in Fig. 31.16. For example, the Southern Ocean is individual studies have provided higher estimates of the
characterized by a lower-than-average sea-level rise, likely projected sea-level rise compared to those reported by the
related to changes in wind stress and heat fluxes. These IPCC (0.52 to 0.98 m for the highest emission scenario).
changes are also responsible for the greater sea-level rise in the The main reasons for this disagreement are related to the
Pacific (Bouttes and Gregory 2014). A dipole pattern with large uncertainties associated with the future ice-sheet evo-
greater sea-level rise in the north and smaller sea-level rise in lution and to the fact that the IPCC only selected the most
the south can also be seen in the North Atlantic and is attributed likely range of sea-level rise estimates and excluded the most
to changes in heat and freshwater fluxes. extreme (considered as less likely) outcomes. All the pro-
While changes in sea level due to thermal expansion can jections together give a range of 0.16–2.54 m in 2100
easily be determined from climate models, we have seen in (Garner et al. 2018) which reflects the large uncertainties in
the previous section that there are significant uncertainties in the maximum contribution of Greenland and Antarctica
the simulation of future continental ice. These uncertainties (DeConto and Pollard 2016) and suggests the possibility for
dominate the projections of sea-level rise resulting from these ice sheets to become the main contributor to global
ice-sheet surface melting, dynamic ice discharges or iceberg mean sea-level rise in the course of the twenty-first century.
calving. As a result, these projections are not very different
from one climate scenario to another. For example, in the
last IPCC report (IPCC 2013), the contribution from the The Climate of the Next Millennium: Towards
rapid dynamics of the ice sheets is estimated to be between Integrated Modeling of the Earth System
0.03 and 0.19 m for the RCP2.6 scenario and between 0.03
and 0.20 m for the RCP8.5 scenario. Climate Change: Anthropogenic Disturbance
To better represent the contribution of polar ice sheets to Versus Variations in Insolation
sea-level rise in the future, an accurate representation of
rapid ice dynamics in ice-sheet models is of primary In Chap. 28, we analyzed in detail the climate variations
importance. The resolution (5–10 km) of the previous gen- induced by the orbital parameter variations over the scale of
eration of three-dimensional large-scale ice-sheet models is tens of thousands of years. We are interested here in much
not fine enough to properly capture the rapid ice streams longer time scales ranging from 100,000 to millions of years.
whose characteristic spatial scale may be smaller than 1 km The amplitude of warming projected by climate models over
in some cases. Moreover, these models are based on the next decades and even the next century raises questions
approximations resulting in simplified equations of ice about the impact of this profound change in the climate
dynamics. While these approximations are crucial to inves- system over several millennia, especially as the acceleration
tigate the ice-sheet behavior at the multi-millennial time of CO2 emissions suggests that the magnitude of the per-
scale, they may cause deficiencies in the representation of turbation may be even greater than that predicted by the
rapid ice dynamics as well as in the representation of most pessimistic scenario.
474 S. Charbit et al.

Data from Vostok and EPICA ice cores in Antarctica, greenhouse gases) to be taken into account in models
showed that atmospheric CO2 varied by about 100 ppm including representations of the atmosphere, ocean, cryo-
during the transition from an ice age to an interglacial period, sphere, lithosphere and vegetation.
and that these variations were correlated with changes in air One of the first models of this type was developed for the
temperatures, suggesting a link between CO2 and climate. northern hemisphere at the University of Louvain-la-Neuve,
Today, this link is commonly accepted and attention is Belgium (Gallée 1992). It successfully reproduced the main
focused on global warming due to the increase in greenhouse characteristics of the current climate, as well as the variations
gases in the atmosphere. However almost forty years ago, in in ice volume during glacial-interglacial cycles. Since the
1972, the climatological community met in the United States model has been shown to provide reasonable simulations of
to discuss the imminence of the next ice age. This question the 100,000-year cycle, it has been applied as a second step
arose because geological data showed that for about a million to the simulation of future climates. Several tests have thus
years, the Earth had alternated between cold episodes cor- been carried out on the climate of the next 130,000 years,
responding to the glaciation phases, and the warmer inter- either by keeping the CO2 constant at different levels
glacial periods, with a pseudo-periodicity of 100,000 years. (290 ppm, 200 ppm and 250 ppm), or by using the CO2
Data available in 1972 showed that the previous two variations of the last glacial-interglacial cycle (Loutre and
interglacial periods had lasted about 10,000 years. Yet, the Berger 2000). The results of these simulations suggest, on
Holocene, the interglacial period we are currently experi- the one hand, that the climate of the next 50,000 years is
encing, has been going on for 10,000 years. It seemed particularly sensitive to the level of atmospheric CO2 con-
therefore reasonable to think that this warm phase would centration and that our current interglacial period will be
soon end and give way to a new ice age. This reasoning was much longer than any other one in the past. It could last
based on the assumption that all interglacial periods are of more than 55,000 years with CO2 levels between 230 and
equal duration. However, although the first numerical 290 ppm: the first glacial stage would appear around 60,000
experiments carried out with statistical models indicated that AD, and the next glacial maximum (in terms of ice volume)
the cooling initiated 6000 years ago (after the Holocene would be around 100,000 AD, followed by a deglaciation
climate optimum) would continue in the future, Oerlemans phase that would end around 120,000 AD. These results
and Van der Veen (1984) of the University of Utrecht suggest that the marked differences between our interglacial
showed, with an ice-sheet model, that the transition to a new period (present and future) and the previous Quaternary
glacial phase would not occur before 50,000 years. warm periods are due to the small insolation variations that
Based on data from marine sediments and ice cores, it is characterize the former. Based on another scenario designed
now well established that the duration of an interglacial to reproduce the natural variations of CO2 and not the
period varies considerably from one climatic cycle to the anthropogenic contribution, other simulations have been
next. The same models used to predict the future were used carried out, either with a more elaborate version of the
to simulate the last interglacial-glacial transition. Indeed, Loutre and Berger’s model, or with a climate model of
taking into account the atmosphere-ocean feedbacks, these intermediate complexity coupled with a more sophisticated
models produce perennial snow in the Canadian archipelago model of the evolution of polar ice sheets (Ganopolski et al.
(Khodri et al. 2001). It is also known that the Earth’s orbital 2016). The results of these simulations are in agreement with
parameters (see Chap. 28) that govern the seasonal and lat- those presented previously, and show that the next glacial
itudinal distribution of solar energy can vary considerably inception will be postponed by at least 100,000 years.
from one cycle to another. For example, geological data tell Although, the variations in summer insolation at 65°N
us that the interglacial period occurring 400,000 years ago have long been considered as the pacemaker of
(marine isotopic stage 11) was exceptionally long. This sit- glacial-interglacial transitions, the above results show that
uation corresponds to a weak eccentricity where the seasonal the level of atmospheric CO2 will be a key parameter in the
and latitudinal distribution of solar energy varied very little. future. While the natural variations of CO2 along with the
Celestial mechanics tell us that a similar situation should insolation forcing exclude the possibility of a glacial episode
recur within 20,000 years, and for this reason the marine in the future for at least 50,000 years, it is possible that the
isotopic stage 11 is often considered to be one of the best impact of the anthropogenic contribution will lead to a
analogues for the future climate. Many internal feedbacks complete disappearance of polar ice sheets, and that a return
generated by the different components of the climate system to a glacial phase could occur only in 100,000 years. Cou-
amplify or reduce the effect of the latitudinal and seasonal pled climate-ice sheet models, validated on the last
distribution of insolation. Thus, climate projections on the glacial-interglacial cycle, will make it possible to explore the
scale of a few hundreds of thousands of years require the threshold values of CO2 which could have long-term effects
variations of both insolation and atmospheric CO2 (and other on the fate of ice sheets.
31 From the Climates of the Past to the Climates … 475

The Long-Term Future of the Polar Ice Sheets: long-term evolution of greenhouse gas emissions to explore
Impact and Irreversibility the sensitivity of the present-day ice sheets with numerous
scenarios. Performing simulations over several centuries or
As seen earlier, the response times of the different compo- several millennia cannot be achieved through the use of
nents of the climate system are extremely variable, ranging general circulation models, which are too expensive in terms
from a few minutes to a few days for the atmosphere and of computational time, and simplified climate models cou-
from a few months to several hundreds of years for the pled with efficient ice-sheet models are required.
ocean. While the ice sheets have long been considered as a In fact, there are three different time frames relevant in the
slow component of the Earth system (with characteristic study of the evolution of ice sheets. The evolution over the
time scales ranging from thousands to hundreds of thousands twenty-first century, the basis of all IPCC analyzes, will very
of years), recent observations provide evidences that they likely bring about a rise in sea level of several tens of cen-
can react to climate change far more quickly than previously timeters (or even more), although there is a great deal of
thought. The last glacial-interglacial cycle demonstrates how uncertainty linked to the emission scenarios, to climate
deeply the climate is influenced by the slow development model biases and to difficulties of ice-sheet models to cap-
and rapid collapse of the ice sheets. ture the processes causing rapid dynamical change. The
The ice core records retrieved from the Vostok and second horizon, beyond the twenty-first and the following
Dome C sites in Antarctica show that, for 800,000 years, the centuries, is one where the CO2 level may stabilize at three
world has alternated between four ice sheets (during glacial or four times that of the pre-industrial level, likely resulting
periods) and only two ice sheets (during interglacial peri- in a massive retreat of Greenland and West Antarctica. These
ods). In other words, Greenland and Antarctica withstood changes are also likely to modify the ocean circulation and
the warming that led to the disappearance of the the global climate. If this high level of CO2 persists for a
Fennoscandian and North American ice sheets. There is no long time in the atmosphere, the ice-sheet melting could be
doubt that Greenland and West Antarctica have not always irreversible, in the sense that there would no longer be any
emerged unscathed from glacial-interglacial cycles. Indeed, perennial snow in Greenland (Charbit et al. 2008). Finally,
during the last interglacial period (i.e. 130–115 ka ago), the there is the much more distant third horizon, which raises the
sea level was 6 to 9 m higher (Kopp et al. 2009). It is following question: for the last million years, our climate has
therefore possible that anthropogenic activity could lead to oscillated between ice ages (long periods of about
the partial or total melting of Greenland (Charbit et al. 2008). 100,000 years) and interglacial periods (short periods of
Recent observations of Greenland, and, more surpris- about 10,000 years). Is it possible that the anthropogenic
ingly, of Antarctica mass balances (Velicogna 2009) show perturbation might cause a switch to another climate mode
that these ice sheets have become one of the main contrib- with strongly reduced ice sheets or even no ice sheet at all,
utors to the increase in sea level (Cazenave et al. 2009). similar to the hot climate mode of the pre-Quaternary era? In
Their complete disappearance would lead to a sea-level rise other words, is it possible that anthropogenic disturbance
of nearly 60 m: 6.6 m for Greenland and 52.8 m for could induce modifications such that the next marked
Antarctica, of which 3.3 m would come from West decline in summer insolation due in about 100,000 years
Antarctica. These figures should be compared to the 120 m (Loutre and Berger 2000) might not bring about an ice age?
sea-level rise corresponding to the disappearance of the Even if this question seems ‘futuristic’, we have good
North American and Fennoscandian ice sheets in response to reasons to believe that the glacial-interglacial cycles of the
very small changes in insolation compared to the additional last million years will no longer occur because the expected
radiative forcing from anthropogenic activities. Also, the decline of insolation will not be large enough to compensate
disappearance of past ice sheets was spread over for the radiative forcing due to the high atmospheric CO2
14,000 years (from the Last Glacial Maximum to the levels. This is what is suggested by the results of simulations
Mid-Holocene). The long-term effects (several hundred carried out for periods such as the Pliocene around 3 Ma
years) of anthropogenic forcing on ice sheets are not easy to where the CO2 level is estimated to have been around
model, partly because of uncertainties in the future 405 ± 50 ppm.
socio-economic pathways over the twenty-first century and In these scenarios, a key factor is the long time required
therefore in the evolution of the greenhouse gas concentra- to reach atmospheric CO2 equilibrium which can be several
tions in the atmosphere. It is therefore even more difficult to tens to hundreds of thousands years (Archer et al. 1997). The
establish scenarios over several centuries (Charbit et al. anthropogenic disturbance is almost instantaneous, but its
2008). However, several assumptions can be made about the consequences will last for a very long time, and the
476 S. Charbit et al.

Milankovitch cycles that have so far allowed glacial- Archer, D., et al. (2009). Atmospheric lifetime of fossil fuel carbon
interglacial transitions through profoundly non-linear pro- dioxide. Annual Review of Earth and Planetary Sciences, 37, 117–
134. https://doi.org/10.1146/annurevearth031208.100296.
cesses, can become inoperative in the future. Is it a bit Bony, S., et al. (2006). How well do we understand and evaluate
presumptuous though to think that a couple of hundred years climate change feedback processes? Journal of Climate, 19(15),
of ‘energy profligacy’ leading to a very large increase in 3445–3482.
atmospheric CO2 could still be felt tens of thousands of Bouttes, N., & Gregory, J. M. (2014). Attribution of the spatial pattern
of CO2-forced sea level change to ocean surface flux changes.
years later? Maybe, but still, it is important to remember that Environmental Research Letters, 9, 034004. https://doi.org/10.
studies of the Earth past climates show that the presence of 1088/1748-9326/9/3/034004.
ice sheets is associated with low CO2 levels. It is therefore Bronselaer, B., et al. (2018). Change in future climate due to Antarctic
logical to think that, in a world that sustains high CO2 meltwater. Nature, 564, 53–58. https://doi.org/10.1038/s41586-018-
0712-z.
concentrations, the behavior of present-day ice sheets will Cazenave, A., et al. (2009). Sea level budget over 2003-2008: A
change in the short and medium term. reevaluation from GRACE space gravimetry; Satellite altimetry and
The world at the beginning of the Cenozoic (with argo. Global and Planetary Change, 65, 83–88.
1120 ppm of CO2 in the atmosphere) was much hotter than Charbit, S., et al. (2008). Amount of CO2 emissions irreversibly leading
to the total melting of Greenland. Geophysical Research Letters, 35,
today. Antarctica was already in its polar position since the L12503. https://doi.org/10.1029/2008GLO33472.
end of the Cretaceous (70 Ma ago), but far from forming an Church, J. A., & White, N. J. (2011). Sea-level rise from the late 19th to
ice sheet, it was covered with forests. There was no sign of the early 21st century. Survey of Geophysics, 32, 585–602.
an ice sheet, until the CO2 level dropped sufficiently to Christiansen, H. H., et al. (2010). The thermal state of permafrost in the
Nordic area during the International Polar Year 2007–2009.
trigger the progressive freeze-up of Antarctica (DeConto and Permafrost Periglacial Processes, 21, 156–181.
Pollard 2003). A planet with no ice sheet is not a figment of Clark, P. U., et al. (2016). Consequences of twenty-first-century policy
the imagination, rather their presence being the exception for multi-millennial climate and sea-level change. Nature Climate
when viewed over geological times. Change, 6, 360–369.
DeConto, R. M., & Pollard, D. (2003). Rapid cenozoic glaciation of
This book shows how the face of our planet changed over Antarctica induced by declining atmospheric CO2. Nature, 421,
these time scales: from plate tectonics (millions of years) to 245–249. https://doi.org/10.1038/nature01290.
the development of huge ice sheets at temperate latitudes DeConto, R. M., & Pollard, D. (2016). Contribution of Antarctica to
(hundreds of thousands of years), with very strong vari- past and future sea-level rise. Nature, 531, 591–597. https://doi.org/
10.1038/nature17145.
ability in glacial climates at the scale of a few thousand Defrance, D., et al. (2017). Consequences of rapid ice-sheet melting on
years. Our own interglacial period, the Holocene, has been the Sahelian population vulnerability. In Proceedings of the
much more stable in terms of climate and has contributed to National Academy of Sciences (Vol. 114, Issue 25). https://doi.
the extraordinary expansion of the human population. With org/10.1073/pnas.1619358114.
Dufresne, J.-L., & Bony, S. (2008). An assessment of the primary
an Earth inhabited by more than 9 billion men and women, it sources of spread of global warming estimates from coupled
will be essential to manage climate change in order to protect atmosphere-ocean models. Journal of Climate, 21(19), 5135–5144.
societies and their environment. The past teaches us that our https://doi.org/10.1175/2008jcli2239.1.
little blue planet has undergone many changes. It is not the Dufresne, J.-L., & Saint-Lu, M. (2016). Positive feedback in climate:
Stabilization or runaway, illustrated by a simple experiment.
planet that is in danger, it is rather the populations, especially Bulletin of American Meteorological. Society, 97(5), 755–765.
as they do not have an equal standing in the face of climate https://doi.org/10.1175/bams-d-14-00022.1.
change. And so, paradoxically, if man has become a major Fourier, J.-B. (1824). Remarques générales sur les températures du
actor in climate change through industrial development and globe terrestre et des espaces planétaires. Annales de Chimie et de
Physique, 2e série, XXVII, 136–167.
massive use of fossil fuels, he could also suffer at his own Gallée, H., et al. (1992). Simulation of the last glacial cycle by a
expense from upheavals and difficult situations that he has coupled, sectorially averaged climate-ice-sheet model 2. Response
created himself. to insolation and CO2 variations. Journal of Geophysical Research,
97(D14), 15713–15740.
Ganopolski, A., et al. (2016). Critical insolation-CO2 relation for
diagnosing past and future glacial inception. Nature, 529, 200–203.
References https://doi.org/10.1038/nature16494.
Garner, A. J., et al. (2018). Evolution of 21st century sea level rise
Adler, R. F., et al. (2003). The Version 2 Global Precipitation projections. Earth’s Future, 6, 1603–1615. https://doi.org/10.1029/
Climatology Project (GPCP) monthly precipitation analysis (1979– 2018EF000991.
Present). Journal of Hydrometeorology, 4, 1147–1167. Gillet-Chaulet, F., et al. (2012). Greenland ice sheet contribution to
Allen, R. J., & Sherwood, S. C. (2008). Warming maximum in the sea-level rise from a new-generation ice-sheet model. The Cryo-
tropical upper troposphere deduced from thermal winds. Nature sphere, 6, 1561–1576. https://doi.org/10.5194/tc-6-1561-2012.
Geoscience, 1, 399–403. Golledge, N. R., et al. (2019). Global environmental consequences of
Archer, D., et al. (1997). Multiple timescales for neutralization of fossil twenty-first-century ice-sheet melt. Nature, 566, 65–72. https://doi.
fuel CO2. Geophysical Research Letters, 24, 405–408. org/10.1038/s41586-019-0889-9.
31 From the Climates of the Past to the Climates … 477

Haywood, A. M., et al. (2011). Pliocene Model Intercomparison Project Nerem, R. S., Beckley, B. D., Fasullo, J. T., Hamlington, B. D.,
(PlioMIP): Experimental design and boundary conditions (experi- Masters, D., & Mitchum G. T. (2018). Climate-change–driven
ment 2. Geoscientific Model Development, 4, 571–577. https://doi. accelerated sea-level rise. Proceedings of the National Academy of
org/10.5194/gmd-4-571-2011. Sciences, 115(9), 2022–2025. https://doi.org/10.1073/pnas.
Haywood, A. M. et al. (2016). The Pliocene Model Intercomparison 1717312115.
Project (PlioMIP) Phase 2: Scientific objectives and experimental Oerlemans, J., & Van der Veen, C. J. (1984). Ice Sheets and Climate,
design. Climate of the Past, 12, 663–675. https://doi.org/10.5194/ Reidel, 217 pp.
cp-12-663-2016IMBIE team (2018). Mass balance of the Antarctic Oerlemans, J. (2005). Extracting a climate signal from 169 glacier
ice sheet from 1992–2017. Nature, 558, 219–222. https://doi.org/ records. Science, 308(5722), 675–677.
10.1038/s41586-018-0179. Peltier, D. F., et al. (2015). ICE-5G and ICE-6G models of postglacial
IPCC. (Ed.). Climate Change 2007: The Physical Science Basis; relative sea-level history a Space geodesy constrains ice age
Contribution of Working Group I to the Fourth Assessment Report terminal deglaciation: The global ICE-6G_C (VM5a) model.
of the Intergovernmental Panel on Climate Change, Cambridge, Journal of Geophysical Research (Solid Earth), 120, 450–487.
United Kingdom, and New-York, USA, Cambridge University Purkey, S. G., & Johnson, G. C. (2010). Warming of global abyssal and
Press. deep southern ocean waters between the 1990s and 2000s:
IPCC. (Ed.). Climate Change (2013): The Physical Science Basis; Contributions to global heat and sea level rise budgets. Journal of
Contribution of Working Group I to the Fifth Assessment Report of Climate, 23, 6336–6351.
the Intergovernmental Panel on Climate Change, Cambridge, Ramanathan, V., & Coakley, J. A., Jr. (1978). Climate modeling
United Kingdom, and New-York, USA, Cambridge University through radiative-convective models. Reviews of Geophysics and
Press. Space Physics, 16(4), 465.
Jevrejeva, S., Moore, J. C., Grinsted, A., & Woodworth, P. L. (2008). Ramillien, G., et al. (2006). Interannual variations of the mass balance
Recent global sea level acceleration started over 200 years ago? of the Antarctica and Greenland ice sheets from GRACE. Global
Geophysical Research Letters, 35, L08715. and Planetary Change, 53, 198–208.
Johnson, G. C., et al. (2007). Recent bottom water warming in the Ray, R. D., & Douglas, B. C. (2011). Experiments in reconstructing
Pacific Ocean. Journal of Climate, 20, 5365–5375. twentieth-century sea levels. Progress in Oceanography, 91, 495–
Keeling, C. D., et al. (1995). Interannual extremes in the rate of rise of 515.
atmospheric carbon dioxide since 1980. Nature, 375, 666–670. Rignot, E., et al. (2019). Four decades of Antarctic ice sheet mass
Khodri, M., et al. (2001). Simulating the amplification of orbital forcing balance from 1979–2017. Proceedings of the National Academy of
by ocean feedbacks in the last glaciation. Nature, 410, 570–574. Sciences, 116(4), 1095–1103. https://doi.org/10.1073/pnas.
Kopp, R. E., et al. (2009). Probabilistic assessment of sea level during 18128883116.
the last interglacial stage. Nature, 462, 863–867. https://doi.org/10. Rigor, I. G., & Wallace, J. M. (2004). Variations in the age of sea ice
1038/nature08686. and summer sea ice extent. Geophysical Research Letters, 31.
Kouketsu, S., et al. (2011). Deep ocean heat content changes estimated https://doi.org/10.1029/2004GL019492.
from observation and reanalysis product and their influence on sea Romanovsky, V. E., Smith, S. L., & Christiansen, H. H. (2010a).
level change. Journal of Geophysical Research (Ocean), 116, Permafrost thermal state in the polar Northern Hemisphere during
C03012. the International Polar Year 2007–2009: A synthesis. Permafrost
Lambert, S. J., & Fyfe, J. C. (2006). Changes in winter cyclone Periglacial Processes, 21, 106–116.
frequencies and strengths simulated in enhanced greenhouse Romanovsky, V. E., et al. (2010b). Thermal state of permafrost in
warming experiments: results from the models participating in the Russia. Permafrost Periglacial Processes, 21, 136–155.
IPCC diagnostic exercise. Climate Dynamics, 26, 713–728. https:// Shepherd, et al. (2012). A reconciled estimate of ice-sheet mass
doi.org/10.1007/s00382-006-0110-3. balance. Science, 338, 1183. https://doi.org/10.1126/science.
Loutre, M.-F., & Berger, A. (2000). Are we entering an exceptionally 1228102.
long interglacial? Climatic Change, 46, 61–90. Stott, P. A., et al. (2004). Human contribution to the European
Luo, D., et al. (2016). Environmental Earth Science, 75, 555. https:// heatwave of 2003. Nature, 432, 610–614.
doi.org/10.1007/s12665-015-5229-2. Szopa, S., et al. (2013). Aerosol and ozone changes as forcing for
Madden, R. A., & Julian, P. R. (1994). Observations of the 40-50 day climate evolution between 1850 and 2100. Climate Dynamics, 40
tropical oscillation: A review. Monthly Weather Review, 122, 814– (9–10), 2223–2250. https://doi.org/10.1007/s00382-012-1408-y.
837. Tarasov, L., & Peltier, W. R. (2007). Coevolution of continental ice
Manabe, S., & Wetherald, R. T. (1967). Thermal equilibrium of the cover and permafrost extent over the last glacial-interglacial cycle in
atmosphere with a given distribution of relative humidity. Journal North America. Journal of Geophysical Research, 112, F02S08.
of the Atmospheric Sciences, 24(3), 241–259. https://doi.org/10. https://doi.org/10.1029/2006jf000661.
1175/1520-0469(1967)024%3c0241:TEOTAW%3e2.0.CO;2. Tschudi, M. A., Stroeve, J. C., & Stewart, J. S. (2016). Relating the age
Meehl, G. A., Tebaldi, C., Walton, G., Easterling, D., & McDaniel, L. of Arctic sea ice to its thickness, as measured during NASA’s
(2009). Relative increase of record high maximum temperatures ICESat and IceBridge campaigns. Remote Sensing, 8(6). https://doi.
compared to record low minimum temperatures in the U.S. org/10.3390/rs8060457.
Geophysical Research Letters, 36, L23701. https://doi.org/10. VandenBerghe, J. (2011). Permafrost during the Pleistocene in North
1029/2009gl040736. West and Central Europe. Permafrost Response on Economic
478 S. Charbit et al.

Development, Environmental Security and Natural Resources Geophysical Research Letters, 36(L19503). https://doi.org/10.
(pp. 185–194). 1029/2009glo4022.
Vial, J., et al. (2013). On the interpretation of inter-model spread in Velicogna, I., et al. (2014). Regional acceleration in ice mass loss from
CMIP5 climate sensitivity estimates. Climate Dynamics, 41(11–12), Greenland and Antarctica using GRACE time-variable gravity data.
3339–3362. https://doi.org/10.1007/s00382-013-1725-9. Geophysical Research Letters, 41, 8130–8137. https://doi.org/10.
Velicogna, I., & Wahr, J. (2006). Measurements of time-variable 1002/2014GL061052.
gravity show mass loss in Antarctica. Science, 311, 1754–1756. WCRP Global Sea Level Budget Group. (2018). Global sea-level
Velicogna, I. (2009). ‘Increasing Rates of Ice Mass Loss from the budget 1993–present. Earth Syst. Sci. Data, 10, 1551–1590. https://
Greenland and Antarctic Ice Sheets Revealed by GRACE’. doi-org.insu.bib.cnrs.fr/10.5194/essd-10-1551-2018.