100% found this document useful (1 vote)

928 views533 pages

Planetary Surface Processes

Uploaded by

naohanomedisponivel

We take content rights seriously. If you suspect this is your content, claim it here.

Available Formats

Download as PDF, TXT or read online on Scribd

100% found this document useful (1 vote)

928 views533 pages

Planetary Surface Processes

Uploaded by

naohanomedisponivel

We take content rights seriously. If you suspect this is your content, claim it here.

Available Formats

Download as PDF, TXT or read online on Scribd

You are on page 1/ 533

Copyright © 2011. Cambridge University Press. All rights reserved.

May not be reproduced in any form without permission from the publisher, except fair uses permitted
under U.S. or applicable copyright law.

Account: s4526441
AN: 399390 ; H. Jay Melosh.; Planetary Surface Processes
EBSCO Publishing : eBook Academic Collection (EBSCOhost) - printed on 11/3/2023 1:50 PM via YDUQS
Copyright © 2011. Cambridge University Press. All rights reserved. May not be reproduced in any form without permission from the publisher, except fair uses permitted

Planetary Surface Processes

Planetary Surface Processes is the first advanced textbook to cover the full range of geo-
logic processes that shape the surfaces of planetary-scale bodies. This comprehensive intro-
duction ranges from microscopic aspects of the soil on airless asteroids to the topography
of super-Earth planets.
Using a modern, quantitative approach, this book reconsiders geologic processes outside
the traditional terrestrial context. It highlights processes that are contingent upon Earth’s
unique circumstances and processes that are universal. For example, it shows explicitly that
equations predicting the velocity of a river are dependent on gravity; traditional geomorph-
ology textbooks fail to take this into account.
This textbook is a one-stop source of information on planetary surface processes, pro-
viding readers with the necessary background to interpret new data from NASA, ESA,
and other space missions. Based on a course taught by the author at the University of
Arizona for 25 years, it is aimed at advanced students, and is also an invaluable resource
for researchers, professional planetary scientists, and space-mission engineers.

h. jay melosh is Distinguished Professor of Earth and Atmospheric Science at Purdue

University. His principal research interests are impact cratering, planetary tectonics, and
the physics of earthquakes and landslides. He is a science team member of NASA’s Deep
Impact mission that successfully cratered comet Tempel 1 on July 4, 2005. Professor
Melosh was awarded the Barringer Medal of the Meteoritical Society in 1999, the Gilbert
prize of the Geological Society of America in 2001, the Hess Medal of the American
Geophysical Union in 2008, and was elected to the US National Academy of Sciences in
2003. He has published over 170 technical papers, edited two books and is the author of
Impact Cratering: A Geologic Process (1989, Oxford University Press). Asteroid #8216
was named “Melosh” in his honor.
under U.S. or applicable copyright law.

EBSCO Publishing : eBook Academic Collection (EBSCOhost) - printed on 11/3/2023 1:50 PM via YDUQS
AN: 399390 ; H. Jay Melosh.; Planetary Surface Processes
Account: s4526441
Cambridge Planetary Science

Series Editors: Fran Bagenal, David Jewitt, Carl Murray, Jim Bell, Ralph Lorenz,
Copyright © 2011. Cambridge University Press. All rights reserved. May not be reproduced in any form without permission from the publisher, except fair uses permitted

Francis Nimmo, Sara Russell

Books in the series

1. Jupiter: The Planet, Satellites and Magnetosphere†

Edited by Bagenal, Dowling and McKinnon
978-0-521-03545-3
2. Meteorites: A Petrologic, Chemical and Isotopic Synthesis†
Hutchison
978-0-521-03539-2
3. The Origin of Chondrules and Chondrites
Sears
978-0-521-83603-6
4. Planetary Rings
Esposito
978-0-521-36222-1
5. The Geology of Mars: Evidence from Earth-Based Analogs
Edited by Chapman
978-0-521-83292-2
6. The Surface of Mars
Carr
978-0-521-87201-0
7. Volcanism on Io: A Comparison with Earth
Davies
978-0-521-85003-2
8. Mars: An Introduction to its Interior, Surface and Atmosphere
Barlow
978-0-521-85226-5
9. The Martian Surface: Composition, Mineralogy and Physical Properties
Edited by Bell
978-0-521-86698-9
10. Planetary Crusts: Their Composition, Origin and Evolution†
Taylor and McLennan
978-0-521-84186-3
11. Planetary Tectonics
Edited by Watters and Schultz
978-0-521-76573-2
under U.S. or applicable copyright law.

12. Protoplanetary Dust: Astrophysical and Cosmochemical Perspectives

Edited by Apai and Lauretta
978-0-521-51772-0
13. Planetary Surface Processes
Melosh
978-0-521-51418-7

†
Issued as a paperback

EBSCO Publishing : eBook Academic Collection (EBSCOhost) - printed on 11/3/2023 1:50 PM via YDUQS
AN: 399390 ; H. Jay Melosh.; Planetary Surface Processes
Account: s4526441
Copyright © 2011. Cambridge University Press. All rights reserved. May not be reproduced in any form without permission from the publisher, except fair uses permitted

Plane tary Sur face P ro ce sse s

H. Jay Me l osh
under U.S. or applicable copyright law.

EBSCO Publishing : eBook Academic Collection (EBSCOhost) - printed on 11/3/2023 1:50 PM via YDUQS
AN: 399390 ; H. Jay Melosh.; Planetary Surface Processes
Account: s4526441
c amb r i dge uni v e r si t y pr e ss
Copyright © 2011. Cambridge University Press. All rights reserved. May not be reproduced in any form without permission from the publisher, except fair uses permitted

Cambridge, New York, Melbourne, Madrid, Cape Town,

Singapore, São Paulo, Delhi, Tokyo, Mexico City

Cambridge University Press

The Edinburgh Building, Cambridge CB2 8RU, UK

Published in the United States of America by Cambridge University Press, New York

www.cambridge.org
Information on this title: www.cambridge.org/9780521514187

© H. Jay Melosh 2011

This publication is in copyright. Subject to statutory exception

and to the provisions of relevant collective licensing agreements,
no reproduction of any part may take place without the written
permission of Cambridge University Press.

First published 2011

Printed in the United Kingdom at the University Press, Cambridge

A catalogue record for this publication is available from the British Library

Library of Congress Cataloguing in Publication data

Melosh, H. J.
Planetary surface processes / H. Jay Melosh.
p. cm. – (Cambridge planetary science)
Includes bibliographical references and index.
ISBN 978-0-521-51418-7 (hardback)
1. Planets–Surfaces. 2. Geomorphology. I. Title. II. Series.
QB603.S95M45 2011
559.92–dc23
2011019269

ISBN 978-0-521-51418-7 Hardback

Cambridge University Press has no responsibility for the persistence or

accuracy of URLs for external or third-party internet websites referred to in
this publication, and does not guarantee that any content on such websites is,
or will remain, accurate or appropriate.
under U.S. or applicable copyright law.

This book is dedicated to the students and colleagues who participated in my class PtyS
554, Planetary Surfaces, and PtyS 594, Planetary Field Geology Practicum, at the Lunar
and Planetary Lab of the University of Arizona and at Caltech and Stony Brook before that.
In the years stretching from 1976 to 2009 and from the classroom to campfires in unearthly
landscapes under star-studded skies, we all learned together.
under U.S. or applicable copyright law.

Contents

Preface page xv
Acknowledgments xix
1 The grand tour 1
1.1 Structure of the Solar System 2
1.1.1 Major facts of the Solar System 3
1.1.2 Varieties of objects in the Solar System 4
1.2 Classification of the planets 5
1.2.1 Retention of planetary atmospheres 6
1.2.2 Geologic processes on the terrestrial planets and moons 7
1.3 Planetary surfaces and history 9
1.3.1 The Moon 10
1.3.2 Mercury 14
1.3.3 Venus 15
1.3.4 Mars 16
1.3.5 Jupiter’s Galilean satellites 18
1.3.6 Titan 20
1.3.7 The Earth 22
Further reading 24
2 The shapes of planets and moons 25
2.1 The overall shapes of planets 26
2.1.1 Non-rotating planets: spheres 26
2.1.2 Rotating planets: oblate spheroids 27
2.1.3 Tidally deformed bodies: triaxial ellipsoids 30
under U.S. or applicable copyright law.

2.1.4 A scaling law for planetary figures? 34

2.1.5 Center of mass to center of figure offsets 34
2.1.6 Tumbling moons and planets 35
2.2 Higher-order topography: continents and mountains 36
2.2.1 How high is high? 36
2.2.2 Elevation statistics: hypsometric curves 38
Box 2.1 Topographic roughness 40

vii

EBSCO Publishing : eBook Academic Collection (EBSCOhost) - printed on 11/3/2023 1:50 PM via YDUQS
AN: 399390 ; H. Jay Melosh.; Planetary Surface Processes
Account: s4526441
viii Contents

2.2.3 Where are we? Latitude and longitude on the planets 41

2.3 Spectral representation of topography 44

Further reading 47
Exercises 47
3 Strength versus gravity 49
3.1 Topography and stress 49
Box 3.1 Collapse of topography on a strengthless planet 51
3.2 Stress and strain: a primer 52
3.2.1 Strain 52
3.2.2 Stress 53
3.2.3 Stress and strain combined: Hooke’s law 55
3.2.4 Stress, strain, and time: viscosity 57
3.3 Linking stress and strain: Jeffreys’ theorem 58
3.3.1 Elastic deformation and topographic support 58
3.3.2 Elastic stress solutions and a limit theorem 60
3.3.3 A model of planetary topography 62
3.4 The nature of strength 64
3.4.1 Rheology: elastic, viscous, plastic, and more 64
3.4.2 Long-term strength 64
Box 3.2 The ultimate strength of solids 65
3.4.3 Creep: strength cannot endure 74
3.4.4 Planetary strength profiles 80
3.5 Mechanisms of topographic support 82
3.5.1 Plastic strength: Jeffreys’ limit again 82
3.5.2 Viscous relaxation of topography 82
3.5.3 The topographic advantages of density differences: isostatic
support 87
3.5.4 Dynamic topography 90
3.5.5 Floating elastic shells: flexural support of topographic loads 91
3.6 Clues to topographic support 93
Box 3.3 Flexure of a floating elastic layer 94
3.6.1 Flexural profiles 96
3.6.2 Anomalies in the acceleration of gravity 97
3.6.3 Geoid anomalies 99
Box 3.4 The ambiguous lithosphere 100
under U.S. or applicable copyright law.

4.2 Sources of tectonic stress 108

4.2.1 External sources of tectonic stress 108

4.2.2 Internal sources of tectonic stress 109
4.3 Planetary engines: heat sources and heat transfer 113
4.3.1 Accretional heat 113
4.3.2 Tidal dissipation in planetary interiors 114
4.3.3 Heat transfer by thermal conduction and radiogenic heat
production 116
4.3.4 Thermal convection and planetary heat transfer 121
4.4 Rates of tectonic deformation 127
4.5 Flexures and folds 128
4.5.1 Compression: folding of rocks 128
Box 4.1 Elastic and viscous buckling theory 130
4.5.2 Folding vs. faulting: fault-bend folds 133
4.5.3 Extension: boudinage or necking instability 135
4.5.4 Gravitational instability: diapirs and intrusions 136
4.6 Fractures and faults 139
4.6.1 Why faults? Localization 139
4.6.2 Joints, joint networks, and lineaments 141
4.6.3 Faults: Anderson’s theory of faulting 143
Box 4.2 Dip angle of Anderson faults 147
4.7 Tectonic associations 154
4.7.1 Planetary grid systems 154
4.7.2 Flexural domes and basins 155
4.7.3 Stress interactions: refraction of grabens by loads 157
4.7.4 Io’s sinking lithosphere 158
4.7.5 Terrestrial plate tectonics 160
Further reading 161
Exercises 162
5 Volcanism 169
5.1 Melting and magmatism 169
5.1.1 Why is planetary volcanism so common? 170
Box 5.1 The adiabatic gradient 173
5.1.2 Melting real planets 175
5.1.3 Physical properties of magma 183
under U.S. or applicable copyright law.

5.1.4 Segregation and ascent of magma 187

Box 5.2 The standpipe model of magma ascent 189
5.2 Mechanics of eruption and volcanic constructs 194
5.2.1 Central versus fissure eruptions 194
5.2.2 Physics of quiescent versus explosive eruptions 195
Box 5.3 A speed limit for volcanic ejecta 200

EBSCO Publishing : eBook Academic Collection (EBSCOhost) - printed on 11/3/2023 1:50 PM via YDUQS
AN: 399390 ; H. Jay Melosh.; Planetary Surface Processes
Account: s4526441
x Contents

5.2.3 Volcanic surface features 204

5.3 Lava flows, domes, and plateaus 208

5.3.1 Lava flow morphology 208
5.3.2 The mechanics of lava flows 210
5.3.3 Lava domes, channels, and plateaus 214
Further reading 218
Exercises 218
6 Impact cratering 222
6.1 History of impact crater studies 222
6.2 Impact crater morphology 223
6.2.1 Simple craters 224
6.2.2 Complex craters 224
6.2.3 Multiring basins 226
6.2.4 Aberrant crater types 228
6.2.5 Degraded crater morphology 229
6.3 Cratering mechanics 229
6.3.1 Contact and compression 230
6.3.2 Excavation 233
6.3.3 Modification 238
Box 6.1 Maxwell’s Z model of crater excavation 242
6.4 Ejecta deposits 244
6.4.1 Ballistic sedimentation 246
6.4.2 Fluidized ejecta blankets 248
6.4.3 Secondary craters 250
6.4.4 Oblique impact 251
6.5 Scaling of crater dimensions 251
6.5.1 Crater diameter scaling 252
6.5.2 Impact melt mass 253
6.6 Atmospheric interactions 254
6.7 Cratered landscapes 255
6.7.1 Description of crater populations 256
6.7.2 Evolution of crater populations 261
6.8 Dating planetary surfaces with impact craters 262
6.8.1 b > 2 population evolution 263
6.8.2 b < 2 population evolution 265
under U.S. or applicable copyright law.

6.8.3 Leading/trailing asymmetry 266

6.9 Impact cratering and planetary evolution 267
6.9.1 Planetary accretion 267
6.9.2 Impact catastrophism 268
6.9.3 Origin of the Moon 269
6.9.4 Late Heavy Bombardment 269

EBSCO Publishing : eBook Academic Collection (EBSCOhost) - printed on 11/3/2023 1:50 PM via YDUQS
AN: 399390 ; H. Jay Melosh.; Planetary Surface Processes
Account: s4526441
Contents xi

6.9.5 Impact-induced volcanism? 270

6.9.6 Biological extinctions 271

Further reading 271
Exercises 272
7 Regoliths, weathering, and surface texture 276
7.1 Lunar and asteroid regoliths: soil on airless bodies 276
7.1.1 Impact comminution and gardening 279
Box 7.1 Growth of the lunar regolith 282
7.1.2 Regolith maturity 285
7.1.3 Radiation effects on airless bodies 286
7.2 Temperatures beneath planetary surfaces 288
7.2.1 Diurnal and seasonal temperature cycles 289
7.2.2 Heat transfer in regoliths 290
7.2.3 Thermal inertia 293
7.3 Weathering: processes at the surface/atmosphere
interface 293
7.3.1 Chemical weathering 295
7.3.2 Physical weathering 300
7.3.3 Sublimation weathering 306
7.3.4 Duricrusts and cavernous weathering 308
7.3.5 Desert varnish 309
7.3.6 Terrestrial soils 310
7.4 Surface textures 311
7.4.1 “Fairy castle” lunar surface structure 311
7.4.2 Stone pavements: why the Brazil nuts are on top 313
7.4.3 Mudcracks, desiccation features 315
Further reading 316
Exercises 316
8 Slopes and mass movement 319
8.1 Soil creep 319
8.1.1 Mechanism of soil creep 320
8.1.2 Landforms of creeping terrain 323
8.2 Landslides 326
8.2.1 Loose debris: cohesion c = 0 327
under U.S. or applicable copyright law.

8.2.2 Cohesive materials c > 0 331

Box 8.1 Crater terraces as slump blocks 336
8.2.3 Gravity currents 339
8.2.4 Long-runout landslides or sturzstroms 340
Further reading 344
Exercises 345

EBSCO Publishing : eBook Academic Collection (EBSCOhost) - printed on 11/3/2023 1:50 PM via YDUQS
AN: 399390 ; H. Jay Melosh.; Planetary Surface Processes
Account: s4526441
xii Contents

9 Wind 348
Copyright © 2011. Cambridge University Press. All rights reserved. May not be reproduced in any form without permission from the publisher, except fair uses permitted

9.1 Sand vs. dust 349

9.1.1 Terminal velocity 349
9.1.2 Suspension of small particles 352
9.2 Motion of sand-sized grains 353
9.2.1 Initiation of motion 354
9.2.2 Transport by the wind 361
9.2.3 The entrainment of dust 363
9.2.4 Abrasion by moving sand 365
9.3 Eolian landforms 365
9.3.1 The instability of sandy surfaces 365
9.3.2 Ripples, ridges, and sand shadows 366
Box 9.1 Kamikaze grains on Mars 368
9.3.3 Dunes 371
9.3.4 Yardangs and deflation 376
9.3.5 Wind streaks 377
9.3.6 Transient phenomena 378
Further reading 379
Exercises 380
10 Water 382
10.1 “Hydrologic” cycles 383
10.1.1 Time, flow, and chance 383
10.1.2 Rainfall: infiltration and runoff 386
10.2 Water below the surface 388
10.2.1 The water table: the piezometric surface 388
10.2.2 Percolation flow 390
10.2.3 Springs and sapping 392
Box 10.1 How long can streams flow after the rain stops? 393
10.3 Water on the surface 395
10.3.1 Overland flow 396
10.3.2 Streamflow 401
10.3.3 Channels 407
Box 10.2 Analysis of stream networks 416
10.3.4 Standing water: oceans, lakes, playas 418
10.3.5 Fluvial landscapes 428
under U.S. or applicable copyright law.

11.2 Flow of glaciers 439

11.2.1 Glen’s law 440

11.2.2 The plastic-flow approximation 442
11.2.3 Other ices, other rheologies 443
11.2.4 Basal sliding 444
Box 11.1 Salt glaciers and solution creep 445
11.3 Glacier morphology 446
11.3.1 Flow velocities in glaciers and ice sheets 447
11.3.2 Longitudinal flow regime and crevasses 448
11.3.3 Ice-sheet elevation profile 449
11.4 Glacial landforms 451
11.4.1 Glacial erosion 451
11.4.2 Glacial deposition 452
11.5 Ice in the ground 454
11.5.1 Permafrost 455
11.5.2 Patterned ground 459
11.5.3 Thermokarst 462
Further reading 462
Exercises 463
References 465
Index 485
Color plates appear between pages 236 and 237
under U.S. or applicable copyright law.

Preface

We are privileged to be living in one of the greatest eras of exploration that humankind has
ever undertaken. Our current Age of Space grew out of the dark struggles of World War II
when large rockets were developed as agents of mass murder. The subsequent Cold War
rivalry between the United States and Soviet Union pushed rocket capabilities to the point
that it became possible to send vehicles into Earth orbit and beyond (even though the stated
aim was to send missiles carrying nuclear weapons over mere continental distances). The
Russians put the first human into Earth orbit. The Apollo missions took American astro-
nauts to the Moon, a target that Russia reached first with its unmanned vehicles: Russia
stopped just short of a manned lunar landing. Somehow, amid all this politically motivated
grandstanding, a few visionary engineers and scientists accomplished the feat that will be
remembered by all future generations: the exploration of our Solar System.
While humans have not yet traveled beyond the Moon, robotic spacecraft with increas-
ingly sophisticated electronic brains and sensory systems have now left the bounds of the
Solar System. Spacecraft have visited all of the major planets, with the exception of Pluto
(although some now argue that it is not really a “major planet”). Many planets and even
asteroids have been flown by, orbited and landed upon by spacecraft. We have yet to bring
back samples of any body other than the Moon, comet Wild 2 and asteroid 25143 Itokawa,
so there is much more to accomplish, but we are learning about the universe outside our
little Earth at a tremendous rate.
When I first started teaching a course in Planetary Sciences in 1977 it was possible to
treat each individual planet as a separate entity. Weeks would be spent talking about the
Moon and its special attributes. Mars was a Moon-like disappointment after the flybys of
Mariners 4, 6, and 7 that all, ironically, imaged nothing but the heavily cratered terrains of
the southern highlands. Mariner 10 had just returned our first views of Mercury, which also
under U.S. or applicable copyright law.

turned out to be very much like the Moon. As time went on, we learned much more about
the planets we had first studied and learned new things about planets that had never before
been visited by spacecraft. Mars blossomed into a new world in the wake of the Mariner 9
orbiter, with giant volcanoes, canyons that dwarfed Arizona’s Grand Canyon and channels
that could only have been cut by gigantic floods. Pioneer Venus and the Soviet Veneras
made it clear that our “sister” planet was a very odd relative indeed, and the Voyagers were
off on their historic tours of the outer Solar System.

EBSCO Publishing : eBook Academic Collection (EBSCOhost) - printed on 11/3/2023 1:50 PM via YDUQS
AN: 399390 ; H. Jay Melosh.; Planetary Surface Processes
Account: s4526441
xvi Preface

Before long it was clear that a planet-by-planet course organization could no longer
Copyright © 2011. Cambridge University Press. All rights reserved. May not be reproduced in any form without permission from the publisher, except fair uses permitted

work. It would be tediously repetitious to talk about craters and volcanoes on the Moon,
then later to talk about craters and volcanoes on Mars, and then craters and volcanoes on
Mercury, adding a new “craters and volcanoes” block for each new planet. While the num-
ber of planets kept multiplying, the number of different geologic processes did not. Pretty
much the same processes, modified a bit for local conditions, act on every body we have
investigated so far. So the modern course organization emphasizes processes, not individ-
ual planets. Furthermore, the body of information about each planet has multiplied to the
point that it is no longer possible to comprehensively cover all that is known about even
one planet within the confines of a one-semester class. If you doubt this, go to a library and
look at the shelf of books about planets in just the University of Arizona’s Space Science
Series. The total collection occupies about two meters of shelf space, and it grows by a few
tens of centimeters (or more!) every year.
This practical limitation accounts for the “process” orientation of this book. Beyond
this, I had to make decisions about which processes to treat and in what order. Textbooks
on terrestrial geomorphology abound and “process orientation” is a buzzword that most
modern books respect, but fluvial processes dominate terrestrial geomorphology. Fluvial
processes, however, are rare in the larger Universe and must take a back seat to more uni-
versal processes, such as impact cratering, in a planetary context. In teaching this class I
have long used an approach that follows the planetary exploration mantra of “first flyby,
then orbit, land, and finally return samples.” I start with those aspects of a planet that you
can see from the greatest distance, even telescopically (a level that we have just attained for
extrasolar planets). Thus, we can ask: what determines a planet’s shape and the topography
of its surface? Deviations from a spheroidal shape must be supported by internal strength,
which motivates a discussion of what strength is and how topographic variations can be
supported.
If topography is limited by strength, then what happens when it is exceeded? The answer
is tectonics: faults and fractures. As we approach ever closer to a planet, the next things
we might notice are craters, just as the first features on Mars and Mercury imaged by
spacecraft were cratered terrains. I had thus planned to make impact craters the subject of
Chapter 5, followed by volcanism in Chapter 6. However, one of the anonymous review-
ers of my original book proposal cogently argued that volcanism is most closely linked to
tectonics so that the order of these two chapters should logically be reversed. I agree, and
so the order is as you now have it – after all, the first things that Mariner 9 saw looming out
of the global dust storm were the summits of Mars’ four great shield volcanoes. The last
under U.S. or applicable copyright law.

five chapters are organized around the principle of most-to-least universal processes. All
bodies have regoliths, although the regolith of airless bodies such as the Moon or asteroids
differs profoundly from the agricultural soil of Earth. Regoliths do not need slopes to form,
but mass movement is a process that acts only on slopes, so that is the subject of Chapter
8. Chapters 9, 10, and 11 are, in the broadest sense, about the processes that involve wind,
water, and ice, even though the “wind” may be blowing carbon dioxide, the “water” liquid
methane, and the “ice” solid carbon dioxide or methane. These chapters are really about

EBSCO Publishing : eBook Academic Collection (EBSCOhost) - printed on 11/3/2023 1:50 PM via YDUQS
AN: 399390 ; H. Jay Melosh.; Planetary Surface Processes
Account: s4526441
Preface xvii

transport by atmospheric gases (universal for large enough planets and moons), liquids
Copyright © 2011. Cambridge University Press. All rights reserved. May not be reproduced in any form without permission from the publisher, except fair uses permitted

(fewer bodies possess flowing liquids on their surfaces), and solids warm enough to flow at
measurable rates (that is, very close to their melting points, which must be pretty unusual
on a planet’s surface).
In teaching this course I try to get through the entire set of processes in one semester (15
weeks of three hours of lecture per week). As my former students well know, I often do
not succeed. New discoveries come up, someone asks a lot of deep questions about some
topic, and I end up spending more time on one topic than the syllabus allows. The result is
that I usually have to rush through the last sections. I have often said, “If only there were a
text for this course, I could have the students read up on this topic and not miss out on an
important idea.” Well, here is the text. Maybe it will solve this problem.
Another note about how I teach this class: I typically assign challenging homework
problems that are meant to encourage the students to think. There are sometimes no strictly
right or wrong answers, just reasonable ones that admit of a lot of interpretation (there are
also some easy problems that just involve substitutions, but I hope the answers are enlight-
ening). I also ask the students to write a research paper on some topic that interests them,
and I base much of the final grade on these research papers. These papers are about ten
pages long and I encourage the students to think independently, not just regurgitate what
they may have read in some published paper. New calculations or even small-scale exper-
iments and field investigations are strongly encouraged. I do not penalize the students,
gradewise, if some initially promising line of research does not work out. Many of these
papers have turned into abstracts presented at the annual Lunar and Planetary Research
Conference. Some have turned into papers published in the scientific journals and a few
have become Ph.D. theses.
Because paper-writing becomes more intense as the semester proceeds, I ease off on
the amount of homework assigned to allow the students time to explore their own ideas.
This has often resulted in surprising bursts of creative activity that I do not wish to smother
under too much “set work.” For that reason you will find that the number and difficulty of
the exercises associated with each chapter falls off toward the end of the book. I do this in
the hope that the early part of the course will serve as a kind of “launch pad” for independ-
ent investigation of this fascinating field.
Anyone who teaches this subject must realize that planetary science is an active and
ever-changing subject. New discoveries are constantly being made. I have tried to incorp-
orate some of the latest discoveries in this text, but I fully realize that by the time this book
appears in print some things I have written will be obsolete (indeed, in my own research
under U.S. or applicable copyright law.

I am doing my best to make that happen). So it is important to supplement this text with
readings from the current literature and even news stories and NASA data releases.
Ah yes, one last piece of advice (and my former students would not forgive me if I failed
to mention this!): The stories. Some of them are here in the book, cleaned up a bit and prop-
erly referenced. Not all of them (some of the good ones I was unable to verify in this way –
they are in a file labeled “dubious stories” until I can find a reliable reference). Stories about
people, about ideas, about what motivated whom to do what and how some great idea came

EBSCO Publishing : eBook Academic Collection (EBSCOhost) - printed on 11/3/2023 1:50 PM via YDUQS
AN: 399390 ; H. Jay Melosh.; Planetary Surface Processes
Account: s4526441
xviii Preface

from something that seemed wholly unrelated. Some of it is the usual scuttlebutt of science,
Copyright © 2011. Cambridge University Press. All rights reserved. May not be reproduced in any form without permission from the publisher, except fair uses permitted

told over coffee or around campfires. But most of my stories are different: Like Aesop’s
fables, they all have a moral. Like all teachers, I am often distressed by how little students
seem to remember about some topic after the lapse of even one semester, let alone a few
years. So I try to wrap the really important ideas into a really good story about someone
or something. I think that makes the idea easier to remember and hope that the idea might
remain mentally accessible long after the equation or intricate train of reasoning has passed
beyond recall. I am not sure this works, but I do meet students who, after many years, still
retain the story, if not the point that it was meant to illustrate. Not everyone who teaches
this course will want to emulate this particular technique, but I do ask you not to drain the
human interest from the science. Science is done by humans, and for humans to continue to
do it they must realize how quirky and illogical the course of discovery can be.
With that, I invite you to move on into this book and make your own discoveries. I hope
you have as much fun learning this stuff as I have had.
October 2010
West Lafayette, Indiana
under U.S. or applicable copyright law.

Acknowledgments

I am grateful, most of all, to the many students who participated in my classes and whose
questions, answers to homework problems, responses to challenges, insightful research
projects, and field trip presentations added immeasurably to my appreciation and know-
ledge of the surfaces of the various strange objects that inhabit our Solar System. Most
of this work, particularly the field component of my classes, would have been impossible
without the support of the Department of Planetary Sciences and the Lunar and Planetary
Lab of the University of Arizona. There were some tense moments and challenging situ-
ations that developed in the field, but overall this support has been exemplary. I had long
dreamed of distilling my class lectures into a book, while adding more material that I never
had time to cover in a semester course, but it was Susan Francis, of Cambridge University
Press, who finally persuaded me to take on this monumental task. The actual writing of
this book has stretched over more years than I like to remember, during which time I
received aid from a large number of people. Virgina Pasek drafted nearly all of the figures
from my rough sketches. Her artistic ability and sense of graphic design shines through
in every chapter, except 1 and 6. She has been a steady and patient collaborator and I am
most grateful that she agreed to join me in this effort and persisted up to the last moment.
As the book neared completion, Francis Nimmo read Chapters 3 and 4 and made numer-
ous helpful suggestions that clarified the accuracy and precision of the material. Several
former students contributed figures, particularly Jason Barnes, Eric Palmer, Ralph Lorenz,
and Ingrid Daubar-Spitale. Steve Squyres contributed the beautiful cover image of Duck
Bay, Mars, through several iterations of re-processing. I am also most grateful to my wife,
Ellen Germann-Melosh, who has borne my preoccupation with writing on too many nights
and weekends with patience and understanding. She is surely tired of hearing about “The
Book” and I am delighted to finally send it forth to whatever fate awaits it in the larger
under U.S. or applicable copyright law.

world of science.

xix

Account: s4526441
AN: 399390 ; H. Jay Melosh.; Planetary Surface Processes
EBSCO Publishing : eBook Academic Collection (EBSCOhost) - printed on 11/3/2023 1:50 PM via YDUQS
1
Copyright © 2011. Cambridge University Press. All rights reserved. May not be reproduced in any form without permission from the publisher, except fair uses permitted

The grand tour

The planets are no longer wandering lights in the evening sky. For cen-
turies man lived in a universe which seemed safe and cozy – even tidy.
The Earth was the cynosure of creation and Man the pinnacle of mortal
life. But these quaint and comforting notions have not stood the test of
time. … No longer does “the World” mean the Universe. We live on one
world among an immensity of others.
Carl Sagan (1970)

From the seventeenth to the middle of the nineteenth century it was customary for the sci-
ons of affluent British families to make a long tour of all the capitals of Europe to acquaint
them with the architecture and culture of their larger world. In the twentieth century NASA
planned a “grand tour” of our Solar System that would visit every planet outside the orbit of
Mars. That tour never happened. Nevertheless, we have just about accomplished its goals,
with the final New Horizons encounter with Pluto scheduled for 2015.
Scientific exploration of the Solar System can be said to have started around 1610, when
Galileo Galilei (1564–1642) applied the newly invented telescope to investigate the world
beyond the Earth. Telescopes have increased greatly in both size and sophistication since
the days of Galileo, but even the best ground-based telescopes are unequal to the task of
detailed exploration of the planets. The beginning of the Space Age, opening with the
launch of Sputnik 1 in 1957, was the next leap forward in planetary exploration. Spacecraft,
carrying instruments and humans, have greatly expanded our knowledge of the planets and
moons around us.
The usual course of exploration of a planetary body, after remote astronomical observa-
tions, has been, first, flyby spacecraft, followed by orbiters, then landers and finally sample
returns and human in situ visitation. The majority of bodies discussed in this book are still
under U.S. or applicable copyright law.

in the orbiter or flyby stage of exploration: Humans have returned multiple large samples
only from our Moon, so far. In addition, very small amounts of material from comet Wild 2
and asteroid 25143 Itokawa have been returned by NASA’s Stardust and Japan’s Hayabusa
missions, respectively. Nevertheless, a great deal has been learned about the other bodies
orbiting our Sun. As this is written, the astronomical exploration of the planetary systems
around other stars has been underway for about a decade.

EBSCO Publishing : eBook Academic Collection (EBSCOhost) - printed on 11/3/2023 1:50 PM via YDUQS
AN: 399390 ; H. Jay Melosh.; Planetary Surface Processes
Account: s4526441
2 The grand tour
Copyright © 2011. Cambridge University Press. All rights reserved. May not be reproduced in any form without permission from the publisher, except fair uses permitted

Planet

a
Sun

Figure 1.1 Planets orbit about the Sun in elliptical orbits that lie in a plane. The Sun lies at one focus
of the ellipse. The size of the ellipse is defined by the semimajor axis a and the semiminor axis b.

1.1 Structure of the Solar System

Planets move around the Sun along paths that are, to a very good first approximation,
elliptical (Figure 1.1). Johannes Kepler (1571–1630), using tables of planetary positions,
deduced that the Sun lies at one focus of each planet’s ellipse and that the square of the
period P of each planet’s orbit is proportional to the cube of its semimajor axis a, a relation
now known as Kepler’s third law (the first law is that the planets move in elliptical orbits
with the Sun at one focus and the second is that, in moving along its orbit, planets sweep
out equal areas in equal times). Isaac Newton (1642–1727) extended Kepler’s empirical
observations with his law of universal gravitation, writing that the period of a small body
moving around a much larger body of mass M is given by:

4π 2 3
P2 = a (1.1)
GM
where G is Newton’s gravitational constant, equal to (6.67259 ± 0.00085) x 10–11 m3/(kg
s2). The detailed position of a planet in its orbit is defined by six numbers, of which we
shall be concerned with only three. The first is the semimajor axis a, which is conven-
tionally expressed in Astronomical Units (AU), equal to the semimajor axis of the Earth’s
orbit around the Sun. One AU is approximately equal to 1.496 x 108 km. The second is the
eccentricity e, defined by the ratio between the semimajor axis and semiminor axis b of
the ellipse:
under U.S. or applicable copyright law.

a 2 − b2
e= . (1.2)
a2
The eccentricity is zero for a circular orbit and equal to one for a parabolic orbit.
Eccentricities larger than one describe unbound, hyperbolic orbits. The last orbital parameter

EBSCO Publishing : eBook Academic Collection (EBSCOhost) - printed on 11/3/2023 1:50 PM via YDUQS
AN: 399390 ; H. Jay Melosh.; Planetary Surface Processes
Account: s4526441
1.1 Structure of the Solar System 3

of interest to us is the inclination of the ellipse to a reference plane, conventionally chosen

to be the plane of the Earth’s orbit around the Sun, the ecliptic plane (another often-used
reference plane is the invariable plane of the Solar System, which is mostly defined by
Jupiter’s orbit). The other three orbital parameters describe the orientation of the ellipse’s
axis in space (two numbers, the longitude of the ascending node and the argument of the
perihelion) and the location of the planet in its orbit (the true anomaly).

1.1.1 Major facts of the Solar System

There are a number of important facts about the Solar System, not all of which are yet
explained by models of Solar System formation, although most are at least consistent with
the current solar nebula model of Solar System formation. Table 1.1 lists the orbital char-
acteristics of the major planetary objects.
Planets and asteroids all formed at the same time. To the best of our ability to date
them, the Earth, Moon, Mars, and the most ancient meteorites all have the same forma-
tion age, about 4.57 Gyr before the present. To less precision, the age of the Sun is also
the same.
Planetary orbits are nearly coplanar. The planetary objects with the largest inclinations
to the ecliptic plane are Mercury, inclined at 7.0°, Venus, inclined at 3.4°, and Saturn,
inclined at 2.5°. Pluto, which is inclined at 17.1° is now believed to be a member of a fam-
ily of similar Trans-Neptunian Objects or TNOs with similarly large inclinations (Eris, the
largest TNO, is somewhat larger than Pluto and is inclined at 44° to the ecliptic).
Planetary orbits lie near the plane of the Sun’s equator. The Sun’s equator is inclined
to the ecliptic at about 7.5°, so that its orbital momentum vector is similar in direction to
those of the planets.
Planetary orbits are nearly circular. The three most eccentric planetary orbits are those
of Mercury, e = 0.2065, Mars, e = 0.0934, and Saturn, e = 0.0560. Pluto, with e = 0.2482,
seemed less like the other planets, until the discovery of many other similar TNO objects
confirmed that it belongs to a different population of objects than the inner planets (Eris’
eccentricity is 0.44, larger than Pluto’s).
The planets all revolve around the Sun in the same direction. There are no exceptions to
this rule, although many long-period comets do travel on retrograde orbits.
The rotation direction of most planets is direct. The obliquity (the inclination of the axis
of rotation to the orbital plane) of most planets is small, with the following major excep-
tions: Venus, whose rotation is retrograde, is inclined at an angle of 177.3° to its orbital
under U.S. or applicable copyright law.

plane. Uranus is also retrograde, although with an obliquity of 97.86° it nearly lies on its
side with respect to its orbit. Neptune is prograde, with an obliquity of 29.6°.
The spacing of planetary orbits follows a rough logarithmic series. Called the Titus–
Bode law, this relation states that the semimajor axis of the nth planet, an, is given by:
an = c1 + c2 2n. (1.3)

EBSCO Publishing : eBook Academic Collection (EBSCOhost) - printed on 11/3/2023 1:50 PM via YDUQS
AN: 399390 ; H. Jay Melosh.; Planetary Surface Processes
Account: s4526441
4 The grand tour

Table 1.1 Planetary orbits

Semimajor Semimajor Orbital

Period axis, a axis a by Eccentricity, inclination, i
Planet (years) (AU) Titus–Bode e (degrees)

Mercury 0.241 0.387 0.55 0.2056 7.004

Venus 0.615 0.723 0.7 0.0068 3.394
Earth 1.000 1.000 1.0 0.0167 0.000
Mars 1.881 1.524 1.6 0.0934 1.850
Ceres 4.60 2.766 2.8 0.0739 10.585
Jupiter 11.862 5.203 5.2 0.0483 1.038
Saturn 29.458 9.539 10 0.0560 2.488
Uranus 84.01 19.191 19.6 0.0461 0.774
Neptune 164.79 30.061 * 0.0097 1.774
Pluto 248.54 39.529 38.8 0.2482 17.148
Eris 557 67.67 77.2 0.4418 44.187

*
Advocates of the validity of this law frequently skip Neptune and list Pluto in its place, noting that
the “agreement is better.” Readers are encouraged to judge for themselves.

If the constants c1 and c2 are chosen to be 1 for Earth (n = 3) and 5.2 for Jupiter (n = 6),
this rule states:

2  3 n
an = + 2. (1.4)
5  40 
Table 1.1 lists the predictions of this “law” and shows that it does agree roughly with
observation. So far, there is no fundamental understanding of why this relation should be
true.

1.1.2 Varieties of objects in the Solar System

Besides the major planets, many different types of object make up the Solar System. In
order of decreasing size, these comprise the satellites of major planets: The larger ones in
the Solar System include Ganymede and Callisto (satellites of Jupiter, 5262 and 4800 km
diameter, respectively), Titan (satellite of Saturn, 5150 km diameter), Triton (satellite of
under U.S. or applicable copyright law.

Neptune, 2700 km diameter) and our Moon (3476 km diameter). Ganymede and Titan are
larger than Mercury and, as bodies in themselves, can be classed as planetary objects. For
the purposes of this book, we shall discuss these large satellites as varieties of planetary
object and make no distinction between objects that orbit the Sun and those that orbit other
planets.
In addition to satellites, we recognize asteroids, the largest of which is Ceres (diameter
950 km), but whose sizes range downward to a few kilometers, grading into objects that

EBSCO Publishing : eBook Academic Collection (EBSCOhost) - printed on 11/3/2023 1:50 PM via YDUQS
AN: 399390 ; H. Jay Melosh.; Planetary Surface Processes
Account: s4526441
1.2 Classification of the planets 5

would be classed as meteoroids (there is no universally recognized size that divides aster-
Copyright © 2011. Cambridge University Press. All rights reserved. May not be reproduced in any form without permission from the publisher, except fair uses permitted

oids from meteoroids: current authors set the dividing line anywhere between 1 km and a
few meters). The total mass of all the asteroids is small, only about 4% of the Moon’s mass,
most of which resides in the largest asteroids, Ceres, Vesta, and Pallas. By definition an
asteroid orbits the Sun and appears “star-like” in a telescopic image: That is, it does not dis-
play a “coma” or regularly emit gas and dust like a comet. Unfortunately for classifications,
there are a few objects that do not possess comas but in every other respect are comet-like,
while other objects long recognized as asteroids have suddenly acquired comas.
Comets are objects that, upon approaching the Sun, emit gas and dust to produce a coma
(literally, “hair” in Latin: Comets are “hairy stars”). The diameters of cometary nuclei range
from about a kilometer up to several tens of kilometers. They contain ices that, upon warm-
ing near the Sun, create their characteristic tails of gas and dust as the ices evaporate. There
are several classes of cometary orbit, ranging from low inclination orbits typical of short-
period comets to long-period comets that approach the Sun from the depths of the Oort
cloud (which ranges out to a large fraction of the distance from the Sun to the nearest star).
In addition to these macroscopic objects, the Solar System also contains dust particles,
whose diameter ranges down to submicron sizes, as well as individual atoms in the form of
plasma and cosmic rays, most of which are emitted by the Sun, although a small compo-
nent comes from interstellar space, as do some tiny dust particles.

1.2 Classification of the planets

The planetary-scale objects in our Solar System can be grouped into three general classes,
with a number of important subclasses.
Terrestrial planets. Planets similar to the Earth are of most direct interest to us as inhab-
itants of such a planet. The main mass of these planets is composed of silicate minerals,
although most have a metallic core rich in iron and nickel. Their densities range from 3000
to 6000 kg/m3 and they all have well-defined surfaces. Members of this class may or may
not possess atmospheres. This class includes many satellites and asteroids as well as major
planets. Examples are Mercury, Venus, Earth, our Moon, Mars, Io, and Ceres.
Icy satellites. Confined to the outer Solar System, this class consists of bodies mainly
composed of water ice, with a possible component of other ices such as carbon dioxide,
ammonia or, in the extreme outer Solar System, methane and nitrogen. These objects may
have cores of silicate minerals that include hydrated silicates and, possibly, metallic inner
cores. Densities of these objects range from 1000 to more than 2000 kg/m3.
under U.S. or applicable copyright law.

Jovian (gas giant) planets. Jupiter and Saturn are composed of nearly the same materials
as the Sun, mainly hydrogen plus helium and an admixture of heavier elements such as car-
bon and oxygen typical of “ices.” These objects do not possess definite surfaces although
they may have dense rocky or metallic cores. Their densities fall in the range of 700 to 2000
kg/m3. These planets are often divided into Jovian Planets (Jupiter and Saturn) and ice-rich
Neptunian Planets (Uranus and Neptune), which contain a higher proportion of carbon and
oxygen than the Sun.

EBSCO Publishing : eBook Academic Collection (EBSCOhost) - printed on 11/3/2023 1:50 PM via YDUQS
AN: 399390 ; H. Jay Melosh.; Planetary Surface Processes
Account: s4526441
6 The grand tour

1.2.1 Retention of planetary atmospheres

All of the large planets and the largest moons possess atmospheres. Planetary atmospheres
are of great interest in themselves, but this book lacks space to discuss them: References to
good sources of information on them are given in the Further reading section at the end of
the chapter. However, the presence of an atmosphere strongly affects surface temperatures
and, when present, permits a wide variety of surface processes to operate that would not be
possible in its absence.
Whether or not a planetary object possesses a substantial atmosphere is dependent upon
two main factors: The planet’s escape velocity and the temperature of its exosphere. The
escape velocity is the minimum speed required for a body initially on the surface to escape
to infinity. If the mass of a planet is M and its radius is R, the escape velocity is given by:

2GM
vesc = = 2 gR (1.5)
R
where g is the surface acceleration of gravity. The second form of this relation can be used
to compute the escape velocity of the objects listed in Table 1.2. The escape velocity of the
Earth is 11.2 km/s, faster than almost any molecule near its surface (except hydrogen and
helium) and so it retains a substantial atmosphere. The Moon, on the other hand, has an
escape velocity of only 2.4 km/s and it has no permanent atmosphere.
The other factor in atmospheric escape is the temperature of the atmospheric gases.
Kinetic theory tells us that the most probable velocity of a gas molecule of mass m at tem-
perature T is given by:

2 kT
vT = (1.6)
m

where k is Boltzmann’s constant. Note the dependence on the inverse molecular mass:
Light molecules escape more easily than heavy ones, which accounts for the near absence
of hydrogen and helium in the atmospheres of the terrestrial planets. There are two subtle-
ties of this relation. The first is that, although Equation (1.6) is the most probable velocity,
the velocity distribution has an exponential tail at high velocities and so a small fraction
of gas molecules moves many times faster than vT. Over geologic time a large fraction of
the atmosphere may slowly leak away even though the most probable velocity is several
times smaller than the escape velocity. The other subtlety is that the relevant tempera-
ture is not the surface temperature but the temperature high in the atmosphere, where the
under U.S. or applicable copyright law.

atmospheric gases are so thin that a molecule moving upward has a good chance of escap-
ing into interplanetary space without colliding with another molecule. This portion of a
planetary atmosphere is called the exosphere. Solar UV radiation tends to heat the upper
reaches of planetary atmospheres to temperatures much higher than the surface, so that
gases that might seem to be stable on the basis of the surface temperature can, in fact, leave
the planet.

Table 1.2 Physical data on terrestrial planets and major moons

Inclination Maximum
Rotation Mean Surface of equator surface
Planet or period Equatorial density gravity to orbit temperature
moon (days) radius (km) (kg/m3) (m/s2) (degrees) (K)

Mercury 58.6 2439 5430 2.78 2 700

Venus 243 6051 5250 8.60 177.3 735
Earth 1.00 6378 5520 9.78 23.45 311
Moon 28 1738 3340 1.62 6.68 396
Mars 1.03 3393 3950 3.72 25.19 293
Ceres 0.38 487 2080 0.27 ~3 ~239
Io 1.77 1821 3530 1.80 2.2 130
Europa 3.55 1569 3010 1.31 0.1 125
Ganymede 7.15 2634 1936 1.43 0–0.33 152
Callisto 16.69 2410 1834 1.24 0 165
Titan 15.95 2576 1880 1.35 0 94
Triton 5.88 1353 2061 0.78 0 38
Pluto 6.4 1153 2030 0.66 119.6 55
Eris > 0.33 ? 1300 2250 ~0.8 ? 55

In addition to this thermal escape mechanism, other erosion processes such as impinge-
ment of the solar wind on atmospheres that are not defended by a planetary magnetic field,
or ejection of atmospheric gases by impacts, may play important roles. One or both of
these mechanisms may have depleted the Martian atmosphere over time. An early phase
of strong UV emission by the newborn Sun may have ejected atmospheric gases from the
early planets in a process known as hydrodynamic escape.

1.2.2 Geologic processes on the terrestrial planets and moons

The gamut of geologic processes that act on the surface of a planet or moon is the princi-
pal subject of this book. In subsequent chapters each process is considered in some detail.
The chapters are arranged in approximate order of the universality of each process, ranging
from processes that act on all planetary objects to processes that may affect only a few
special bodies. Different processes act on different bodies, but if there is any overall organ-
under U.S. or applicable copyright law.

ization to how different processes affect different planets (and there are many exceptions
to any rule one might try to make), it is that planetary complexity increases with planetary
size. Small bodies are home to only a limited variety of processes, while large planets are
much more diverse.
Figure 1.2 is an attempt to capture this progression in a simple diagram that lists pro-
cess as a function of planetary diameter, with a few examples added for definiteness.

EBSCO Publishing : eBook Academic Collection (EBSCOhost) - printed on 11/3/2023 1:50 PM via YDUQS
AN: 399390 ; H. Jay Melosh.; Planetary Surface Processes
Account: s4526441
8 The grand tour

Geologic Process: Planetary Diameter*:

1 km 10 km 100 km 1000 km 10 000 km

Meteorites Asteroids Moon Mars Earth
Vesta Mercury Venus

Tectonics _ _ _____________________

(bent and broken rocks)

Volcanism _ __________________________

(liquids and gases venting from the interior)

Impact Cratering _________________________________________________ _

(collisions with space junk)

Weathering ______________________________________________________
(surface interactions with space or atmosphere)

Mass Wasting _________ _

(landslides, downslope movement, undermining)

Interaction with Surface Fluids: ______

Eolian Processes
(wind, sand and dust)
Fluvial Processes ____
(springs, streams, lakes and oceans)
Glacial Processes ____
(slow flow of “thick water”, ice in the ground)

Biologic Processes _
(critters on and below the surface)

Surface Fluids: Airless Trace Dense

Atmosphere

Internal Heat Transfer Conduction Convection

* For silicate bodies: Size ranges are generally smaller for icy bodies.

Figure 1.2 The activity of different geological processes is a function mainly of the size of the
planetary body. The horizontal lines in this figure indicate the importance of each of the processes
listed along the left side of the figure.

Impact cratering is probably the most universal process, although it is less important on
large bodies than on small ones, mainly because other, more rapid, processes are more
effective. Mass movement and surface modification of various kinds are similarly univer-
sal, although for small bodies both are strongly coupled with impact cratering. Tectonics,
the process of rock deformation and fracture, is also important across the entire scale of
sizes, but it is somewhat more effective on larger bodies where stresses can more easily
approach the limit of rock strength. Volcanism might seem to be a mostly large-body
under U.S. or applicable copyright law.

process, yet evidence of melting has been found on even the smallest objects in the Solar
System, remnants of an early era in which radioactive heat sources were more effective
than now.
The cluster of processes that require active fluids on the surface of planets, including
“wind” (the movement of any atmospheric gas), flowing “water” (which could be any li-
quid, such as methane on Titan), and “ice” (again, any highly viscous material near its
melting point) are exclusively large-planet processes, because it requires a large body to

hold an atmosphere. Finally, biologic processes are, so far as we know, confined to the lar-
Copyright © 2011. Cambridge University Press. All rights reserved. May not be reproduced in any form without permission from the publisher, except fair uses permitted

gest and most diverse planet in the Solar System, Earth.

1.3 Planetary surfaces and history

Not all of the planets in the Solar System have well-defined surfaces. By surface, I mean
a thin zone (thickness under a centimeter or less) across which physical and mechanical
properties such as density, strength, sound speed, etc., show an abrupt change. The gas giant
planets Jupiter, Saturn, Uranus, and Neptune probably do not show such abrupt changes
and so can be considered to be without surfaces in this sense.
In this book we are mostly concerned with the surfaces of the terrestrial planets, silicate
and icy moons. Much of what we will learn can also be applied to the surfaces of asteroids
and comets. The surfaces of silicate-rich bodies are mostly composed of the refractory oxides
of Si, Mg, Fe, Al, and Ca. These minerals have melting points far above ambient temperatures
and so the physical state of the terrestrial planet’s surfaces is solid: For most purposes we
can treat the crust of these planets as a brittle elastic solid (something we loosely call rock).
Because rocks do not deform until applied stresses exceed some yield point, it is difficult
to remove all traces of past events that may have acted upon them. Rocks can record many
aspects of the history of their planets. This fortunate aspect also makes planetary surfaces
complex. A given planetary surface not only shows the effects of forces acting on it at the
present time, but its structure also carries traces of forces that may have long ceased to exist.
Similar considerations apply to the icy moons, so long as they remain well below the melting
point of their ices. In the outer Solar System, beyond the orbit of Jupiter, temperatures are far
below the melting point of water ice, and so water ice behaves in most respects like rock.
The gas giant planets are different in this respect from the solid planets: Although the
atmospheres of the giant planets show complex structures, these structures are the conse-
quence of presently acting forces: History is not a major player in their form. Even though
the Great Red Spot on Jupiter has been visible for centuries, it would dissipate within hours
if the vortex maintaining it were to die out.
The Earth’s surface presents a minor exception to our emphasis on solid surfaces:
Roughly ¾ of the Earth’s surface is underlain by liquid water. The sea surface cannot main-
tain a record of the forces acting upon it for more than a few hours or days. The study of
the sea surface is thus one of current or very recent forces acting upon the liquid: There is
no paleontological aspect to sea surfaces. Of course, beneath the sea we find rocks that do
maintain a record (although still of limited length due to plate tectonic recycling), so that
under U.S. or applicable copyright law.

we can learn much about the Earth’s history by studying the sea floor.
Our emphasis in this book will be more on the response of rocky silicate or ice surfaces
to applied forces than on the historical goal of using planetary surfaces to unravel the his-
tory of the planet. A clear understanding of these physical processes is a prerequisite to
the interpretation of some particular surface. Moreover, we do not yet have sufficient data
about most planets to make unambiguous statements about their history. Thus, at the pre-
sent time, the most fruitful approach is the study of physical processes.

EBSCO Publishing : eBook Academic Collection (EBSCOhost) - printed on 11/3/2023 1:50 PM via YDUQS
AN: 399390 ; H. Jay Melosh.; Planetary Surface Processes
Account: s4526441
10 The grand tour

We begin our study of planetary surfaces with a review of the gross properties of the
Copyright © 2011. Cambridge University Press. All rights reserved. May not be reproduced in any form without permission from the publisher, except fair uses permitted

terrestrial planets and moons, and undertake a brief description of their present surfaces
and what is known of their past. Table 1.2 lists some of the basic physical properties of a
selection of Solar System objects with solid surfaces.

1.3.1 The Moon

Two easily recognizable terrain types dominate the surface of the Moon: The maria and
the terrae (or highlands). The maria are dark (with a normal albedo of about 0.08), smooth
plains that are found predominantly on the Moon’s nearside. They are lightly cratered and
are mantled with a layer of comminuted rock (the regolith) to a depth of a few meters. Lava
flow fronts are observed in them, with scarps up to 100 m high (many times higher than
typical for terrestrial lava flows). Samples returned by the Apollo astronauts show that they
are composed of fine-grained basalt. The maria are volcanic plains formed by the extrusion
of large volumes of highly fluid basalt. These flows may be equivalent to terrestrial flood
basalts such as those that form the Columbia Plateau or the Deccan Traps. The mare basalts
sampled so far were extruded over a ca. 700 Myr interval from about 3.2 to 3.9 Gyr ago.
Some small central volcanic features are also observed, although there is little indication of
silica-rich explosive volcanism.
The terrae represent an older surface than the mare. The typical terra is a rugged-looking
hilly surface in which many circular, rimmed pits (impact craters) form at many different
size scales. The terrae surfaces are typically bright (normal albedo about 0.16) and are best
developed in the Moon’s southern latitudes on the nearside. They are the predominant ter-
rain type on the farside. The terrae are covered with a layer of broken rock tens of meters
thick that may itself overlie a megaregolith of broken rock ten or more kilometers thick.
All lunar samples that have been returned from terrae are breccias (that is, composed of
broken angular rock fragments that may have been cemented by heat and pressure from
impacts). The terrae may represent areas where the original crust of the Moon is exposed:
They have, however, been subjected to such intense meteoritic bombardment that crater has
obliterated crater to the point that no remnant of the original surface remains. The entire
surface is in a “saturated” or “equilibrium” condition with respect to cratering. Although
the terrae were also once composed of igneous rock, they are richer in Al than mare basalts.
The composition of the terrae is often described as “anorthositic gabbro” (70% plagioclase,
20% orthopyroxene, 9% olivine, 1% ilmenite). Anorthositic gabbro is mafic (Fe and Mg
rich and poor in SiO2) compared to terrestrial granite.
under U.S. or applicable copyright law.

Because the terrae all formed since the crust of the moon crystallized at about 4.6 Gyr,
their high crater density, compared to that on the more lightly cratered mare, implies that
an era with an especially high cratering rate must have occurred in the interval between 4.6
and 3.9 Gyr. A controversy is currently raging about whether this era of heavy bombard-
ment was the tail end of a high cratering flux extending from the time of the Solar System’s
origin, about 4.6 Gyr ago, or whether it was part of a relatively short spike in the cratering
rate, the “Late Heavy Bombardment.” Current models of Solar System formation relate

such a spike to a resonant interaction between Jupiter and Saturn that destabilized orbits in
Copyright © 2011. Cambridge University Press. All rights reserved. May not be reproduced in any form without permission from the publisher, except fair uses permitted

the asteroid belt and led to a shower of asteroids that affected all of the terrestrial planets,
including the Moon.
Multiring basins dominate the regional geology of the Moon. The type example of such
basins is Orientale, just over the Moon’s western limb. Two or more inward-facing ring
scarps surround its inner mare basin, each 2 to 8 km high and 600 to 900 km in diameter.
Outside the rings a radially lineated and grooved terrain is composed of terra material (the
Hevelius Formation). Associated with the lineated terrain are chains of secondary impact
craters. These features, along with direct sampling of the lineated unit surrounding the
Imbrium basin (the Fra Mauro Formation) imply that these basins were formed by very
large impact events. Imbrium, with a ring diameter of 1340 km (nearly equal to the radius
of the Moon itself) is the largest fresh basin. On the farside, the heavily degraded South
Pole-Aitken basin is even larger, with a diameter of 2600 km and a depth of about 13 km.
Grimaldi, the smallest multiring basin, is only about 200 km in diameter.
The surface of the Moon seems to be saturated with multiring basins (Figure 1.3), just as
the terrae are saturated with craters. Every point on the Moon’s surface is thus either inside
an old multiringed basin or on the ejecta blanket of one. The surface of the Moon is thus
layered to a depth perhaps exceeding 10 km, the stratification consisting of overlapping
ejecta blankets from nearby basins. These basins are so large that the formation of each of
them affected the entire Moon.
There is a widely held misconception (which has even prompted the writing of papers
to “explain” it) that there are more large impact basins on the nearside than on the farside.
Figure 1.3 should dispel this notion. It is not even true that the largest basin is on the near-
side: South Pole-Aitken, on the southern farside, is nearly twice as large as Imbrium.
The source of this misconception is probably the very real asymmetry in maria between
the nearside and farside. A multiring basin that has been flooded with basalt becomes a
very striking circular feature, of which there are many on the nearside and essentially none
on the farside. The coincidence between mare basalt flooding and a multiring basin also
produces an anomalously large gravity field over the basin, a “mascon” that may locally
increase the acceleration of gravity by 0.003 m/s2 (or about 0.2% of the total field), com-
parable to the largest gravity anomalies observed on Earth.
Although there is no nearside–farside asymmetry of impact basins, there is a strong
asymmetry in crustal thickness between the nearside and farside. The farside crust is
almost twice as thick as that on the nearside. The major consequence of this asymmetry is
the lack of dark maria on the farside, perhaps because of the difficulty lavas experienced
under U.S. or applicable copyright law.

in rising through the thicker crust. Another obvious consequence of this asymmetry is that
the Moon’s center of figure is displaced several kilometers away from its center of mass.
This displacement, as might be expected, is away from the Earth, roughly along the line
connecting the centers of the Earth and Moon.
In summary, the most common landforms on the moon are impact craters (Figure 1.4), cir-
cular rimmed pits that range in size from submicroscopic to multiring basins with diameters
comparable to the Moon’s own diameter. Each size range of crater has its own distinctive

EBSCO Publishing : eBook Academic Collection (EBSCOhost) - printed on 11/3/2023 1:50 PM via YDUQS
AN: 399390 ; H. Jay Melosh.; Planetary Surface Processes
Account: s4526441
NORTH
80° POLE
(a) 70°
80°
70°
60° 60°

HU
? ?
Copyright © 2011. Cambridge University Press. All rights reserved. May not be reproduced in any form without permission from the publisher, except fair uses permitted

M
?
50°

BO
50°

LD
? ?

T
?

IA
40° 40°

NU
?

M
?
30° ? 30°

?
?
? IMBRIUM

?
Apollo SERENITATIS
20° 20°
15 + Ma

?
P1
P2 P3 +Apollo CRISIUM

?
10° 17
? 10°

?
In Tr ?
? Luna
+
?

?
0° ? Apollo 14 20 0°
Grimaldi ? + Fe Smythii
FB

?
? +? ?

?
?Apollo
NECTARIS
16 10°
10°
ORIE

?
HUMORUM Nu WA
?
NTAL

20° 20°

?
?
?
E

?
30° 30°
Sc MV
40° Zu hille
cch r- 40°

a le
M ius

str
RY EN 30° 20° 10° 0° 10° 20° 30

Au
PH Bailly 0°

?
°4 50
DB DE ° °4
60 50 0° 6° 50°
?

ER L- 0°

70
°
°

80
South Pole-Aitken
70
80

G 60°

°
60°

°
70° AG 70°
NEARSIDE 80°SOUTH POLE 80°
WEST EAST
LONGITUDE LONGITUDE
NORTH POLE
(b) 80° ? 80°
70° 70°
60° IMB 60°
RIU
?
?

50° HUMBOLDTIANUM M
? 50°
Coulomb
40° ?
?

BIRKHOFF Sarton 40°

30° 30°
V
ONO

MOSCOVIENSE
LORENTZ
20° CRISIUM
HAR

20°
?
H-S

?
?

? ?
?

10° LF ?
LIC

10°
G
?

N
UND
?

RU
?
?

?
EV

AK
Smythii

0° ? 0°
FRE

MENDELEEV
RO

Z
RT

?
KO

Keeler- ORIENTALE
?

HeavisideAitken 10°
?

10° TS -
l e ?
Po
?
h
ut

20° MILNE 20°

Ingenii
?

30° GW APOLLO 30°

Au
str
a
40° le Planck Poincaré 40°
under U.S. or applicable copyright law.

L-
?

° 0° 0°170°180°
170°16 E G
ND ER
0°150 140° 1
° 40 15 16
?

50° SR1201310° ? ° 1
30 20°
° ME DB
10
11

SCHRODINGER 60° RY
0°
11 °
0

0°
10

0°

60°
70° AG 70°
80° 80° FARSIDE
SOUTH POLE
EAST WEST
LONGITUDE LONGITUDE

Figure 1.3 The major multiring basins of the Moon and the extent of their ejecta deposits are
indicated. Curved lines indicate major rings. Panel (a) is the Moon’s nearside and (b) is its farside.
See color plate section for full detail. Blue indicates the deposits of the youngest (Imbrian) basins,
yellow-orange Nectarian, dark brown Pre-Nectarian. After Plates 3A and 3B in Wilhelms (1987).

EBSCO Publishing : eBook Academic Collection (EBSCOhost) - printed on 11/3/2023 1:50 PM via YDUQS
AN: 399390 ; H. Jay Melosh.; Planetary Surface Processes
Account: s4526441
1.3 Planetary surfaces and history 13
Copyright © 2011. Cambridge University Press. All rights reserved. May not be reproduced in any form without permission from the publisher, except fair uses permitted

Figure 1.4 The topography of the Moon referenced to a sphere with a radius of 1737.4 km. Data
were obtained from the Lunar Orbiter Laser Altimeter (LOLA) that was flown on the mission Lunar
Reconnaissance Orbiter (LRO). The color-coded topography is displayed in two Lambert equal
area images projected on the near and farside hemispheres, see color plate section. Courtesy Mark
Wieczorek, August 7, 2010.

morphology and in each size range we see a gradation from “fresh,” recently formed craters
to degraded, barely recognizable battered remnants of the pristine form. While the large
impact craters dominate the tectonics and stratigraphy of the lunar surface, the small to
medium-size impacts make their contribution in forming the lunar soil or “regolith.”
The regolith is a mantle of broken rock and glass which is 4 to 10 m thick on the mare
and which covers the entire surface of the Moon. The regolith is slowly churned (gardened)
by small meteorite impacts, which are also responsible for the bulk of lateral transport of
material both on the plains and down slopes.
Except for the basaltic mare plains, there are few indications of internal tectonic activ
ity on the Moon. The maria, especially the mascon basins, are cut by tectonic faults, with
mainly extensional graben concentric to the basins and compressional mare ridges within
the basins. These features are linked to subsidence of the Moon’s lithosphere under the
under U.S. or applicable copyright law.

immense weight of the lava filling the basins. Lunar seismicity is weak by terrestrial stand-
ards, and slaved to alternating tidal stresses. Recent high-resolution observations by the
Lunar Reconnaissance Orbiter do reveal a few small compressive thrust faults in the lunar
highlands, suggesting a small amount of geologically recent global contraction. The major
possible exception to this lack of tectonics is the lunar “grid,” a preferential trend of line-
ations on the Moon’s surface that may reflect ancient fracturing of the Moon’s crust as the
Moon receded from the Earth.

EBSCO Publishing : eBook Academic Collection (EBSCOhost) - printed on 11/3/2023 1:50 PM via YDUQS
AN: 399390 ; H. Jay Melosh.; Planetary Surface Processes
Account: s4526441
14 The grand tour
Copyright © 2011. Cambridge University Press. All rights reserved. May not be reproduced in any form without permission from the publisher, except fair uses permitted

Figure 1.5 Oblique view of the surface of Mercury from the MESSENGER spacecraft showing
volcanic flows of the intercrater plains. NASA/JPL/APL/Carnegie image PIA11773.

We can thus look upon the Moon as an essentially primitive body, chemically differen-
tiated, but carrying a nearly unchanged record of the heavy meteorite bombardment that
affected all of the planets inside the orbit of Jupiter. The Moon’s surface has changed only
slightly since about 3 Gyr ago, after the last lava flows had cooled. Only a few large young
craters such as Copernicus (ca. 1 Gyr) and Tycho (ca. 100 Myr) were added to complete the
present picture. Interestingly, the Moon’s surface is only slightly younger than the oldest
rocks preserved on the Earth’s surface, so that the Moon preserves a record of events for
which we have no comparable information in terrestrial history.

1.3.2 Mercury
Upon first glance, Mercury appears to be very Moon-like (Figure 1.5). Although its mean
density of 5430 kg/m3 indicates that it is internally different from the Moon (bulk density
3340 kg/m3), the surface seems nearly indistinguishable from that of the Moon. Mercury’s
dominant landform is impact craters, ranging from the limit of resolution (a few 100 m at
under U.S. or applicable copyright law.

present) to the 1540 km diameter Caloris Basin.

Mercury possesses mare-like plains that are probably volcanic in origin, although
Mercury’s mare are not dark like the Moon’s. The crater density on Mercury, however, is
lower than that of the lunar highlands. There are extensive light intercrater plains separat-
ing individual large craters.
Unlike the Moon, Mercury shows some signs of internal tectonic activity early in its
history. Long thrust fault scarps, up to a few kilometers high, wind hundreds of kilometers
across the surface of the planet, cutting craters and plains alike. These scarps may have

formed as Mercury’s interior cooled and contracted after an elastic lithosphere developed.
Copyright © 2011. Cambridge University Press. All rights reserved. May not be reproduced in any form without permission from the publisher, except fair uses permitted

Recent images from NASA’s MESSENGER mission flybys reveal at least one volcanic
center located on the periphery of the large, volcanically flooded Caloris Basin.
Mercury’s rotation is locked into a resonance with its orbital period: It orbits the Sun in
88 days as it rotates on its axis once every 58.6 days, in a 3:2 resonance that appears stably
maintained by mass anomalies within the body of the planet. This resonance coordinates
with its highly elliptical orbit so that one point on its equator (the hot pole) alternately
faces toward and away from the Sun at perihelion. The warm poles, 90° away from these
scorching locations (reaching about 700 K), are slightly more temperate but still very hot
by terrestrial standards. On the other hand, radar reflections seem to indicate the presence
of large masses of ice in permanently shadowed craters near Mercury’s north pole.
In spite of its slow rotation Mercury possesses a small magnetic field, suggesting that
its core is still molten and convecting. The magnetic field prevents the solar wind from
impinging directly on its surface, which may have important consequences for regolith
processes that are still not fully understood. It is unclear why Mercury should have such
a large metallic core: It makes up a much larger fraction of Mercury’s total mass than the
core of any other terrestrial planet. Perhaps a large, nearly head-on collision early in Solar
System history ejected a large fraction of its silicate mantle. In spite of its large iron-rich
core, remote chemical measurements of Mercury’s surface rocks show a surprising defi-
ciency of iron oxide, which may account for its uniformly bright surface compared to the
Moon.
The MESSENGER spacecraft is about to enter its orbital phase as this book goes to
press, at which time a great increase in our understanding of Mercury’s geology is antici-
pated. Readers should watch the news and subsequent papers for more information on this
surprisingly interesting body.

1.3.3 Venus
Although Venus, of all the terrestrial planets, is the most similar in size to the Earth, it is
most emphatically not Earth’s twin. Venus’ surface is searingly hot, baking at a nearly
constant temperature of 730 K, hotter than the surface of Mercury. It is crushed under a
CO2 atmosphere 100 times more massive than the Earth’s, which maintains the high sur-
face temperature through a runaway greenhouse effect. Although wind velocities in this
massive atmosphere are low, aeolian dune fields and wind streaks are abundant. Venus may
once have possessed oceans, but it has lost its water through dissociation and preferential
under U.S. or applicable copyright law.

loss of hydrogen to space.

Venus’ dense atmosphere supports a thick layer of high clouds composed mainly of a
very stable sulfuric acid aerosol that reflects 67% of the incident sunlight and completely
obscures the surface from visual imaging. Only radar from Earth-based or spacecraft plat-
forms has finally revealed the details of its surface. Venus’ rotation is retrograde, with a
period of 243 days. This period puts is very close to a near-resonance with the Earth, such
that Venus shows the same face to the Earth whenever conjunction occurs, which happens
every five Venusian days.

EBSCO Publishing : eBook Academic Collection (EBSCOhost) - printed on 11/3/2023 1:50 PM via YDUQS
AN: 399390 ; H. Jay Melosh.; Planetary Surface Processes
Account: s4526441
16 The grand tour

(a) (b)
Copyright © 2011. Cambridge University Press. All rights reserved. May not be reproduced in any form without permission from the publisher, except fair uses permitted

Figure 1.6 Topographic elevations from the Magellan radar altimeter. Panel (a) is centered on 0°
Longitude, panel (b) is centered on 180°. The surface of Venus is occupied by seemingly randomly
spaced rises and plains with a few highlands such as the Lakshmi Plateau in panel (a), near Venus’
north pole. Maxwell Montes, the highest point on Venus, rises above the plateau. Note the extensive
chain of circular coronae extending across the lower half of panel (b). This chain ends in the large
incomplete circle of Artemis Chasma. Very few impact craters are visible. Panel (a) is NASA/JPL/
USGS PIA00157 and panel (b) is PIA00159. See also color plate section.

Venus does not lose internal heat through plate tectonics, but may suffer episodic global
spasms of volcanic resurfacing. The most recent such event obliterated any pre-existing
surface of the planet about 700 Myr ago, covering the entire surface with thick sheets of
fluid basaltic lava. Large volcanic/tectonic complexes called coronae are the most char-
acteristic feature (Figure 1.6), along with central volcanoes, folded mountain belts and
extensional plains. Recent evidence from ESA’s Venus Express mission suggests that vol-
canism may be currently active. The surface is warped and broken by myriads of faults of
all varieties: Venus is a tectonic paradise. A small number of craters (about 940) punctuate
the surface, the largest of which, Mead, is about 280 km in diameter: There is no sign of the
ancient large basins that sculpture the crust of the Moon, Mercury or Mars.
The most surprising fact about Venus is that it possesses substantial topography: Its high-
est mountain, Maxwell Montes, rises 11 km above its mean elevation. It is a major puzzle,
how rocks at the high temperatures of the Venusian surface can support such high topog-
raphy for geologic periods. Perhaps a total absence of water in the Venusian crust adds an
under U.S. or applicable copyright law.

unearthly degree of strength to the near-surface material.

1.3.4 Mars
Mars has long been the center of attention of spaceflight enthusiasts and is still considered
one of the most likely planets, other than the Earth, to harbor life. It appears Earth-like

EBSCO Publishing : eBook Academic Collection (EBSCOhost) - printed on 11/3/2023 1:50 PM via YDUQS
AN: 399390 ; H. Jay Melosh.; Planetary Surface Processes
Account: s4526441
1.3 Planetary surfaces and history 17
Copyright © 2011. Cambridge University Press. All rights reserved. May not be reproduced in any form without permission from the publisher, except fair uses permitted

Figure 1.7 False color topography of Mars from the MOLA instrument aboard the Mars Global
Surveyor spacecraft. The left hemisphere is dominated by the Tharsis Rise with its enormous
volcanoes. Olympus Mons rises to the upper left. The gigantic trough of Valles Marineris extends to
the right center. The northern lowlands and the Borealis plains dominate the upper half of the right
hemisphere. The deep circular basin to the lower left is Hellas and the smaller basin near the center
is Utopia. NASA/JPL image PIA02820. See also color plate section.

in its possession of an atmosphere (albeit very thin, with a surface pressure only 0.6% of
Earth’s and composed almost entirely of CO2), polar caps, and clouds. Although Mars does
not presently possess a magnetic field, detection of a magnetized ancient crust suggests
that it once did.
Like the Moon, the Martian crust is divided into asymmetric halves: The crust beneath
the northern plains is thinner than that beneath the southern highlands: A relatively narrow
escarpment separates the two. Mars is heavily cratered. Its surface is dominated by several
large impact basins, the largest of which, Hellas, is 2300 km in diameter and 7 km deep
(Figure 1.7).
Mars stands as an intermediate between Moon-like bodies and the Earth. In addition to
under U.S. or applicable copyright law.

heavily cratered surfaces that presumably date from the time of heavy bombardment, Mars
has been resurfaced by much younger lava flows and shows abundant tectonic activity.
Olympus Mons, the largest volcano in the Solar System, sits astride the crustal dichotomy
between the northern lowlands and southern highlands, as does the large volcanic Tharsis
Rise that supports another three major volcanoes. The Tharsis volcanic system dominates
the volcanic and tectonic framework of the entire planet. The grand canyon of Mars, Valles
Marineris, is radial to the oval Tharsis uplift.

EBSCO Publishing : eBook Academic Collection (EBSCOhost) - printed on 11/3/2023 1:50 PM via YDUQS
AN: 399390 ; H. Jay Melosh.; Planetary Surface Processes
Account: s4526441
18 The grand tour
Copyright © 2011. Cambridge University Press. All rights reserved. May not be reproduced in any form without permission from the publisher, except fair uses permitted

Figure 1.8 The Galilean satellites of Jupiter as imaged by the Galileo spacecraft. In order from left to
right are Io, Europa, Ganymede, and Callisto. Volcanoes dominate Io, Europa is covered with an ice shell,
Ganymede’s surface is a patchwork of bright young and dark old terrain, and Callisto is an undifferentiated
mixture of ice and rock. NASA/JPL/DLR image PIA01400. See also color plate section.

The most intriguing aspect of Mars is evidence of fluvial activity on its surface. Ancient
valley networks appear to be fossil stream valleys, some of which may have been formed
by rain running off the surface. Enormous channels cutting thousands of kilometers across
the surface imply huge catastrophic floods in the distant past, while recent gullies may
indicate contemporary fluvial activity. Features suggestive of shorelines now seem to line
up at the same elevation, perhaps indicating an ancient global ocean, and glacial landforms
suggest that extensive icecaps once covered parts of the southern highlands.
Evidence for water now appears to be everywhere: It is frozen in the polar caps and in
permafrost beneath the high-latitude plains. Fluvially deposited sediments discovered by
the MER rovers suggest shallow lakes on the surface, although concretions in the same
sediments indicate highly acidic sulfur-rich waters. The puzzle in all of these observations
is that liquid water is not presently stable on the surface of Mars: Not only is its surface too
cold for liquid water, the present atmospheric pressure is below the triple point of water so
that liquid water is never stable, except perhaps on hot summer days at the bottom of the
Hellas Basin. If Mars did once possess oceans and icecaps, where did all that water go?
And if Mars was once warm and wet, how were those high temperatures maintained?
Mars today is a cold, dry planet whose surface is dominated by aeolian activity, but the
evidence for a very different climate in its past grows stronger with every new investiga-
tion. How this early climate was maintained and what happened to bring Mars to its current
inhospitable state are major questions that remain to be answered.
under U.S. or applicable copyright law.

1.3.5 Jupiter’s Galilean satellites

Although Jupiter possesses many satellites, the four large ones first discovered by Galileo
are the most interesting from a planetological perspective. Although satellites, they are
comparable in size to Mercury and can be treated as individual worlds (Figure 1.8). They
are also surprisingly dissimilar, although their peculiarities seem to be strongly correlated
with their distance from Jupiter. All of these satellites are tidally locked to Jupiter and they,

thus, rotate on their axes with the same period as they orbit about Jupiter. Their orbits are
Copyright © 2011. Cambridge University Press. All rights reserved. May not be reproduced in any form without permission from the publisher, except fair uses permitted

also regularly spaced and they interact strongly with each other.
Io, the innermost Galilean satellite, has a density suggestive of a silicate planet. It has a
small orbital eccentricity, which is maintained by gravitational interactions with Europa.
Because of this eccentricity its distance from Jupiter alternates with every orbit, flexing
Io by the alternating tidal stresses. This continual flexing dissipates large amounts of heat
in Io, whose surface heat flow is approximately 25 times larger than the Earth’s. Io is the
most volcanically active body in the Solar System. Its surface is mantled with volcanic-
ally extruded sulfur and sulfur dioxide, producing a wide range of lurid yellow and orange
colors dotted with occasional black pools of molten sulfur. No impact craters are known
anywhere on Io, whose surface is renewed at the rate of about 1 cm per year. Instead, vol-
canic calderas dot the surface, from which flows of sulfur-rich and silicate lavas proceed
over the surface while plumes of sulfur dioxide spray more than 100 km upward into space
before raining back onto the surface. The ultimate source of this volcanism appears to be
exceedingly hot silicate lavas that mobilize the more volatile sulfur compounds.
In addition to the volcanic calderas, Io also possesses about one hundred mountains, the
tallest of which towers 17 km above the surface. Sulfur is not strong enough to support such
high topography, so these mountain masses indicate the presence of stronger silicate rocks
in Io’s thin lithosphere.
Europa is also a mostly silicate body, but its surface is covered by water ice, not rock or
sulfur, although some albedo variations on the surface may be caused by sulfur implanted
from Jupiter’s magnetosphere. Europa has a few large impact craters, but its principal sur-
face features are ridges and long curving fracture systems of several types. Like Io, Europa
derives its internal heat mainly from tidal flexing, but it is almost twice as far from Jupiter
as Io and its tidal heating is much less intense.
In addition to the ubiquitous double ridges, which reach heights of a few hundred meters
and widths of a few kilometers, a substantial fraction of Europa’s surface is pocked with
irregular shallow depressions known as chaos regions, where the crust has apparently been
disrupted and broken into polygonal blocks that were once mobile. The crust of Europa is
believed to be mostly water ice that floats on a briny water ocean that may be 100 km deep.
The thickness of the ice shell is currently controversial: Estimates of its thickness range
from about 20 km to as little as a few kilometers. Topographic elevations or depressions
seldom exceed 1 km anywhere on Europa.
The small number of craters on its surface indicates an average age of about 60 Myr for
the surface, which must thus be continually renewed by internal activity. Europa, possess-
under U.S. or applicable copyright law.

ing a deep water ocean beneath a thin ice shell, has fueled speculation about its potential
for supporting life, which may be nurtured by warm hydrothermal plumes ejected from the
silicate floor of the subsurface ocean.
Ganymede is the next outer Galilean satellite. It circles Jupiter with a period twice that
of Europa. Its density of 1940 kg/m3 implies a body composed of about 60% rock and 40%
ice, which is substantially more ice-rich than Europa. Ganymede possesses two principal
terrain types: dark terrain that is heavily cratered and apparently very old, and light terrain

EBSCO Publishing : eBook Academic Collection (EBSCOhost) - printed on 11/3/2023 1:50 PM via YDUQS
AN: 399390 ; H. Jay Melosh.; Planetary Surface Processes
Account: s4526441
20 The grand tour

that is much younger. Ganymede appears to be differentiated and may also have a small
Copyright © 2011. Cambridge University Press. All rights reserved. May not be reproduced in any form without permission from the publisher, except fair uses permitted

metallic core, as it surprisingly possesses a small magnetic field of its own. It also shows
an induced magnetic field that suggests the presence of a thin liquid ocean buried some
170 km below the surface.
Craters on Ganymede show both fresh and viscously relaxed morphologies. A crater
type endogenous to Ganymede is the palimpsest crater, a relaxed circular form that may
either have been completely relaxed by viscous flow in relatively warm ice, or have formed
in a fluid, slushy surface. Central pit craters are also abundant on Ganymede and Callisto.
The younger bright regions are traversed by several generations of kilometer-wide furrows
and grooves that seem to be extensional in origin, because all of the recognizable faults
are extensional normal types. A substantial population of small craters overlies even the
bright terrain, so these surfaces are all older than the surface of Europa. Although evidence
for tectonic resurfacing is abundant on Ganymede, there is little sign of volcanic activity,
although some smooth low areas and circular pits called paterae may be of cryovolcanic
origin. Evidence for small-scale mass wasting and sublimation is abundant. Ganymede also
possesses polar caps that extend down to latitudes near 40° in both hemispheres. These
polar caps are evidence for the mobility of some water ice on the surface.
Callisto is the outermost Galilean satellite. Unlike the other three, Callisto does not
appear to have completely differentiated into a crust, mantle and core. Tidal heating is not
important and its surface seems to date from ancient times. Callisto is mainly notable for a
number of very large multiring basins in which a central bright patch containing the crater
and its ejecta blanket is surrounded by dozens of rings. Valhalla, 1800 km in diameter, is
the type example of this class. Callisto’s low density suggests that it is mainly composed
of ice, with only a small rocky component, although its dark surface seems to be largely
mantled by hydrated silicate dust.
Callisto apparently lacks tectonic features other than those associated with the multiring
basins. There is also no sign of volcanism. At high resolution, however, its surface shows
a bizarre alternation of bright, steep-sided peaks and mesas and dark, level plains that is
attributed to the dominance of sublimation and mass wasting of the mixture of silicate dust
and ice.

1.3.6 Titan
Titan, Saturn’s largest satellite, has rapidly become one of the most interesting objects in
the outer Solar System. It has long been known that, unique among moons, it possesses a
under U.S. or applicable copyright law.

dense atmosphere (Triton, the large satellite of Neptune, does have a thin N2 atmosphere).
Titan’s atmosphere is mostly composed of N2 with a small admixture of methane. The
pressure at its surface, however, is about 1.6 times larger than the Earth’s. Titan’s upper
atmosphere is filled with organic aerosols that obscure the surface almost completely at
visible wavelengths, although infrared wavelengths do penetrate to the surface and reveal
broad surface details (Figure 1.9).

EBSCO Publishing : eBook Academic Collection (EBSCOhost) - printed on 11/3/2023 1:50 PM via YDUQS
AN: 399390 ; H. Jay Melosh.; Planetary Surface Processes
Account: s4526441
1.3 Planetary surfaces and history 21
Copyright © 2011. Cambridge University Press. All rights reserved. May not be reproduced in any form without permission from the publisher, except fair uses permitted

Figure 1.9 Global view of Titan’s surface from the VIMS spectrometer aboard the Cassini spacecraft.
This false-color composite is constructed from three wavelengths in the infrared that penetrate Titan’s
hazy atmosphere (1.3 μm is shown in blue, 2 in green and 5 in red). The dark region in the center of
the image is named Xanadu and may be the site of a large ancient impact. NASA/JPL/University of
Arizona image PIA09034. See also color plate section.

Titan’s density indicates a large admixture of ice in its composition and the surface is
mainly composed of water ice, which at Titan’s surface temperature of 94 K is as hard and
strong as silicate rock on the surface of the terrestrial planets. Methane can exist as either a
gas in the atmosphere or as a liquid on the surface, so Titan possesses a “hydrologic” cycle
based on methane. It is apparent that methane rain occasionally beats down on its surface.
Broad, shallow methane lakes accumulate in low areas near its poles. Radar images from
the Cassini spacecraft show an eerily Earth-like landscape with lakes, river valleys, and
mountain ranges. Images from the Huygens lander could easily be mistaken for a stream
bed on the Earth.
Titan does have a few large impact craters, but small craters are nearly absent: Like the
Earth, erosional processes on Titan are active enough to remove most craters. Titan pos-
sesses an extensive equatorial belt of “sand” dunes that resemble the Libyan sand seas of
under U.S. or applicable copyright law.

the Sahara.
There are many puzzles concerning Titan, such as what kind of weathering processes
act to disintegrate its frigid ice bedrock and how active are the methane rivers that have
carved out deep valleys into its crust. Titan provides many fine examples of how the same
processes can act on vastly different materials to produce similar landforms. The abun-
dant organic chemistry acting in its atmosphere, lakes, and below its surface also offers an

EBSCO Publishing : eBook Academic Collection (EBSCOhost) - printed on 11/3/2023 1:50 PM via YDUQS
AN: 399390 ; H. Jay Melosh.; Planetary Surface Processes
Account: s4526441
22 The grand tour
Copyright © 2011. Cambridge University Press. All rights reserved. May not be reproduced in any form without permission from the publisher, except fair uses permitted

Figure 1.10 Topographic map of the Earth from NOAA. ETOPO1 is a 1 arc-minute global relief
model of Earth’s surface that integrates land topography and ocean bathymetry. It was built from
numerous global and regional data sets (Amante and Eakins, 2009). See also color plate section.

attractive laboratory for the study of whether some kind of carbon-based life could have
gotten a start in an environment so different from our familiar Earth.

1.3.7 The Earth

The Earth presents the opposite extreme to the Moon and rounds out our brief tour of the
planets (home is always the last stop on any long journey). It is the largest terrestrial planet.
Probably as a result of its size, the Earth cannot rid itself of heat generated by the decay of
long-lived radioactive elements (chiefly U and Th at the present time) by means of conduc-
tion alone. Only the convective transfer of heat by currents deep in the Earth’s interior can
cool the planet effectively. The surface of the Earth is profoundly affected by this internal
convection and a slow (1–10 cm/yr) relative motion of its large surface structures is the
result. Thus, the Earth’s surface structure is largely generated by internal tectonics.
The process of plate tectonics determines the large-scale structures of the Earth
(Figure 1.10). The surface of the Earth is divided into about a dozen large, semi-rigid plates
that slowly glide over hotter, more fluid rock at depths of about 100 km. New basaltic crust
under U.S. or applicable copyright law.

is created at mid-ocean spreading centers. This crust is only 5 to 10 km thick, compared to

the continental rafts that range from 30 to 70 km thick. The dense basaltic crust stands low,
an average of about 4 km below the level of the continents that are composed of less dense
silica-rich rocks. Earth’s free surface water fills these low basins and, thus, exposes a large
area of mostly dry land to the atmosphere above.
The oceanic crust is almost neutrally buoyant, so it is easily created and later reabsorbed
in subduction zones where it sinks back into the mantle. In the process of being recycled,

however, it partially melts and differentiates, creating long lines of volcanoes that rise over
Copyright © 2011. Cambridge University Press. All rights reserved. May not be reproduced in any form without permission from the publisher, except fair uses permitted

the subducting crust. The continental crust is too buoyant to subduct to great depths. It thus
remains near the surface and has accumulated a history nearly as long as the Earth’s own
existence, although it is continually fragmented and the pieces rejoined to create a patch-
work of amazing and often baffling complexity.
Linear mountain ranges rise where continental fragments collide, while erosion quickly
reduces their elevations (at rates averaging about 10 cm/1000 yr) and distributes their mater-
ial to lower elevations. Geologists divide continental rafts into young mountain ranges at
the sites of recent slow collisions (tens to 100 Myr), platforms of middle age (less then
about 1 Gyr), and ancient shields (more than 1 Gyr) that have not suffered fragmentation or
collision during this long interval. Old continents that have escaped breakup may, in fact,
possess cold and strong roots that resist fragmentation and thus preserve such undisturbed
regions intact for long intervals.
This continental process of erosion, deposition and re-collision to raise new mountains
led early geological theorists like James Hutton (1726–1797) to posit a universal rock cycle
in which the rocks of the Earth are eroded, deposited as sediments, metamorphosed after
burial, melted to form igneous rocks, uplifted, then eroded again to continue the cycle. This
rock cycle has been one of the foundations of terrestrial geology, but it now appears to be a
uniquely terrestrial process, one that requires both plate tectonics and active fluvial erosion.
There is little hint of such a cycle on any other planet in the Solar System (although Io may
undergo a weird variant of crustal recycling).
Free water is present on the Earth’s surface under conditions where it can change phase
freely from solid to liquid to gas. This creates a hydrologic cycle that can readily transport
solid silicate debris from place to place. Even more damaging to the integrity of surface
rock, life has produced an atmosphere of which 20% consists of a toxic and corrosive gas –
oxygen (even living systems go to great lengths in their internal biochemistry to avoid
being damaged by this reactive substance). Oxygen, in conjunction with water, can corrode
and disintegrate nearly every silicate, except quartz, which is otherwise stable in the crust,
leading to rapid degradation (weathering) of rocks that are brought to the surface. Thus,
on Earth, probably more so than on any other planet, crustal structures are weathered and
eroded by atmospheric processes, so further affecting the surface morphology and engen-
dering relatively rapid changes of the landscape.
It is, thus, no surprise that the Earth’s crust contains essentially no trace of the early meteor-
itic bombardment to which the Moon was subjected. Indeed, 20 km craters only a few tens of
millions of years old are obliterated almost beyond recognition unless special circumstances
under U.S. or applicable copyright law.

preserve them. The giant impact basins that form the most ancient structures on most of the
other planets have vanished from the Earth without leaving any presently recognized record.
The study of the Earth’s surface therefore requires a wholly different orientation from the
study of the Moon’s surface. On Earth, geologists are primarily concerned with internally
generated tectonics and its modification by weathering and erosion. The effect of external
agencies is so slight that for many years geologists refused to accept the opposite situation
on the Moon and other planets.

EBSCO Publishing : eBook Academic Collection (EBSCOhost) - printed on 11/3/2023 1:50 PM via YDUQS
AN: 399390 ; H. Jay Melosh.; Planetary Surface Processes
Account: s4526441
24 The grand tour

Because the surface of the Earth is treated at great length in other books on geology and
Copyright © 2011. Cambridge University Press. All rights reserved. May not be reproduced in any form without permission from the publisher, except fair uses permitted

geomorphology, in this book we emphasize those processes which are either universal, act-
ing on all the planets including the Earth, or those that hardly affect the Earth at all, such
as impact cratering. Compared with other texts this one says little about fluvial processes
or even plate tectonics. On the other hand, as Carl Sagan said, the Earth is not the center of
the Universe and its geology is not that of our neighboring planets.

Further reading
Readers wishing to learn more about orbital mechanics and its application to the Solar
System should consult the book by Bate et al. (1971), which is my personal favorite for
learning orbital mechanics in spite of its use of English units, or the comprehensive treatise
by Murray and Dermott (1999). The best general overview of the planetary system is still
the excellent collection of topical articles in Beatty et al. (1999). The subject of planet-
ary atmospheres has lacked an up-to-date text for many years. However, a recent, well-
reviewed book on this subject has just appeared (Sanchez-Levega, 2010).
The individual planets of the Solar System are well treated in the many volumes of the
University of Arizona’s Space Science Series. The series includes a book about Mercury,
but it is badly out of date. Instead, I refer you to Robert Strom’s summary article of what
was learned from the Mariner 10 mission (Strom, 1984) or his popular but authoritative
book (Strom and Sprague, 2003). The MESSENGER mission will shortly make anything
now in print out of date. For Venus, see the collection of papers in Bougher et al. (1997).
There are many geology books that describe the Earth and its tectonics. The latest edi-
tion of an old, reliable standard that has gone through many editions is Grotzinger et al.
(2006a). A personal favorite of mine (I often recommend this as an introduction for non-
geologists) is the visually oriented book Geology Illustrated (Shelton, 1966). Michael Carr
has been writing the best summaries of Martian geology for years. His latest effort is Carr
(2006). Asteroidal bodies are treated well in Bottke et al. (2002). The Jovian system after
the Galileo mission is well summarized in Bagenal et al. (2004). Titan is the subject of
a book that just appeared (Brown et al., 2009), although Cassini’s investigation of the
Saturnian system is ongoing. In the outer Solar System, Neptune (Cruikshank, 1995) and
Uranus (Bergstralh et al., 1991) have not been visited since Voyager and so there is little
change to the summary articles on each in the books cited. Pluto (Stern and Tholen, 1997)
will still be something of an enigma until the New Horizons spacecraft visits that system in
2015. Trans-Neptunian objects have been recently reviewed (Cruikshank and Morbidelli,
under U.S. or applicable copyright law.

2008), as have comets (Festou et al., 2004).

EBSCO Publishing : eBook Academic Collection (EBSCOhost) - printed on 11/3/2023 1:50 PM via YDUQS
AN: 399390 ; H. Jay Melosh.; Planetary Surface Processes
Account: s4526441
2
Copyright © 2011. Cambridge University Press. All rights reserved. May not be reproduced in any form without permission from the publisher, except fair uses permitted

The shapes of planets and moons

The equal gravitation of the parts on all sides would give a spherical fig-
ure to the planets, if it was not for their diurnal revolution in a circle ….
I. Newton, Principia, Theorem XVI

Modern space exploration has made everyone familiar with the idea that planets are mostly
spherical. From a great distance a casual observer might not even notice that rotating plan-
ets and moons are not quite perfect spheres. However, careful examination reveals depar-
tures from perfection. Rotating planets are slightly oblate spheres, while tidally locked
satellites are triaxial. Furthermore, once these bodies are approached closely, it becomes
clear that nearly every planet and moon possesses topographic variations. Mountains, val-
leys, plains, and craters create landscapes that, up close, can challenge attempts to traverse
them by mechanical rovers or human explorers.
The forces that create and maintain the topography of planetary bodies depend on the
scale of the feature. The gravitational self-attraction that tends to make planets spherical
operates differently on the scale of individual mountains. It is thus useful to distinguish
several orders of relief that categorize different scales of topographic feature. This notion
can be made mathematically precise through the use of spherical harmonics, a concept that
will be discussed later in this chapter.
The tendency of large masses of material to take on a spherical shape was first recog-
nized by Isaac Newton (1643–1727) in 1686. His brilliant insight into universal gravitation
showed that, in the absence of other forces, the attraction of matter for other matter tends
to mould all bodies into spheres. Gravity is weak compared to other forces so, on a human
scale, bodies must be very large for gravity to dominate the electromagnetic forces that
give atomic matter its strength to resist deformation. The conflict between strength and
gravitation is the subject of Chapter 3. This chapter concentrates on the largest-scale fea-
under U.S. or applicable copyright law.

tures of planetary topography and its geometric properties.

EBSCO Publishing : eBook Academic Collection (EBSCOhost) - printed on 11/3/2023 1:50 PM via YDUQS
AN: 399390 ; H. Jay Melosh.; Planetary Surface Processes
Account: s4526441
26 The shapes of planets and moons

Table 2.1 Physical characteristics of large bodies in the Solar System

Mean Equatorial Sidereal

Equatorial density acceleration of rotation
Name radius (km) Mass (kg) (kg/m3) gravity (m/s2) period

Mercury 2 439 3.303 x 1023 5430 2.78 58.75 days

Venus 6 051 4.870 x 1024 5250 8.60 243.01 days
Earth 6 378 5.976 x 1024 5520 9.78 1.00 days
Moon 1 738 7.349 x 1022 3340 1.62 27.322 days
Mars 3 393 6.421 x 1023 3950 3.72 1.029 days
Ceres 424 8.6 x 1020 1980 0.32 9.08 hours
Vesta 234 3.0 x 1020 3900 0.37 5.34 hours
Jupiter 71 492a 1.900 x 1027 1330 22.88 9.925 hoursb
Io 1 815 8.94 x 1022 3570 1.81 1.769 days
Europa 1 569 4.80 x 1022 2970 1.30 3.551 days
Ganymede 2 631 1.48 x 1023 1940 1.43 7.155 days
Callisto 2 400 1.08 x 1023 1860 1.25 16.689 days
Saturn 60 268a 5.688 x 1026 690 9.05 10.675 hoursb
Titan 2 575 1.35 x 1023 1880 1.36 15.945 days
Uranus 25 559a 8.684 x 1025 1290 7.77 17.240 hoursb
Neptune 24 764a 1.02 x 1026 1640 11.0 16.11 hoursb
Triton 1 350 2.14 x 1022 2070 0.78 5.877 days
Pluto 1 150 1.29 x 1022 2030 0.658 6.3872 days
Charon 604 1.52 x 1021 1650 0.278 6.3872 days

a
Measured at the 1-bar pressure level.
b
Internal rotation period derived from the magnetic field.

2.1 The overall shapes of planets

2.1.1 Non-rotating planets: spheres

The lowest order of relief, which in this book we call the “zeroth order” because it cor-
responds to the zeroth-order spherical harmonic, is the overall diameter of the body. The
shape of a non-rotating, self-gravitating mass of material that has no intrinsic strength is
a perfect sphere. We can describe such an object by its radius or diameter, and with this
under U.S. or applicable copyright law.

single datum our description of its shape is complete. Of course, the smaller bodies in
the Solar System, such as asteroids or small satellites, may depart considerably from a
perfect spherical shape, but it is still useful to describe them by the radius of a sphere of
equal volume. Most tabulations of the physical properties of the planets and moons give
their diameter, along with their mass (see Table 2.1). From the mass and diameter other
physically interesting parameters, such as density or the surface acceleration of gravity,
can be computed.

The volume V of a sphere of radius r is, as everyone learns in high school, V = 3 π r .

4 3
Copyright © 2011. Cambridge University Press. All rights reserved. May not be reproduced in any form without permission from the publisher, except fair uses permitted

Fewer readers may remember that the volume of a triaxial ellipsoid with semiaxes a, b and
c is given by the similar formula V = 4 π abc. Thus, the mean radius (that is, the radius of
3
a sphere of equal volume) of a triaxial body is given by rmean = 3 abc . In many cases a, b,
and c differ only slightly from the mean. When this happens we can use the approximate
formula rmean ≈ (a + b + c)/3. These simple formulas will prove useful in interpreting the
relationships between various radii commonly encountered in tabulations.

2.1.2 Rotating planets: oblate spheroids

Newton himself recognized the first level of deviation from a perfectly spherical shape.
Rotating bodies are not spherical: They are oblate spheres with larger equatorial radii
than polar radii. In the language of spherical harmonics this is the second order of relief.
Historically, the Earth’s oblateness was not at all obvious to Newton’s contemporaries and
sparked a debate between Newton and contemporary astronomer Jacques Cassini (1677–
1756) that resulted in one of the first major scientific expeditions, a project to precisely
measure and compare the length of a degree of latitude in both France and Lapland. The
result, announced in 1738, was the first direct evidence that the Earth is shaped like an
oblate spheroid, whose equatorial radius is about 22 km longer than its radius along its axis
of rotation. This apparently tiny deviation, of only 22 km out of 6371 km, is expressed by
the flattening of the Earth, f, defined as:

a−c
f = (2.1)
a

where a is the equatorial radius of the Earth and c is its polar radius (Figure 2.1). The cur-
rently accepted value for the Earth’s flattening, 1/298.257, is substantially different from
Newton’s own theoretical estimate of 1/230.
Newton derived his estimate of Earth’s oblateness from a theory that assumed that the
Earth’s density is uniform. Under this assumption he was able to show that the flattening
is proportional to a factor m, the ratio between the centrifugal acceleration at the equator
(a consequence of rotation) and the gravitational acceleration at the equator. This ratio
expresses the tendency of a planet to remain spherical: Smaller m implies that rotation is
less important and the planet is more spherical.

ω 2 a3 3 ω2
m= ≅
under U.S. or applicable copyright law.

. (2.2)
GM 4π G ρ

In this equation ω is the rotation rate of the Earth (in radians per second), G is Newton’s
gravitational constant, 6.672 x 10–11 m3/kg-s2, M is the mass of the Earth and ρ is its mean
density. Even in Newton’s time it was known that m ≈ 1/290. Newton used a clever argu-
ment involving hypothetical water-filled wells drilled from both the pole and equator that
join at the center of the Earth. Supposing that the wellheads are connected by a level canal

EBSCO Publishing : eBook Academic Collection (EBSCOhost) - printed on 11/3/2023 1:50 PM via YDUQS
AN: 399390 ; H. Jay Melosh.; Planetary Surface Processes
Account: s4526441
28 The shapes of planets and moons
Copyright © 2011. Cambridge University Press. All rights reserved. May not be reproduced in any form without permission from the publisher, except fair uses permitted

Figure 2.1 Oblate spherical shape of a rotating fluid planet with no strength. The equatorial radius is
a, the polar radius is c, and the rotation rate is ω.

at the surface, he used the impossibility of perpetual motion to argue that the pressure at the
bottom of the water columns had to be equal at the Earth’s center, from which he derived
the expression:

5
f = m. (2.3)
4

Although not quite correct for the Earth, this formula gives an excellent first approxima-
tion to the flattening.
Implicit in Newton’s derivation is the idea that water tends to assume a level surface.
That is, the surface that water naturally attains coincides with a surface on which the gravi-
tational potential energy is constant. If the surface of a body of water, or any other strength-
under U.S. or applicable copyright law.

less fluid, did not follow a constant gravitational potential, it could gain energy by flowing
downhill. Thus, in the absence of currents or other imposed pressure gradients, the surface
of a fluid must coincide with an equipotential. This is sometimes called an equilibrium
surface. This proposition holds equally well for planetary atmospheres, which also tend
to have equal pressures on equipotential surfaces. It is important to note that on a rotat-
ing planet the equilibrium surface is not a sphere but an oblate spheroid (on very rapidly

rotating bodies this surface becomes still more complex, but such surfaces do not seem to
Copyright © 2011. Cambridge University Press. All rights reserved. May not be reproduced in any form without permission from the publisher, except fair uses permitted

be important on any currently known planets). In real planets, lateral variations in density
lead to further distortions of the equilibrium surfaces. An arbitrarily chosen equilibrium
surface, called the geoid, forms the primary reference from which topographic elevations
are measured on Earth. Deviations of a planet’s surface from an equilibrium surface must
be supported by strength.
Several hundred years of further effort by mathematical physicists were needed to extend
Newton’s simple flattening estimate to a more comprehensive form. In 1959 Sir Harold
Jeffreys established a formula for the flattening of a rotating, strengthless body in hydro-
static equilibrium (for which contours of constant density coincide with equipotentials at
all depths within the body). This more complicated formula is:
5
m
f = 2 (2.4)
2
25  3  C  
1 + 1 − 
4  2  Ma 2  

where C is the planet’s moment of inertia about the polar axis. The moment of inertia is
defined as the integral:
M
C = ∫ r 2 dm (2.5)
o
where r is the radial distance of the infinitesimal mass element dm from the axis about
which C is computed. In the case of the polar moment of inertia C this is the rotation axis.
The dimensionless moment of inertia ratio, C/Ma2, expresses the concentration of
mass toward the center of the planet. This ratio is equal to zero for a point mass, equals
2/5 (= 0.4) for a uniform density sphere and is measured to be 0.33078 for the Earth.
Earth’s moment of inertia ratio is less than 0.4 because mass is concentrated in its dense
nickel–iron core. In the uniform density case, C/Ma2= 2/5, and Equation (2.4) reduces
exactly to Newton’s estimate.
Newton’s contemporaries also noted the rather large flattening of the rapidly rotating
planets Jupiter and Saturn. Modern measurements of the flattening of the planets in our
Solar System are listed in Table 2.2.
Mars is a special case. Its observed shape flattening, fMars = 1/154 is considerably larger
than that estimated from Jeffreys’ formula above, which gives f = 1/198. This is because
under U.S. or applicable copyright law.

Mars is far from hydrostatic equilibrium, and so violates the assumptions of Jeffreys’ der-
ivation. The Tharsis Rise volcanic complex is so large and so massive that it dominates the
gravitational field of the planet and warps its shape well out of hydrostatic equilibrium. Its
equatorial radius varies by almost 5 km, depending on longitude. Geophysical models of
Mars have yet to fully separate the effects of Tharsis from the radial concentration of mass
towards its core. The shapes of small bodies in the Solar System – comets, asteroids and

EBSCO Publishing : eBook Academic Collection (EBSCOhost) - printed on 11/3/2023 1:50 PM via YDUQS
AN: 399390 ; H. Jay Melosh.; Planetary Surface Processes
Account: s4526441
30 The shapes of planets and moons

Table 2.2 Deviations of Solar System bodies from spheres

Moment of Center of mass –

inertia factor, Topographic Dynamical center of figure
Name C/MR2 flattening flattening offset (km)

Mercury 0.33 ~1/1800. 1/1.03 x 106 ?

Venus 0.33 0 1/1.66 x 107 0.280
Earth 0.33078 1/298.257 1/301 2.100
Moon 0.394 1/801.6 1/1.08 x 105 1.982
Mars 0.366 1/154. 1/198. 2.501
Jupiter 0.254 1/15.42a 1/15.2 –
Saturn 0.210 1/10.2a 1/10.2 –
Uranus 0.225 1/43.6a 1/50.7 –
Neptune 0.24 1/58.5a 1/54.4 –

Data from Yoder (1995).

a
At 1-bar pressure level in the atmosphere.

small moons – do not obey Jeffreys’ formula for similar reasons; their inherent strength
produces large departures from hydrostatic equilibrium.
Mercury is another interesting planet from the flattening perspective. Although we do
not know its flattening very accurately (not to better than 10%), it seems to be very small,
about 1/1800. This is consistent with its current slow rotation period of about 57 days.
However, Mercury is so close to the Sun that it is strongly affected by solar tides. It may
originally have had a much faster rotation rate that has since declined, placing the planet
in its current 3/2 spin-orbital resonance with the Sun (that is, Mercury rotates three times
around its axis for every two 88-day trips around the Sun). If this is correct, then Mercury’s
flattening may have changed substantially over the history of the Solar System. Chapter 4
will discuss the tectonic consequences of this global shape change.

2.1.3 Tidally deformed bodies: triaxial ellipsoids

After rotation, the next degree of complication in planetary shapes includes the effects of
tidal forces. Tides are the result of the variation of the gravitational potential of a primary
attractor across the body of an orbiting satellite, along with the centripetal potential due
under U.S. or applicable copyright law.

to its orbit about the primary. The consequence of these varying potentials on a mostly
spherical satellite is that it becomes elongated along the line connecting the centers of the
primary and satellite, compressed along its polar axis, and compressed by an intermediate
amount along the axis tangent to its orbit (Figure 2.2).
If the satellite (and this includes the planets themselves, which are satellites of the Sun)
spins at a rate different than its orbital period, then the elongation of the equipotential
surface in a frame of reference rotating with the satellite varies with time. A point on the

EBSCO Publishing : eBook Academic Collection (EBSCOhost) - printed on 11/3/2023 1:50 PM via YDUQS
AN: 399390 ; H. Jay Melosh.; Planetary Surface Processes
Account: s4526441
2.1 The overall shapes of planets 31
Copyright © 2011. Cambridge University Press. All rights reserved. May not be reproduced in any form without permission from the publisher, except fair uses permitted

R a
b
Mearth

Mmoon

Figure 2.2 Tidal deformation of a synchronously rotating satellite such as our Moon, orbiting about
its primary at a distance R. The tidal forces stretch the satellite along the line connecting its center to
that of its primary such that the radius a along this line is larger than any of its other principal radii.
The smallest radius c is perpendicular to the orbital plane and the intermediate radius b is parallel to
the orbit’s tangent.

equator is alternately lifted and dropped as it rotates from the line between its center and
that of the primary, to the line tangent to the orbit. These periodic equipotential changes,
which on Earth we recognize as the force responsible for oceanic tides, create motions that
dissipate rotational energy and, for strong enough tides or long enough times, may even-
tually slow the satellite’s rotation. The Earth is subject to both solar and lunar tides, which
have gradually lengthened our day from 18 hours, 1.3 billion years ago, to its present 24
hours. The Moon, being less massive than the Earth, long ago lost any excess rotation and
is now synchronously locked to the Earth, rotating once on its axis for each orbit around the
Earth. Many other satellites in the Solar System are similarly synchronously locked to their
primaries, including the four large Galilean satellites of Jupiter and Saturn’s large satellite
Titan. Mercury itself is an exception. Although astronomers long believed that Mercury is
synchronously locked to the Sun, we now understand that its highly elliptical orbit led it to
be trapped in the present 3/2 spin-orbit resonance.
Tidal forces produce a characteristic pattern of deformation on synchronously rotating
satellites. Harold Jeffreys, in his famous book The Earth (Jeffreys, 1952), showed that the
equipotential surface of a tidally locked body is a triaxial ellipsoid with three unequal axes
a>b>c. The lengths of these axes are:
 35 
a = rmoon  1 + Ω
 12 
under U.S. or applicable copyright law.

 10 
b = rmoon  1 − Ω
 12 

 25  (2.6)
c = rmoon  1 − Ω
 12 

EBSCO Publishing : eBook Academic Collection (EBSCOhost) - printed on 11/3/2023 1:50 PM via YDUQS
AN: 399390 ; H. Jay Melosh.; Planetary Surface Processes
Account: s4526441
32 The shapes of planets and moons

where rmoon is the mean radius of the Moon. It is clear that the average of these three axial
Copyright © 2011. Cambridge University Press. All rights reserved. May not be reproduced in any form without permission from the publisher, except fair uses permitted

distances just equals the mean radius. These distortions depend on the dimensionless factor
Ω, given by:
3
M earth rmoon
Ω= (2.7)
M moon R 3

where Mearth and Mmoon are the masses of the Earth and Moon, respectively, and R is the dis-
tance between the Earth and the Moon.
Naturally, for satellites other than the Moon orbiting about some other primary, the analo-
gous quantities must be inserted in Ω. The distortion is larger for a larger ratio between
primary mass and satellite mass and smaller for increased distance between the two bodies.
Jeffreys’ equations above are strictly valid only for a Moon of uniform density. The book
by Murray and Dermott (1999) describes how to treat the general case where the Moon’s
density is not uniform. Table 2.3 lists the ratios between the different axes of the satellites
in the Solar System for which they are known.
According to the formulas (2.6) and (2.7), the maximum difference between the axes, a –
c, on Earth’s Moon is presently about 66 m. The Moon is 10 to 20 times more distorted than
this, a fact that was known even in Harold Jeffreys’ day. The current best estimates indicate
that the Moon is at least 10 times more distorted than the hydrostatic prediction. This obser-
vation prompted Jeffreys to propose that the present figure of the Moon is the fossil remnant
of a formerly much larger distortion. The Moon is presently receding from the Earth at the
rate of about 3.8 cm/year. This recession rate was surely higher in the past when the Moon
was closer to the Earth and tides were, therefore, higher (the full story of the evolution of the
Moon’s orbit is a complicated one, involving changes in tidal dissipation over the age of the
Earth as continents and seas shifted). Nevertheless, it is clear that the Moon was once much
closer to the Earth than it is now. Because the lengths of the axes depend on the Earth–Moon
distance to the inverse cube power, Jeffreys postulated that the present figure could have
been frozen-in at a time when the Moon was about 1/2.7 times its present distance from the
Earth. At this distance the Moon would have circled the Earth in only 6.1 days and, by angu-
lar momentum conservation, a day on Earth would have lasted only 8.2 hours.
The major problem with Jeffreys’ proposal (which he recognized himself) is that the
hydrostatic formulas (2.6) make a definite prediction that that the ratio ( a -c )/( b – c ) = 4,
independent of the Earth–Moon distance. The latest value of this ratio from Japan’s Kaguya
mission (Araki et al., 2009) is 1.21, far from the hydrostatic value of 4. The general opin-
ion at the present time is that, although the Moon clearly departs from a hydrostatic shape,
under U.S. or applicable copyright law.

the present shape is more a consequence of geologic forces that sculpted the lunar surface
rather than the remnant of a former hydrostatic figure.
In examining Table 2.3 the ratio of axes differences in the last column is more often quite
different from the theoretical value of 4 than close to it. The larger satellites approach most
closely to this ideal ratio, while the smaller ones are clearly dominated by strength rather
than gravitational forces.

Table 2.3 Triaxial shapes of small satellites

Density Mean radius (a–c) (a–c)/

Name (kg/m3) (km) (km) (b–c)

Earth’s Satellite
Moona 3340 1737 0.713 1.21
Martian Satellites
Phobosb 1900 11.1 4.0 2.2
Deimosb 1760 6.3 2.1 2.6
Jovian Satellites
Metisc 3000 21.7 13.0 4.3
Adrasteac 3000 8.2 3.0 3.0
Almatheac 862 83.6 61.0 6.8
Thebec 3000 49.2 16.0 2.3
Iod 3528 1818.1 14.3 4.1
Europad 3014 1560.7 3.0 3.8
Ganymeded 1942 2634.1 1.8 4.5
Callistod 1834 2408.3 0.2 2.0
Saturnian Satellites
Mimase 1150 198.1 16.8 2.7
Enceladuse 1608 252.1 8.3 2.7
Tethyse 973 533.0 12.9 3.6
Dionee 1476 561.7 3.5 5.0
Rheae 1233 764.3 4.1 -6.8
Titanf 1881 2574.5 0.410 3.8
Iapetuse 1083 723.9 35.0 n/a
Uranian Satellites
Mirandag 1200 235.7 7.1 5.5
Arielg 1670 578.9 3.4 17.0

Data from:
a
Araki et al. (2009)
b
Thomas (1989)
c
Thomas et al. (1998)
d
Davies et al. (1998)
e
Thomas et al. (2007)
f
Iess et al. (2010)
under U.S. or applicable copyright law.

g
Thomas (1988)

Most of the other moons in the Solar System depart substantially from a hydrostatic
shape. However, it appears that the shape of Jupiter’s large, tidally heated moons Io and
Europa may be close to equilibrium ellipsoids. If that is the case, then the maximum dis-
tortion, a–c, should be 14.3 km for Io and 3.9 km for Europa. The Galileo spacecraft

EBSCO Publishing : eBook Academic Collection (EBSCOhost) - printed on 11/3/2023 1:50 PM via YDUQS
AN: 399390 ; H. Jay Melosh.; Planetary Surface Processes
Account: s4526441
34 The shapes of planets and moons

confirmed these expectations with an accuracy of 0.12 and 0.65 km, respectively (Davies
Copyright © 2011. Cambridge University Press. All rights reserved. May not be reproduced in any form without permission from the publisher, except fair uses permitted

et al., 1998).

2.1.4 A scaling law for planetary figures?

William Kaula (1926–2000), who contributed extensively to understanding the figure of
the Earth from satellite measurements, proposed a scaling law that seems to approximately
give the deviations of planetary figures from a hydrostatic shape. The law is based on the
idea that all planets have about the same intrinsic strength, and that stress differences are
proportional to the surface acceleration of gravity, g (Kaula, 1968, p. 418). In the absence
of any other information about a planet, this law may give a useful first estimate of how far
a “normal” planet’s figure departs from the hydrostatic shape.
This departure is given by Kaula’s first law:

C−A 1 (2.8)
2
∝ 2
Ma non-hydrostatic g

where C and A are the moments of inertia about the shortest and longest axes of the
figure, respectively. The difference in moments of inertia is proportional to the normal-
ized difference of the lengths of the axes, (a-c)/a, so that this rule also implies that the
deviations of this ratio from its hydrostatic value should depend on the inverse square
of the gravitational acceleration. For constant density, g is proportional to the planetary
radius, so this ratio should equally depend on the inverse square of the planetary radius.
Looking ahead to Chapter 3, we will see in Section 3.3.3 and Figure 3.5 that this relation
does seem to hold approximately in our Solar System for the larger bodies, but it fails
badly for small objects for which strength is controlled by frictional forces that depend
on pressure.
The possibility that planets may have substantial non-hydrostatic contributions to their
figures plays an important role in studies of rotational dynamics and the tidal evolution of
bodies in the Solar System. For example, the present orientation of the Moon with respect
to the Earth may be partly due to the distribution of dense lavas in the low-lying basins on
the nearside. The orientation of Mercury may be controlled by mass anomalies associated
with the Caloris Basin. And if Europa has too large a non-hydrostatic figure, then its puta-
tive slow non-synchronous rotation cannot occur.
under U.S. or applicable copyright law.

2.1.5 Center of mass to center of figure offsets

One of the widely publicized results of the Apollo missions to the Moon was the discovery
that the Moon’s center of mass is about 2 km closer to the Earth than its center of figure. For
many years this offset was known only in the Moon’s equatorial plane, as all of the Apollo
flights circled the Moon’s equator. Now, as a result of the unmanned Clementine mission,
we know more precisely that the offset is 1.982 km.

Although, in hindsight, it is not really surprising that such an offset should exist,
Copyright © 2011. Cambridge University Press. All rights reserved. May not be reproduced in any form without permission from the publisher, except fair uses permitted

the possibility that the center of figure of a planet might not correspond with its cen-
ter of mass was never considered in classical geodesy. All harmonic expansions of the
gravity field of the Earth are made about its center of mass and so the reference geoid
and all other equipotential surfaces are also centered about this point. In fact, the Earth
itself has a substantial center of mass – center of figure offset if the water filling the
ocean basins is neglected. The floor of the Pacific Ocean is about 5 km below sea level,
whereas the opposite hemisphere is dominated by the continental landmasses of Asia,
Africa, and the Americas. The waterless Earth’s center of figure is thus offset from its
center of mass by about 2.5 km at the present time. Of course, as the continents drift
around over geologic time this offset gradually changes in both direction and magni-
tude. The fundamental reason for the offset is the difference in density and thickness
between oceanic crust and continental crust. Table 2.2 lists these offsets for the bodies
where they are known.
For reasons that are still not understood, most of the terrestrial planets show striking
asymmetries on a hemispheric scale. The nearside of the Moon looks quite different from
the farside, and lies at a lower average elevation with respect to its center of mass. It is
generally believed that this is due to a thicker crust on the farside, although what caused
the thickness variation is unknown. Mars also possesses a strong hemispheric asymmetry.
The northern plains of Mars lie an average of 5 km lower than the southern highlands. Here
again the immediate cause may be a difference in crustal thickness or composition, but the
ultimate reason for the difference is presently unknown.

2.1.6 Tumbling moons and planets

Most rotating bodies in the Solar System spin about an axis that coincides with their max-
imum moment of inertia, the C axis in our terminology. The moment of inertia is actually a
second-rank tensor, written Iij, which can be defined for any solid body as a generalization
of the definition (2.5) for C:
Iij = ∫ rirj dm (2.9)
where the subscripts i and j run from 1 to 3 and denote, respectively, the x, y, and z axes of
a Cartesian coordinate system. The symbol ri for i = 1, 2, and 3, thus, denotes the x, y, and
z coordinates of a mass element dm with respect to the origin around which the moment of
inertia is computed, generally taken to be the center of mass of a body.
under U.S. or applicable copyright law.

A fundamental theorem of tensor mathematics states that a suitable rotation of the coord-
inate axes can always be found in which the tensor (2.9) is diagonal (that is Iij= 0 unless
i = j) about three perpendicular axes. The moments of inertia about these special axes are
called the principal moments of inertia and are labeled C, B, and A for the maximum, inter-
mediate, and minimum principal moments, C ≥ B ≥ A. The lengths of the corresponding
principal axes are conventionally written in lower case and, perhaps confusingly, the c axis
is usually the shortest of the three: c ≤ b ≤ a. The reason for this inverse order is that the

EBSCO Publishing : eBook Academic Collection (EBSCOhost) - printed on 11/3/2023 1:50 PM via YDUQS
AN: 399390 ; H. Jay Melosh.; Planetary Surface Processes
Account: s4526441
36 The shapes of planets and moons

moment of inertia is largest about that axis for which the mass elements are most distant
Copyright © 2011. Cambridge University Press. All rights reserved. May not be reproduced in any form without permission from the publisher, except fair uses permitted

and, thus, the ri perpendicular to the axis are largest.

The reason that the C axis is special is that a body that rotates about this axis has the
lowest kinetic energy possible for a fixed angular momentum. Angular momentum is con-
served for an isolated body, but kinetic energy can be converted into heat. A body rotating
about the A axis, say, spins relatively quickly but this is the highest-energy rotational state,
so if it exchanges kinetic energy for heat it must spin about another axis, or a combin-
ation of axes in a complex tumbling motion. When the body finally spins about its C axis
it attains a minimum energy configuration and cannot change its rotation further (unless
some external torque acts on it).
This lesson was learned the hard way in the early days of space exploration when the
first US satellite, Explorer 1, which was shaped like a long narrow cylinder, was stabilized
by spinning it around its long axis. A broken antenna connection flexed back and forth as
the satellite rotated, dissipating energy. Within hours the satellite was spinning about its C
axis – the short axis, perpendicular to the cylinder. Because the spacecraft was not designed
to operate in this configuration radio contact was soon lost.
Most Solar System bodies, therefore, spin about their C axes. A few small asteroids
have been discovered that are in “excited” rotation states in which the object is not in its
minimum-energy rotational state and thus tumbles, but these exceptions are rare because
nearly every object has some means of dissipating energy internally and thus eventually
seeks out the lowest energy configuration.
Mars is a prime example of the importance of this process. Mars’ major positive gravity
anomaly, the Tharsis Rise, is located on its equator. This location puts the excess mass as
far as possible from its rotational axis, maximizing the moment of inertia. The opposite
extreme is illustrated by the asteroid Vesta, which suffered an impact that gouged out a
crater (a mass deficit) nearly as large in diameter as Vesta itself. The central peak of this
crater is now located at its rotational pole (which happens to be the south pole), the most
stable configuration.

2.2 Higher-order topography: continents and mountains

2.2.1 How high is high?

As silly as this question may seem, it highlights a common assumption that underlies our
thinking when we consider the elevation of some topographic feature. An elevation is a
under U.S. or applicable copyright law.

number or contour that we read off a map (or, more commonly in modern times, a color
code on an image). But what is that elevation relative to? Does it directly give the distance
from the center of the planet? Or the height above an arbitrary spheroid? Or, more com-
monly on Earth, the elevation above mean sea level? The answer to all these questions is,
“it depends.”
The fundamental reference surface for elevations, or geodetic datum, is established
empirically. Historically, it has varied with the technology for measuring topography,

completeness or accuracy of information and, sometimes, pure convenience. The era of

space exploration has brought great changes in this type of measurement, as well as bring-
ing new planets under close scrutiny. Each new planet has presented unique challenges in
the apparently simple task of measuring the elevations of its surface features.
In the case of the Earth, detailed mapping began in the eighteenth century with trigono-
metric surveys utilizing telescopes, carefully divided angle scales, rods, chains, levels, and
plumb bobs. The primary referent for elevations was mean sea level, already a problematic
concept in view of changes of water level in response to tides, currents, and meteorological
pressure changes. Individual countries established topographic surveys and used astro-
nomical measurements to locate prime coordinate points. Nations with access to an ocean
established elaborate gauging stations from which a time-average or mean sea level could
be defined. Survey crews could carry this elevation reference inland by the use of levels
and plumb bobs. Elevations of high and relatively inaccessible mountains were estimated
barometrically, by comparing the pressure of the air at the top of the mountain to that at
sea level. All of these methods, on close examination, amount to referring elevations to an
equipotential surface. The equipotential that corresponds to mean sea level is called the
geoid, and all elevations are, ideally, referred to this surface.
The geoid is quite hard to measure accurately. Although it roughly corresponds to a
flattened sphere, as described previously, slight variations in density from one location to
another gently warp it into a complex surface. Determination of the geoid thus requires
precision measurements of the acceleration of gravity, as well as accurate leveling. Much
of both classical and satellite geodesy is devoted to determining the geoid and, thus, per-
mitting elevations to be defined with respect to a level surface (level in the sense of an
equipotential, down which water will not run). Historically, each nation with a topographic
survey created its own version of the geoid, although thanks to satellite measurements these
are now knit into a consistent global network.
Elevations referred to the geoid are very convenient for a variety of purposes. Besides
engineering applications, such as determining the true gradient of a canal or railway line,
they are also essential to geologists who hope to estimate water discharges from the slope
of a river system. The “upstream” ends of many river systems (the Mississippi is one) are
actually closer to the center of the Earth than their mouths. And yet the water still flows
from head to mouth because water flows from a higher to a lower gravitational potential.
The modern era of the Global Positioning System (GPS) is ushering in new changes. The
orbiting GPS satellites really define positions with respect to the Earth’s center of mass,
so converting GPS elevations to elevations with respect to the geoid requires an elaborate
under U.S. or applicable copyright law.

model of the geoid itself. It has become common to refer elevations to a global average
datum that locally may not correspond to the actual geoid.
The determination of elevations on other planets is becoming nearly as complex as that
on Earth, thanks to a flood of new data from orbiting spacecraft. The first body to be
orbited by a spacecraft capable of determining elevations precisely was the Moon. The
Apollo orbiters carried laser altimeters that measured the elevations of features beneath
their orbital tracks. Although it was many years before the Clementine spacecraft expanded

EBSCO Publishing : eBook Academic Collection (EBSCOhost) - printed on 11/3/2023 1:50 PM via YDUQS
AN: 399390 ; H. Jay Melosh.; Planetary Surface Processes
Account: s4526441
38 The shapes of planets and moons

this method beyond the equatorial swaths cut by the Apollo orbiters, a prime reference
Copyright © 2011. Cambridge University Press. All rights reserved. May not be reproduced in any form without permission from the publisher, except fair uses permitted

surface for lunar elevations had to be chosen. Rather than trying to use the very slightly
distorted equipotential surface, the reference surface was chosen to be a perfect sphere of
radius 1738 km centered on the Moon’s center of mass. All lunar topographic elevations
are relative to this sphere. Since this sphere is not a geoid (although it is close enough for
many purposes) care must be taken when, for example, estimating the slopes of long lava
flows over the nearly level plains of the maria.
The topographic reference levels chosen for different planets varies depending on the
rotation rate (and direction) of the planet, its degree of deviation from a sphere, and the
technology available. The reference surfaces of Mercury and Venus are spheres. The mean
diameter of Mercury has yet to be established: At the moment, the only global data set
comes from radar measurements of equatorial tracks, but the MESSENGER laser altimeter
should soon make a better system possible. The reference surface for Venusian elevations is
a sphere of diameter 6051 km, close to the average determined by the Magellan orbiter.
Mars offers mapmakers serious problems when it comes to elevations. The Martian
geoid is far from a rotationally symmetric spheroid, thanks to the large non-hydrostatic
deviations caused by the Tharsis gravity anomaly. The geoid is intended to coincide with
the level at which the mean atmospheric pressure equals 6.1 mb, the triple point of water.
However, the atmospheric pressure varies seasonally by a substantial fraction of the entire
pressure, so locating this point is not straightforward.
Some Martian elevation maps are referenced to a spheroid with flattening 1/170, a sys-
tem recommended by the International Astronomical Union (IAU) (Seidelmann et al.,
2002). However, much of the high-precision data currently available is referenced to a
more complex and realistic geoid, so that the user of such information must be alert to the
system in use. Elevations with respect to a geoid are most useful in determining what direc-
tions are really downhill (that is, toward lower gravitational potential), which determines
the expected flow direction of water or lava. Because geoids improve with time, no map of
elevations is complete without a specification of the reference surface in use.
The surfaces of small asteroids such as Eros or Ida, comet nuclei such as Tempel 1, and
many other small bodies that will be mapped in the future, present new problems. They are
too irregular in shape to approximate spheres. At the moment their surface elevations are
defined in terms of the distance from their centers of mass.
The gas giant planets in the outer Solar System lack solid surfaces and so elevations are
especially difficult to define. The convention is now to refer elevations on these bodies to
a spheroid at the 1-bar pressure level, which is a good approximation to an equipotential
under U.S. or applicable copyright law.

surface on such planets.

2.2.2 Elevation statistics: hypsometric curves

Elevation data can be processed and interpreted in many ways. A map of elevations, a topo-
graphic map, is certainly the most familiar and contains a wealth of data. However, one can
extract more general features from such data that tell their own stories. On a small scale,

roughness is important for safely landing on and roving over planetary surfaces. Some
Copyright © 2011. Cambridge University Press. All rights reserved. May not be reproduced in any form without permission from the publisher, except fair uses permitted

of the methods for analyzing roughness are described in Box 2.1. On the large scale, an
elevation plot known as the hypsometric curve has proven useful in highlighting general
properties of planetary crusts.
A hypsometric curve is a plot of the percentage of the area of a planet’s surface that
falls within a range of elevations. Curves of this type have been constructed for the Earth
ever since global topographic data sets became available and they show one of the Earth’s
major features. Figure 2.3c illustrates the Earth’s hypsometric pattern, binned into eleva-
tion intervals of 500 m. The striking feature of this curve is the two major peaks in areas
that lie between the maximum elevation of 7.83 km and the minimum of –10.376 km
below mean sea level (note that these are not the highest and lowest points on the Earth –
they are the highest and lowest elevations averaged over 5′ x 5′, 1/12-degree squares). A
sharp peak that encompasses about 1/3 of the area of the Earth lies close to or above sea
level. The second peak lies about 4 km below mean sea level and accounts for most of the
rest of the Earth’s area. These two peaks reflect the two kinds of crust that cover the sur-
face of our planet. The low level is oceanic crust, which is thin (5–10 km), dense (about
3000 kg/m3), basaltic in composition and young, being created by mid-oceanic spreading
centers. The second principal topographic level is continental crust, which is much thicker
(25–75 km) than oceanic crust, less dense (about 2700 kg/m3), granitic in composition and
much older than the ocean floors. Plate tectonics creates and maintains these two different
crustal types.
Although Figure 2.3a shows a hypsometric curve for Mercury, the data set from which
this is derived is sparse at the moment, consisting of a number of mostly equatorial radar
tracks. At least with this data, however, there is no indication of an Earth-like dichotomy
of crustal thickness.
Figure 2.3b illustrates the hypsometric curve of Venus, for which an excellent data set
exists from the Magellan radar altimeter. This curve is an asymmetric Gaussian, skewed
toward higher elevations. There is no indication of a double-peaked structure, from which
we must infer that, whatever processes are acting to create the crust of Venus, they must
differ profoundly from those that affect the Earth.
The Moon’s hypsometric curve is illustrated in Figure 2.3d. Like Venus, the Moon lacks
a dichotomy of crustal types, although there are important differences in elevation between
the nearside and farside, illustrated by the thin lines that show separate hypsometric curves
for the two hemispheres. This is generally attributed to systematic differences in crustal
thickness between the nearside and farside, rather than compositional differences. The
under U.S. or applicable copyright law.

role of the large basins, especially the gigantic South Pole-Aitken basin, is still not fully
understood.
Finally, we come to Mars in Figure 2.3e. Mars, surprisingly, shows a double-peaked
distribution similar to that of the Earth. As shown by the light lines, the lower peak is
accounted for by the Northern Lowlands, while the high peak represents the contribution
of the Southern Highlands. The two terrains are divided by the Martian Crustal Dichotomy,
an elliptical region tilted with respect to the north pole that may represent the scar of an

EBSCO Publishing : eBook Academic Collection (EBSCOhost) - printed on 11/3/2023 1:50 PM via YDUQS
AN: 399390 ; H. Jay Melosh.; Planetary Surface Processes
Account: s4526441
40 The shapes of planets and moons
Copyright © 2011. Cambridge University Press. All rights reserved. May not be reproduced in any form without permission from the publisher, except fair uses permitted

Box 2.1 Topographic roughness

The roughness of a surface is a concept that everyone is familiar with. Surprisingly, although
there are many ways to measure roughness, there is no standard convention. Intuitively, rough
surfaces are full of steep slopes, while smooth surfaces lack them. One measure of roughness
would, thus, cite some statistical measure of the frequency of occurrence of a particular slope,
extended over a range of slopes. For example, Root Mean Square (RMS) measures have often
been used because they can be easily extracted from radar backscatter data. This statistic
implicitly assumes that the distribution of slopes approximately follows a Gaussian curve, an
assumption that needs to be tested. One might also cite mean or median slopes.
These simple statistical measures, while perhaps capturing our intuitive idea of roughness,
do miss an important aspect of the concept, and that is the scale of the roughness. A surface
that is smooth on a scale of 100 m might be very rough on the scale of 10 cm, a difference
that is of overwhelming importance when one is trying to set a 1 m lander down safely onto a
planetary surface. We thus need to define a baseline, L, in addition to the slope of a surface and
to describe the statistics of slopes with respect to a range of baselines.
We could, for example, define the slope s(L) of a surface for which we measure the
elevations z(x, y), where x and y are Cartesian coordinates that define a location on the surface.
The surface slope is then given by:
z ( x , y ) − z ( x ′, y ′ )
s( L ) = . (B2.1.1)
L
The points where the slope is evaluated, (x, y) and (x′,y′), are separated by the distance
L = ( x − x ′ )2 + ( y − y′ )2 . On a two-dimensional surface we need some understanding of
how to locate the different points at which elevations are evaluated, and many methods have
been devised for this. At the present time, elevation data is often acquired by laser altimeters
on orbiting spacecraft. The tracks along which elevations are measured are linear or gently
curved, simplifying the decision process. New techniques may make this problem more acute,
however: The laser altimeter aboard the Lunar Reconnaissance Orbiter now collects elevation
data simultaneously from an array of five non-collinear spots, so that a full two-dimensional
array of slopes can be defined.
A promising statistic is derived from fractal theory, the Hurst exponent, and has been
applied to the analysis of Martian slopes in the MOLA dataset (Aharonson and Schorghofer,
2006). This statistic, at present, is limited to linear sets of elevation data, z(x), in which the
y coordinate is ignored. The variance v(L) of the elevation differences along the track is
computed for a large number N of equally-spaced locations:

1 N 2
v( L ) = ∑  z( x ) − z( xi + L ) . (B2.1.2)
N i =1  i
under U.S. or applicable copyright law.

The roughness is thus given by the slope s(L):

H −1
v( L )  L
s( L ) = = s0   . (B2.1.3)
L  L0 

EBSCO Publishing : eBook Academic Collection (EBSCOhost) - printed on 11/3/2023 1:50 PM via YDUQS
AN: 399390 ; H. Jay Melosh.; Planetary Surface Processes
Account: s4526441
2.2 Higher-order topography 41
Copyright © 2011. Cambridge University Press. All rights reserved. May not be reproduced in any form without permission from the publisher, except fair uses permitted

Box 2.1 (cont.)

It is often found that the variance v(L) is a power function of the baseline L, so that if the
variance is compared to the slope s0 at some particular baseline L0, a relation expressed by the
second term in (B2.1.3) is found with an exponent H known at the Hurst exponent. Whether this
exponent is constant over a broad range of baselines or depends in some simple way upon the
scale is not yet known, nor is it understood how surface processes are related to this exponent,
although some proposals have been made (Dodds and Rothman, 2000). More progress can be
expected as more closely spaced elevation data is collected on a number of different planets.
A disadvantage of the definition (B2.1.2) is that a long, smooth slope also contributes to
the variance because the elevations z(x) on a straight slope differ by a constant amount along
the track. One would really like to filter out all elevation differences except those close to the
scale L. One solution to this problem is the median differential slope, which is based upon four
points along the track. The two extreme points are used to define a regional slope, which is
then subtracted from the elevation difference of the two inner points to achieve a measure of
slope that is not affected by long straight slopes, but responds to short wavelength variations on
the scale of the distance between the inner points (Kreslavsky and Head, 2000). The array of
four points is located at –L, -L/2, L/2, and L around an arbitrary zero point that slides along the
spacecraft track. This differential slope sd is given by:

z( L / 2) − z( − L / 2) z( L ) − z( − L )
sd ( L ) = − . (B2.1.4)
L 2L
The first term is the slope between the inner points and the second term is the larger scale
slope. The differential slope is zero for a long straight slope, as desired.
In the past, the study of roughness tended to focus on landing-site safety, but current efforts
are also making progress on extracting information on the surface processes creating the
roughness. One may expect to hear more about this in the future.

ancient giant impact. Do these two peaks represent two types of crust, as on the Earth, or
are these just areas with very different crustal thickness? We do not believe that Mars pos-
sesses plate tectonics, although it has been suggested that some plate processes may have
acted in the distant past.
The Martian hypsometric curve offers an interesting lesson in the importance of referen-
cing elevations to the geoid. Earlier plots of Mars elevations showed a Gaussian-like dis-
tribution of elevations similar to that of Venus or the Moon. Only after a good gravity field
was measured and elevations referenced to a true geoid did the double-peaked character of
Martian elevations become apparent.
under U.S. or applicable copyright law.

2.2.3 Where are we? Latitude and longitude on the planets

Latitudes and longitudes are the conventional means for locating features on the surface of
a planet. However, before such a system can be defined, the pole of rotation must be estab-
lished. All systems of latitude and longitude are oriented around the north pole, which must

EBSCO Publishing : eBook Academic Collection (EBSCOhost) - printed on 11/3/2023 1:50 PM via YDUQS
AN: 399390 ; H. Jay Melosh.; Planetary Surface Processes
Account: s4526441
42 The shapes of planets and moons

Mercury Venus
Copyright © 2011. Cambridge University Press. All rights reserved. May not be reproduced in any form without permission from the publisher, except fair uses permitted

(a) 60 (b) 35

50 30
Percent of sampled area

Percent of total area

40
20
30
15
20
10
10 min max min max
–3.3 3.2 5 –2.46 9.70
0 0
–15 –10 –5 0 5 10 15 20 25 –15 –10 –5 0 5 10 15 20 25
Elevation above mean, km Elevation above datum, km

Earth Moon
(c) 20 (d) 15
Whole Moon
Nearside
15 Farside
Percent of total area

Percent of total area 10

5
5 min max
–10.376 7.830 min max
–7.979 8.722
0 0
–15 –10 –5 0 5 10 15 20 25 –15 –10 –5 0 5 10 15 20 25
Elevation above datum, km Elevation above datum, km
Mars
(e) 15
Entire Mars
Northern Lowlands
Percent of total area

10 Southern Highlands

min max
–8.068 21.134

0
–10 0 10 20 30
Elevation above datum, km

Figure 2.3 Hypsometric curves of the terrestrial planets and the Moon. Maximum and minimum
elevations are also shown. These elevations refer to the maximum and minimum elevations averaged
over square sample areas of different sizes, not the highest and lowest points on the planet’s surface.
under U.S. or applicable copyright law.

(a) Mercury data from 9830 radar elevations in PDS file allMerc.txt. (b) Venus hypsometric curve
derived from a 1° x 1° Magellan map of Venus from PDS dataset MGN-V-RDRS-5-TOPO-L2-V1.0,
file TOPOGRD-DAT. (c) Earth, data is binned in 1/12 degree squares, from National Geophysical Data
Center file TBASE.BIN. (d) Moon, data is ¼ degree data from Clementine LIDAR from PDS dataset
CLEM1-L-LIDAR-5-TOPO-V1.0, file TOPOGRD2.DAT. Light lines show separate curves for the
nearside and farside. (e) Mars, ¼ degree MOLA gridded data from PDS dataset MGS-M-MOLA-
5-MEGDR-L3-V1.0, file MEGT90N000CB.IMG. Light lines show data separately for the Northern
and Southern crustal provinces. For this purpose, the planet was divided into two hemispheres by a
great circle whose pole is located at 53° N and 210° E longitude.

be determined in some absolute system of coordinates. The locations of the Earth’s poles
Copyright © 2011. Cambridge University Press. All rights reserved. May not be reproduced in any form without permission from the publisher, except fair uses permitted

are determined astronomically, by the position of the extrapolated rotation axis among the
stars in the sky. Although the Earth’s axis precesses slowly with a period of about 26 000
yr, this gradual change is predictable and can be taken into account when referring to the
pole position by citing the time, or epoch, at which the position is cited. The pole positions
of the other planets are defined in a similar way, in terms of the celestial coordinates, the
declination and right ascension, of their projected northern axis of rotation.
The choice of prime meridian for different planets is entirely arbitrary, but must be a
definite location. On the Earth, we use the longitude of a point in Greenwich, UK, as the
zero of longitude. On Mercury a crater known as Hun Kal defines the location of the 20°
longitude. The choice of the 0° longitude on Venus fell to the central peak of a crater known
as Ariadne, while on Mars the 0° longitude passes through a small crater known as Airy-0.
Pluto’s 0° of longitude (at present) passes through the mean sub-Charon point. As new
bodies are mapped and their rotation axes determined, new choices for the prime meridian
have to be made.
The prime meridians of the fluid gas giant planets in the outer Solar System are much
harder to define and are based on the rotation rates of their magnetic fields rather than the
shifting patterns of clouds in their atmospheres. Because the clouds rotate at different rates
depending on latitude, they do not yield definite rotation rates for the entire planet. For
these planets an accurate rotation rate must be determined and that rotation rate, plus the
epoch at which it was established, defines the prime meridian.
Venus presents an interesting cartographic problem because its spin is retrograde. Its
north pole, nevertheless, lies on the north side of the ecliptic by convention and, also by
cartographic convention, longitudes increase eastward from 0° to 360°.
The International Astronomical Union adopted a convention in 2000 that defines the
latitude and longitude coordinate system used in locating features on the surface of planets.
The first principle defines the north pole of a Solar System body:
(1) The rotational pole of a planet or satellite that lies on the north side of the invariable
plane will be called north, and northern latitudes will be designated as positive.
The second principle is more controversial and there is some disagreement between geo-
physicists and cartographers about the most sensible way to present longitudes:
(2) The planetographic longitude of the central meridian, as observed from a direction
fixed with respect to an inertial system, will increase with time. The range of longi-
tudes shall extend from 0° to 360°.
under U.S. or applicable copyright law.

Thus, west longitudes (i.e., longitudes measured positively to the west) will be used
when the rotation is prograde and east longitudes (i.e., longitudes measured positively
to the east) when the rotation is retrograde. The origin is the center of mass. Also,
because of tradition, the Earth, Sun, and Moon do not conform with this definition.
Their rotations are prograde and longitudes run both east and west 180°, or east 360°
(Seidelmann et al., 2002).

EBSCO Publishing : eBook Academic Collection (EBSCOhost) - printed on 11/3/2023 1:50 PM via YDUQS
AN: 399390 ; H. Jay Melosh.; Planetary Surface Processes
Account: s4526441
44 The shapes of planets and moons

This convention means that Mars presents a left-handed coordinate system, a conse-
Copyright © 2011. Cambridge University Press. All rights reserved. May not be reproduced in any form without permission from the publisher, except fair uses permitted

quence not favored by geophysicists. This debate is not resolved at the present time, so the
user of cartographic data must carefully check what conventions are in use – it is very easy
to download geophysical data for Mars and find that one is working on a mirror image of
the actual planet Mars.

2.3 Spectral representation of topography

Maps showing contours of elevation are not the only way of codifying topographic infor-
mation. Just as any function of spatial coordinates can be broken down into a Fourier
series in an inverse space of wavenumbers, topography can be represented as a sum of
oscillating functions on a sphere. This mode of representation is known as harmonic
analysis or spectral analysis and for many data sets is preferred over a purely spatial
representation.
Spectral analysis averages elevations over all positions on a sphere and presents the
information as a function of wavelength, not position. No information is lost in this pro-
cess: With the appropriate mathematical tools one can freely transform from space to wave-
length and back again.
The full details of harmonic analysis are too specialized for full presentation in this
book. The interested reader is referred to a fine review of the entire subject by Wieczorek
(2007). In this book it is enough to note that any function of latitude ϕ and longitude λ,
such as elevation H(ϕ,λ), can be expressed as:
∞ l
H (ϕ , λ ) = ∑ ∑ HlmYlm (ϕ , λ ) (2.10)
l = 0 m =− l
where l and m are integers and the Ylm(ϕ,λ) are spherical harmonic functions of order l and
degree m. They are given in terms of more standard functions as:

 Plm (sin ϕ )cos mλ if m ≥ 0

Ylm (ϕ , λ ) =  (2.11)
 Pl m (sin ϕ )sin m λ if m < 0

where the Plm are normalized associated Legendre functions given by:

(l − m)!
Plm (sin ϕ ) = (2 − δ 0 m )(2l + 1) P (sin ϕ ) (2.12)
(l + m)! lm

under U.S. or applicable copyright law.

where δij is the Kronecker delta function and Plm are the standard, unnormalized Legendre
functions. These functions are tabulated in standard sources and, more importantly, can
be computed with readily available software. There are many issues about the conven-
tional normalizations of these functions, which are not standardized across all scientific
disciplines.

EBSCO Publishing : eBook Academic Collection (EBSCOhost) - printed on 11/3/2023 1:50 PM via YDUQS
AN: 399390 ; H. Jay Melosh.; Planetary Surface Processes
Account: s4526441
2.3 Spectral representation of topography 45

The functions Ylm are the spherical analogs of sines and cosines for Fourier analysis.
Copyright © 2011. Cambridge University Press. All rights reserved. May not be reproduced in any form without permission from the publisher, except fair uses permitted

They are simple for small-order l and become more complex, with more zero crossings, as
l increases. They possess 2|m| zero-crossings in the longitudinal direction and l – |m| in the
latitudinal direction. Thus Y00 is a constant, Y10, Yl-1, and Yl1 correspond to the displacement
of a sphere from the center of mass (for example, a center-of-figure, center-of-mass offset)
and the five l = 2 terms describe an oblate or triaxial tidally distorted sphere. In general,
as l increases the wavelength of the feature that can be represented by these harmonics
decreases. This is made more precise by an approximate relation between the wavelength
w of features that can be represented by spherical harmonics of order l: w ≈ 2π a / l (l + 1) ,
where a is the planetary radius.
The expansion of topography in spherical harmonics makes the idea of orders of
relief precise: The zeroth-order harmonic is just the radius, the first is the center-of-mass
center-of-figure offset, the second is the rotational or tidal distortion, etc. Harmonic
coefficients for the topography of the planets to degree and order 180 are becoming
common, and still higher degrees exist for the Earth and are planned for the other plan-
ets as sufficiently precise data becomes available. In addition to topography, the geoid
and gravity fields are also represented by spherical harmonics, a format that makes
many computations of, for example, global isostatic compensation, much simpler than
it is for spatial data.
The spherical harmonic functions are orthogonal after integration over the com-
plete sphere, so that the harmonic coefficients Hlm can be obtained from the topography
H(ϕ,λ) by:

1 2π π / 2
Hlm = ∫ ∫ H (ϕ , λ ) Ylm (ϕ , λ ) cos ϕ dϕ d λ. (2.13)
4π 0 −π / 2

One can freely pass from the spatial representation of topography to the harmonic
representation and back again. There is no loss of information, nor any saving in the amount
of data to be stored for one or the other representation.
Just as the hypsometric function attempts to distil useful global information from the
map of elevations, a similar extraction of data is often made from the harmonic coefficients.
This data contraction is called the spectral power and is a measure of how much of the top-
ography is due to a particular wavelength. The RMS spectral power density collapses a full
set of l2+2l+1 numbers for harmonic coefficients to degree and order l down to a set of only
l numbers by summing over all orders for each degree:
under U.S. or applicable copyright law.

l
Sl = 2
∑ Hlm . (2.14)
m =− l

The RMS spectral power densities given by Equation (2.14) are plotted in Figure 2.4 for
each of the terrestrial planets (excluding Mercury) and the Moon. Except for the Moon,
most of the RMS power densities on this plot rise with increasing wavelength, so each body

EBSCO Publishing : eBook Academic Collection (EBSCOhost) - printed on 11/3/2023 1:50 PM via YDUQS
AN: 399390 ; H. Jay Melosh.; Planetary Surface Processes
Account: s4526441
46 The shapes of planets and moons

10 000
Copyright © 2011. Cambridge University Press. All rights reserved. May not be reproduced in any form without permission from the publisher, except fair uses permitted

Mars Earth

RMS topography, m
1 000

100 Venus
Moon
10 Kaula rule slope

1
10 100 1 000 10 000 100 000
Wavelength, km

Figure 2.4 Topographic power spectra of the terrestrial planets and the Moon, excluding Mercury for
which the necessary data does not yet exist. Lunar spectral data are from the Kaguya data set (Araki
et al. 2009). Spectral data on Earth, Venus, and Mars are from Mark Wieczorek’s website, http://
www.ipgp.fr/~wieczor/SH/SH.html, files SRTMP2160, VenusTopo719.shape and MarsTopo719.
shape, respectively.

has more power in longer-wavelength topography. Another way of saying this is that the
slopes of the surfaces are approximately independent of their scale. It is popular to call this
a fractal relationship, but it is unclear, at present, exactly what this means. The Earth and
Venus are comparably smooth at short wavelengths, while Mars is rougher than both and
the Moon is rougher still. The Moon’s roughness does not rise with increasing wavelength
as fast as that of the larger planets.
William Kaula worked extensively on harmonic representations of topography and his
studies of the Earth’s topography led him to formulate what we will call here “Kaula’s
second law,” which is that the RMS topography depends on the inverse order, 1/l. Because
wavelength depends on the inverse order as well, his law states that the power is directly
proportional to the wavelength. The prediction of this law is shown on Figure 2.4 and it
does seem to hold fairly well for the major planets, but not for the Moon. The deviation
shown here for the Moon is relatively new: It was not known before the data from the
Kaguya laser altimeter were analyzed.
Spectral representations are, at present, difficult to interpret (see, e.g. the discussion by
Pike and Rozema, 1975). Spectral data at a very small scale is widely used for computa-
under U.S. or applicable copyright law.

tions of “trafficability” of vehicles across terrains and for designing vehicle suspension
systems, but its use in geologic interpretation has been limited. The reason for this may
be that the spectral method averages over a wide variety of different terrain types that are
shaped by different processes and so loses the signatures characteristic of individual proc-
esses. Whatever the reason, it is currently an analysis technique in search of an interpret-
ation, although some suggestive models have provided more insight into the interpretation
of such data (Dodds and Rothman, 2000).

EBSCO Publishing : eBook Academic Collection (EBSCOhost) - printed on 11/3/2023 1:50 PM via YDUQS
AN: 399390 ; H. Jay Melosh.; Planetary Surface Processes
Account: s4526441
Exercises 47

Further reading
Copyright © 2011. Cambridge University Press. All rights reserved. May not be reproduced in any form without permission from the publisher, except fair uses permitted

Newton’s Principia is tough going, but full of surprising results. It is astonishing how far
Newton went with the theory of the Earth’s figure (Newton, 1966). You do not need to know
Latin to read the translation of the Principia, but you do need to be patient and resourceful.
Harold Jeffreys was deeply interested in the figure of the Earth, the Radau approximation,
and the theory of the Moon’s triaxial figure. There are six editions of his famous book The
Earth, but the third (Jeffreys, 1952) and fourth present the apex of his insight into this
problem. The problem of determining the shape of the Earth has been of major interest to
astronomers and mathematicians since Newton. The early history of investigation of the
figure of the Earth is exhaustively told in the full language of mathematics by Todhunter
(1962). More modern extensions are well covered in Chandrasekhar (1969) and Jardetzky
(1958). Kaula’s book (Kaula, 1968) is now very dated in its facts, but he covered many of
the methods of planetary geophysics, particularly geodesy, in great detail. The nature of the
geoid on Earth and its determination are well discussed in Lambeck (1988). The details of
modern planetary cartography are described in book form by Greeley and Batson (2000).
Spectral analysis of both topography and gravity are the subjects of a very recent and very
clear review by Mark Wieczorek (2007) that offers the simplest introduction to spherical
harmonics that I am aware of. He also goes to some trouble to explain the different normal-
ization conventions in the geophysical literature.

Exercises
2.1 A whirling moon
Saturn’s moon Iapetus is currently synchronously locked to Saturn, with a rotation (and
orbital) period of 79.3 days. In spite of its slow rotation, Iapetus has a considerable equator-
ial bulge, a−c ≈ 35 km (Table 2.3). Iapetus’ density is not very different from that of water
ice, so it can be treated as an approximately homogeneous body. If Iapetus’ equatorial
bulge is a fossil remnant from a time when it was spinning faster than at present, estimate
the minimum initial period of Iapetus’ rotation (explain why this is a minimum estimate).
What do you think may have happened to Iapetus?

2.2 Hot Jupiteus shaped like water melons

Planet WASP-12b circles a Sun-like star about 600 light years from Earth in the constella-
tion Auriga. It is a hot Jupiter planet, with a mass equal to 1.41 times that of Jupiter, radius
under U.S. or applicable copyright law.

1.83 times larger than Jupiter, but circles only 0.0229 AU (Astronomical Units) from its
star with a period of 1.0914 days. Use Equations (2.6) and (2.7), suitably generalized for
a planet orbiting a star, to compute the tidal distortion of this planet, assuming that it is
synchronously locked to its star (which is almost certainly true). Tabulate the lengths of the
three principal axes a, b, and c. What do you think this implies for the planet? For more on
this system, see Li et al. (2010).

EBSCO Publishing : eBook Academic Collection (EBSCOhost) - printed on 11/3/2023 1:50 PM via YDUQS
AN: 399390 ; H. Jay Melosh.; Planetary Surface Processes
Account: s4526441
48 The shapes of planets and moons

2.3 The axis of least effort

The kinetic energy of rotation of a body with a principal moment of inertia I about some
1
axis is given by E = Iω 2 , where ω is the angular rotation rate (radians/s). The angular
2
momentum L of a rotating body is given by L = Iω. For fixed angular momentum, show
that the kinetic energy of a rotating body is a minimum if it rotates about the axis with the
maximum moment of inertia C of the three principal moments C ≥ B ≥ A.
Extra Credit: If a body is rotating stably about its C axis and some internal process in
the body redistributes its internal mass and switches the C and B principal axes, what hap-
pens to this body? Note that a process of this kind has been proposed for, among others,
Enceladus (Nimmo and Pappalardo, 2006).
under U.S. or applicable copyright law.

EBSCO Publishing : eBook Academic Collection (EBSCOhost) - printed on 11/3/2023 1:50 PM via YDUQS
AN: 399390 ; H. Jay Melosh.; Planetary Surface Processes
Account: s4526441
3
Strength versus gravity

The existence of any differences of height on the Earth’s surface is

decisive evidence that the internal stress is not hydrostatic. If the Earth
was liquid any elevation would spread out horizontally until it disap-
peared. The only departure of the surface from a spherical form would
be the ellipticity; the outer surface would become a level surface, the
ocean would cover it to a uniform depth, and that would be the end of us.
The fact that we are here implies that the stress departs appreciably from
being hydrostatic; …
H. Jeffreys, Earthquakes and Mountains (1935)

3.1 Topography and stress

Sir Harold Jeffreys (1891–1989), one of the leading geophysicists of the early twentieth
century, was fascinated (one might almost say obsessed) with the strength necessary to
support the observed topographic relief on the Earth and Moon. Through several books and
numerous papers he made quantitative estimates of the strength of the Earth’s interior and
compared the results of those estimates to the strength of common rocks.
Jeffreys was not the only earth scientist who grasped the fundamental importance of rock
strength. Almost fifty years before Jeffreys, American geologist G. K. Gilbert (1843–1918)
wrote in a similar vein:
If the Earth possessed no rigidity, its materials would arrange themselves in accordance with the laws
of hydrostatic equilibrium. The matter specifically heaviest would assume the lowest position, and
there would be a graduation upward to the matter specifically lightest, which would constitute the
entire surface. The surface would be regularly ellipsoidal, and would be completely covered by the
ocean. Elevations and depressions, mountains and valleys, continents and ocean basins, are rendered
possible by the property of rigidity.
G. K. Gilbert, Lake Bonneville (1890)

By rigidity Gilbert meant the resistance of an elastic body to a change of shape. He was
well aware that this rigidity has its limits, and that when some threshold is exceeded Earth
Copyright 2011. Cambridge University Press.

materials fail to support any further loads. We call this threshold strength and recognize
that this material property resists the tendency of gravitational forces to erase all topo-
graphic variation on the surface of the Earth and the other solid planets and moons.

EBSCO Publishing : eBook Academic Collection (EBSCOhost) - printed on 11/3/2023 9:53 AM via YDUQS
AN: 399390 ; H. Jay Melosh.; Planetary Surface Processes
Account: s4526441.main.estacio
50 Strength versus gravity

The importance of strength is highlighted by a simple computation that Jeffreys included

in his masterwork, The Earth (1952). This computation is summarized in Box 3.1, where
it is shown that, without strength, a topographic feature of breadth w would disappear from
the surface of a planet in a time tcollapse given by:

π w
tcollapse = (3.1)
8 g

where g is surface gravitational acceleration. Without strength, a mountain 10 km wide on
the Earth would collapse in about 20 seconds, and a 100 km wide crater on the moon would
disappear in about 3 minutes. Clearly, such features can and do persist for much longer
periods of time.
Planetary topography, and the material strength that makes it possible, lend interest
and variety to planetary surfaces. However, when seen from a distance, it is clear that
the shapes of planets are, nevertheless, very close to spheroids. Only very small aster-
oids and moons (Phobos and Deimos are examples) depart greatly from a spheroidal
shape in equilibrium with their rotation or tidal distortion. Thus, although the strength of
planetary materials (rock or ice) is adequate to support a certain amount of topography,
it is evidently limited. Such things as 100 km high mountains do not exist on the Earth
because strength has limits. The ultimate extremes of altitude on a planet’s surface are
regulated by the antagonism between the strength of its surface materials and its gravi-
tational field.
Although everyone has an intuitive idea of strength, the full quantification of this property
is both complex and subtle. Many introductory physics or engineering textbooks present
strength as if it were a simple number that can be looked up in the appropriate handbook.
This impression is reinforced by handbooks that offer tables of numbers purporting to
represent the strength of given materials. But further investigation soon reveals that there
are different kinds of strength: crushing strength, tensile strength, shear strength, and many
others. Strength sometimes seems to depend on the way that forces or loads are applied to
the material, and upon other conditions such as pressure, temperature, and even its history
of deformation. The various strengths of ductile metals, like iron or aluminum, typically do
not depend much on how the load is applied, or how fast it is applied, but common planet-
ary materials behave quite differently.
Quantitative understanding of the relation between topography, strength, and gravity
requires, first, some elementary notions of stress and strain and, second, a more detailed
understanding of how apparently solid materials resist changes in shape. This chapter intro-
duces the basic concepts of stress, strain, and strength before failure, and applies them to
the limits on possible topography. It also introduces the role of time and temperature in
limiting the strength of materials and the duration of topographic features. The next chapter
examines deformation beyond the strength limit and the tectonic landforms that develop
when this limit is exceeded.

EBSCOhost - printed on 11/3/2023 9:53 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
3.1 Topography and stress 51

Box 3.1 Collapse of topography on a strengthless planet

Consider a long mountain ridge of height h, width w and effectively infinite length L standing
on a wide, level plain. For simplicity suppose that the profile of the mountain is rectangular,
with vertical cliffs of height h bounding both sides (Figure B3.1.1). The surface gravitational
acceleration of the planet on which this mountain lies is g, and ρ is the density of the material
from which both the mountain and planetary surface are composed.
The weight of the mountain is ρghwL. If there is no strength, this weight (force) can only
be balanced by the inertial resistance of material accelerating beneath the surface, according to
Newton’s law F = ma. The driving force F equals the weight of the mountain, F = ρghwL. The
d 2h
acceleration a is equal to the second time derivative of the mountain height, a = 2 . The
dt
mass being accelerated is less easy to compute exactly, but it is approximately the mass
enclosed in a half cylinder of radius w/2 beneath the mountain (this neglects the mass of the
mountain itself, which is not strictly correct, but if h is small compared to w, the mountain
π
mass is only a small correction). The mass is then m ≈ w 2 L ρ . This yields a simple, second-
8
order differential equation for the mountain height h as a function of time, t:

d 2 h( t ) 8 g
= h(t ). (B3.1.1)
dt 2 π w

This equation has the solution

− t / tcollapse
h(t ) = ho e (B3.1.2)

where h0 is the initial height of the mountain and the timescale for collapse is given by:

π w
tcollapse = . (B3.1.3)
8 g

w
g
h

Figure B3.1.1 The dimensions and velocity of a linear collapsing mountain of height h and
width w on a strengthless half space of density ρ that is compressed by the surface gravity g
on a fluid planet. As the mountain collapses vertically it drives a plug of material of mass m
underneath it that flows out through the dashed cylindrical surface.

EBSCOhost - printed on 11/3/2023 9:53 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
52 Strength versus gravity

Fs
l l x V
V
Fl
b p
Ac Ab

(a) (b) (c)

Figure 3.1 Three varieties of strain. (a) Longitudinal strain, in which a block of material of original
length l and basal area Ac is extended an amount Δl by a force Fl. (b) Shear strain, in which the top
of a block of height b is sheared a distance Δx relative to its base (to an angle θ) by a differential
force Fs. (c) Volume strain, in which a block of original volume V is compressed an amount ΔV by
a pressure p.

3.2 Stress and strain: a primer

A full exposition of the continuum theory of stress and strain is beyond the scope of
this book. For the intimate details, the reader is referred to sources such as Turcotte and
Schubert’s excellent book Geodynamics (2002). A few simple concepts will suffice for a
general understanding of planetary surface processes, although the actual computation of
stresses under the different loading conditions illustrated later in this chapter requires an
application of the full theory of elasticity.

3.2.1 Strain
Strain is a dimensionless measure of deformation. It is a purely geometric concept that is
meaningful only in the limit where solids are approximated as continuous materials: All
relevant dimensions must be much larger than the atoms of which matter is composed.
Historically, the concept of strain was derived from measurements of the change in length
of a rod that is either stretched or compressed. When a force is applied parallel to a rod of
length l, its length changes by an amount Δl. The length change Δl is observed to be propor-
tional to the length l itself, so Δl depends on the size of the specimen being tested. A measure
of deformation that is independent of the specimen size is obtained by taking the ratio of
these two quantities to define a dimensionless longitudinal strain as (see Figure 3.1a):
∆l
εl = . (3.2)
l

A full description of extensional strain in a three-dimensional body requires three per-

pendicular longitudinal strains, one for each direction in space.
In addition to stretching or compression, a solid can also be deformed by shear, in which
one side of a specimen shifts in a direction parallel to the opposite side. In the special case

EBSCOhost - printed on 11/3/2023 9:53 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
3.2 Stress and strain: a primer 53

of simple shear the top of a layer of thickness b is displaced by a horizontal distance Δx

from the bottom, while its thickness b remains constant. In this case the shear strain is
defined as (Figure 3.1b)
∆x
εs = ≈θ (3.3)
b
where θ is the slope angle of the sheared material. This angle becomes exactly equal to
Δx/b as Δx approaches zero. Again, because space is three-dimensional there are three
independent shear strains.
Mathematically sophisticated readers may note that the six strains are not vector quan-
tities, but form components of a 3 × 3 symmetric tensor. The three perpendicular lon-
gitudinal strains are the diagonal components and the shear strains are the off-diagonal
components. An important theorem states that the coordinate axes can always be rotated
to a system in which the strain tensor is diagonal. In this coordinate system all strains are
longitudinal, although some may be compressional while others are extensional. A gen-
eral 3 × 3 matrix has 9 components, not 6. The extra three (which form an antisymmetric
tensor) correspond to pure rotations, which, because they do not cause distortions of the
material, are wisely excluded from the definition of the strain tensor.
Finally, if all the dimensions are shrunk or expanded equally, the shape is preserved,
but the volume V changes, and the resulting deformation is described by the volume strain
(Figure 3.1c):
∆V
εV = . (3.4)
V
There is only one volume strain and it depends entirely on the longitudinal strains,
because it can be expressed as the sum of the three perpendicular longitudinal strains.

3.2.2 Stress
Stress is a measure of the forces that cause deformation. In the limit of small deformations
it is linearly proportional to strain for an elastic material. Just as the strain is expressed as
a ratio of the change in length divided by the length, to make it independent of the size
of the test specimen, stress is expressed as the ratio between the force acting on the spe-
cimen and its cross-sectional area. Defined in this way, stress is independent of the size of
the test specimen and has dimensions of force per unit area, the same as pressure. Thus, if
the cross-sectional area of a rod is Ac, and a force Fl is acting to stretch or compress it, the
normal stress in the rod is defined as:
Fl
σl = . (3.5)
Ac

Similarly to longitudinal strain, there are three normal stresses, one for each perpendicu-
lar direction of space.

EBSCOhost - printed on 11/3/2023 9:53 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
54 Strength versus gravity

Stress is defined as positive when a rod is extended. This makes stress proportional to
strain times a positive number. This is a sensible procedure and is used without further
comment in engineering texts, in which positive stress is tensional. However, in geologic
applications stresses are nearly always compressional. Even when stretching does occur, it
is often under conditions of an overall compressional background stress, so that the stress
in the extended direction is simply less compressive than the other directions (in this case,
the stress is often said to be extensional as opposed to tensional). For such applications it
would obviously be simpler if compressional stress is taken as positive. However, such a
convention complicates other simple relations in the full theory of stress and strain. Various
geological authors have tried special definitions to deal with this problem, although few
have gone so far as to make the constants relating stress and strain negative. Turcotte and
Schubert, in their otherwise excellent book, actually switch conventions halfway through,
and other authors recommend changing the sign of the strain definition. The least drastic
convention, and the one followed in this book, is to define pressure as the negative of the
average of the three perpendicular stresses, so that compressive (negative) stress always
give rise to positive pressure. This means that a compressional stress acting on a rock mass
is negative.
In close analogy to shear strains, the three shear stresses are defined as the ratio between
a deforming force Fs and, in this case, the basal area of the sheared layer Ab:

Fs
σs = . (3.6)
Ab

Just as for strains, stresses are components of a 3 × 3 tensor whose diagonal components
are the normal stresses and the off-diagonal components are the shear stresses. (The three
antisymmetric components of the full 3 × 3 tensor are torque densities, which almost never
arise in practice. We do not consider them further.) Stresses are not vectors: The forces are
vectors, but because the forces are divided by an area that also has a direction in space, the
stresses are components of a tensor. Stresses, thus, do not point in some direction in space.
However, it is always possible to rotate the coordinate axes such that the off-diagonal shear
stresses are zero in the new coordinate system, and stresses are sometimes graphically rep-
resented as triplets of arrows of different lengths pointing in perpendicular directions. But
beware! Such arrows cannot be added or subtracted in the same fashion as vectors!
Finally, in the special case where the stresses are equal in three perpendicular spatial
directions, the negative of the force per unit area (all directions are equivalent in this case)
is defined as the pressure:

F
P = −σ vol = − . (3.7)
A

Because stresses, and stress differences in particular, play a major role in determining
the ability of a solid to resist deformation, it is often convenient to single out the three
perpendicular normal stresses in the special coordinate system in which the shear stresses

EBSCOhost - printed on 11/3/2023 9:53 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
3.2 Stress and strain: a primer 55

vanish. These special stresses are called principal stresses and are frequently denoted σ1,
σ2, and σ3 for the maximum (most tensional), intermediate, and minimum (most com-
pressive) normal stress directions – but be careful of stress conventions here: in geologic
applications the maximum stress is often taken as the most compressive. So long as
this is understood, it causes little difficulty. In the case of hydrostatic stress (pressure)
these principal stresses are all equal. When there are three unequal deviatoric stresses the
definition of pressure in Equation (3.7) is generalized so that p is equal to the negative
average of the three principal stresses. This quantity plays a special role in the tensor
description of stress because it is a rotational invariant, the (negative) trace of the stress
tensor, divided by 3.
Because of the qualitatively different dependence of strength on pressure and shear, the
stress is often separated into a component that depends only on differential stresses, called
the deviatoric stress (often written as σ ′ – thereby forming a test of the readers’ attentive-
ness) plus the (negative) pressure. The principal stresses are then written as σ1′-p, σ2′-p and
σ3′-p, whereas the shear stresses are the same as before.
The ultimate strength of many materials is often found to depend on the magnitude of the
difference between the maximum and minimum principal stresses, |σ 1 − σ 3|, without any
dependence on the intermediate principal stress. A somewhat more complicated measure
of the total distortional stress that does take the intermediate principal stress into account is
called the second stress invariant Σ2 (pressure is the first invariant):

1
Σ2 =
6 
(σ 1 − σ 3 ) + (σ 1 − σ 2 ) + (σ 2 − σ 3 )  .
2 2 2


(3.8)

The factor of 1/6 under the square root is a conventional part of the definition. There is
also a third invariant, whose role in failure mechanics is more complex, and is not consid-
ered further in this text. These quantities are called invariants because their magnitude does
not depend on the orientation of the coordinate system. Once their values are established in
one coordinate system, they are the same in all.
It may seem surprising that there is no shear stress term in either of these formulas: after
all, it is common experience that solids break more readily in shear than under compres-
sion. However, shear actually is incorporated, although this may not be apparent. The rea-
son is that shear is one of those off-diagonal components that are intentionally eliminated
by the coordinate rotation that brings the stress tensor to its diagonal form. It can be shown
that a state of pure shear stress σs is equivalent to one in which the coordinate axes are
rotated 45° and the principal stresses are σ 1 = −σ 3 = σs.

3.2.3 Stress and strain combined: Hooke’s law

English scientist (and Newton’s arch-rival) Robert Hooke (1635–1703) recorded some of
the first observations of the relation between stress and strain in 1665. Working mainly with
springs (Hooke was really interested in clocks) that produce visible deformations under

EBSCOhost - printed on 11/3/2023 9:53 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
56 Strength versus gravity

relatively small loads, Hooke hypothesized a linear relation between longitudinal stress
and strain, now known as Hooke’s law:
σ l = E ε l (3.9)
where the proportionality constant E has dimensions of pressure and is generally known as
Young’s modulus, after a much later researcher who studied the extension of elastic rods.
Although it was once believed that a single elastic constant is sufficient to describe the
stress–strain relation for a given material, it was finally demonstrated in the early 1800s
that at least two constants are necessary to characterize an isotropic solid (in fact, for a
single crystal, up to 21 elastic constants may be necessary, but here we consider only the
minimum required). The second constant is often taken to be the shear modulus μ that
relates shear stress to shear strain:
σ s = 2 μ ε s. (3.10)
The factor of 2 is a conventional part of the definition that derives from the way shear
strain is defined. Because there are two elastic constants they can be, and often are, com-
bined in various ways. For example, pressure and volume strain are related by a constant K
usually known as the bulk modulus:
p = −K ε V (3.11)
(note the minus sign because of the way pressure is defined). Because there are only two
independent stress–strain constants, one of these three must obviously be a function of the
others: It can be shown that E = 9Kμ/(3K + μ).
Another useful combination is called Poisson’s ratio ν. In Figure 3.1a the extended rod
is illustrated as having contracted in the direction perpendicular to its extension. This is a
real, observed effect (indeed, the case of pure extension, without lateral contraction, is very
difficult to realize in practice as it requires tensional loads perpendicular to the extension
axis to maintain a constant cross section). The dimensionless Poisson’s ratio is defined
as the ratio between the amount of lateral contraction and the longitudinal extension of a
laterally unconstrained rod. The deformation illustrated in Figure 3.1a actually involves
both a volume change and shear (change of shape), so that the Young’s modulus contains
contributions from both the bulk modulus and shear modulus. In terms of Poisson’s ratio,
ν, the Young’s modulus is E = 2(1 + ν)μ.
Relations between stress and strain are generally known as constitutive relations.
Hooke’s law was simply the first of what is now understood to be a large class of possible
relationships between deformation (strain) and applied force (stress). Such relations may
also involve time: We will shortly meet the concept of viscosity (invented by Newton) that
relates the strain rate (the derivative of strain with respect to time) to applied stress. In
modern times the study of the relation between deformation and stress has reached a high
degree of sophistication. This field is now known under the name of rheology. Because the
materials that make up planets are complex, the rheologic properties of materials as diverse
as rock, air, ice, and lava are crucial for an understanding of how the surfaces of planets and
moons formed and continue to evolve.

EBSCOhost - printed on 11/3/2023 9:53 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
3.2 Stress and strain: a primer 57

ductile

c
sti
Stress
brittle

pla
stic
ela

Strain

Figure 3.2 In a real solid, stress is linearly proportional to strain only for small stresses and strains
(typically only up to a strain of about 0.001). Beyond this limit the relationship becomes non-linear. In
this regime the flow deformation may be reversible (non-linear elasticity) or non-reversible (plastic).
At even larger strains the material may fracture, losing its strength suddenly in a brittle fracture, or
continue to deform to large strains in ductile flow.

The mathematically convenient linear relation between stress and strain does not hold in
all, or even in most, real situations: Although stress and strain are always proportional for
sufficiently small deformations, when the deformation becomes large enough (and large
may be a strain of only 0.001 – not even visible to the human eye!) the relation becomes
non-linear and catastrophic failure of various kinds may occur (Figure 3.2). Nevertheless,
the combination of simple constitutive laws, such as that of Robert Hooke, and the require-
ment that both internal and external forces are in balance (often known under the name
stress equilibrium) has been immensely fruitful in explaining the ability of planets to sup-
port topographic loads.

3.2.4 Stress, strain, and time: viscosity

Just as ideal elasticity is a useful limit describing the deformation of materials at small
strains, so too is the concept of ideal viscosity. Isaac Newton first recognized viscosity
on the basis of his extensive experimental studies, and proposed an ideal generalization
of his experiments (in fact, Newton proposed this property mainly to undermine his rival
Descartes’ vortex theory of planetary motion). Ideal elasticity relates shear stress σs and
shear strain εs by a linear equation. Similarly, ideal (or Newtonian) viscosity relates the
shear stress and shear strain rate ε̇ s through a single constant η, the viscosity:
σ s = 2ηε ̇ s. (3.12)
Viscosity has dimensions of stress × time, or Pa-s in SI units. The rules for viscous
flow are somewhat more complicated than those of elasticity because the volume strain εV

EBSCOhost - printed on 11/3/2023 9:53 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
58 Strength versus gravity

cannot be a function of time: If it were, the volume of a viscous substance under pressure
would gradually decrease to zero! Discussions of viscous flow must, therefore, pay careful
attention to the difference between volume strain and shear strain. In most ideal models the
volume strain is set equal to zero; this is called the incompressible limit. A more realistic,
but mathematically more complex, approximation is to treat the volume strain as elastic
and the shear strain as viscous.

3.3 Linking stress and strain: Jeffreys’ theorem

3.3.1 Elastic deformation and topographic support

The earliest and simplest models of topographic support are derived from applications of the
classic theory of elasticity. This theory combines the full tensor definitions of stress and strain
with a linear Hooke-type relation between stress and strain (with just two elastic constants, the
minimum number) and the stress equilibrium equations to derive a closed mathematical system.
Within the context of this theory, one can show that, starting from an unstressed initial solid, the
stress and strain throughout the solid are uniquely determined by the forces and displacements
acting on its surface. Thus, if we approximate a planet, or some well-defined portion of it, as an
elastic solid, and treat the weight of topography as a load acting on its surface, the stress differ-
ences induced by the topography can be accurately computed throughout its interior.
Of course, this is an unrealistically rosy picture of what is actually possible: The
troubles come from the detailed conditions under which elastic theory is valid. Harold
Jeffreys, to whom we owe many of the results that follow, was painfully aware of the
limitations of the elastic model, and he devoted much effort to understanding both its
successes and its failures. The first difficulty is the obvious limitation of elastic behavior
to small deformations. Once failure or flow occurs, elastic theory becomes invalid. In
principle this can be addressed by numerical methods and is thus inconvenient but not
insurmountable. The second, more insidious difficulty stems from the condition of an
unstressed initial solid. All planetary surfaces with which we are familiar exhibit a long
history of change, of repeated events that certainly exceeded the limits of linear elasti-
city. So to what extent can the near-surface material be considered initially unstressed?
All planetary materials have mass and all are subject to gravity, so at a minimum, the rocks
beneath the surface must develop sufficient stresses to support their own weight. However,
even a liquid, without resistance to deformation (but still resisting volume change!) can
support its own weight. It does this by compressing slightly and thus balancing the gravi-
tational force of the overlying material against the much stronger quantum mechanical
forces that resist the close approach of atoms (gravity eventually wins this struggle in the
stellar collapse to a black hole, but this is far outside the range of planetary processes). The
stresses are hydrostatic in this case, and the pressure p a distance h below the surface of a
body with uniform density ρ and surface gravitational acceleration g is given by:
p = ρgh. (3.13)

EBSCOhost - printed on 11/3/2023 9:53 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
3.3 Linking stress and strain 59

Although such lithostatic pressures may be very large compared to the stress differences
needed to cause rock failure, the large value of the bulk modulus K for most substances
ensures that the associated volume strain is small. In this case, we can simply add the
lithostatic stress and strain of the subsurface rock to that caused by other loads. This is a
consequence of the linearity of the theory of elasticity: Two solutions can always be added
to give a third solution, so long as the boundary conditions of the third solution are the sums
of those of its components.
If the rock beneath a planet’s surface crystallizes from a deep liquid mass, or is heated
to such a high temperature that all differential stresses relax after some time, then the
lithostatic stress state described above can be accurately considered to be the initial state
and the response to any subsequent loads can be computed as elastic additions to this
basic state. Unfortunately, most planets are not so cooperative: In most cases one can-
not assume that all differential stresses were erased just before the latest episode of
topographic loading.
Another elastic solution useful for describing an initial state is derived from the stresses
that develop in an initially unstressed and very wide elastic sheet that is suddenly subjected
to the force of gravity. The elastic sheet cannot expand laterally; it can only compress ver-
tically. In this case the principal stresses are not all equal (lithostatic), but the vertical stress
σV and horizontal stresses σH differ in magnitude:

σ V = − ρ gh
ν
σH = − ρ gh (3.14)
1 −ν

where ν is Poisson’s ratio, which can be no larger than 0.5. Poisson’s ratio for most solid
rocks is close to 0.25, although it can approach 0.0 for loosely consolidated sediments. In
this solution the magnitude of the horizontal stress is smaller than the magnitude of the ver-
tical stress. The difference between the horizontal stresses and the vertical stress increases
linearly with depth and so, at some large enough depth failure must occur, but this is often
so deep that the solution has great practical value.
Alert readers may wonder that this solution has any practical value at all: the idea that
a mass of rock might be assembled in the absence of gravity, which is afterwards magic-
ally turned on, seems so artificial that it could not apply to any real situation. However, as
demonstrated by Haxby and Turcotte (1976), this is precisely the stress state that develops
in a rock mass assembled from the gradual accumulation of a stack of thin, broad and ini-
tially stress-free layers. Thus, the stresses that develop in a thick pile of lava flows, or in an
accumulating sedimentary basin, are well described by this model. Compilations of verti-
cal and horizontal stress measurements in the Earth (McGarr and Gay, 1978) show that, in
many places, such as southern Africa or in sedimentary basins in North America, stresses
are bounded between the lithostatic and infinite-layer results (this is not true everywhere:
In Canada and much of Europe horizontal stresses are much larger than suggested by these
solutions).

EBSCOhost - printed on 11/3/2023 9:53 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
60 Strength versus gravity

Although the two basic states just described are frequently useful, they are certainly not
unique: Through all six editions of The Earth, Jeffreys invariably emphasized that, due to
the generally unknown history of previous deformation, there are an infinite number of
stress and strain configurations that are compatible with the presently observed topography.
So why did he devote so much time and effort to obtaining elastic solutions when he did not
believe that such solutions could be accurate? Jeffreys frequently cited a theorem he called
Castigliano’s principle, which asserts: “Of all states consistent with given external forces,
the elastic one implies the least strain energy” (Jeffreys, Ed. 6, Appendix C). Thus, to the
extent that the forces acting below a planetary surface tend toward a minimum of energy,
the elastic solution delineates the favored minimum. A second reason is that, although a
given elastic solution may not represent the complete stress state, it does often indicate how
the stresses change in response to a small change in the applied loads. For example, the
formation of a distant impact crater or a change in planetary spin rate or tidal stresses may
cause stress changes that are accurately described by an elastic deformation. In either case,
the elastic solutions are of greater significance than the limitations of the strictly conceived
elastic model would suggest.

3.3.2 Elastic stress solutions and a limit theorem

Using the full theory of elasticity, stresses can be computed beneath various surface loads,
assuming an initially hydrostatic initial state. Contour plots of the second invariant Σ2 for
four of these configurations are shown in Figure 3.3a–d. Figures 3.3a–c apply to long
loads intended to represent idealized mountain profiles, originally computed by Jeffreys.
Figure 3.3d shows the stress differences underneath an axially symmetric idealized impact
crater with a depth/diameter ratio of 0.3.
Although the patterns illustrated by these various solutions are diverse in detail, there
are a number of similarities. Most obvious is that the maximum stress differences are not at
the surface, but occur some distance below. Thus, most of the weight of a sinusoidal series
of mountain ridges is not supported by the strength of the material in the mountains them-
selves, but by material some distance below. This is an important lesson (one ignored by
the builders of the Tower of Pisa): Foundations are important! The second important lesson
is that the maximum stress difference is about 1/3 of the total load itself for all four cases
illustrated. These results are summarized in Table 3.1, where the depth to the maximum
stress and the maximum stress differences for Figures 3.3a–d are listed.
The first lesson from these solutions, the isolation of the maximum stress region below
the surface, is not strictly valid outside the domain of elastic solutions. More sophisti-
cated analyses, using the theory of plasticity described below, show that, although first
failure upon loading does, indeed, occur where the elastic solution predicts the max-
imum stress differences, once this failure has occurred the failure zone may work its
way toward the surface, especially if the load has sharp edges, as for a cliff or steep
surface slope. The final, visible failure may, thus, involve a surface landslide localized at
one of these sharp edges. However, the region over which the strength of the material is

EBSCOhost - printed on 11/3/2023 9:53 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
3.3 Linking stress and strain 61

(a) (b) 1
1

0 0
+ + + + +

–1 –1

–2 –2

–3
–3
–3 –2 –1 0 1 2 3 –3 –2 –1 0 1 2 3

0.0
+
–0.5

1 –1.0

0 –1.5
+

–1 –2.0

–2 –2.5

–3 –3.0
–3 –2 –1 0 1 2 3 0.0 0.5 1.0 1.5 2.0 2.5 3.0

Figure 3.3 Stresses below various loads placed on an originally unstressed elastic half space. Contours
are of the second invariant Σ2 and are drawn at intervals of 0.05, 0.1, 0.15, 0.2, 0.25, 0.3, 0.35, and
0.4 of the maximum load. These plots were constructed by summing the Fourier components of
the Airy stress function that satisfies the load boundary conditions. (a) Shows the differential stress
magnitudes beneath a series of very long mountains with sinusoidal hills and valleys. (b) Stresses
beneath a vertical-sided strip mountain. (c) Stresses beneath a long mountain with a triangular profile
and (d) Stresses beneath a circular impact crater with depth/diameter ratio 0.3. Plots are not vertically
exaggerated; horizontal dimensions are in units of the load width. The + sign marks the position of
the stress maximum in each plot.

exceeded is far broader than such a surface manifestation and is well delineated by the
elastic solution.
The second lesson from the elastic analysis is more enduring. Generations of struc-
tural engineers have devoted their ingenuity to ways of extending their ability to analyze
the maximum stresses that develop in any given structure. The results of this effort (and
the subject of a huge literature of its own) are the so-called limit theorems. Although
theorems of this type do not give the user the detailed distribution of stresses in some
complex structure (this must be done on a case-by-case basis using a full knowledge of
the structure and its history of loading), they do give some overall constraints on how

EBSCOhost - printed on 11/3/2023 9:53 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
62 Strength versus gravity

Table 3.1 Elastic stress differences, Poisson’s ratio ν = 0.25

Maximum stress Depth of maximum

Load shape difference Σ2/ρgh below surface

Sinusoidal strip, 0.384 0.289 λ

wavelength λ
Rectangular strip, 0.352 0.865 w
width w
Triangular strip, 0.305 0.388 w
basal width w
Axisymmetric crater, 0.359 0.305 D
depth/diameter=0.3, diameter D

strong materials must be to support some given load, independent of structure and history
of construction.
As summarized by Jeffreys, structural limit theorems assure us that to support a surface
load of order ρgh, somewhere in the body stresses between ½ and 1/3 of this load must be
sustained. Furthermore, this stress is generally supported at a depth comparable to the load
width (exceptions to this depth rule, such as loads supported by strong, thin plates, usually
imply stresses greatly in excess of the minimum).
This fundamental theorem is so important (and so often overlooked in the planetary lit-
erature!) that I set it out by itself for emphasis:
Jeffreys’ Theorem: The minimum stress difference required to support a surface load
of ρgh is (1/2 to 1/3) ρgh. This stress is usually sustained over a region comparable in
dimensions to the load.
Of course, this theorem does not prevent much larger stresses from developing in specific
situations, but a given topographic load cannot be supported by any smaller stress diffe-
rence. The value of this theorem is that it can be linked to specific strength models to obtain
quick estimates of the maximum topographic variation to be expected on any given Solar
System body, even when the specifics of interior structure and history are unknown. An
example of this procedure is given in the next section.

3.3.3 A model of planetary topography

Consider a generic planetary body (Figure 3.4) of mass M, average radius R̅ and average
density ρ̅. The surface acceleration of gravity g is:
GM 4
g=− 2
=− πGρR (3.15)
R 3
where G is Newton’s gravitational constant.

EBSCOhost - printed on 11/3/2023 9:53 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
3.3 Linking stress and strain 63

R max

∆h

ρc CM
g R min
M, ρ

Figure 3.4 A simple model of the gravitational forces in an irregular self-gravitating body such as an
asteroid. The average radius is R̅ and the maximum and minimum radii for points on the surface are
Rmax and Rmin from the center of mass CM. The mean density of the object is ρ̅.

This relation is exact for a spherical body, and approximate for any other shape. If
the surface has topography of order Δh, and its material is of density ρc, the surface load
imposed by this topographic variation is about Δσ = ρcgΔh. Applying Jeffreys’ theorem, a
minimum stress of magnitude Y must be present somewhere in the body’s interior:
1 2
Y≈ ∆σ = π G ρ ρc R ∆h. (3.16)
2 3
Rearranging, we obtain an equation that relates the maximum topographic variation, Δh,
to some measure of strength, Y.
3 Y 1
∆h ≈ . (3.17)
2π G ρc ρ R

Applying this equation to the Earth, take ρ̅ = 5200 kg/m3, ρc = 2700 kg/m3, R̅ =
6340 km. We find:
ΔhEarth (m) ≈ 80.4 Y (MPa). (3.18)
Taking Y ≈ 100 MPa, which is about the crushing strength of granite, we see that the
Earth can support abut 8 km of topography – not far off the 8850 m height of Mount
Everest or the 11 000 m depth of the Marianas trench, when the buoyancy of submerged
rock is taken into account. However, the dependence of Δh on 1/R̅ means that, if Y is the

EBSCOhost - printed on 11/3/2023 9:53 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
64 Strength versus gravity

same for all the terrestrial planets, we should expect 8 km high mountains on Venus, 24
km high mountains on Mars and 50 km high mountains on the Moon. As shown in Figures
2.3b and 2.3e, this is not far off for Venus and Mars, but is more than twice the observed
topographic range on the Moon in Figure 2.3d. Evidently strength is not the major factor
limiting the Moon’s topography: History must play a role, too.
Applying this model for topography to the smaller bodies of the Solar System, such as
Phobos, this rock strength limitation leads to ridiculous conclusions about the topographic
ranges on these bodies (see Problem 3.1 at the end of the chapter). One might be tempted
simply to give up and look for factors other than strength that limit topography. However,
as we shall see in the next section, a better appreciation of the concept of strength lets us go
considerably farther down the strength limitation path. In particular, we need to appreciate
the laws that govern the strength of broken rock.

3.4 The nature of strength

3.4.1 Rheology: elastic, viscous, plastic, and more

Rheology is the study of the response of materials to applied stress. Although stemming
from roots in prehistory, E. C. Bingham (of whom we will learn much more in Chapter
5) first established it as a scientific discipline in the 1930s. It is not a simple science: Real
materials are complex and so is their detailed description. However, much of this com-
plex behavior can be understood in terms of the properties of a number of simple ideal
materials, which are then compounded to approximate real substances. We have already
described ideal elastic and viscous substances. A third ideal behavior is implicit in the idea
of strength: An ideal plastic substance is one which does not undergo any strain at all until
the strength reaches some limiting value, after which the strain increases to any extent con-
sistent with other constraints on the material. Of course, no real material behaves in this
way, but many materials do not undergo any very large strains until some limiting stress
is reached, after which strain increases rapidly. A slightly more realistic model is to com-
pound elastic behavior with plastic yielding to arrive at an elastic-plastic substance that
responds to applied stress as an ideal elastic material until the stress exceeds some limit,
after which its strain is limited only by system constraints. Then we could add materials
whose elastic strain depends on a non-linear function of stress. We can add time depend-
ence by coupling elastic and viscous behavior. And so on.
This section explores some examples of such compound behavior relevant to understand-
ing planetary topography and its long-term evolution. The first topic we examine is the
ultimate limits to topographic heights, after which we will look at more realistic limits.

3.4.2 Long-term strength

The ultimate strength of atomic matter. A full understanding of the strength of matter was
achieved only in the mid-twentieth century. Despite the triumphs of quantum mechanics
in explaining the bulk properties of matter in the early twentieth century, an explanation of

EBSCOhost - printed on 11/3/2023 9:53 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
3.4 The nature of strength 65

strength came much later. The earliest modern attempt to compute the strength of materi-
als from basic principles was a mitigated disaster: Yakov Frenkel (1894–1952), in 1926
(Frenkel, 1926), constructed a simple model of shear resistance (see Box 3.2 for his deriv-
ation) that relates the ultimate strength, Yultimate, of a material to its shear modulus m:
Yultimate = μ/2π. (3.19)

Box 3.2 The ultimate strength of solids

The first estimate of the theoretical upper limit to the strength of a solid was formulated by
Yakov (a.k.a. Jacov or James) Frenkel (1926). Frenkel started from the fact that atoms in a
crystal lattice are uniformly spaced at the interatomic distance a. When a solid is subjected to
shear strain, each plane of atoms parallel to the direction of the strain shifts a small distance
u with respect to the plane immediately above or below. The net shear strain is thus given
by εs = u/a, and is numerically the same at both the atomic and macroscopic scales (see
Figure B3.2.1). The force resisting this deformation increases as one plane of atoms shifts
over the adjacent plane, because the length of the bonds between each atom and its neighbor
increases. However, when the deformation becomes so large that the atoms of adjacent planes
are midway between lattice sites (that is, at a strain εs equal to ½), the attraction to the next
atom in the adjacent plane equals the attraction from the shifting atom’s previous neighbor and
the resistance to deformation drops to zero. Further deformation brings each atom into closer
proximity to its new neighbor. New bonds form: The atomic plane snaps into a new position,
jumping forward by one atomic step.
The force between adjacent atomic planes of a strained crystal is thus periodic, with a repeat
distance equal to the interatomic spacing. Frenkel assumed that this periodic function would
be the simplest that he could think of: A sine function. He set the force resisting deformation
equal to a constant times sin (2πu/a). Because the maximum value of the sine function is
1 (when u = a/4), the constant equals the ultimate strength of the crystal, YFrenkel. Thus, he
supposed that the shear stress is given by:

 2π u 
σ s = YFrenkel sin  = YFrenkel sin(2πε s ). (B3.2.1)
 a 

To determine the constant, he noted that very small deformations are elastic, and in this
limit σs = μεs. Expanding the sine function for very small arguments yields Frenkel’s relation
for the ultimate strength of a solid in terms of the shear modulus μ,

µ
YFrenkel = . (B3.2.2)
2π

Although defect-free solids such as fine whiskers and carbon microtubules can approach this
limit, Table 3.2 shows that Frenkel’s limit greatly overestimates the strength of real materials,
even for rocks at high confining pressures.
Accurate computation of the actual strength of materials is not yet possible, so that
measurement and empirical estimates are still necessary to determine the strength of a real
substance under conditions of interest to planetary science.

EBSCOhost - printed on 11/3/2023 9:53 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
66 Strength versus gravity

Box 3.2 (cont.)

Force / Yield Stress

2
peak
stress
0 1 3

symmetrical new
–1 position position

0 0.25 0.50 0.75 1

Strain, u/a

1 undeformed 2 small strain 3 large deformation --

bonds reform after
plastic flow

Figure B3.2.1 The theoretical limit to the strength of a solid, based on the model of Yakov
Frenkel. The graph on the top shows the sinusoidal dependence of shear force on shear strain,
indicating that it is a periodic function of lattice displacement. The lower part of the figure shows
the deformation of a lattice at three different strains, correlated with points on the force–strain
plot above by the circled numbers: (1) is the undeformed solid, (2) has been subjected to a small
strain, while (3) indicates a strain so large that the atoms in the solid are again in register with
their neighbors, so that the shear force vanishes.

The shear modulus has been measured for a large variety of materials. It is a bulk
property that can now be computed from first principles for many single crystals.
Although Frenkel’s formula is elegantly simple, it is also grossly inadequate: As shown
in Table 3.2, the actual measured strength of most materials is a factor of 100 or more
smaller than the Frenkel limit. Nevertheless, the Frenkel limit is not wholly wrong or
useless: The strength of a few materials, such as carefully prepared single crystals or
fine carbon fibers, does approach this limit. However, the Frenkel limit clearly does
not capture the factors controlling the strength of the materials we are likely to meet in
planetary interiors.
The principal shortcoming of Frenkel’s strength estimate is its neglect of defects. Rocks
are composed of crystals of individual minerals. While the crystals themselves might be
strong, they are bonded through weaker surface interactions. Most igneous rocks, such as
granite or basalt, have cooled through a large range of temperatures and, because of the
different thermal expansion coefficients of their constituent minerals, tiny grain-boundary
cracks develop in abundance. Sedimentary and metamorphic rocks also contain vast num-
bers of microscopic cracks and weak bonds between individual grains. All rocks contain

EBSCOhost - printed on 11/3/2023 9:53 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
3.4 The nature of strength 67

Table 3.2 Theoretical vs. observed material strength

Yultimate Yobserved
Solid material = μ /2π (GPa)a At p = 1 and 5 (GPa)b

Iron, Fe 13.0 0.11–1.0

Aluminum, Al 4.14 0.10–0.30
Corundum, Al2O3 25.9 0.26–0.92
Periclase, MgO 20.9 0.14–1.07
Quartz (Opal), SiO2 7.08 0.35–1.8
Forsterite, Mg2SiO4 12.9 1.13 (p = 0.5 GPa)c
Calcite, CaCO3 5.09 0.27–0.84
Halite, NaCl 2.34 0.09–0.29
Ice, H2O 0.54 0.20–1.0d

a
Elastic moduli from Bass (1995).
b
At 23°C from Handin (1966) Table 11–9, except as noted.
c
At 24°C Handin (1966), Table 11–3, Dun Mtn., NZ, peridotite.
d
At 77–115 K; extrapolated from Beeman et al. (1988).

macroscopic cracks in the form of joints. In addition to cracks between mineral grains, the
minerals themselves inevitably contain arrays of a peculiar sort of strength-related line
defect called dislocations. First described in the 1950s by engineers studying the creep
elongation of turbine blades in high-temperature jet engines, dislocations flow under
stresses far below the Frenkel limit. It is only by studying the properties and interactions of
entities such as cracks and dislocations that progress has been made in understanding the
practical limitations on the strength of materials.
Although the strength of materials is a large field of endeavor in itself, one too vast to
cover in this book (references for this literature are provided at the end of this chapter), the
basic take-away lesson is that defects rule the macroscopic strength properties of materials.
One cannot expect planetary materials to be stronger than a small fraction of the Frenkel
limit. And, in spite of a half-century of progress in understanding the fundamental basis of
strength, there are so many complex contributing factors that the strength of a particular
material under given conditions of pressure, temperature, and chemical environment is still
best determined by experiment.
Traditional material science focuses on the strength properties of metals. Only recently
have the much more complex problems presented by the strength of ceramics and geologic
materials, such as rocks, become amenable to rational explanation. Naturally, experiment-
ers did not wait for theoreticians to make up models of the strength of rock, so that much
of our present understanding is based upon empirical observations.
Built upon sand: The strength of broken rock. Most experts on asteroids now believe that
all but the very smallest asteroids (bigger than a few tens of meters in diameter) are better
described as fragmented rubble piles than as solid chunks of rock. Unlike solid rock, rubble

EBSCOhost - printed on 11/3/2023 9:53 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
68 Strength versus gravity

piles have no tensile strength. Their entire ability to resist changes in shape depends on the
frictional forces acting across the rock–rock contacts between their components.
Coulomb in 1785 first formulated the laws governing the mechanical behavior of a mass
of broken rock (or a pile of sand). Because the frictional resistance at a rock–rock con-
tact is proportional to the force pushing the rocks together, the strength of a mass of bro-
ken rock is proportional to the pressure. This fact was first clearly stated by Leonardo da
Vinci (1452–1519) in the fifteenth century, but not published by him. Guillaume Amontons
(1663–1705) in 1699 resurrected this relation from da Vinci’s codices. This behavior is in
stark contrast to the strength of ductile metals, such as aluminum or steel, which is nearly
independent of pressure. Many experimental studies of the strength of sand or soil show
that the mass begins to yield when the applied shear stress σs reaches a constant fraction of
the overburden pressure p:
|σ s| = ff p = tan φ f p (3.20)
where ff is the coefficient of friction and φ f is the related angle of internal friction. This
angle is also closely related to φ r, the angle of repose, which is the maximum steepness of a
slope composed of this material (See Section 8.2.1 and Table 8.1 for more on internal fric-
tion). This coefficient is typically about 0.6 for most geologic materials (including water
ice well below its freezing point), making φ f about 30°.
Applying this formula to a model of small-body topographic support, the most obvious
evidence of topography on small bodies is the difference between their longest and shortest
dimensions, Rmax − Rmin (refer back to Figure 3.4) This out of roundness corresponds to a
load of breadth comparable to the mean radius of the body itself, R̅. The stress support-
ing this load is, thus, localized deep within the body. The average pressure in the center
1
of a homogeneous body (ρc = ρ̅) is pctr = ρ g R , so that the strength, Y, or resistance to
2
yield, is Y ≈ ff pctr. Inserting this into the equation for Δh, we find that a small-body model
of strength implies:
Δhsmallbody ≈ ff R.̅ (3.21)
Another way of deriving the same result is to note that a constant coefficient of friction
implies a constant angle of repose, which is nearly equal to the angle of internal friction.
Imagine a hypothetical, maximally out-of-round asteroid constructed in such a way that
every slope on its surface is at the angle of repose in its local gravitational field (such a
shape has now been constructed by Minton, 2008). Although the precise shape is complex,
it is clear that, in traversing the surface of the asteroid from equator to pole, a distance of
(π/2)R̅, up (or down) a constant slope of angle φr, an elevation change of the order of (π/2)R̅
tan φr must take place. This yields essentially the same Δhsmallbody as above.
This small-body topography model predicts that the maximum fractional deviation from
sphericity, (Rmax − Rmin)/R̅, is actually independent of size. This is in strong contrast to the
constant-strength model derived for the Earth, which suggests that, as a body becomes lar-
ger, its shape becomes relatively closer to a spheroid because ( Rmax − Rmin ) / R ∝1 / R , so
2

that the ratio decreases as R̅ increases.

EBSCOhost - printed on 11/3/2023 9:53 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
3.4 The nature of strength 69

10
10 MPa 1 GPa
0.1 GPa
= 30°

(Max-Min)/Mean Radius
1

0.1

0.01

0.001
1 10 100 1000 10000
Mean Radius, km

Figure 3.5 The ratio of the maximum elevation difference to the radius for various Solar System
bodies as a function of diameter. Up to a diameter of about 200 km, this ratio is nearly constant, as
expected for rubble piles supported only by frictional strength. Above this diameter the ratio falls off,
consistent with an ultimate planetary crustal strength of about 0.1 GPa. The solid dots are silicate
bodies and the open circles are icy. The data suggests that icy bodies are weaker than silicate objects
although they have similar friction coefficients.

How do these model predictions fare against reality? Figure 3.5 plots the maximum frac-
tional deviation from sphericity against mean radius for a variety of Solar System objects.
It is clear that the topography of the smaller bodies does, indeed, follow a law that suggests
the dominance of frictional strength. There is no obvious tendency for the fractional topo-
graphic deviation to decrease with increasing size. However, at a radius of about 200 km
the frictional relationship breaks off and the maximum topographic deviations of the larger
planets and moons decrease sharply with increasing diameter, following an approximate
1/R2̅ dependence on the log–log plot. For these large objects greater size does imply greater
smoothness. The trend of the curve for larger planetary objects suggests that the ultimate
strength of planetary crusts is about 0.1 GPa.
The constancy of the maximum fractional deviation for small objects is a direct con-
sequence of the ability of pressure to increase the strength of broken rock materials.
Obviously, however, this frictional increase in strength has its limits. This fact is also clear
from laboratory measurements of rock strength: As shown in Figure 3.6, the frictional
regime holds up to some maximum stress, generally a few GPa, when the intrinsic strength
of the rock is reached and yielding occurs in spite of increasing overburden pressures. As
in the large–planet topography model, it seems that the ultimate limit to topography lies
in the ultimate ability of matter to resist deformation. It is thus worth inquiring just what
determines this resistance.
David Griggs and the strength of rocks. The most obvious feature of the rocks outcrop-
ping on the surface of the Earth is that they are pervaded by fractures at all scales. How
these fractures actually form, however, is much less obvious. It took many years before
experimenters could reproduce the pressures and temperatures existing in the Earth’s

EBSCOhost - printed on 11/3/2023 9:53 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
70 Strength versus gravity

75
Tension Compression

MPa
50

s,
Shear Stress Intact rock
Y = 100 MPa
M
25
Y = 20 MPa
Y Fractured rock
0
0
f = 0.6
F

0
–50 0 50 100 150
–Y /2
0
Pressure p, MPa

Figure 3.6 Yield stress of a typical intact rock specimen (heavy line) described by the Lundborg
strength envelope, Equation (3.23). Note the substantial tensional strength (equal to Y0 / 2 by the
Brace construction, which is, nevertheless, weaker than the extrapolation of the Lundborg strength
envelope, shown by the dotted line, would suggest) indicated on the negative pressure axis. Shown
also as a heavy dashed line is the yield curve for a fractured rock specimen for which the shear
resistance is entirely due to friction.

interior and come to an understanding of how rocks break. Indeed, this is still an active area
of research in the earth sciences. David Griggs (1911–1974) was one of the first people to
systematically investigate rock fracturing under high pressures and temperatures. Griggs’
interest in geologic processes began as a boy, when he accompanied his father, geologist
Robert Griggs, on a National Geographic expedition to study the deposits of the fam-
ous 1912 eruption of Mount Katmai in Alaska (Griggs, 1922). From his experience in
the field, David decided to study how rocks break deep within the Earth. He sought out
Percy Bridgeman at Harvard University and signed on as his graduate student in 1933.
Bridgeman’s laboratory was one of the few places in the world where pressures approach-
ing those deep in the Earth’s crust could be attained.
Griggs eventually perfected an apparatus widely known as a Griggs’rig that could both
compress and heat a small rock sample, typically a cylinder a few centimeters in length and
diameter, while subjecting it to controlled differential stresses. Continuing his work after
World War II at UCLA, and accompanied by a growing number of similarly motivated
experimenters, he showed that, unlike metals, the fracture strength of rock is a strong func-
tion of both pressure and temperature.
It has long been known that metals and alloys, such as iron or steel, fail at similar stresses
under both compression and tension. Ideal plasticity is a useful approximation to metal
failure, in which half the stress difference at failure (equivalent to the shear stress through
a coordinate rotation) is assumed to be a constant Y, the yield stress:

σ1 − σ 3
σs = = Y. (3.22)
2

EBSCOhost - printed on 11/3/2023 9:53 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
3.4 The nature of strength 71

Table 3.3 Lundborg strength parameters for representative rocks

Friction Cohesion, Von Mises plastic

Rock coefficient, ff Y0(MPa) limit, YM, (MPa)

Granite I 2.0 60 970

Granite II 2.5 50 1170
Quartzite 2.0 60 610
Gray slate 1.8 30 570
Black slate 1.0 60 480
Limestone I 1.2 30 870
Limestone II 1.0 20 1020
Sandstone 0.7 20 900

Data from Lundborg (1968).

The yield stress of metals is, to a good approximation, independent of pressure and
strain, although it declines with increasing temperature. Because of its utility in engin-
eering, the theory of failure of ideally plastic materials is highly developed, in spite of
serious mathematical difficulties that stem from this very lack of dependence on strain
(Hill, 1950).
Experimental studies of rock fracture show, however, that the strength of rock depends
very strongly on pressure, at least up to pressures approaching 5 GPa (50 kilobars).
Many analytic representations of the failure strength of rock have been proposed; among
them, one that seems to fit many materials was suggested by Lundborg (1968) for
unfractured rock:
ff p
σ s = Y0 + (3.23)
 ff p 
1+ 
 YM − Y0 

where Y0 is the strength at zero pressure, often called cohesion, and YM is known as the von
Mises plastic limit of the material. YM limits the maximum stress that can be achieved at
arbitrarily high pressure. The Lundborg form of the failure law is illustrated in Figure 3.6
and some representative values of the parameters are listed in Table 3.3.
Although the Lundborg law, and others like it, gives a good description of the failure of
rock over the full range of pressures from very low to very high, much more data has been
collected in the low pressure regime where a linear version is generally adequate. Thus,
when p << YM,
σs Y0 + f f p for p << YM . (3.24)

Table 3.4 lists representative values of Y0 and ff for a small number of materials, ranging
from a hard igneous rock (at crustal temperatures) to weak sedimentary rock.

EBSCOhost - printed on 11/3/2023 9:53 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
72 Strength versus gravity

Table 3.4 Low-pressure failure envelope for representative rocks

Rock Friction coefficient, ff Cohesion, Y0(MPa)

Westerly granite @ 500°C 0.6 50

Pennant sandstone @ 25°C 0.97 35
Limestone @ 25°C 0.75–1.6 3.5–35
Siltstone @ 25°C 0.55 21
Chalk @ 25°C 0.38 0.9

Data from Handin (1966).

The sloping, low-pressure portion of the failure law illustrated in Figure 3.6 is superfi-
cially similar to that of sand. However, in this case the pressure coefficient ff is less obvi-
ously related to friction, although it is often referred to as a coefficient of internal friction,
presumably because it is dimensionless and relates strength linearly to overburden pressure,
as does the true friction coefficient. Numerically, it is also similar to the coefficient of rock-
on-rock friction, although the reader should not confuse the two: ff is the (approximate)
linear slope of the strength envelope that defines the stress conditions under which intact
rock fails, whereas fB is the (static or starting) coefficient of friction of a pre-existing planar
rock fracture sliding over another. The difference between these two curves is responsible
for the brittle–ductile transition that gives rise to discrete faults in rock, as will be discussed
in more detail in Section 4.6.1.
Extensive tables of the strength envelopes of rocks under various conditions can be found
in Handin (1966) and Lockner (1995). The ultimate strength limit of about 0.1 up to 1 GPa
for real rocks is in fair agreement with the observed trend of topographic deviations on the
larger planets illustrated in Figure 3.5. It, thus, appears that we presently have a good first-
order understanding of the strength properties of planetary bodies, although many details
remain to be worked out.
The presence of pre-existing fractures in most large rock masses greatly complicates ana-
lyses of the strength of rock. The actual strength of a large volume of rock generally lies some-
where between that of intact rock and that defined by the coefficient of friction (the dashed
line in Figure 3.6). A constant value of the friction on a pre-existing fracture, fB  0.85 (up to
a mean pressure p of about 100 MPa; the slope is somewhat less at larger pressure) is often
known as Byerlee’s law after the researcher who showed that this value describes the friction
of a wide variety of rock surfaces (Byerlee, 1978). In its exact form Byerlee’s law states:
0.85 σ n σ n < 200 MPa
σs =  (3.25)
50 + 0.6 σ n σ n ≥ 200 MPa

where σn is the normal stress across a fracture, σs is the shear stress and all stresses are in
megapascals.

EBSCOhost - printed on 11/3/2023 9:53 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
3.4 The nature of strength 73

Note that the mean pressure, p, in Equation (3.23) is somewhat confusingly equal to the
negative of either one-half of the sum of the maximum and minimum principal stresses,
or (more correctly, if less frequently seen) to one-third of the sum of all three principal
stresses. A similar equation is often written in which, in the location occupied by the term
p in Equation (3.23), a term for the normal stress acting across the failure plane appears
instead. Byerlee’s law is strictly valid only for this normal stress. The disadvantage of this
formulation is that the failure plane must be known before the equation can be applied.
Thus, for the present goal of defining a strength envelope, a formulation in terms of stress
invariants (pressure and shear stress) is preferable. The wary user of data tables is careful
to make sure which definitions are in use before accepting a given coefficient of internal
friction at face value!
The mean pressure, p, in the Equation (3.23) must be modified by subtracting the pore
fluid pressure, p → p − pf, when the rock is pervaded by a fluid that itself is at some hydro-
static pressure pf. This modification is very important when a fluid such as water or oil on
Earth, or methane on Titan, is present. It was first introduced by Terzaghi (1943) for soils,
and by Hubbert and Rubey (1959) for rocks. Its detailed implications are the subject of a
large literature. It will be discussed further in Section 8.2.1, but suffice it to say now that
high fluid-pore pressures cause substantial weakening of rock through this pressure sub-
traction effect.
The coefficient Y0 in Equation (3.23) is the zero-pressure strength or cohesion.
Mathematically, it is the intercept of the strength envelope with the zero-pressure axis (see
Figure 3.6). Physically, it represents the adhesion of crystals in the rock to one another and
can range from only a few megapascals for weak sedimentary rocks to several tenths of a
gigapascal for intact granite. It is strongly affected by pre-existing cracks in the rock and
drops to zero in a fully fractured rock mass. An extrapolation of this line to negative values
of p intercepts the pressure axis (zero shear stress) at pT = −Y0 /ff. This intercept corresponds
to the tensile strength of the rock. The linear extrapolation yields an overestimate of the
actual yield stress by a factor of two to three: More sophisticated models based on crack
theory (Brace, 1960) give a different, and more accurate, analytic form for tensile stresses
that is indicated by the heavy yield curve on Figure 3.6.
The slope of the failure curve decreases at large values of the average pressure, and the
maximum shear stress that the rock can sustain approaches a constant YM, independent
of pressure. This rollover occurs when the frictional stress of sliding on inter- and intra-
crystalline cracks approaches the intrinsic strength of the individual crystals. A full under-
standing of this process is still under development, but the general outlines are now in fairly
good agreement with observations (Ashby and Sammis, 1990). This change in the depend-
ence of the strength on pressure is known as the brittle–ductile transition, for reasons that
will be discussed in more detail in the next chapter, Section 4.6.1. It occurs at, or near, the
point where the failure curve for fractured rock crosses that for intact rock in Figure 3.6.
The ultimate yield stress YM in Equation (3.23) is, as shown in Figure 3.6, still far
below the Frenkel limit because of intra-crystalline defects such as dislocations. Although
independent of pressure, by definition, it does depend strongly on temperature. There is

EBSCOhost - printed on 11/3/2023 9:53 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
74 Strength versus gravity

no universal law for this temperature dependence, which must be determined empirically,
but it is clear that the strength must vanish at the melting temperature, Tm. Using this hint,
a widely used approximation to the temperature dependence is to multiply both Y0 and YM
by the same factor:
2
 T − Tm 
FT =  (3.26)
 Tm 

which assures that the strength falls to zero as the temperature approaches the melting
point. The exponent in this relation is purely empirical, chosen to fit a large body of data
on both metals and rocks.

3.4.3 Creep: strength cannot endure

David Griggs and the flow of rocks. When David Griggs began his now-classic work in
1933 he was already the veteran of many geologic field excursions and knew from personal
experience that the rocks of the Earth’s crust often show signs of large amounts of deform-
ation without fracture. This fluid-like deformation had long been attributed to the high
pressure and temperature within the inaccessible depths of the Earth, but no one understood
the rates or conditions under which this flow occurred. Griggs began his lifework with a
relatively simple apparatus that measured the slow deformation of rocks under an applied
load as a function of time, initially working at room temperature and pressure (Figure 3.7).
Although he found that most rocks deform elastically only for periods of time less than a
year, he discovered a few that exhibited slow pseudoviscous flow or creep according to a
simple law relating the strain ε and time t:
ε = A + B log t + C t (3.27)
where the constant A represents instantaneous elastic deformation, B a kind of decelerating
creep now often called primary creep, and C is the rate of steady, long-term flow. Although
the primary creep term is important for short-term flow processes, such as the response to
fluctuating tidal stresses or the small strains that accompany planetary reorientation and
spin changes, most geologic interest centers on the third, steady-state term, because it rep-
resents deformation that increases steadily with increasing time, apparently without limit.
In this respect the flow of rocks resembles that of more familiar viscous liquids, such as
honey, motor oil or tar.
Sixty years of subsequent research by Griggs and a large cadre of laboratory geologists
who recognized the importance of this research has shown that the rate of steady-state
creep is a function of stress, temperature, and pressure, as well as rock composition, grain
size, presence or absence of water, trace elements, and a host of other factors. Most creep
experiments can be fit by a formula of the form:
Q*
−
C = εsteady = Ac σ n e RT
(3.28)

EBSCOhost - printed on 11/3/2023 9:53 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
3.4 The nature of strength 75

(a)
strain
indicator

weight

test
l specimen

(b)
fracture
Strain

steady creep

0
Time

Figure 3.7 Schematic representation of a creep experiment on rock, similar to Griggs’ 1933 room-
temperature measurements. (a) The test specimen, of original length l, is mechanically loaded (by a
weight and a lever) while its deflection is measured on a sensitive scale. (b) Schematic creep curve,
showing strain as a function of time after loading. The curve shows three distinct portions after the
initial elastic deflection: A period of decelerating creep, a long period of steady creep and, for lab
specimens, a final acceleration just before rupture.

where Ac is a constant with dimensions (stress)–n time–1, σ is deviatoric stress, n a dimen-

sionless constant, Q* is activation enthalpy (this term incorporates most of the pressure
dependence because Q* = E* + pV*, where p is pressure and E* and V* are constants),
R the gas constant, and T is absolute temperature. The dot over the strain ε, following
Newton’s fluxion notation, indicates differentiation with respect to time.
It is often convenient to express the rate of steady-state creep, Equation (3.28), in terms
of an effective viscosity, even though it depends on the stress level. Adapting the definition
of viscosity, Equation (3.12), the effective viscosity ηeff is defined as:
Q*
σs e RT
ηeff = = . (3.29)
2 εsteady 2 Acσ n −1

EBSCOhost - printed on 11/3/2023 9:53 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
76 Strength versus gravity

This definition of viscosity generalizes Newton’s original definition, which applies to the
case n = 1. It has now become common to refer to the case n = 1 as “Newtonian viscosity”
and to use the term “viscosity” in the broader sense for any value of n, as long as it refers
to a flow law in which the strain rate is a function of stress.
Unlike viscous liquids, the power n relating stress and strain rate is usually larger than
1 for creeping rocks and minerals, justifying the use of the term “pseudoviscous” for this
kind of flow. Doubling the stress on materials such as ice or olivine may cause the creep
rate to increase by a factor of 10, in strong contrast to ideally viscous materials in which
the creep rate only doubles. It is also important to realize that creep rate depends exponen-
tially on the temperature. Although rocks deform very slowly at low temperatures, as the
temperature climbs toward the melting point the creep rate increases rapidly (by as much
as a factor of 10 for each 100°C increase in temperature for many rocks). A useful approxi-
mation is that for most materials, creep rates become important over geologic time periods
(millions of years, which implies ε steady
̇ ≈ 10−13 s−1 or less) when the temperature reaches
one-half the melting temperature, T ~ 1/2Tm. A useful simplification of the temperature
dependence of the creep rate is to absorb the activation energy and melting temperature
into a constant g and express the temperature as the dimensionless ratio T/Tm, the homolo-
gous temperature:
Tm
−g
C = εsteady = Ac σ n e T
. (3.30)

Table 3.5 gives typical values for Ac, n, Q*, Tm, and g for a few materials of geologic and
planetary interest.
Extensive tables, such as that of Kirby and Kronenberg (1987a, b) and Evans and
Kohlstedt (1995), have been compiled to categorize the creep of rocks, and theoretical
models have been developed to explain this flow behavior in terms of diffusion and dis-
location motion (e.g. Evans and Kohlstedt, 1995; Poirier, 1985). However, for the purposes
of this book the principal concept to remember is that at high temperatures rocks can flow
like liquids over geologic timescales.
J. C.Maxwell and the viscosity of “elastic solids.” Observation and experiment have
taught us that cool materials (that is, materials at temperatures well below their melting
point) deform elastically under applied loads, while hot materials gradually flow. Elastic
behavior is mostly recoverable: that is, when the load is removed the deformation reverses
itself; while viscous flow is not recoverable: when the load is removed the deformation
remains. The alert reader might wonder how these very different types of behavior can be
reconciled at intermediate temperatures: At what point does the elastic response stop and
viscous flow take over?
This important question received a definitive answer from an unlikely source. Most
people who recognize the name of nineteenth-century physicist J. C. Maxwell (1831–1879)
think immediately of Maxwell’s equations that describe electric and magnetic fields, or
perhaps of his contributions to thermodynamics and statistical mechanics. In fact, it was
during his 1867 study of the viscosity of gases that Maxwell faced the puzzling dichotomy

EBSCOhost - printed on 11/3/2023 9:53 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
3.4 The nature of strength 77

Table 3.5 Creep properties of selected materials

Ac Q* Tm g
Material (MPa–n s) n (kJ/mol) (K) = Q/RTm

Olivine 2200 27
Dry 1.2 × 102 3.0 502
Wet 2.0 × 103 3.0 420
Diabase 1100 53
Drya 5.4–347 4.7 485
Wetb 6 × 10 –2 3.05 276
Quartz 1996 8.1
Dry 1.3 × 10 –6 2.7 134
Wet 2.0 × 10 –2 1.8 167
Granite (Westerly) 1320 12.7
Dryc 2.5 × 10 –9 3.4 139
Wetc 2.0 × 10 –4 1.9 137
Anorthosite 3.2 × 10 –4 3.2 238 1400 20.5
Halite, NaCl 6.3 5.3 102 1074 11.4
Water ice, Ih,d T > 258 K 6.3 × 1028 4 181 273 80
σ > 1 MPa
Solid CO2, 150 < T < 190e 4.4 × 103 4.5 31 217 17
Limestone, Dry, Solenhofen ls 2.5 × 103 4.7 298 1520 23.6

Data is from Evans and Kohlstedt (1995), except as noted:

a
Mackwell et al. (1998)
b
Caristan (1982)
c
Kirby and Kronenberg (1987b)
d
Durham and Stern (2001)
e
Durham et al. (1999)

between the elastic and viscous behavior of solids (Maxwell, 1867). His insight came from
what might seem like an annoying detail: The steel wire supporting the torsion pendulum
he was using to measure gas viscosity exhibited viscous behavior of its own. He invented
a theory of what are now known as viscoelastic materials to separate the viscosity of the
pendulum wire from that of the gas.
Maxwell proceeded by postulating that the total deformation of his wire is the simple
sum of the elastic plus the viscous strain, εtotal = εelastic + εviscous. He supposed that each strain
would develop under the influence of the same stress, obeying the equations previously
stated for ideal elastic and viscous behavior. His equation, however, suffers a serious math-
ematical problem, because the viscous strain is not determined directly from the stress:
The stress determines only the strain rate. It is possible to write the viscous strain as the
time integral of the strain rate, but it is more straightforward to differentiate both sides of
Equation (3.10) with respect to time and sum the result to obtain the fundamental equation

EBSCOhost - printed on 11/3/2023 9:53 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
78 Strength versus gravity

for a Maxwell viscoelastic substance, ε total

̇ = ε ̇elastic + ε ̇ viscous. Inserting the definitions of
each term:
ε total
̇ = σ̇ /2μ + σ/2η. (3.31)
This equation embodies both an elastic response for loads applied quickly and viscous
flow for long sustained loads. Its full solution is complex because volume strain and shear
strain must be treated differently in each term of the full tensor equation. However, it is not
necessary to actually solve this equation to attain an insight of major importance. Simple
dimensional analysis shows that the ratio of the viscosity η to the elastic shear modulus μ
has the dimensions of time. This ratio is known as the Maxwell timeτ M and it plays a funda-
mental role in the transition from elastic to viscous behavior. Its definition is:
η
τM ≡ . (3.32)
µ
If a load is applied instantaneously to a Maxwell viscoelastic material, then held constant,
the Maxwell time is equal to the length of time that passes before the accumulated viscous
strain equals the instantaneous elastic strain. Thus, for times shorter than the Maxwell time,
the material response is dominated by the elastic deformation. For times longer than the
Maxwell time, the response is essentially viscous. Maxwell supposed that even water must
act as an elastic material on a short enough timescale, but he computed this time as about
10–13 s – unobservably small in the late 1800s. However, he did later succeed in observing
both elastic and viscous behavior in Canada balsam (pine tree sap).
Although Equation (3.32) was derived from the equations for ideal elastic and viscous
substances, a generalization of the idea of Maxwell time can be applied even to pseudovis-
cous materials that do not obey the equation of ideal viscosity: The generalized Maxwell
time is the length of time over which creep must act for the total creep strain to equal the
elastic strain. In the form of an equation:
(elastic strain ) ε
τM = = elastic . (3.33)
(creep strain rate) εcreep

The Maxwell time is often surprisingly short. This is because the elastic strain in most
geologic materials is invisibly small – typically only about 0.0001, even for stresses near
fracture. For this reason Australian geologist S. Warren Carey invented a term,which he
called rheidity (Carey, 1953), and for which he proposed a timescale of exactly 1000 τ M.
Although this rheidity concept adds nothing fundamental to the idea of Maxwell time, it
does give an estimate of the time necessary for viscous or pseudoviscous flow to become
visible to the human eye.
Most children are familiar with the high-polymer material known as Silly PuttyTM, which
behaves as a brittle elastic material on a short timescale – it can be fractured by a hammer
blow – but flows like a liquid when left undisturbed for a long period. It is less widely
appreciated that all materials behave this way, if only the timescale is chosen appropriately.
Water ice is another example: ice cubes in common experience are brittle elastic materials,

EBSCOhost - printed on 11/3/2023 9:53 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
3.4 The nature of strength 79

Table 3.6 Maxwell time and rheidity time for various materials

Shear modulus, Viscosity, Maxwell time, Rheidity time,

Material μ (GPa) η (Pa–s) τM τR

Soda-lime glass @ 250°C 25 4.3 × 1011 17 s 4.8 hr

Glacier ice @ 0°C 4 ~1013 42 min 29 days
Halite @ 200°C 20 3 × 1016 17 days 48 yr
Earth mantle from glacial 50 1020 66 yr 66 000 yr
rebound

but it is obvious from glaciers that ice flows like a liquid over long timescales. Table 3.6
lists the Maxwell and rheidity times for a number of geologic materials.
Maxwell viscoelasticity neatly resolves other apparent paradoxes of earth science.
William Thomson, later Lord Kelvin, used the difference between solid Earth tides and
ocean tides to show that the Earth’s elastic modulus is similar to that of steel. Kelvin him-
self, and Harold Jeffreys after him, never accepted the idea that over long intervals of time
the Earth’s mantle could flow like a liquid (England et al., 2007). However, our modern
understanding of mantle convection and plate tectonics requires just that. The resolution of
this conundrum is through Maxwell viscoelasticity: Table 3.6 shows that the Earth’s mantle
(which is mainly composed of the mineral olivine) has a Maxwell time of about 100 years.
Thus, the mantle behaves as an elastic solid with respect to the month-long tidal deform-
ation (or even the 22-month Chandler wobble of its axis), and yet flows like a liquid during
the 100 Myr timescale of mantle convection.
Ironically, Lord Kelvin himself provided one of the most graphic illustrations of the
role of viscoelastic flow in the Earth and other planets. Kelvin loved mechanical models,
often stating that he could “never satisfy myself until I can make a mechanical model of a
thing” (Kargon and Achinstein, 1987). In his famous Baltimore Lectures of 1884, Kelvin
described a classroom model in which he floated a layer of “Scottish shoemaker’s wax” on
a beaker of water. He submerged a number of corks underneath the wax and set a few lead
bullets on top (Figure 3.8). Over the course of a semester, the bullets sank into the visco-
elastic wax while the corks burrowed upward into it. By the semester’s end, the bullets had
dropped to the bottom of the beaker and the corks had emerged on top. While he could not
have found a better analogy for the geologic behavior of the Earth, Kelvin himself used
this model to illustrate his concept of the hypothetical aether, to show how the Earth could
move through the all-pervading aether apparently without friction, while light waves trav-
eled like elastic waves in this universal substance.
Maxwell’s model of viscoelastic flow turns out to be only one of many possible vari-
ations. Depending upon how the elastic and viscous strains combine (coupled, more
generally, with the possibility of plastic flow), a variety of viscoelastic (or elasto-visco-
plastic) responses to stress are possible. Kelvin himself proposed a model in which elastic
and viscous stresses are summed and the strains are then set equal. Now known as the

EBSCOhost - printed on 11/3/2023 9:53 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
80 Strength versus gravity

Figure 3.8 Lord Kelvin’s class demonstration. Over the course of a semester, Kelvin showed that
dense bullets would sink and light corks would rise through a layer of viscoelastic wax on top of a
beaker of water. Although Kelvin himself did not intend it as such, it provides an apt illustration of
the long-term flow properties of a planetary mantle.

Kelvin–Voigt model, it provides a better description of short-term flow, such as primary

creep (the B term in Equation (3.27)), than does the Maxwell model (the C term in Equation
(3.27)). The Kelvin–Voigt model may provide a good description of flow under short-term
oscillatory stresses, such as tidal flexing, while the Maxwell model is more appropriate for
steady, long-term deformation in which the total strain can increase without limit.

3.4.4 Planetary strength profiles

It should now be clear to the reader that the “strength” available to support topographical
features on a planet is a complex issue. Topographic loads can be supported by the resist-
ance to deformation exerted by cold solids, as well as by slow viscous or pseudoviscous
deformation of warm solid materials. “Strength,” thus, depends on pressure, temperature,
and the duration of the load, among many other modifying factors such as the pressure of
included fluids, presence or absence of chemical weakening agents such as water, and even
the history of previous deformation.
Given this complex response of “solid” materials to differential stresses, can one make
any simple generalizations at all about the ability of planets to support topographical fea-
tures? One simple observation is that strength generally decreases as temperature rises (and
vanishes at the melting temperature). Most planets are warmer inside than on their out-
sides, although exceptions to even this apparently obvious situation occur during planetary
accumulation and very large impact events. Thus, a simple generalization is that most of a
planet’s strength resides near its surface. This observation gives rise to the idea of a litho-
sphere, a relatively thin shell near the surface of planets large enough to have hot interiors,
which embodies most of its long-term strength. The outermost part of the lithosphere is
usually cool enough to exhibit brittle strength, while deeper portions resist loads by slow
deformation (this definition of the lithosphere is oversimplified: It will be made more pre-
cise in the next chapter when the concept of Maxwell time is applied).
These ideas are used to construct strength profiles; envelopes that show the maximum
differential stresses that can be supported as a function of depth in any given planet. Besides

EBSCOhost - printed on 11/3/2023 9:53 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
3.4 The nature of strength 81

Venus Earth
(b) 0
(a) 0

100 100

Depth, km
Depth, km

200 200

300 300
0 0.25 0.5 0.75 0 0.25 0.5 0.75
Shear Strength, GPa Shear Strength, GPa

Moon Mars
(c) 0 (d) 0

100 100
Depth, km

Depth, km
200 200

300 300
0 0.25 0.5 0.75 0 0.25 0.5 0.75
Shear Strength, GPa Shear Strength, GPa

Figure 3.9 Strength profiles for the lithospheres of (a) Venus, (b) the Earth, (c) the Moon, and (d)
Mars. The upper parts of the curve are controlled by friction on pre-existing fractures and, thus, follow
Byerlee’s law, Equation (3.25). The lower portions are cut off by creep in olivine, with parameters
listed in Table 3.5. Temperatures are computed from mantle heat flow on the Earth and by assuming
an average chondritic composition for the other planets. Thermal conductivity is taken to be 3.0
W/m-K. The strain rate is 10–13 s–1, although the curves are only slightly different at 10–15 s–1.

the failure laws themselves, these profiles require knowledge (or estimates) of the tempera-
ture and pressure as a function of depth. Because many factors influence strength, these
curves are oversimplifications of the actual facts, and are useful only for general guidance
to the strength levels available. There are also many variants of this kind of curve, each
designed to show some especially pertinent relationship. To show the effect of strain rate,
the strength profiles constructed here are assumed to reflect deformation at some particular
strain rate. The cohesive strength of cold rocks is neglected because it is assumed that
beyond some small strain the rock will fracture, so that only frictional strength continues
to act.
Figure 3.9 shows computed strength profiles for Venus, the Earth, the Moon, and Mars.
In all of these curves the upper cold portion is assumed to follow Byerlee’s law, while
the lower portion is controlled by the rheological properties of the common mantle min-
eral, olivine. Each assumes that the lithosphere is stretched at a strain rate of 10–13 s–1, a
typical plate-tectonic strain rate on the Earth. Very similar curves would result for litho-
spheric compression, with slightly higher frictional strengths. Lower strain rates decrease

EBSCOhost - printed on 11/3/2023 9:53 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
82 Strength versus gravity

the stresses in the lower part of the lithosphere and push the cusp marking the transition
between friction and pseudoviscous flow to shallower depths. The sharpness of the cusp is
artificial: In reality the transition is probably gradual, but the flow laws are not known well
enough to represent this accurately.
The main lesson from these curves is that the maximum strength in a planet’s interior
resides neither at its surface, due to the pressure dependence of rock friction, nor at great
depths, due to the weakening effect of high temperatures. The maximum strength is at an
intermediate depth, and it is at this depth that most of the forces that support long-term
topography are exerted.

3.5 Mechanisms of topographic support

3.5.1 Plastic strength: Jeffreys’ limit again

Short-wavelength loads on a planetary lithosphere are supported by plastic strength, as
described in Section 3.3.2. Stress differences reach approximately 1/3 of the vertical load
and are supported at a depth comparable to the width of the load. The meaning of “short
wavelength” is defined by reference to the thickness of the lithosphere. If the breadth of
the load is comparable to or larger than the lithosphere’s own thickness, then new factors
come into play and more sophisticated models, such as the flexural models discussed later,
in Section 3.5.5, must be brought into play. These new factors generally decrease the ability
of the lithosphere to support the load: Jeffreys’ theorem must always hold, but it does not
guarantee that the stresses are not much larger than the minimum given by his limit. Direct
support of a load by a strong material right underneath is always the most effective way to
carry the weight of a topographic feature.

3.5.2 Viscous relaxation of topography

Just as a mound created on the surface of a dish of honey gradually relaxes to a flat sur-
face, so topographic features formed on the surface of a planet whose interior materials
obey a viscous or pseudo-viscous flow law will eventually relax to a flat plain. Because of
the complexity of the full non-linear pseudoviscous creep law determined for real rocks,
most analyses of the viscous relaxation of topography approximate the actual flow law as
Newtonian (at present, numerical methods are rapidly superseding such crude approxima-
tions, but there is still much to be learned from “back of the envelope” computations using
Newtonian viscosity). The viscosity determined from such an analysis is then termed an
“effective viscosity,” ηeff, and its value must be accompanied by an estimate of the stress at
which it is determined. Although such a procedure is not exact, and in some special cases
may be seriously misleading, it often yields useful insights into the mechanical behavior of
a planetary body, so long as the user understands what the effective viscosity really is, and
does not mistake it for what it is not.

EBSCOhost - printed on 11/3/2023 9:53 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
3.5 Mechanisms of topographic support 83

Viscous relaxation acts to gradually erase any deviation from a “level” planetary sur-
face (that is, from a surface coinciding with a gravitational equipotential surface). Thus,
both elevations and depressions will gradually fade away with time. How much time this
requires depends on the viscosity. If the viscosity is large enough, even a few billion years
is not enough to erase the topography and we can speak of the surface elevations as “per-
manent,” even though, in principle, there is no such thing as a solid and all materials even-
tually creep to relax their deviatoric stresses.
The first estimates of the Earth’s viscosity derived from the early nineteenth-century
observation by Swedish naturalist Celsius that some shorelines around the Baltic Sea are
rising as rapidly as one meter per century. Hotly contested at the time, it is now accepted
that central Scandinavia, formerly depressed by the weight of continental ice sheets, is
gradually rebounding to its pre-ice age position. A still larger area in North America is cur-
rently rebounding from the former weight of the Laurentide ice sheets, which melted away
about 11 000 yr ago. Although detailed analyses of the implications of this uplift have been
ongoing for the past 60 yr, it is easy to perform a first-order estimate of the viscosity of the
Earth’s interior that gives a value for its effective viscosity close to the most sophisticated
modern determinations.
Following in the spirit of Jeffreys’ computation in Box 3.1, it is possible to balance the
stress created by a depression (or elevation: the analysis is identical except for the sign)
of time-dependent depth h(t) against the rate of deformation implied by the ideal viscous
stress relation. Jeffreys’ theorem tells us that this stress difference is of order 0.3ρgh. The
strain rate ε ̇s is of order h ̇/w, where w is the breadth of the depression. Inserting these fac-
tors into the definition of viscosity, Equation (3.12), yields a first-order differential equa-
tion for h(t), h ̇(t) = −[0.3 ρghw/ηeff]h(t), whose solution is:
h(t ) = h0 exp[ − t / τ R ]
ηeff
where τ R = . (3.34)
0.3 ρ gw

In this equation h0 is the initial depression due to the weight of the ice and τ R is the time-
scale for relaxation. It is this relaxation timescale that yields an estimate of the effective
viscosity, which is thus given by:
ηeff = (0.3 ρgw) τR. (3.35)
As one might intuitively expect, the relaxation time grows longer as the viscosity
increases, and it decreases as the crustal density or gravity increases. Perhaps the least
intuitive result is that the relaxation time depends inversely on the width of the load w. A
physically intuitive way of appreciating this result is to realize that as w increases, the depth
over which the flow occurs also increases. For a given pressure gradient, the flow is faster
in a wider channel, leading to a faster relaxation rate. Thus, for a given viscosity, broad
loads relax faster than narrow ones. The important implications of this result will shortly
be highlighted in more detail.

EBSCOhost - printed on 11/3/2023 9:53 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
84 Strength versus gravity

In addition to the viscous half-space assumed in the relaxation computation just outlined,
a second important limit is that of a thin viscous channel underlying the load. Following
through a derivation similar to that above yields an equation similar to (3.34), except that
the inverse load width 1/w is replaced by w2/d3, where d is the depth of the thin channel
(this derivation requires the equation for the parabolic velocity profile driven by a pressure
gradient in a thin layer, a topic of so-called lubrication theory). In this case the relaxation
time is proportional to the load width, squared. A more general analysis of the relaxation
of an axisymmetric crater of arbitrary profile on a substrate whose viscosity is a more com-
plex function of depth can be found in Section 8.4 of Melosh (1989).
Performing an actual estimate of the viscosity beneath the Canadian shield, take 2700
kg/m3 as the average crustal density, 9.8 m/s2 as the acceleration of gravity, suppose the
load is 3000 km across and that it relaxes over a timescale of 6000 yr. This yields an order-
of-magnitude viscosity estimate of 5 x 1021 Pa-s, nearly identical to the current best esti-
mate for the Earth’s lower mantle. To interpret this estimate, remember that most of the
stress generated by a broad load is supported at a depth of about 1/3 the load width; that is,
about 1000 km deep in this case, or near the top of the Earth’s lower mantle. Furthermore,
this is an effective viscosity that applies to a stress level of about 0.3 ρgh0 or around
20 MPa, assuming that h0 was about 2 km.
Although the idea of using the duration of topographic support to estimate planetary
viscosity was first applied to the Earth, planetary geologists were quick to apply this idea
to the planets. Ralph Baldwin, in his epochal 1963 book, The Measure of the Moon, made
the first estimates of the Moon’s viscosity based on the persistence of its non-hydrostatic
tidal bulge and on the depths of lunar basins. In 1967 Ron F. Scott, a soil mechanics
engineer at Caltech, was inspired to show how lunar surface viscosities could be estimated
from the shape of relaxed lunar craters. He created a number of model crater shapes in a
pan of viscous tar and allowed them to relax, recording how their shapes changed with
time. Three of his time steps are shown in Figure 3.10. The most prominent characteristic
of these changes is the dependence of relaxation rate on the size scale of the feature. Thus,
large craters relax faster than small ones, so long as the viscous substrate is deeper than
the diameter of the crater. Furthermore, the small-scale crater rims persist long after the
larger-scale crater bowls have relaxed, just as Equation (3.34) suggests (other factors, such
as the presence of a shallow lithosphere, may account for the persistence of crater rims on
real planets, as opposed to craters in pans of uniform-viscosity tar).
Although Scott and others thus showed how viscous relaxation affects crater morph-
ology, it has not yet been conclusively demonstrated that viscous relaxation has actually
occurred in craters on any of the terrestrial planets or moons. Processes such as impact ero-
sion or lava infilling often obscure any depth changes caused by viscous flow. The absence
of relaxation does give useful lower limits to the viscosity, but this does not constitute a
numerical measurement. However, the icy moons of the outer Solar System tell a different
story. Figure 3.11 shows a 500 km wide crater, Odysseus, on the Saturnian satellite Tethys,
contrasted with an unrelaxed 130 km wide crater, Herschel, on Mimas. Odysseus’ floor has
clearly relaxed to conform to the equipotential surface of the satellite, while its still-sharp

EBSCOhost - printed on 11/3/2023 9:53 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
(a) (b)

(c)

Figure 3.10 Viscous relaxation of model craters produced in asphalt of viscosity about 105 Pa-s.
The largest crater is about 10 cm in diameter and the smaller craters about 2 and 0.2 cm. (a) 0.1
minute after the craters were molded into the surface. (b) After 30 minutes the larger crater floor has
rebounded and the middle-sized crater floor is beginning to rise. All of the crater rims are still sharp.
(c) After 18 hours the large and middle crater and their rims have relaxed, while the smallest crater is
still evident. Image selection from Scott (1967).

EBSCOhost - printed on 11/3/2023 9:53 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
86 Strength versus gravity

(a) (b)

Figure 3.11 Large craters on moons of Saturn (a) The floor of the 500 km diameter crater Odysseus
on Tethys has mostly relaxed to conform with the spherical shape of the satellite. NASA Cassini
image PIA 08400. (b) The 130 km diameter crater Herschel on Mimas shows little sign of viscous
relaxation. NASA Cassini image PIA 12570.

rim and central peak attest to the size-dependence of viscosity. If the age of the crater were
known, these two observations would produce a tight bound on the viscosity of the moon,
and by implication (from laboratory measurements of the flow law of ice) give an estimate
of the internal temperature of Tethys. Much effort by planetary scientists is currently being
expended on viscosity estimates of this kind, with the ultimate goal of estimating internal
temperatures and even temperature gradients.
An important, but often overlooked, point refers back to Jeffreys’ theorem: Topographic
loads are typically supported at a depth comparable to the width of the load. Thus, the vis-
cosity deduced from the relaxation of a feature of breadth w applies to a depth comparable
to w itself (or, slightly better, about w/3, as indicated in Table 3.1). The fact that a narrow
crater rim relaxes more slowly than the crater bowl itself is, thus, due to both the scale
dependence of the relaxation time in Equation (3.34) and also to a possibly different (gen-
erally larger) viscosity at shallower depths below the narrow rim. In effect, topography of
breadth w “probes” the viscosity at a depth of about w/3. When sufficiently detailed data on
crater relaxation profiles are available this effect can be used to invert for the depth depend-
ence of viscosity and, by inference from creep measurements on the (presumed known)
underlying material, for the subsurface temperature gradient.
One of the major surprises of the past few decades is the existence of substantial top-
ography on the planet Venus. Shortly after its 730 K surface temperature was discovered,
but before its surface had been imaged by spacecraft-borne radar systems, material science
expert J. Weertman (1979) predicted that any mountains on Venus would have long since
relaxed away and that its surface must be a vast, gently undulating plain. On Earth, the
temperature contour that defines the bottom of the elastic oceanic lithosphere is similar to

EBSCOhost - printed on 11/3/2023 9:53 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
3.5 Mechanisms of topographic support 87

Venus’ surface temperature, so this prediction seemed very reasonable. The discovery of
large topographic variations on Venus, first by the Soviet Venera 15 and 16 radar missions
and then by the US Magellan mission, was thus greeted with consternation. To this day we
do not fully understand why the crust and upper mantle of Venus are so strong. The most
common assumption is that high temperatures have cooked all of the water out of its near-
surface rocks, thus eliminating the major weakening agent affecting terrestrial and Martian
rocks. However, even the total elimination of water and the assumption of a low thermal
gradient can barely explain the existence of the 13 km high Maxwell Montes, the highest
elevation on Venus.

3.5.3 The topographic advantages of density differences: isostatic support

Most of the long-wavelength topography on the silicate planets and moons is a direct con-
sequence of the difference in density between a crust and underlying mantle. Where no
density differences exist, such as on the icy moons of Jupiter or Saturn, elevation differ-
ences tend to be of short wavelength. The resulting concept of isostasy has long been a
staple of geological explanation on Earth. It has found broad application to the Earth-like
planets. The basic idea of isostatic equilibrium is that high topography is high because it is
underlain by rocks that are less dense than average. Elevation correlates with either density
itself (Pratt isostasy) or with the thickness of a layer of lesser density (Airy isostasy). The
crust is supposed to be in floating equilibrium, so that at some depth below the surface (the
depth of isostatic compensation) the pressure of the overlying rock layers is the same along
an equipotential surface.
A key component of the idea of isostatic equilibrium is that at the depth of isostatic
compensation, deviatoric stresses vanish and pressure is the only force available. This
concept accords well with the observational facts indicating that as the temperature rises,
rock strength declines and creep rates increase. Initial stresses, even those applied nearly
instantaneously by, say, the formation of an impact crater, relax rapidly on a geologic time-
scale and bring topography into a state of isostatic equilibrium. This idea puts a premium
on determining the depth of this level of compensation. If its depth can be determined, for
example, using the methods discussed in the next section, this information can be con-
verted to an estimate of the planet’s interior temperature.
Geodesist Colonel George Everest accidentally initiated the discovery of isostasy in
1847, while he was triangulating the “Great Arc” in India. As he approached the massive
Himalayan mountains he found that he could not get good agreement between his trian-
gulated positions and astronomical measurements of latitude. J. H. Pratt, archdeacon of
Calcutta, who was familiar with Newton’s law of universal attraction, suggested that the
mountains deflected the vertical, although the observed deflection was much less than what
he first calculated. Pratt then supposed that the rocks underlying the Himalayas might be less
dense than those underlying the Indian peninsula. Pratt announced his conclusions in 1855,
the same year that G. B. Airy, the Astronomer Royal of Great Britain, suggested that varia-
tions in the thickness of a low-density crust floating on a denser substratum could account

EBSCOhost - printed on 11/3/2023 9:53 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
88 Strength versus gravity

for Everest’s observations. Four years later Pratt published his own theory of isostasy in
which he attributed variations in the elevation of surface features to lateral variations in the
density of the crust above a level of “compensation,” at which the density is uniform.
Geodetic observations in the late 1800s could not discriminate between the Pratt and Airy
models. In the early twentieth century the US Coast and Geodetic survey officially adopted
the Pratt model because of its computational simplicity, but when the developing field of
seismology revealed deep roots beneath the Alps and Himalayas, the weight of opinion
swung in favor of Airy isostasy for most of the twentieth century. Most recently, however,
it has been shown that Pratt isostasy dominates California’s southern Sierra Nevada. The
elevation of the western US’s Colorado Plateau now appears to be due to low densities in
the mantle, not the crust. Furthermore, precise gravity measurements from the Magellan
spacecraft have shown that the Pratt mechanism, with the low densities supplied by some
combination of high temperature and a low-density mantle residuum, may support the vol-
canic uplands of Venus (Smrekar et al., 1997). Evidently, the Pratt and Airy mechanisms
are end members of a continuum and the determination of crustal and mantle density and
thickness must be pursued independently, insofar as that is possible.
The application of the idea of isostasy to planetary topography is simple, which is part of
its appeal. Figures 3.12a and 3.12b illustrate the idea of isostatic balance between two crustal
columns in both the Pratt and Airy limits. For simplicity, these examples assume constant
densities for both the crust and mantle, but it is easy to generalize these examples by inte-
grating a depth-dependent density from the surface down to the depth of compensation.
For the Pratt hypothesis, Figure 3.12a, the pressure at the depth of compensation, dc, be-
neath the plains and highlands crustal blocks is given by:
[ρcpt + ρm(dc − t)]g = [ρch (t + hP) + ρm (dc − t)]g (3.36)
The contribution from the depth of compensation, as well as the mantle density and the
acceleration of gravity all cancel out, so that the topographic elevation on the Pratt hypoth-
esis, hP, is given in terms of the crustal thickness below the plains, t, by:

 ρcp − ρch 
hP =  t. (3.37)
 ρch 

For the Airy hypothesis, Figure 3.12b, the density of the crust ρc is the same in both
crustal columns, but the thicker crust sinks with respect to the thinner crust and produces
a root beneath the highlands of thickness tR. Performing the same type of pressure balance
as for the Pratt case,
[ρct + ρm(dc − t)]g = [ρc(t + hA + tR) + ρm(dc − t − tR)]g. (3.38)
Again benefiting from many cancellations, including the crustal thickness itself, the final
expression for the elevation in terms of the depth of the root is:

 ρ − ρc 
hA =  m tR . (3.39)
 ρc 

EBSCOhost - printed on 11/3/2023 9:53 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
3.5 Mechanisms of topographic support 89

a) b)

plains highlands plains highlands

hP hA
t + hP t + h A + tR
ρ ch
ρ cp t ρc t
dc dc ρc
ρm dc – t tR
d c – t – tR

Figure 3.12 Isostatic compensation of topography is possible where a low-density crust overlies
a higher-density mantle. (a) Pratt isostatic compensation, in which highlands are underlain by less
dense crustal material than lowland plains. (b) Airy isostasy, in which the crust is the same density
everywhere, but is thicker under highlands than plains. The dimensions defined in this figure are used
in equations described in the text. The horizontal dashed line near the bottom of both figures is the
depth of compensation, below which no stress differences are postulated to exist.

Because of these cancellations, the depth of isostatic compensation does not contribute
directly to the topography for either Pratt or Airy isostasy. This is both a blessing and an
annoyance: Insofar as the depth of compensation cannot be directly measured, we lose an
important piece of information about the flow in a planet’s interior. We must thus resort to
indirect methods to learn about internal temperatures.
Isostatic equilibrium, although often assumed to be the final state of topographic
relaxation in much of the geological literature, is not, in fact, the most stable end state: The
minimum energy of any self-gravitating body is attained only when the density decreases
monotonically outwards from the center of mass. Thus, even when the topography is fully
compensated there is a tendency for low-density material to spread over adjacent denser
rocks, and for mountain roots to spread laterally. Stress differences, thus, still exist in a
state of full isostatic compensation. Calculation of these stress differences and their impli-
cations for flow in the crust of the Earth is now a part of terrestrial geodynamics (Sonder
and Jones, 1999).
Although isostasy does contribute greatly to topographic support, it cannot evade the
principal part of Jeffreys’ theorem: Stresses of order ρgh must still develop somewhere
beneath the load. The advantage of the isostatic support mechanism is that it shifts the lo-
cation of the maximum stress differences from a depth comparable to the width of the topo-
graphic load to the region above the depth of compensation, where the rocks are cooler,
stronger and, thus, more capable of bearing the load. Vertical topographic loads are con-
verted into horizontal loads acting near density or thickness gradients. It is even possible to
convert topography into detailed horizontal stress maps, assuming that isostasy is strictly
valid (Artyushkov, 1973; Fleitout and Froidevaux, 1982; Molnar and Lyon-Caen, 1988).
Thus, planet-wide elevation differences, such as the hemispheric dichotomy of Mars or the

EBSCOhost - printed on 11/3/2023 9:53 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
90 Strength versus gravity

center of mass-center of figure offset of the Moon can be supported by modest stresses in
the cooler, stronger outer layers of the planet, rather than by long-term strength at great
depths.

3.5.4 Dynamic topography

Slow viscous or pseudoviscous deformation can do more than just eliminate pre-existing
topographic features. Stresses generated by forced flow can actually create topography.
When this occurs, the resulting elevations and depressions are referred to as dynamic top-
ography. Slow flows in the interior of planets may be generated by a number of proc-
esses, but the most common are driven by density differences, where the density deviations
from the mean are due either to temperature differences (the process is then referred to as
thermal convection) or to compositional differences (compositional convection). The slow
convective flows that drive plate tectonics in the Earth are a combination of both types of
difference.
The most striking dynamic topographic features on the Earth are the deep submarine
trenches that mark the sites of subduction zones. At subduction zones the cold, dense, and
relatively stiff tectonic plates sink into the warmer, less dense mantle at rates up to about
10 cm/yr. As the plates sink, they undergo a sharp bend, changing their attitude from nearly
level to plunging at angles that may exceed 45°. The cool, highly viscous material of the
plates thus undergoes a large amount of stretching on the upper part of the bend, coun-
terbalanced by compression at depth. This stretching creates stresses that literally suck
the overlying surface downward, resulting in the observed topographic troughs. The depth
of the trough is readily estimated from the definition of the effective viscosity, Equation
(3.29), along with Jeffreys’ theorem:

 η   η  v
hdynamic =  eff  ε =  eff  (3.40)
 0.3 ρ g   0.3 ρ g  w

where v is the velocity of motion (subduction, in this case) and w is the distance scale over
which bending occurs.
Unfortunately, we cannot accurately determine the effective viscosity from first princi-
ples, but we can invert the formula and determine how large it must be to give the observed
ca. 5 km of trench depth as the plate bends through a radius w ~ 200 km. The result, about
2 x 1021 Pa-s at a stress of about 50 MPa, is at least reasonable – it is almost two orders
of magnitude greater than that of the underlying asthenosphere and in moderately good
agreement with extrapolated laboratory measurements of the creep rate of olivine at this
stress level.
One of the major complications in making this kind of estimate precise for subduction
zones is that the much stiffer brittle-elastic plate that tops the tectonic plates interferes with
the viscous flow deeper within the plate. Indeed, many models for subduction zones focus
exclusively on the elastic plate and neglect viscous flow entirely. Such models, which were

EBSCOhost - printed on 11/3/2023 9:53 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
3.5 Mechanisms of topographic support 91

among the earliest explanations of subduction zone topography, suffer from the predic-
tion of enormous extensional stresses in the strongly bent elastic plate (up to 5 GPa, far in
excess of any measured rock strength) and neglect seismic data that indicate that the elastic
plate is extensively fractured and, thus, unlikely to support any extensional loads at all.
Nevertheless, this kind of elastic-viscous coupling is common in planetary tectonics and
will be discussed in more detail in the next chapter.
The other likely source of dynamic topography on the Earth (with its unique plate tec-
tonics) and the other planets is associated with rising (or descending) convective plumes.
Arising from deep within a planetary interior, buoyant plumes approach the surface and
exert viscous stresses on the overlying cool rock layers. These stresses account for a sub-
stantial portion of the uplift associated with the plume’s arrival (the rest is associated with
the plume’s low density). Equation (3.40) can also be used to estimate the dynamic portion
of the topography associated with a plume of horizontal dimension w rising at a velocity v,
provided an estimate of the effective viscosity can be made.
Because dynamic topography can develop even when temperatures are too high to per-
mit much static rock strength, it has been suggested that a vigorous plume rising from
deep within the Venusian mantle might cause the astonishingly high elevations of Maxwell
Montes and Beta Regio on Venus. The estimated plume velocities must be quite high, on
the order of meters per year, and if the flow fluctuates with time, one might expect to see
the elevation of Maxwell Montes fluctuate in concert with the flow. Pursuing this idea, the
Magellan radar altimeter repeatedly measured the height of Maxwell throughout the dur-
ation of the mission, seeking for measurable fluctuations. Unfortunately, none were found
and the reason for Maxwell’s high elevation remains unresolved.

3.5.5 Floating elastic shells: flexural support of topographic loads

A small, cool planet or moon may possess considerable long-term strength right down to
its center. However, as interior temperatures rise, strength declines and the ability of a large
planet to support long-duration, non-hydrostatic loads comes to reside exclusively near its
surface. This gives rise to the concept of a lithosphere, a cool outer rind whose strength
is controlled by increasing pressure near its top and by slow viscous creep near its base.
The mechanical behavior of such a lithosphere can be very complex: Its upper portion
responds to loads both by elastic deformation and plastic failure, while its underside flows
on long timescales. However, a drastic but surprisingly effective approximation neglects
the viscous deformation altogether and treats the lithosphere as an elastic plate floating on
a perfectly fluid substratum. Loads on the surface flex the lithosphere downward and are
supported by a combination of elastic stress from the lithosphere itself and the buoyancy
of the displaced fluid below. The lithosphere thus supports loads in the same way that a
skater on a frozen pond is supported by the flexure of the layer of ice. Indeed, Heinrich
Hertz, otherwise renowned for his discovery of radio waves, first published the equations
describing the effect of a point load on floating ice in 1884 and so initiated the mathemat-
ical study of lithospheric support.

EBSCOhost - printed on 11/3/2023 9:53 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
92 Strength versus gravity

Flexural models of topographic support were first proposed around 1900, when most
scientists supposed that the interior of the Earth is literally molten and that the continents
simply float on a liquid interior like ice on a frozen pond. Although it is now clear from
the propagation of seismic shear waves and the slow rate of post-glacial rebound that the
Earth’s interior is actually a hot, viscoelastic solid, the relaxation of differential stresses
at high temperatures still makes the elastic flexure approximation a good one for loads of
long duration.
The bottom of the elastic lithosphere is now understood to be the depth at which the
Maxwell time equals the duration of the load to be supported. Surface rocks behave elas-
tically above this depth and flow gradually below it. Because the duration of the load enters
into the lithosphere thickness, this concept is a bit fuzzy: For a load lasting only a few min-
utes, as might be applied by a meteorite impact, the entire mantle of the Earth is the litho-
sphere. For glacial rebound over 10 000 yr, the effective lithosphere is about 100 km thick,
whereas for a mountain chain built over 100 Myr the lithosphere thickness might be only a
few tens of kilometers. However, because the creep rate of most rocks is a strong function
of temperature, the effective lithosphere thickness varies only by a small amount for loads
lasting from a few million to a few billion years. Under these circumstances the lithosphere
can be approximated as having a constant thickness determined by its composition and the
near-surface thermal gradient.
The equations describing the response of such a floating elastic shell are very complex.
In their simplest form, for a thin flat plate of uniform thickness, they obey a fourth-order
partial differential equation called the biharmonic equation. However, these equations need
not be solved to attain a qualitative idea of how topographic loads are supported by an
elastic lithospheric shell. The most important concept deriving from these equations is
embodied in a factor with dimensions of length called the flexural parameter, α, which is
defined as:
1/ 4
 1 Et 3 
α=  (3.41)
 3 (1 − ν ) ρm g 
2

where t is the thickness of the lithosphere and ρm is the density of the mantle underlying
the lithospheric plate. E is Young’s elastic modulus, ν is Poisson’s ratio, and g is the accel-
eration of gravity. A representative value of α for the Earth’s oceanic lithosphere is about
53 km, derived from the deflection caused by the Great Meteor Seamount in the Atlantic
Ocean (Watts et al., 1975). It is generally a few times larger than the lithosphere thickness
itself, which in this case is about 16 km.
The flexural parameter describes the tradeoff between elastic flexure and buoyancy
in supporting a concentrated load. Loads of breadth smaller than the flexural parameter
are mainly supported by elastic stresses that develop in the warped lithosphere, whereas
broader loads must be supported by buoyancy; that is, by isostatic forces. Flexure thus fills
the gap between topographic loads much narrower than the thickness of the lithosphere,
which are supported essentially on an elastic half-space, and very broad topographic loads
that are supported by isostasy.

EBSCOhost - printed on 11/3/2023 9:53 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
3.6 Clues to topographic support 93

There is, however, a price to be paid for the advantages of flexural support. Plate
flexure creates bending stresses, and for broad loads these stresses are usually much
larger than the minimum required by Jeffreys’ theorem. In the absence of isostatic sup-
port, for a sinusoidal load of wavelength λ elastic plate flexure theory gives a maximum
stress of:

3 λ2
σ max = ρc gh. (3.42)
2π 2 t 2

Thus, because of the factor λ/t, squared, stresses build rapidly when the width of the
load becomes substantially broader than the plate thickness. This equation suggests that as
the load breadth increases, the flexural stress increases without limit. However, this does
not occur when a low-density crust overlies a denser mantle: At long wavelengths isostasy
takes over and the stresses actually decline as 1/λ2 after the stress peaks at a wavelength
of 2 π α .
The surprisingly high stresses that develop in an elastically flexed plate are reduced
somewhat when plastic yielding occurs and spreads the stresses over a larger volume, but
the lesson is that flexure cannot support very broad loads. Indeed, in locations where the
topography suggests that flexural support is important, it is common to observe tectonic
evidence of rock failure.
The flexural parameter α is often directly observable. The size of the region depressed
by a concentrated load is governed by the flexural parameter. Thus, the island of Hawaii is
surrounded by a broad shallow moat where the elastic lithosphere of the Pacific Ocean floor
is flexed downward by the weight of the volcanic pile. Similarly, the ice shell of Europa
is flexed downward by the weight of the ridges crisscrossing its surface, creating shallow
troughs flanking the ridges (Figure 3.13). The gigantic Artemis Corona on Venus is par-
tially surrounded by a moat similar to that around Hawaii, and also may be due to a flexing
(or, perhaps, subducting and flexing) lithosphere.
One of the goals of planetary surface studies is to find evidence for such flexural depres-
sions flanking topographic loads and, from their breadth, use Equation (3.41) to determine
the thickness of the lithosphere. This thickness, in turn, can be used to estimate the near-
surface temperature gradient, and, hence, planetary heat production.

3.6 Clues to topographic support

With all of the different mechanisms that can contribute to topographic support, the ques-
tion naturally arises, how can we tell which mechanism, or what combination of mecha-
nisms, is actually supporting the topography of a given planet? Some first-order guesses are
easy: Long-wavelength loads are generally supported by isostasy, short-wavelength loads
by flexure. Limits to the strength of materials provide some clues. However, how can we
know, in a particular case, what mechanism is actually in play?

EBSCOhost - printed on 11/3/2023 9:53 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
94 Strength versus gravity

Box 3.3 Flexure of a floating elastic layer

Determining the deformation of the surface of a floating elastic plate under a given load is both
an old problem and a difficult one. Heinrich Hertz (of radio-wave fame) first offered a solution
in 1884 (Hertz, 1884). His interest was not planetary lithospheres, but rather the form of the
surface of a frozen pond under the weight of an ice skater. The weight of the skater (what we
would now call a concentrated load) is supported both by the bending of the ice layer itself and
by the buoyancy of the underlying water displaced by the deflection of the lower surface of the
ice layer. Hertz’s solution relied on centuries of mathematical study of the deflection of beams
and the creation of an effective theory of elasticity. His work on the deflection of a floating
plate found immediate application to railway engineering, where the bed of the tracks forms a
support similar in many ways to a dense liquid layer.
In more recent times, the well-known geophysicist Don Anderson relates (personal
communication, 1971) that his first introduction to plate tectonics took place courtesy of the
United States Air Force, which required him to use Hertz’s theory to determine how close
together airplanes might be parked on an ice floe before the ice ruptured. Currently, the theory
of plate flexure is widely applied in geodynamics to investigate the structure and evolution of
planetary lithospheres. It is the principal subject of at least one modern monograph (Watts, 2001)
and is discussed as part of thousands of papers on both terrestrial and planetary geophysics.
A serious student of terrestrial or planetary geophysics should, thus, be familiar with both the
derivation and many applications of the flexural equations. However, the passage of time has not
made this subject much easier than it was for Hertz, and a full derivation would be out of place in
a broad overview (indeed, the correct application of the lower boundary condition between elastic
and fluid materials presents a subtlety so obscure that, of all the books I know, only one (Cathles,
1975) treats it correctly!). The equations are, happily, linear for small vertical deflections w of
the centerline of the plate. They are, however, fourth-order differential equations that, thus, have
four parameters that must be determined from the boundary conditions. The most frequently
used version of the full equations assumes that both the plate and the load are uniform in the y
direction, hence, the solution depends only on the horizontal distance x:

∂4w ∂2 w
D 4
+ N 2 + ρa g w = q( x ) (B3.3.1)
∂x ∂x
where N is a horizontal force applied in the x direction (taken to be positive in compression), ρa
is the density of the underlying fluid layer, g is the acceleration of gravity and q is an applied
surface load (force per unit area). D is the flexural rigidity, defined in terms of the Young’s
modulus E, Poisson’s ratio ν and plate thickness t as:

E t3
D= . (B3.3.2)
12 (1 − ν 2 )

A typical solution to this equation is given by Figure B3.3.1, which shows the deflection
of the lithosphere under a load of uniform thickness a that extends arbitrarily far to the left of
the center, x = 0 and infinitely far perpendicular to the page. This might represent the edge of
a very broad plateau with a straight edge. The density of both the load and the flexed plate is
ρl, while that of the underlying fluid layer is ρa. The formula for the vertical deflection of the
center of the lithosphere in this case is:

EBSCOhost - printed on 11/3/2023 9:53 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
3.6 Clues to topographic support 95

Box 3.3 (cont.)

1 trough bulge

0
weight
Elevation

–1

–2

–3

–4
–4 –2 0 2 4
x, units of α

Figure B3.3.1 Deflection of a floating elastic plate by a sharp-edged load with the same density
as the plate. The horizontal dimension is in units of the flexural parameter α, while the elevation
is in units of half of the lithosphere thickness. The dashed line is the neutral sheet of the plate.
The load uniformly depresses the plate on the left, but near the edge of the load the plate is flexed
down with a curvature comparable to the flexural parameter. A very low bulge develops beyond
the main flexure. Equation (B3.4.3) in Box 3.3 describes the deflection of the neutral sheet.

 ρl
ρ

a
2
{ }
2 − e x / α cos( x / α ) x≤0
w ( x) =  a (B3.3.3)
 ρl a − x /α
e cos( x / α ) x > 0.
 ρa 2

In these equations the horizontal scale of the deflection is determined by the flexural
parameter, α, defined in the text. To find the actual topography one must be careful to add
in the thickness of the overlying half of the plate, plus the load. The lithosphere is deflected
ρ a
downward by a distance of l right under the edge of the load, while far to the left it
ρa 2
ρ
achieves the deflection required by isostatic equilibrium, l a .
ρa
Note, in Figure B3.3.1, the very slight reversal of the vertical deflection that crests at a
distance of 3πα/4 (labeled “bulge” in the figure). Noted by Hertz in his solution, this small-
amplitude reverse deflection is characteristic of flexural solutions. On Earth, this slight bulge
is readily apparent on topographic maps of the great oceanic trenches where the oceanic
lithosphere is subducted into the mantle. A few hundred kilometers seaward of every trench
there is a small rise, termed the “outer rise,” that seems to represent the flexure of the
oceanic lithosphere. The flexural trough surrounding each of the Hawaiian islands is likewise
accompanied by a slight outer rise farther from the island loads.
The above results are valid only for a flat elastic plate. When the lithosphere has a
substantial curvature it is technically called a shell and the solutions for topographic support
must include membrane stresses from the stretching or compression of the plate in addition to
flexural stresses. This case is examined in some detail by Turcotte et al. (1981).

EBSCOhost - printed on 11/3/2023 9:53 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
96 Strength versus gravity

100

Height, m 60

20
0
3000 6000
Horizontal
distance, m
–40

Figure 3.13 Flexure of the lithosphere adjacent to a ridge on Europa. The plot shows the topography
determined by the method of photoclinometry applied to a Galileo image of Europa. Note the
prominent trough flanking the ridge and the low bulge beyond the trough. This trough and bulge
topography is the expected shape for a loaded floating elastic plate of thickness 350 ± 50 m. After
Figure 4 in Hurford et al. (2005).

This question would be easy to answer if we could directly determine the stresses acting
in a planet’s interior. Unfortunately, stresses are difficult to measure even in the laboratory.
In situ stress measurement techniques do exist, however, and are sometimes used at shal-
low depths in the Earth (Engelder, 1993), but such data do not yet exist for any other planet.
Earthquakes also give clues about stress magnitudes and directions, but even in the Earth
their interpretation is presently somewhat controversial, and for other planets the necessary
seismic data sets do not exist (apart from some intriguing, but limited-term, data on lunar
seismicity). Associations of tectonic features, to be discussed in more detail in the next
chapter, may give clues about regional stress distributions and this is an important source
of information. However, the most direct information on topographic support comes from
measurements of the acceleration of gravity.

3.6.1 Flexural profiles

The most direct method of determining lithospheric thickness is to observe a topographic
flexural profile. As discussed above, such profiles are commonly observed around vol-
canic islands in the Earth’s oceans (Fig 3.14), and have been noted on Venus and Europa.
Figure 3.13 is a good example of such a profile, first discovered in 2005 (Hurford et al.,
2005). Upon observing new images of the surface of a terrestrial planet or satellite, one of
the first efforts of any geophysically oriented planetary scientist is to seek evidence of such
surface deformation. Once found, the wavelength of the observed topographic flexure is
readily expressed in terms of the flexural parameter α, Equation (3.41), and with a few add-
itional assumptions it yields the thickness of the elastic lithosphere. Still more assumptions
give an estimate of the planet’s heat flow, a number that is difficult to determine remotely
in almost any other way. The amplitude of the flexural deflection yields the magnitude of
the load via the full flexural equations described in Box 3.3.

EBSCOhost - printed on 11/3/2023 9:53 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
3.6 Clues to topographic support 97

topographic
load
bulge
depression
elastic flexure

neutral
sheet
buoyancy

Figure 3.14 Flexure of a floating elastic plate subjected to a topographic load. The weight of the load
is supported by a combination of flexural stresses developed by bending of the plate and buoyancy
generated by the depression of the lithosphere into the fluid mantle below. This schematic drawing
indicates the neutral sheet in the plate by a dashed line.

3.6.2 Anomalies in the acceleration of gravity

To first order, the gravitational acceleration of a large planet or satellite is directed toward
its center of mass and varies as the inverse square of the distance from that center, out-
side of the body itself. However, small but significant deviations from an ideal spheroidal
field (for a rotating body) are readily detected either on the surface or from orbiting satel-
lites. This is not the place for a detailed discussion of this large area of research (see, e.g.
Lambeck, 1988), but a few important results require brief discussion.
Deviations of the gravitational acceleration of a body from an ideal assumed field are
known as anomalies. Free-air anomalies are probably the easiest to understand: they are
simply the difference between the expected gravitational acceleration and the observed
acceleration at some conventional elevation above the surface. Gravity anomalies are gen-
erally reported in units of gal (short for galileo: 1 gal = 1 cm/s2). The excess gravitational
acceleration produced by an infinitely wide sheet of density ρ and thickness h is given
by:
Δg = 2π Gρh. (3.43)
Thus, the gravitational acceleration exerted by a broad sheet of basalt 1 km thick is about
125 mGal. Although this is a very small acceleration compared with the surface acceler-
ation of the Earth (9.8 m/s2, or 980 000 mGal), such anomalies are readily detected by the
orbital perturbations of low-flying satellites. Indeed, accuracies of 0.1 mGal are presently
achievable.
The importance of free-air gravity anomalies is that, to first order, isostatically com-
pensated topography has no anomaly. Because the extra mass of surface topography is
compensated by a low density at depth, the net excess mass below the surface of isostati-
cally compensated topography is zero, and so is the gravity anomaly. This simple conclu-
sion is not quite true for loads of limited lateral extent, in which case detailed modeling is
required, but it works as a general rule of thumb. Topography supported by strength in the
form of flexure or supported dynamically is not compensated by underlying low densities

EBSCOhost - printed on 11/3/2023 9:53 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
98 Strength versus gravity

and so exhibits free-air gravity anomalies, which may be quite substantial. For example,
free-air gravity anomalies over the dynamic topography of subduction zone trenches may
drop below –300 mGal. Flanked by smaller positive anomalies along the volcanic arc, they
constitute the largest free-air gravity anomalies on Earth.
Anomalies in the acceleration of gravity (both excesses and deficits) can be interpreted
as equivalent topographic loads (positive or negative) and, thus, sources of stress in the
lithosphere. A rough-and-ready method of estimating such stresses is to replace the gravity
anomaly, Δg, by a broad layer of sufficient mass to create the same anomaly, then to equate
the weight of this extra load to the applied stress. This yields an equation for the equivalent
stress σgrav of the gravity anomaly:

g ∆g
σ grav = (3.44)
2π G

where G is Newton’s gravitational constant. In the case of the 300 mGal anomalies observed
in terrestrial oceanic trenches, this corresponds to a stress of about 70 MPa – close to the
crushing strength of rock. On the moon, the same anomaly would only imply a stress of
about 12 MPa: Although the mass anomaly is the same on the moon, its lower acceleration
of gravity implies a smaller stress.
An important episode in the history of lunar exploration was the unexpected (and, at
the time, unwelcome!) discovery of strong mass concentrations on the lunar nearside.
Termed “mascons” after they were discovered in 1968, they are circular anomalies of up
to about 500 mGal that are associated with basalt-filled impact basins. The Apollo mis-
sion planners went to great trouble to compensate for the effects of these anomalies in the
manned Apollo orbits. The mascons’ effect on satellite orbits is so strong that they crashed
the Apollo 16 subsatellite PFS-2 onto the lunar surface only a month after the astronauts
released it.
It is now understood that isostatically uncompensated lava within the circular nearside
impact basins creates most of the mascon anomalies, with an additional contribution from
an uplifted mantle plug underneath the basins (Neumann et al., 1996). The effect of the
lava’s enormous weight is clearly visible in the Humorum basin, where adjacent craters tilt
inward toward the sagging basin center and the stresses generated by the load have frac-
tured the crust in great circumferential faults.
A second commonly employed type of gravity anomaly is the Bouguer anomaly. It is
computed from the free-air anomaly by subtracting an estimate of the gravitational attrac-
tion of the topography above (or below) the reference geoid. The resulting anomaly reflects
the mass deficit (or excess) below (or above) the topographic elevation (or depression). For
more details on this and other types of gravity anomaly, see the book by Garland (1965).
Isostatically compensated mountainous terrain, such as the Alps in Europe, exhibits small
free-air anomalies and strong negative Bouguer anomalies, reflecting the low-density root
compensating the weight of the mountains. If it is further assumed that the density anom-
alies are entirely due to variations in the thickness of a constant-density crust, then grav-
ity anomalies can be inverted to create maps of crustal thickness. Such maps have been

EBSCOhost - printed on 11/3/2023 9:53 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
3.6 Clues to topographic support 99

produced for the Earth, Moon, and Mars, but it should be understood that a great many
assumptions enter into such maps unless they are constrained by seismic data.
Because isostatic compensation by a floating lithosphere depends on the breadth of the
load, a useful approach is to compute the Fourier transforms of both gravity anomalies and
topography. Comparison of the amplitude of the gravity anomaly to topographic height as
a function of wavelength on a so-called admittance diagram may then be used to estimate
the thickness of the elastic lithosphere. Unfortunately, subsurface loads easily confuse this
method; more recent work focuses on correlations between gravity anomalies and topog-
raphy as a function of wavelength. This kind of study yields maps of lithosphere thickness,
which have been compiled to date for the Earth, Venus, and Mars.

3.6.3 Geoid anomalies

A somewhat different type of gravity anomaly is the ratio between geoid height and topo-
graphic elevation (or depression). The geoid, by definition, coincides with a gravitational
equipotential surface, whereas the acceleration of gravity is proportional to the potential
gradient perpendicular to this surface. The geoid and gravitational acceleration anomalies,
thus, contain different types of information. It can be shown that geoid height variations
depend on the near-surface density gradient, rather than the density itself. Thus, the Geoid
to Topography Ratio, or GTR, allows estimates of the depth of isostatic compensation as
well as of the type of compensation, whether of the Airy, Pratt, or mixed type.
Over the Earth’s oceans the geoid can be measured directly by precision observations of
the shape of the sea surface. Over the land areas of the Earth and over planets with solid
surfaces it can be constructed from careful tracking of gravitational perturbations of satel-
lite orbits.
The geoid of Mars is utterly dominated by the huge Tharsis dome, which affects a region
about 5000 km in diameter and appears to overlie an enormous lens of low-density material
in the Martian mantle. Because of the uncompensated load, the Tharsis dome causes such
a large geoid distortion that it obscures efforts to determine that planet’s moment of inertia
and thus the size and mass of its core (Neumann et al., 2004). On Venus the GTR of high-
land features is much larger (tens of m/km; Smrekar et al., 1997) than those of terrestrial
features (typically less than 5 m/km in the oceans). These large ratios indicate much larger
depths of isostatic compensation than observed on the Earth (a paradox, considering the
much thinner lithosphere on Venus because of its high surface temperature!). Furthermore,
the GTR varies widely from one highland feature to another, suggesting a highly variable
depth of compensation. The major geoid anomalies on Earth are associated with subduc-
tion zones. However, unlike the acceleration anomalies, which are negative in the trenches
and positive over the flanking volcanic arcs, the geoid anomalies are broad positive welts
that follow the trend of the subduction zones. These anomalies are generally attributed to
the cold, dense subducted slabs slowly sinking into the mantle.
The great importance of gravity anomalies for understanding topographic support is the
principal reason that planetary geophysicists are eager to establish polar orbiters about as
many planets and satellites as possible. The Earth, Venus, and Mars have been well covered

EBSCOhost - printed on 11/3/2023 9:53 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
100 Strength versus gravity

Box 3.4 The ambiguous “lithosphere”

The term “lithosphere,” as introduced in this chapter, refers to that part of a planet’s interior
that responds elastically to applied loads. Because of the time-dependent response of rock
materials to applied stress, the lithosphere’s size and location is somewhat ill defined, as it
depends on the duration of the load under consideration and the rheology of the material of
which it is composed. It is, nevertheless, a useful concept because the extreme variations in
the effective viscosity of most planetary materials make the uncertainties in the lithosphere’s
boundaries small in relation to the size of the elastic region itself for timescales of geologic
duration.
However, the ambiguities of the term “lithosphere” only begin with this definition.
Numerous geophysicists over the past 70 years have complained that the same term is
promiscuously applied to three disparate concepts (Anderson, 1995), but to little avail: The
word “lithosphere” is employed by large segments of the geophysical community to mean
either the elastic portion of a planet’s interior (the “elastic lithosphere”), the portion of the
Earth above the seismic low-velocity zone (the “seismic lithosphere”), or the cold boundary
layer of a thermally convecting cell (the “thermal lithosphere”). It is wise to be cautions when
encountering the term “lithosphere” and to ask oneself which usage is intended!
Another frequent confusion is between “lithosphere” and “crust.” The outer regions of
planets are frequently differentiated into an outer, less dense crust that overlies a deeper
interior zone often called the “mantle,” in analogy to the divisions of the Earth’s interior. The
distinction between the crust and mantle is purely chemical: They are composed of materials
with different average densities. In contrast, the elastic lithosphere is a mechanical division.
In the Earth’s ocean basins, the elastic lithosphere comprises both the oceanic crust and upper
mantle, while on the continents the elastic lithosphere may include only the upper portion of
the crust (and this is often underlain by a second lithosphere at the top of the mantle).

by such orbiters. Our Moon is not so well understood because we cannot directly measure
the range to a satellite over its farside. High-precision lunar gravity measurements of the
farside will require missions that incorporate at least two satellites tracking one another.
Recently, the successful Japanese Kaguya mission has produced the best and most com-
plete lunar gravity field, including the farside. At the time of writing, the MESSENGER
mission is on its way to orbit Mercury in 2012 and we are eagerly awaiting the data on
Mercurian gravity that will result from successful completion of that mission. In the future
we can hope for missions that orbit, and track, spacecraft around the major moons in the
outer Solar System.

Further reading
G. K. Gilbert achieved a great deal of insight into the relationship between temporary
loads on the Earth’s surface and the viscous response of the underlying mantle in his
famous monograph on Lake Bonneville (Gilbert, 1890), which can still be read with
profit. The history of the investigation of the strength of the Earth and gravity anomalies,

EBSCOhost - printed on 11/3/2023 9:53 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
Exercises 101

among other things, is well told by Greene (1982). Long a classic in the field, the book
Fundamentals of Rock Mechanics by J. C. Jaeger has gone through many editions. It was
out of print for many years, but a new edition has recently appeared that updates its nota-
tion and includes many new measurements (Jaeger et al., 2007). The ideas behind our
modern understanding of material strength are engagingly told in a semi-popular book
(Gordon, 2006), while the details of brittle fracture theory are explored by Lawn and
Wilshaw (1993) and for rocks by Paterson and Wong (2005). The nature and theory of
dislocations is well described in Hull and Bacon (2001) and Weertman and Weertman
(1992). Gilman (1969) applies dislocation mechanics to the plastic deformation of solids.
Harold Jeffreys devoted much of his life to understanding the relation between strength
and topography of the Earth. He wrote a fine, although now somewhat dated, popular
book (Jeffreys, 1950), but his enduring masterpieces are the third and fourth editions of
The Earth (Jeffreys, 1952, 1962). Later editions of this book exist, but by the fifth edition,
the aging Jeffreys was on a campaign to stamp out the upstart theory of plate tectonics and
these later editions are rather polemic. The best treatment of the relation between grav-
ity, the geoid, and the shape of the Earth is Lambeck (1988). A clear, detailed discussion
of rheology applied to the Earth is Ranalli (1995), although a more recent book focused
on the detailed mechanisms of deformation and flow is Karato (2008). The now-classic
book on the application of theories of elasticity and viscosity to geodynamic problems is
Turcotte and Schubert (2002). This book has become a standard text for advanced courses
in geodynamics. The flexure of the lithosphere and its relation to isostatic support is now
well covered at book length by Watts (2001).

Exercises

3.1 Strength vs. gravity

a) Phobos, the innermost satellite of Mars, is an irregular, potato-shaped body with
extremes of radius, rmin ≈ 10 km and rmax ≈ 14 km, and mean density 1900 kg/m3. If
these extremes are the maximum that Phobos’ strength could support, how large is the
strength of its rock? If the strength of Phobos’ rock is similar to that of the Moon’s,
about 10 MPa, how large could the extremes of Phobos’ radii be? What do you think
this means?
b) If asteroids are incoherent “rubble piles”, the maximum slope that can exist on their
surface is the angle of repose for rock debris, about 30˚ for most types of rock. Estimate
the maximum difference in elevations possible on a non-rotating rubble-pile asteroid
and compare this to the actual difference in dimensions of known asteroids.
Extra Credit: Suppose the asteroid is rotating at the limit for breakup. Now estimate
(crudely: to do this exactly is a very hard problem) the maximum possible difference in
the asteroid’s dimensions. For a more sophisticated approach to this problem see Minton
(2008).

EBSCOhost - printed on 11/3/2023 9:53 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
102 Strength versus gravity

3.2 Viscous flow

Use the formula, similar to the one derived in Section 3.5.2, for the relaxation time τ of a
disk-shaped load on a viscous half-space with viscosity η:
τ = 5η/ρgR
where ρ is the rock density, g is the surface acceleration of gravity, and R is the radius of
the disk.
a) The Imbrium mascon (R = 500 km) is not isostatically compensated (relaxed). The last
lavas on its surface are ca. 3 × 109 yr old. Derive a lower limit for the Moon’s present-
day viscosity.
b) Lake Bonneville (Ancestral Great Salt Lake in Utah) relaxed almost completely in the
1500-year interval between the Bonneville and Provo stages, when much of its water
drained out to the northern Columbia River drainage. Its radius R ≈ 100 km. What is the
viscosity of the mantle beneath Utah? Compare this to the viscosity (1021 Pa-s) of the
average mantle. What does this mean?
c) Over the last 3 × 109 yr (probably), 100 km diameter crater basins on Ganymede
have relaxed completely, but their 10 km wide rims are still clearly visible. What
does this imply about the viscosity of Ganymede? Is there more than one possible
interpretation?

3.3 Warmed-over Uranian moons

Tiny Miranda, radius 236 km and surface temperature about 70 K, has a shape that is in-
distinguishable from that of an equilibrium tidal ellipsoid, with a maximum tidal bulge of
about 7.1 km. Using the fact that Miranda must have relaxed into this shape over the past
4.5 × 109 yr, derive an upper limit for the viscosity of its interior (you may need to know
G = 6.67 × 10–11 Nm2/kg2 and the mean density of Miranda is 1200 kg/m3).

3.4 The ultimate limit to core formation

Use Frenkel’s estimate of the ultimate strength of a solid, YF = m/2π, to estimate the max-
imum radius, rmax, of an iron sphere that can be supported in the Earth’s mantle. Some
relevant data are ρ (iron) = 8000 kg/m3 (at mantle pressures), ρ (mantle) = 5000 kg/m3,
and m = 2.5 × 1011 Pa.
Reference (consult this after you have solved the problem!): G. F. Davies (1982).

3.5 Global isostasy

a) Suppose that the Moon’s center of mass (CM) is 1.6 km closer to the Earth than its
center of figure (CF), as was determined by the Clementine mission. Model the Moon’s
interior as a mantle of density 3300 kg/m3 and a crust of density 2800 kg/m3. If the
crustal thickness on the nearside is 60 km (determined by the Apollo seismic experi-
ment), how thick must the farside crust be to explain the CM–CF offset?

EBSCOhost - printed on 11/3/2023 9:53 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
Exercises 103

b) Another way to estimate the Moon’s crustal thickness is to note that the floor of the
gigantic, 2600 km diameter, South Pole-Aitken basin lies about 8 km below the best-
fitting sphere representing the lunar mean elevation. If we make the reasonable assump-
tion that this impact cleared away all of the overlying crust, leaving bare mantle on the
floor of the basin, use isostasy to estimate the Moon’s mean crustal thickness. If this
answer differs from (a) above, what may be the reason?

3.6 Supporting Maxwell

Maxwell Montes is the highest elevation on Venus, rising 11 km above the planet’s mean
radius and extending over a 500 km diameter region. It is possible that Maxwell is sup-
ported dynamically by viscous stresses induced by a rising mantle plume impinging on the
overlying lithosphere. If this is correct, use order-of-magnitude estimates to deduce the
velocity of the plume necessary to support Maxwell. You may assume the mean viscosity
of the mantle is 1019 Pa-s, similar to the Earth’s asthenosphere. If this velocity fluctu-
ates by 10% over the year that Magellan observed the altitude of Maxwell’s summit, how
large would the variations in elevation be? Do you think these elevation changes would be
detectable?

3.7 Flexed Venusian lithosphere

The northern edge of Ishtar Terra on Venus is an enormous scarp, 4 km high, that stands
near 30°, the angle of repose. Just north of the plateau edge is a deep trough that is bounded
still farther away by a low rise that crests about 50 km away from the deepest part of the
trough. Use the theory of a sharp-edged load on a floating elastic plate to estimate the
Venusian elastic lithosphere thickness. How does this agree with other estimates of litho-
sphere thickness?
You may need to recall that the acceleration of gravity on Venus is 8.6 m/s2, and may
assume that the elastic constants of the Venusian crust are approximated by E = 1.6 × 1011
Pa and ν = 0.25.

EBSCOhost - printed on 11/3/2023 9:53 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
4
Tectonics

Though the primary direction of the force which thus elevated them [the
All rights reserved. May not be reproduced in any form without permission from the publisher, except fair uses permitted under U.S. or applicable copyright law.

strata] must have been from below upwards, yet it has been so combined
with the gravity and resistance of the mass to which it was applied, as to
create a lateral and oblique thrust, and to produce those contortions of
the strata, which, when on the great scale, are among the most striking
and instructive phenomena of geology.
John Playfair, Illustrations of the Huttonian Theory of the Earth
(1802, p. 45)

4.1 What is tectonic deformation?

During NASA’s Mariner 10 spacecrafts first encounter with the planet Mercury on March
29, 1974, planetary scientists at both Caltech and JPL (the Jet Propulsion Laboratory in
Pasadena, California) were initially disappointed by the Moon-like appearance of Mercury’s
surface. They had hoped for something more exotic than a monotonous, gray, heavily cra-
tered planetary landscape. But closer examination revealed a major difference: dozens of
kilometer-high scarps wound hundreds of kilometers across the surface, cutting plains and
craters alike. These “lobate scarps” indicated a planetary history quite distinct from that of
our moon and implied that powerful compressive stresses fractured Mercury’s lithosphere
sometime after the period of heavy meteorite bombardment.
Planetary surface features created by internal stresses that fracture or deform the
lithosphere are collectively termed tectonic. Not all tectonic features are as dramatic as
Mercury’s lobate scarps: subtle long-wavelength topographic warping is also classified
as tectonic. On Earth, such warping was first recognized by changes in the position of the
land and the sea. The word “tectonic” is derived from the Greek word for “builder” and has
long been identified with the forces that produce mountains. Terrestrial geologists use this
term to describe earth movements and constructional landforms not created by volcanism
or impact.
Copyright 2011. Cambridge University Press.

Analysis of the scarps, ridges, and troughs created by tectonic processes reveals the
nature of the forces that have acted within a planet’s lithosphere. Stress directions and mag-
nitudes can often be inferred from topography coupled with knowledge of the rheology of

104

EBSCO Publishing : eBook Academic Collection (EBSCOhost) - printed on 11/3/2023 9:54 AM via YDUQS
AN: 399390 ; H. Jay Melosh.; Planetary Surface Processes
Account: s4526441.main.estacio
4.1 What is tectonic deformation? 105

the surface materials. Supplemented by information on the age of such features, valuable
insights into the internal workings and history of a tectonically deformed planet may be
gained.

4.1.1 Rheologic structure of planets

Small planetary bodies lacking a substantial internal heat source are rheologically diverse:
Depending on the condition of the materials composing them, they may be described as
rubble piles, monolithic blocks or fractured remnants of larger bodies. However, if the
internal temperatures rise to about half the melting temperature of the planet’s principal
constituents, a simple generalization becomes possible. Inside such an internally heated
body creep relaxes differential stresses over geologic time, while its exterior remains
cold. The actual numerical values of the temperatures of the near surface and interior
depend upon the heat sources available, size of the body, mode of heat transport, surface
temperature, and many other factors. However, for present purposes, it is sufficient to
characterize a large planet as a roughly spherical mass of material with a soft, stress-free
interior enclosed within a brittle outer shell. Gradations of the strength and response of
the shell to stress certainly exist, and are functions of depth, composition, and tempera-
ture, as will be discussed later in this chapter. Nevertheless, to first order, the vision of a
large planet or satellite as a soft, strengthless mass beneath a thin brittle (or sometimes
ductile) shell captures much of the mechanical behavior of the rocky and icy objects in
our Solar System.
The concept of an elastic lithosphere, discussed in the last chapter, is central to under-
standing the tectonic processes that are manifested in the basins, plateaus, scarps, and
troughs that diversify planetary landscapes. The thickness of the lithosphere determines
the size scale for many of these phenomena: The horizontal scale of lithospheric warping
is determined by the flexural parameter α, which itself depends more sensitively on the
lithospheric thickness t than on any other parameter, satisfying the functional relation α
∼ t3/4, Equation (3.41). Similarly, the width of fault-bounded troughs is controlled by the
thickness of the broken elastic layer and, thus, often reflects the thickness of the underlying
lithosphere.
In addition to providing long-term topographic support to a planet, the lithosphere trans-
mits horizontal stresses to long distances away from the actual source of applied loads.
On the Earth, stresses arising from the negative buoyancy of sinking slabs in subduction
zones are broadcast throughout planetary-scale tectonic plates. On smaller planets, such
as Mercury, changes in the rotation rate that affect the shape of the equator alter stresses
acting at the poles. On Mars, the development of the Tharsis Rise affected tectonic patterns
on the opposite side of the planet. The lithosphere is, thus, not only the seat of long-term
strength in planetary interiors, but also an effective stress guide that may integrate the tec-
tonic fabric of an entire planet. Understanding the nature and origin of the lithosphere is
thus of prime importance to understanding the tectonic features observed on the surface of
a planet.

EBSCOhost - printed on 11/3/2023 9:54 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
106 Tectonics

Ts dT
0 qs=k
dz 0

elastic

Thermal boundary
h layer: Conduction
depth

viscous

δ
convection dT
z dz adiabatic

T
Tm
~
~ 2

Figure 4.1 Schematic illustration of temperature in the lithosphere and below. The temperature
near the surface of a planet grows approximately linearly with depth, increasing from its surface
temperature TS to approximately half the melting temperature of the planet’s dominant material,
Tm/2. Material above this depth responds elastically over geologic time to applied stress. This region
defines the lithosphere. At greater depths viscous creep relaxes stress differences, but the temperature
gradient is still usually linear until the viscosity becomes so low that thermal convection alters the
temperature gradient from a conductive profile to a less-steep adiabatic gradient where convection
dominates.

To first order, the thickness of the elastic lithosphere is determined by its composition
and the planet’s surface temperature and heat flow qS. If the surface temperature of the
planet is TS, and the temperature at the base of the lithosphere is about half its melting
point, Tm/2, then the thickness of the lithosphere t is given by (Figure 4.1):

T 
k  m − TS 
 2 
t≈ (4.1)
qS

where k is the thermal conductivity of the material that makes up the lithosphere. For rocks,
k is about 3 W/m-K, while for ice it is roughly 2.2 W/m-K (see Table 4.2 for other values).
Note that temperatures must be expressed in degrees K! Equation (4.1) assumes that the
temperature gradient is linear in the lithosphere. This equation must be amended if there
are layers of varying thermal conductivity or if heat is generated internally. Nevertheless,
this simple formula gives a good first approximation to the thickness of the elastic litho-
sphere on most planets. For example, the thickness of the Earth’s oceanic lithosphere var-
ies widely, from near zero at spreading centers to about 35 km in the oldest plates, but its
bottom coincides nearly everywhere with the 750 K isotherm, which is just about half of

EBSCOhost - printed on 11/3/2023 9:54 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
4.1 What is tectonic deformation? 107

Table 4.1 Elastic lithosphere thickness estimates for the terrestrial planets

Planet or Surface heat flow Surface Elastic lithosphere

moon (mW/m2) temperature (K) thickness (km)

Mercury 13 440 172

Venus 56 730 22
Earth 81 273 34
Moon 27 273 103
Mars 23 218 126

Assumes k = 3.3 W/m-K, Tm = 2200 K. Surface heat flow for Venus and Mars
assumes a bulk heat production equal to carbonaceous chondrite material. Bulk
heat production for Mercury is reduced by a factor of 0.55 to account for the fact
that its core is about 65% of its total mass, whereas the Earth’s core is only 31.5%
of its total mass.

the melting point of depleted peridotite, its primary constituent (Watts, 2001). Table 4.1
lists estimates of the lithosphere thickness for the terrestrial planets and the Moon, derived
from Equation (4.1).

4.1.2 One- and multiple-plate planets

The simple rheologic structure discussed so far applies well to planetary bodies such as our
Moon, Mercury, Mars, and many of the larger icy satellites of Jupiter and Saturn. A single
lithospheric shell encloses all these objects. This shell transmits stresses horizontally from
one part of its surface to every other part. Such objects are aptly termed “one-plate planets.”
Their lithospheric shells are relatively strong and they respond globally to applied stresses.
Geographic features on such planets are fixed with respect to one another and often reflect
events early in Solar System history.
On the other hand, the lithospheres of larger objects such as the Earth and perhaps
Venus and Europa are segmented into multiple lithospheric plates. On Earth, these tec-
tonic plates are composed of cooler, stronger domains separated by narrow zones of hotter
and, thus, weaker material. Stresses cannot easily cross these boundaries and each plate
acts, to some extent, independently of the others. This structure, probably a consequence
of Earth’s high surface heat flow (and perhaps the weakening effect of water), is known as
plate tectonics. Much of Earth’s tectonic history can be understood in terms of the motion
and interactions of these lithospheric shells. Tectonic plates are created at spreading cent-
ers, glide past each other, and sink or are “subducted” back into the mantle in regions of
negative buoyancy. This arrangement enhances heat transfer through the recycled oceanic
lithosphere, while the more buoyant continental rafts are pushed to and fro, alternately
colliding and being rifted apart in the slow dance that has given Earth a highly variable
geography and geology.

EBSCOhost - printed on 11/3/2023 9:54 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
108 Tectonics

Venus, Europa, and Io also have high heat surface flows, but these objects do not seem
to respond to this heat flow in the same manner as the Earth. Venus appears to have been
catastrophically resurfaced about 700 Myr ago, when all of its then-existing impact craters
were erased. Tectonic and volcanic processes have heavily altered its surface, but it is not
organized into recognizable plates similar to those of the Earth. This may be a consequence
of its high surface temperature, producing a very thin elastic lithosphere that deforms so
readily that it cannot transmit stresses any appreciable distance. Europa and Io may also
possess such thin lithospheres that recognizable plates do not form, although Europa pos-
sesses what appears to be spreading centers in some of its broad ridge bands. Even on Earth
there are some regions, such as the Mediterranean and perhaps eastern Asia, that must be
described in terms of a multitude of microplates, for which the plate tectonic paradigm is
not particularly apt.
Perhaps we should recognize a third category of “deformable-plate” planets for objects
similar to Venus, Europa or Io, but is not clear that such an awkward designation would be
of much use. A better organization might focus on the horizontal distance to which stress
perturbations can be transmitted: globally on one-plate planets, regionally on Earth and
only locally on Venus and similar objects. Such a “coherence length” is closely linked to a
combination of lithospheric thickness, strength, and viscosity, which themselves are func-
tions of the temperature gradient, mean temperature, and composition.

4.2 Sources of tectonic stress

Tectonic deformation cannot occur without forces to drive it. There are many sources of
tectonic stress: Nearly every possible source that has been suggested theoretically has by
now been found (or is believed to be) acting on some planetary body. A widely used classi-
fication divides these forces into either external or internal. As in most attempts at exclusive
classification, however, this is not an absolutely clean separation, but it is at least useful for
a starting point.

4.2.1 External sources of tectonic stress

Forces that act on a planetary object from outside and lead to a change in shape may prod-
uce tectonic stress. Such forces include tides induced by adjacent massive bodies, changes
of rotation rate, and spin-axis reorientation due to true polar wander. Internal temperature
variations, although they might ultimately be caused by external changes in heat received
from a star or nearby planet, are classed as internal (such is the limitation of exclusive
classifications!).
The relation between planetary shape and tidal forces or rotation rate was discussed in
Chapter 2. Translation of these shape changes into stresses requires a rheologic model,
which itself depends on the particular circumstances of each planet. Nevertheless, a
Dutch geophysicist, Felix Vening-Meinesz (1887–1966) created a very useful approxi-
mation (Vening-Meinesz, 1947). Although Vening-Meinesz made major contributions to

EBSCOhost - printed on 11/3/2023 9:54 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
4.2 Sources of tectonic stress 109

understanding subduction zones, he lived in the era before plate tectonics and believed that
the Earth could be described as a one-plate planet, with a very thin elastic lithosphere float-
ing on a strengthless substratum. He postulated that global shear patterns would arise as the
Earth’s day lengthened due to tidal friction. To support this idea, he derived an equation for
the stresses that develop in a thin, initially unstressed, elastic shell as the Earth’s flattening
changes by an amount Δf. Because the shell is thin, there are no vertical stresses σrr(other
than lithostatic stresses, which must be added to the following equations to get the total
stress). The horizontal stresses in the east-west direction σθθ and north–south direction σφφ
are given by:

∆f  1 + ν 
σ θθ = µ 5 + 3 cos ( 2θ ) (4.2)
3  5 + ν  

∆f  1 + ν 
σ φφ = − µ 1 − 9 cos ( 2θ ) (4.3)
3  5 + ν  
where μ is the shear modulus of the lithosphere, ν is its Poisson’s ratio and θ is the
colatitude. Note that these stresses are equal, as they must be, at the pole (θ = 0°) and are
mostly extensional for a decrease in flattening (spindown). Their largest difference is at
the equator, where the east–west stress is more compressive than the north–south stress
for spindown. These stresses are uniform through the depth of the lithosphere in the thin
lithosphere approximation.
Although it may seem that a change in rotation rate is a very special situation, Equations
(4.2) and (4.3) can be applied to both tidal deformation and to planetary reorientation by
adding together multiple sets of stresses obeying similar equations after rotating the solu-
tions about an appropriate axis (Melosh, 1980a, 1980b). Such solutions have been applied to
compute the stresses in the Moon, Mercury (Melosh, 1977), Europa (Greenberg et al., 1998;
Helfenstein and Parmentier, 1983), Mars, and, most recently, in Enceladus (Matsuyama
and Nimmo, 2008) from spindown, tidal deformation or spin-axis reorientation.

4.2.2 Internal sources of tectonic stress

Any load, positive or negative, added to the surface of a planet creates stress in its litho-
sphere. These stresses are required to support the additional topography and were discussed
in the last chapter. It should be obvious that the emplacement of volcanic plains applies
loads to the underlying lithosphere, as for example, did the lavas that created the lunar mas-
cons. Magma that does not erupt on the surface is frequently intruded into the lithosphere
and creates tectonic deformation as it displaces the surrounding rock layers. Impact craters,
likewise, remove material and serve as an immediate source of negative loading. It may be
less obvious that on planets with an atmosphere or hydrosphere, such as the Earth, Mars
or Titan, erosion and sedimentation by surface fluids also redistribute mass and produce
varying surface loads. On the Earth, loading by sediments deposited in enclosed basins, on

EBSCOhost - printed on 11/3/2023 9:54 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
110 Tectonics

continental shelves, and by large river deltas has been extensively analyzed in the geophys-
ical literature (Watts, 2001). Similarly, erosion of rising mountain masses, such as the Alps
or Himalayas, unloads the crust and usually leads to substantially larger uplift in the core
of the mountain ranges than would otherwise occur. The tectonic changes induced by the
growth and decay of Pleistocene ice sheets on the Earth provide unique insights into the
rheologic structure of our planet.
The stresses produced by loads on or beneath a planetary surface are as various as the
loads and the underlying lithospheric structures themselves. There are no simple rules for
computing the exact sizes and directions of the stresses: Each case must be addressed on
its own terms. Fortunately there are now many examples that have been examined and
there are methods, including sophisticated numerical techniques such as finite element ana-
lysis, for deducing tectonic stresses and rheologic response from given initial conditions
(Turcotte and Schubert, 2002). Although these methods yield results that often cannot be
anticipated by simple rules, the theory of floating elastic plates described in Section 3.5.5
does provide some guidance. Immediately beneath a positive load the surface stresses are
generally compressive as the lithosphere bows downward under the load. The stresses at
the bottom of the lithosphere are correspondingly extensional due to stretching. Outside
the boundary of the load itself, at a distance comparable to the flexural parameter α, these
stresses are much smaller, and reverse in sign. Thus, the centers of lava-filled mare basins
on the Moon are generally deformed in compression, while extensional structures surround
them. Naturally, the same description applies to uplifted regions (negative loads), but with
all signs reversed.
A very important, but often overlooked, role is played by the curvature of the planetary
surface. Most published analyses of stresses in floating plates adopt a flat-plate approxima-
tion, in which the curvature of the surface is ignored. This is often an excellent approxima-
tion and only small errors are incurred by adopting it. However, when curvature of the plate
is important (in this case the “plate” is called a “shell” in the engineering literature) the
directions of the principal stresses may shift 90°, changing, for example, the radial exten-
sion surrounding an uplift to concentric extension (Janes and Melosh, 1990). The effect
of planetary curvature is appreciable for loads of surprisingly small dimensions. A simple
criterion for estimating the breadth of the load for which curvature is important was derived
by Landau and Lifshitz in their famous book on elasticity (Landau and Lifshitz, 1970, p.
62). If the breadth of the load is L and the radius of the planet is R, then shell curvature is
important when:

L > R t. (4.4)

Thus, on the Moon, R = 1738 km, if the lithosphere is 100 km thick, then loads broader
than about 400 km will be strongly affected by planetary curvature. In particular, this
means that stresses from mascon loads cannot be correctly computed using a flat-plate
model (Freed et al., 2001).
Aside from surface loads, any internal process that changes the shape or volume of a
planet can produce tectonic stresses. The largest scale change is an alteration in planetary

EBSCOhost - printed on 11/3/2023 9:54 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
4.2 Sources of tectonic stress 111

volume. This can be caused by heating or cooling (which is probably the source of Mercury’s
crustal compression), phase changes such as the crystallization of ice from water in the icy
satellites or of liquid iron to solid iron in the cores of the terrestrial planets, or even rear-
rangements such as the separation of iron from silicates during core formation. All changes
in planetary volume cause a change in radius (moreover, this also alters the rotation rate
slightly and thus produces stresses that we have just classified as “external”). For a thin
elastic shell, the stresses due to a change in radius ΔR are computed on the same basis
as Vening-Meinesz’s estimate of stresses caused by flattening changes. Both horizontal
stresses are equal, and the vertical stress change is, again, zero to first order. Thus:

 1 + ν  ∆R
σ θθ = σ φφ = 2 µ  . (4.5)
 1 − ν  R

Expansion produces isotropic extensional stress and contraction produces isotropic

compression, both of which are uniform through the lithosphere and thus cannot cause
preferred orientations of tectonic features.
In addition to global changes in radius, local processes produce stresses through volume
changes. Those most important on the Earth today are probably temperature variations,
although changes in chemical composition or mineral phase are also often cited as rea-
sons for uplift or downwarping. Such changes are often linked to internal convection and
so cannot be easily separated from convective stresses. The buoyancy stresses due to a
local change in volume are easy to estimate: If the local density alteration is Δρ and the
vertical dimension of the region over which the alteration occurs is L, then the buoyancy
stress is Δσ ∼ gLΔρ. The density alteration due to a temperature change ΔT is simply
Δρ = −ραV ΔT, where αV is the volume coefficient of thermal expansion when pressure is
maintained constant:

∆V ∆ρ
= − = α V ∆T . (4.6)
V P ρ P

For substances that expand uniformly in all directions the volume coefficient of expan-
sion is exactly three times larger than the linear coefficient of expansion. Because of the
importance of this parameter for tectonic deformation, Table 4.2 lists the size of the volume
expansion coefficients for a number of substances important in planetary interiors.
Simple density estimates can be made for chemical and phase changes, such as may
occur during hydration of silicate minerals, freezing of liquids, or solid-state phase changes
due to alterations of pressure or temperature in a planetary interior. Many liquids, upon
freezing, decrease their volume by about 10% (water is an important exception!). Such
volume changes do not, however, translate directly into tectonic stress in the lithosphere,
as we must know how the volume or density change interacts with the lithosphere. If the
volume of the elastic lithosphere itself changes, then the change is resisted by the elastic
modulus (Turcotte and Schubert, 2002, Section 4–22). Such changes usually give rise
to enormous stresses because of the large size of the elastic modulus compared with the

EBSCOhost - printed on 11/3/2023 9:54 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
112 Tectonics

Table 4.2 Selected thermal conductivity and volume thermal expansion

coefficients at constant pressure and standard temperature and pressure

Material k (W/m-K) αV x 105

Rocks:
Sandstone 1.5–4.2 3
Dolomite 2–3.4 2.4
Granite 2.4–3.8 2.4
Gabbro (basalt) 1.9–4.0 1.6
Peridotite (Earth mantle) 3–4.5 2.4
Minerals:
Water ice, H2O 2.2 5
Enstatite, MgSiO3 4.47b 2.3a
Forsterite, Mg2SiO4 4.65b 2.5a
Metallic iron, Fe 0.84c 1.5a

After Turcotte and Schubert (2002, Appendix II.E) unless otherwise noted:
a
Poirier (1991) p. 23
b
Clauser and Huenges (1995)
c
Weast (1972)

stresses caused by topographic loads. However, in most cases the volume change occurs in
the more mobile material underlying the lithosphere and then the nature of coupling to the
lithosphere becomes important, particularly whether it is by viscous shear stresses or by
vertical uplift of the surface.
When heat is transferred by convection in the interior of a planet, the subsurface con-
vective flow exerts forces on the overlying lithosphere and may manifest itself by tectonic
deformation of the surface. This is the principal cause of tectonic deformation on the Earth
and probably Venus. Convection, whether continuous or episodic, may also have affected
the surfaces of Europa, Ganymede, and Enceladus, to name just a few. The size of the
stresses can be crudely estimated from the equations above: For thermally driven convec-
tion, substitution of the equation relating density and temperature into the buoyancy equa-
tion tells us that Δσ ∼ ρgαV LΔT, where d is a relevant distance scale, often the thickness of
the convecting layer. But what determines the temperature difference ΔT? And what is the
spatial pattern of the convecting region and the pattern of stresses that convection induces
in the overlying lithosphere? A great deal of effort in both planetary science and geophysics
is currently being expended to answer these questions, but these answers seem to require
elaborate numerical simulations of convection and stress coupling. Nevertheless, we can
make a few simple statements about the onset of convection in the first place and about its
rate and ability to transfer heat out of the interiors of planets, as well as the approximate
size of ΔT. This will be discussed in the next section in the context of internal heat sources
and thermal convection.

EBSCOhost - printed on 11/3/2023 9:54 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
4.3 Planetary engines 113

4.3 Planetary engines: heat sources and heat transfer

4.3.1 Accretional heat

The tectonic engine that wracks the surfaces of planets and satellites is powered by heat.
Internal motions and differential stresses in the lithosphere are created by the flow of in-
ternal heat toward the cooler surface. A fundamental understanding of planetary tectonics
thus requires understanding the fundamentals of heat deposition and transport in the inte-
riors of planetary-scale bodies.
A potent source of internal heat, at least for larger planets, is their gravitational energy of
formation. When an object of mass m strikes the surface of a planet at velocity v, its kinetic
energy ½ mv2 is released, a portion of which may remain buried in the planet (most of the
balance is radiated rapidly to space). Some of this impact velocity is due to the gravitational
attraction of the planet, for which the energy increment is GMm/a, where a is the radius
of the planet, M is its mass and G is Newton’s gravitational constant. If the planet grows
from the accumulation of much smaller bodies, the total gravitational energy released upon
accretion is a constant of order one times GM2/a. For a uniform density sphere the “gravi-
tational binding energy” is:

3 GM 2 16 2 2 5
Ebind = = π ρ Ga (4.7)
5 a 15
where ρ̅ is the mean density. This binding energy clearly grows rapidly as planetary size
increases. The Earth’s accretional heat, if all of it were retained, would raise the entire
planet’s temperature by about 38 000 K. For the Moon, this figure is only 1700 K. These
numbers make it clear that the initial heat of formation may play a major role in the thermal
evolution of planetary-sized bodies.
The original heat of accretion is usually negligible for asteroidal bodies, but it is of major
significance for the larger planets and moons. The principal uncertainty in its effectiveness,
however, is how large a fraction of this heat is lost by radiation and how much is buried
deep within the body. At the present time a full accounting of the heat deposited by impacts
of various sizes has not yet been established, largely due to large contributions from factors
of opposite sign. It has become common practice to assume a “burial efficiency” for impact
heat, to which various writers have assigned values between 0.3 and 0.5. Most recently,
however, the recognition of the role of truly giant impacts, in which bodies of comparable
size collide, has greatly complicated this type of computation and much work still remains
to be done on the heat retained during accretion (Melosh, 1990a).
One of the peculiar consequences of accretional heating is that, in the earliest stages
of accretion when the mass of the growing planet is small, collisions are gentle and little
heating takes place. But as the planet grows and its mass increases, collisions release cor-
respondingly more heat. This produces a nascent planet with a cold center whose internal
temperature increases roughly as the square of distance from its center. Such an “upside-
down” thermal profile is stable against thermal convection until other heat sources, such as

EBSCOhost - printed on 11/3/2023 9:54 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
114 Tectonics

radiogenic heating, compositional convection, or planetary-scale impacts rearrange the in-

terior into the more common pattern of a hot core overlain by a mantle whose temperature
decreases towards the surface. The inverted temperature profile caused by accretional heat-
ing has occasioned much discussion over when and how planetary cores form (Stevenson,
1981).

4.3.2 Tidal dissipation in planetary interiors

One of the major discoveries in planetary science is that the tectonic, and even volcanic,
evolution of many satellites in the Solar System is driven by heat created from tidal dissi-
pation. Just as vigorous flexing of a squash ball warms it and renders it more flexible, so the
inexorable tidal flexing of satellites, such as Jupiter’s Io, dissipates energy in their interiors
and may strongly affect their surfaces. Tidal dissipation also causes orbital changes, such
as the circularization of elliptical orbits, the expansion of our Moon’s orbit, or the gradual
de-orbiting of Phobos (currently scheduled to impact Mars’ surface in about 100 Myr).
Tides have slowed the rotation of many planets or satellites, leading to the synchronization
of rotation and orbital periods or capture into stable resonances, such as the 3:2 spin state of
Mercury. Iapetus’ oblate figure, for example, suggests a primordial 16-hour period, instead
of its current 79-day synchronous rotation about Saturn.
A full discussion of tidal energy dissipation and its orbital consequences is beyond the
scope of this text (for a full account, see the text by Murray and Dermott, 1999), but a few
simple ideas will aid in understanding how tidal energy is dissipated in planets. Just as in
our humble squash-ball analogy, tidal flexing of a planet dissipates energy by internal fric-
tion. The type of friction varies a great deal from one body to another. In very small objects
that are solid throughout, a variety of solid-state deformation mechanisms dissipate energy
by shear between atoms or atomic structures such as dislocation arrays. This type of dissi-
pation is generally inefficient and energy losses from such mechanisms are relatively small.
In our Earth, tidal currents in shallow oceans dissipate most of the energy from lunar and
solar tides. The energy dissipated by these tides is not small: It amounts to about 10% of the
total surface heat flow of the Earth. If this energy were dissipated deep within our planet (as
was once believed) it would contribute more than enough energy to drive the geodynamo.
This energy comes ultimately from the Earth’s rotation and its loss causes the day to grad-
ually lengthen (900 Myr ago the day was only 18 hours long: Sonnett et al., 1996) and,
through the conservation of angular momentum, causes the Moon to recede from the Earth
at about 3.7 cm/yr.
Planets or moons with internal liquid–solid, or even slipping solid–solid interfaces can
also lose large amounts of tidal energy at these junctures, a mechanism that is currently
the subject of much calculational effort. Entirely gaseous planets such as Jupiter, Saturn or
even stars like the Sun, have fluid-dynamical loss mechanisms that cause tidal dissipation,
although such mechanisms are usually even less efficient than solid dissipation.
The rate of energy loss in a periodically disturbed system is conventionally parameter-
ized by a quantity called simply “Q,” short for “quality factor.” The concept of Q originated

EBSCOhost - printed on 11/3/2023 9:54 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
4.3 Planetary engines 115

in the study of oscillating electrical circuits, but it is an apt way of describing the loss of
energy in any periodic system. Q is defined as the inverse ratio between the energy lost
per cycle ΔE, to the periodically oscillating energy present in the system, Emax, times the
conventional factor of 2π (this seemingly arbitrary factor converts the energy loss from per
cycle to per radian and usually eliminates many 2π factors elsewhere):

2 π Emax
Q≡ . (4.8)
∆E
As the energy dissipated per cycle decreases, the value of Q rises – the less loss, the
higher the “quality” of an oscillator. Another way of thinking of Q is that it is the number
of cycles it takes for the energy to decrease by 1/e, so that the higher the Q, the larger the
number of cycles that pass before the oscillation is damped.
The energy stored in a tidally distorted body is not easy to compute: For a solid body it
requires a coupled application of full elasticity theory and gravitational potential theory for
a distorted body. However, the result of such a computation for a uniform, elastic, incom-
pressible satellite distorted by an adjacent massive primary around which it travels in a cir-
cular orbit is easy to state (Goldreich and Soter, 1966):

9π GM 2 a 5 1
Emax = (4.9)
2  19 µ  r
6
1 +
 2 g ρ a 

where G is Newton’s gravitational constant, M the mass of the primary, and r is the distance
between the primary and the orbiting object. The tidally distorted satellite is character-
ized by its radius a, mean density ρ, surface acceleration of gravity g, and average shear
modulus μ. Note that the distortional energy is a very strong function of distance from the
primary: It falls off as the sixth power of r, so that tidal dissipation is generally only im-
portant for very close satellite–primary pairs. The factor in parentheses describes the bal-
ance between gravitational and elastic distortional energy. If gravity alone dominates, as in
very large planets, this factor is essentially 1, whereas in small objects dominated by elastic
distortion the second term dominates.
In general, Q must be computed from detailed models of dissipation. As might be
expected, it depends on the frequency of flexing. For extremely strong tides it also depends
on the amplitude of the flexing, but this is usually neglected in current models. Q is readily
computed for idealized rheologic models, such as the Maxwell model or the Kelvin–Voigt
model described in Chapter 3. The dependence of Q on angular frequency ω is very differ-
ent for these two models (Knopoff, 1964):
ωη
Q= Maxwell solid
µ
µ (4.10)
Q= Kelvvin–Voight solid
ωη

EBSCOhost - printed on 11/3/2023 9:54 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
116 Tectonics

where η is the viscosity and μ is the shear modulus of the material. Dissipation in liq-
uids follows the Kelvin–Voigt equation quite accurately, whereas it is expected that the
slow deformation of hot solids should approximate the Maxwell rule. The dissipation
of seismic waves in the Earth enigmatically shows very little frequency dependence
(Knopoff, 1964). This is generally attributed to the simultaneous operation of many dif-
ferent mechanisms of seismic energy loss, but the subject is still controversial (Karato,
1998).
The time rate of energy dissipation is related to Q and Emax through the frequency:

dE ω Emax (4.11)
= .
dt Q

In a Maxwell solid the rate of energy dissipation is independent of frequency, whereas in a

Kelvin–Voigt solid it depends on the square of the frequency. The situation is more compli-
cated than this discussion suggests when applied to planets, because the maximum stored
tidal energy Emax in a Maxwell solid is itself frequency-dependent: The overall dissipation
depends on 1/ω at high frequencies and ω at low frequencies.
The observed rates of energy dissipation derived from rates of orbital evolution can be
used to estimate Q for various Solar System bodies. Q is known only for a few bodies at
present, but its value serves as a useful probe of the internal state of planetary objects. For
example, the surprisingly low Q of 27 for the Moon suggests frictional losses at the solid–
liquid interface of a small liquid core, whereas the much higher Q = 86 of Mars suggests
that its core is solid. It is common practice, when a value for Q is not known for a solid or
icy body, to assign it a nominal value of 100. Table 4.3 lists known values of Q for bodies
in the Solar System.

4.3.3 Heat transfer by thermal conduction and radiogenic heat production

It is currently believed that planets and satellites are born hot, heated by the gravitational
energy of infalling material as they accreted from smaller bodies in the planetary nebula
around their nascent star, or by heat generated from the decay of radioactive elements.
When growing planets and satellites are not so hot as to actually melt, their accretional heat
is initially lost by solid-state conduction to their cooler surfaces. As they cool, the associ-
ated volume changes can drive tectonic deformation.
Joseph Fourier (1768–1830) established the laws governing heat conduction around
1810. Fourier had a deep interest in geophysics and spent much time and effort descending
into deep mines or visiting geothermal spas to measure the Earth’s internal heat. Fifty years
before Fourier, his countryman Leclerc de Buffon measured the cooling rate of different-
sized balls of white-hot iron in an effort to determine the age of the Earth (Sigurdsson,
1999). For a long time it was believed that the Earth’s age could, in fact, be determined
by the rate at which it cooled by conduction from a presumed incandescent initial condi-
tion (Burchfield, 1990). Nineteenth- and early twentieth-century geologists thought that

EBSCOhost - printed on 11/3/2023 9:54 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
4.3 Planetary engines 117

Table 4.3 Q in the Solar System

Approximate Dissipation
Planet or satellite tidal period location Q

Moona 1 month Core-mantle 27

boundary?
Eartha 1 day Shallow oceans 12
Solid Earth 300
Marsa 1 day Solid interior? 86
Iob 1.8 days Hot mantle? ~36
Jupiter (Io)c 1.8 days Atmosphere ~3 x 104
Saturn (Mimas)c 0.9 day Atmosphere ≥ 1.8 x 104
Uranus (Ariel)c 2.5 days Atmosphere ≥ 2 x 104

a
Yoder (1995)
b
Murray and Dermott (1999)
c
Segatz et al. (1988) Model A

terrestrial tectonics was caused by compression of the crust as the Earth gradually cooled
(Greene, 1982). Folded mountain chains were likened to the shriveled skin of a shrinking,
drying apple (Bucher, 1933).
Fourier (1955) derived a simple, linear differential equation for heat conduction and
devised a clever series method for its solution. In modern vector notation, he started with
the equation governing the flow of heat down a temperature gradient:

q = −k∇T (4.12)
where q is the vector heat flux and k is the thermal conductivity. Adding the conservation
of energy to this equation (actually, Fourier wrote his equation for the hypothetical fluid
“caloric,” but we now recognize that energy is the conserved quantity), he obtained the
general equation governing the temperature in a solid body:

∂T H
= ∇ • (κ ∇T ) + (4.13)
∂t cP

where H is internal heat generation per unit mass, and κ is the thermal diffusivity. The dif-
fusivity is related to the thermal conductivity k by κ = k/ρcP , where cP is the heat capacity
of the solid at constant pressure (cP is typically about 1000 J/kg-K for both rock and ice).
Equation (4.13) includes internal heat generation and the form written above, with the dif-
fusivity κ inside the divergence operator, also applies to the case where the diffusivity is a
function of position.
Fourier and many authors after him devoted extensive efforts to finding solutions to his
equation. A widely used reference containing analytical solutions to many problems of heat
conduction is the monograph by Carslaw and Jaeger (1959).

EBSCOhost - printed on 11/3/2023 9:54 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
118 Tectonics

One of the simplest, but widely useful, solutions to Fourier’s equation is for the steady-
state temperature in an infinite plane slab with uniform conductivity k and no internal heat
generation. If the surface temperature is TS and the surface heat flux is qS, then the tempera-
ture is a linear function of depth in the slab z:

qS
T ( z ) = TS + z. (4.14)
k
This solution was used above in Equation (4.1) for the thickness of the elastic litho-
sphere. Adding a slightly higher level of complication, if the slab contains a uniform source
of heat H, the solution acquires a third, quadratic, term:

qS H 2
T ( z ) = TS + z− z . (4.15)
k 2 cP

According to this equation, internal heat generation actually lowers the subsurface tem-
perature compared to a slab without a source of heat. This is because heat from sources
within the slab does not pass through the entire thickness of the slab, thus finding a shorter
path to the surface and lowering the temperature at depth. This is one of the principal
mechanisms by which the interior temperatures of planets are reduced: By differentiation
of heat-producing radioactive elements upward toward the surface, internal temperatures
are lowered.
Because the current major sources of radioactive heat in the Solar System, 238U, 235U,
232
Th, and 40K, are large lithophile ions, they tend to concentrate near the surfaces of silicate
planets and thus cause less internal heating than if they were uniformly distributed. Table 4.4
illustrates this effect by listing the current radiogenic heat generation for primitive Solar
System material (carbonaceous chondrites), differentiated rocks, and typical mantle rocks.
This cooling strategy does not help the icy satellites, however. Because differentiation of
ice-rich bodies puts their heat-generating silicates beneath their icy mantles, their interiors
are strongly heated. This may partially explain why the small icy satellites are so tecton-
ically active compared to their rocky compatriots. Similarly, in the early Solar System the
short-lived radioactive isotope 60Fe was active for a few million years. Differentiated into
the cores of planetesimals, this heat source would have been particularly effective in con-
vulsing the interiors and surfaces of these small bodies, and 26Al was briefly but fiercely
effective in heating their mantles. Table 4.5 lists the rate at which heat is produced by the
decay of various radioactive species important in our Solar System.
A crucially important aspect of thermal conduction is its control over the rate at which
planets can heat or cool. At the present time, nearly any problem that can be posed involv-
ing heat conduction can be readily solved on a computer to any desired degree of accuracy;
many computer programs are designed to do just this. Such numerical methods can also
incorporate temperature-dependent thermal diffusivity as well as many other more realistic
aspects of planetary materials. However, despite the present-day ease and sophistication for
solving problems in heat conduction, one simple consequence of Equation (4.13) stands
out and can be widely applied without the use of any computers at all. This result follows

EBSCOhost - printed on 11/3/2023 9:54 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
4.3 Planetary engines 119

Table 4.4 Radiogenic heat production in silicate rocks, current Solar System

Heat production
Material H (10–12 W/kg)

Bulk Earth (does not include primordial heat) 4.7

Bulk Moon (based on only two heat flow measurements) ~14
Average granite 1050
Alkali basalt 180
Venus surface rocks (Vega 1 and 2, Veneras 8, 9, 10)a 117–531
Tholeliite (Earth oceanic basalt) 27
Eclogite 9.2
Peridotite (Earth mantle) 1.5
Ordinary chondrite 5.85
Carbonaceous chondrite 5.23

Data from Stacey (1992), except:

a
Schubert et al. (2001) Table 14.3

Table 4.5 Radiogenic heat production in the Solar System

Heat generation Concentration in CI chondrites

(exclusive of neutrino 4567 Myr before present (10–9
Isotope Half-life (years) losses) (W/kg) kg/kg)
238
U 4.47 x 109 9.46 x 10–5 1.93
235
U 7.04 x 108 5.69 x 10 –4 0.610
232
Th 1.40 x 1010 2.64 x 10 –5 4.47
40
K 1.25 x 109 2.92 x 10 –5 583.
26
Al 7.4 x 105 3.55 x 10 –1 430.
60
Fe 1.5 x 106 6.3 x 10–2 840.

Data on heat generation and half-lives of U, Th, K from Van Schums (1995), element abundances
from Newsom (1995), 26Al energetic data from Schramm et al. (1970), initial abundance from
McPherson et al. (2010). 60Fe data from Quitté et al. (2005).

from dimensional analysis of this equation in the case of no internal heat generation and
uniform thermal diffusivity. In this case the order of magnitude of the cooling (or heating)
time of an initially hot (or cold) body of dimension L is simply given by:

L2
τ cool = . (4.16)
κ
This simple formula has powerful implications for planetary timescales and is even cru-
cial for understanding thermal convection. The fact that the size of the cooling body is
squared in Equation (4.16) means that increasing the size of a body leads to a seemingly

EBSCOhost - printed on 11/3/2023 9:54 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
120 Tectonics

Table 4.6 Conductive thermal time constantsa

Object Conductive cooling/

dimensions Typical example heating time

1 mm Fire fountain droplet 1 second

1 cm Large chondrules 2 minutes
10 cm Lava flow chilled crust 3 hours
1m Thin pahoehoe flow 12 days
10 m Typical lava flow 3.2 yr
100 m Rhyolite lava flow 320 yr
1 km Impact crater ejecta blanket 32 000 yr
10 km Basin ejecta blanket, planetesimal 3.2 Myr
100 km Typical lithosphere, asteroid 320 Myr
1000 km Planetary mantle, large moon 32 Gyr
10 000 km Entire terrestrial planet 3200 Gyr

a
Assumes a typical thermal diffusivity for rock, κ = 10–6 m2/s.

disproportionate increase in the cooling time. To emphasize this important point, Table 4.6
lists the cooling times for rocky bodies of a variety of dimensions of planetary interest,
from 1 mm (small chondrules or droplets from volcanic fire-fountains) to 10 000 km (an
entire planet), assuming a thermal diffusivity appropriate for rock.
Thermal conduction timescales quickly become longer than the age of the Solar System
(indeed, even longer than the age of the universe!) for planetary-size bodies. However,
for objects the size of the larger asteroids or the smaller moons of Saturn, conductive
cooling probably determined the pace of their thermal evolution. Volume changes due to
conductive cooling are typically small: The volume expansion coefficient αV is around 3 x
10–5 K–1 for most materials. Nevertheless, if a 100 km diameter moon cools by, say, 1000 K
(over a timescale of a few 100 Myr, according to Table 4.6), its volume decreases by about
3% and, hence, it circumference decreases by about 6 km, a change that, if translated to
offsets on compressional faults, would be quite visible in planetary images. Because of the
large size of elastic moduli, this strain implies compressional stresses far in excess of the
strength of any planetary material, so material failure is inevitable.
Although strains due to temperature changes may seem small, the most consistent topo-
graphic variation on Earth is almost entirely due to this effect. Thus, oceanic plates are born
hot at mid-ocean ridges and cool, following Equation (4.16), as they move away from the
ridge at nearly constant velocity. As the plates cool, they contract and the sea floor subsides
from a depth of about 2.5 km at mid-oceanic ridges to about 6 km for the oldest plates
(about 1/3 of this 3.5 km subsidence is due to isostatic adjustment to the loading by the
weight of ocean water). The resulting age–depth relation is one of the most regular topo-
graphic relations in the Solar System. The depth of the sea floor plotted versus the square
root of its age accurately follows a straight line over about 100 Myr of cooling, before it
levels out when convective heat transfer overcomes conduction.

EBSCOhost - printed on 11/3/2023 9:54 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
4.3 Planetary engines 121

The cooling and contraction of Earth’s oceanic plates nicely illustrates the point that
planetary heating and cooling events need not affect the entire body: Local thermal events
that affect only a limited region often occur. In all cases, however, Equation (4.16) provides
guidance to the tempo of these changes, given only that L is chosen appropriately.

4.3.4 Thermal convection and planetary heat transfer

On sufficiently large bodies the density variations caused by temperature differences cause
bulk motions deep below the surface. Heat is carried efficiently along with the material
in a process known as advection. When advection is driven by temperature differences it
is called thermal convection. In addition to the heat exchanged by this process, stresses
develop whose tectonic consequences often provide the only visible manifestation of con-
vection in a planetary interior. The velocity at which material is advected sets the tempo for
these tectonic processes.
The mathematical analysis of the process of thermal convection has a huge litera-
ture and has been the subject of a large amount of sophisticated analysis. For the inter-
ested (and mathematically inclined) reader, the standard treatment is found in a book by
Chandrasekhar (1961). However, the basic principles governing convection are not com-
plex and a great deal can be captured using ideas that have already been presented, aug-
mented by the numerical results of more exact analyses.
Rayleigh number. When an initially stagnant fluid is heated from below (or cooled
from above), thermal expansion generally lowers the density of the deeper material. This
produces a mechanically unstable configuration, where more dense material overlies less
dense material. The natural tendency of the less dense material is to rise and displace the
more dense material. The timescale for this exchange is governed by the viscosity η of the
fluid. Referring to the last chapter, Equation (3.34), the timescale for relaxation of topog-
raphy on a viscous substrate also gives the rate at which a density anomaly can rise through
a viscous fluid. We need only replace the topographic density ρ by the density anomaly Δρ
and write (ignoring the numerical constant):

η
τ overturn ≈ (4.17)
∆ρ g L

where τoverturn is the timescale for the density anomaly to turn over a mass of fluid of di-
mension L in the gravity field g. Because the density anomaly is caused by a temperature
difference ΔT, a substitution using Equation (4.6) for thermal expansion gives the overturn
time in terms of the temperature difference between the hot region and the overlying colder
region:

η
τ overturn ≈ . (4.18)
α V ∆T ρ g L

Unlike the density anomaly due to a permanent, compositional, difference of materi-

als, a thermal density anomaly lasts only as long as the temperature difference persists.

EBSCOhost - printed on 11/3/2023 9:54 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
122 Tectonics

Thus, if the overturn time is so long that thermal conduction can eradicate the temperature
difference, no motion will take place and the heat will be lost by conduction, in spite of
the density anomaly. We can see that there must be a threshold below which heat is trans-
ported by conduction. Only if the threshold is exceeded will overturn occur and can heat
be transported by convection. This threshold is defined by the ratio between the viscous
overturn timescale and the cooling timescale, Equation (4.16). The criterion for the onset
of convection was first derived by Lord Rayleigh (1916) and is, therefore, known as the
Rayleigh number Ra:

τ cool α V ρ g L3
Ra = = ∆T . (4.19)
τ overturn κη

Because the temperature difference ΔT between the top and bottom of a planetary mantle
is usually not known, a closely related formulation based on the surface heat flow qS and
Fourier’s relation (4.12) is usually more practical:

α V ρ g L4
Ra q = qS . (4.20)
κ kη

The definitions (4.19) and (4.20) are equal only at the onset of convection. Above the
convection threshold they differ by a factor known as the Nusselt number, Nu: Raq = Nu Ra.
Nu, which is a measure of the net heat transport, will be defined more exactly below.
For internally heated convecting systems, qS in Equation (4.20) can be replaced by the
equivalent heat generation per unit surface area, ρL H, assuming that heat generation
and loss through the surface are in balance and that the heat generation H is uniformly
distributed.
If the conductive cooling time is much shorter than the overturn time, then Ra << 1 and
convection will not occur. But in the opposite extreme, when Ra >> 1, convective over-
turn will occur and most heat will be transferred through material exchange between the
underlying hot region and the surface cool zone. The “critical” Rayleigh number, at which
convection just becomes possible, would thus seem to have a value close to one. However,
sometimes quantities that physicists dismiss as “of the order of unity” involve high pow-
ers of π or other “small” numbers. The critical Rayleigh number RaC for convection is one
of these: It ranges between 654 and 1709, depending on the boundary conditions for fluid
flow at the top and bottom of the convecting region. Such precision is only possible for the
restrictive case in which the viscosity and likewise the thermal expansion coefficient are
constants, independent of temperature and stress, and the geometry is a simple plane layer.
Lord Rayleigh also determined that the convecting material organizes itself into cells,
where sinking cool material separates upwelling warm regions. The horizontal breadth λ
of these cells is roughly three times larger than the thickness of the convecting region at the
onset of convection (Figure 4.2). At higher Rayleigh numbers the primary distance scale is
set by the thickness of the boundary layers, as discussed below.

EBSCOhost - printed on 11/3/2023 9:54 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
4.3 Planetary engines 123

λ/2 Ts T
conduction δc

L ν convection

conduction δH
Tb

Figure 4.2 Schematic illustration of the flow pattern and temperature in a convecting layer of thickness
L bounded between a cool surface layer at temperature TS and a hot basal layer at temperature Tb.
Fluid in the center of the layer rises and sinks in convective cells of horizontal dimension λ at velocity
v. The upper and lower surfaces of the convecting layer are conductive boundary layers of thickness
δc at the cool upper boundary and δH at the hot lower boundary. The temperature in these thin layers
changes rapidly, as shown in the right panel, while that in the central convecting region changes
slowly, approximately following an adiabatic gradient.

The perceptive reader may be dismayed by the simplifications introduced by the

Rayleigh analysis and, in view of the fact that the viscosity of planetary materials is strongly
dependent on both temperature and stress and that planetary mantles are hardly infinite
planes, wonder whether this type of analysis has any practical use at all. Professional
geoscientists have shared the same doubts and a great deal of effort has been expended
over the past half-century to lift these restrictions. The answer is not simple: The inter-
ested reader should consult the massive (940 pages) monograph by Schubert et al. (2001)
for a comprehensive survey. But the general conclusion is that, if the viscosity and other
parameters are defined appropriately, then the simple Rayleigh criterion for the onset of
convection does work. For example, in a fluid with a strongly temperature-dependent vis-
cosity, a meaningful Rayleigh number is computed from the value of the viscosity at the
fluid’s mean temperature.
Furthermore, a number of simple generalizations give a good first-order description of
convection well beyond the threshold. The thermal structure of a vigorously convecting
region is divided into three regions (Figure 4.2): A relatively narrow, cool top zone where
heat is transported only by conduction (the “cold thermal boundary layer”), a similar
region at the bottom (the “hot thermal boundary layer”) and an intervening region where
the temperature variations are small. The hot thermal boundary layer may be absent if
the layer is strongly heated internally, as by the decay of radioactive elements, and the
cool boundary layer may be highly mobile, as in plate tectonic convection. The interven-
ing zone is traversed by hot, buoyant masses rising to transport heat upward, which are
counterbalanced by cool sinking masses that similarly transport heat upward by cooling
the lower regions of the convecting zone. The average temperature in this central region
approximates an “adiabat,” where the temperature and pressure are related by the condi-
tion that the entropy of the material is constant. Although the derivation of the equation
describing this gradient is deferred to Chapter 5 on volcanism, the equation for the adia-
batic temperature gradient is:

EBSCOhost - printed on 11/3/2023 9:54 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
124 Tectonics

dT α V gT
= . (4.21)
dz adiabat cP

This gradient is typically about 0.36 K/km in the Earth, which is very small compared
to the average surface temperature gradient of 30 K/km. Because the adiabatic gradient
scales with the gravitational acceleration g, it is smaller in smaller planets or moons. This
relation must not be applied through phase changes, where the constant entropy constraint
may require a temperature jump. Nevertheless, the temperature change across an adiabatic,
vigorously convecting region is generally small compared to that across the top thermal
boundary layer.
Nusselt number. The thickness δ of the boundary layers is related to the total amount of
heat transported, given by Fourier’s Equation (4.12). Because the boundary layer is thinner
than the convecting region itself, the total heat transported by conduction through this layer
is larger than if it were conducted through the thickness of the entire convecting region, by
the ratio L/δ. A dimensionless number known as the Nusselt number expresses this fact:

total heat transport qS L

Nu = ≡ = . (4.22)
heat transport by conduction ∆ T δ
k
L
The Nusselt number is equal to 1 if no heat is transported by convection, and it rises
above 1 as convection becomes more vigorous. The mean convective velocity νconv ̅ can now
be estimated from the convective heat flux. The heat content of a unit volume of rising hot
material is of order ρcp ΔT, and the volume arriving per unit time and per unit area at the
̅ . Add this to the rate at which heat is transported
base of the cold boundary layer is just νconv
by conduction per unit area through the entire layer, kΔT/L, and use the definition of the
Nusselt number,
∆T
k + vconv ρ cP ∆T
Nu = L (4.23)
∆T
k
L

to derive the mean velocity of convection:

κ
vconv ≈
L
( Nu − 1). (4.24)

Applying this equation to the Earth, the only planet for which seismology gives us inde-
pendent information on the boundary layer thickness, δEarth ~ 100 km (this is often stated to
be the thickness of the tectonic plates, which is correct if the thermal boundary layer thick-
ness is meant – it is much larger than the elastic plate thickness), yields a mean convective
velocity of about 0.2 mm/yr. This is much smaller than the plate velocities themselves, but
it does agree rather well with the relative velocities between the mantle plumes and, thus,

EBSCOhost - printed on 11/3/2023 9:54 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
4.3 Planetary engines 125

reflects the speed of gross motions deep in the mantle. Equation (4.24) is only an average:
Rising or sinking plumes may be much smaller than the convecting region as a whole and
may, thus, move faster by the inverse ratio of their size to that of the convecting layer.
Taking this smaller scale to be that of the boundary layer itself, we have

L κ
vplume ≈ v = Nu ( Nu − 1). (4.25)
δ conv L
Parameterized convection. What is still missing is a means of estimating the Nusselt
number, or equivalently, the boundary layer thickness. A simple means of doing this, which
seems to hold up very well even in rather complex rheologies and with complicated planet-
ary structures, is through parameterized convection models (Schubert, 1975). Such models
relate the Nusselt number to the Rayleigh number through a simple power law:
β
 Ra 
Nu =  (4.26)
 Ra C 

where RaC is the Rayleigh number at the onset of convection and β is a power that is close
to1/3 (the best measurements for high Rayleigh number suggest that it is actually about
0.29).
The convective velocity, Equation (4.24), allows us to estimate the rates of convective
processes. In addition, we need an estimate of the stresses generated by convective motions.
The buoyancy stress of a rising (or sinking) mass of material is given by ραVΔTgd, where
d is some “appropriate” distance scale. For most tectonic processes this is the thickness of
the boundary layer δ, not the depth of the convecting region itself. Furthermore, the tem-
perature difference ΔT can be computed from the surface heat flow qS (assuming that most
of the temperature drop occurs across the boundary layer) to give an estimate of the stress
generated by plume-scale convective processes:

ρ α V qS g 2  Ra q  η κ
σ plume ≈ ( ρ α V ∆T g ) δ = δ = 2 2 . (4.27)
k  Nu  L

The crucial parameter missing before the above equations can be used to obtain numer-
ical estimates of convective velocity, strain rate, and stress is the viscosity, η. Because
the viscosity of hot, creeping solids is a strong function of composition, temperature, and
stress, it might seem impossible to make a meaningful estimate of the viscosity of any
given planetary body. On the Earth, some viscosity estimates are available from the rate of
post-glacial rebound, from which it appears that the upper mantle, at least, has a viscosity
in the neighborhood of 1021 Pa-s. However, aside from some rather crude bounds on vis-
cosity from the rate of impact crater relaxation on a few extraterrestrial bodies, the mantle
viscosity of other planets and satellites is not constrained by observations.
A recipe for planetary viscosity. Although this situation may seem dismal, there is an
argument that may make meaningful viscosity estimates possible. This argument dates

EBSCOhost - printed on 11/3/2023 9:54 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
126 Tectonics

Table 4.7 Convection in the terrestrial planets

Quantity: Mercury Venus Earth Moon Mars

Mantle depth, km 640 3000 3000 1500 2000

Surface heat flow, mW/m2 13a 56a 81 27 23a
Rayleigh number 7.1 × 104 5.3 × 107 7.7 × 107 9.3 × 105 3.9 × 106
Nusselt number 3.4 23.4 26.1 7.3 11.0
Thermal boundary layer thickness, km 186 130 110 210 180
Mean convection velocity, mm/yr 0.09 0.18 0.20 0.10 0.12
Plume velocity, mm/yr 0.31 4.2 5.3 0.72 1.3
Plume-scale convective stresses, MPa 38 190 240 43 66

Assuming: k = 3.3 W/m-K, κ = 7.5 x 107 m2/s, η = 1021 Pa-s, ρ = 3200 kg/m3.
a
Estimated by assuming a bulk heat generation equal to that of carbonaceous chondrite material.
This is multiplied by 0.55 for Mercury to account for its excess iron abundance.

back to the dawn of plate tectonics in the late 1960s when an eccentric geophysicist,
David Tozer, was impressed that thermal convection in strongly temperature-depend-
ent materials tends to be self-regulating (Tozer, 1965, 1970). Because of the exponen-
tial temperature dependence of viscosity, a small change in the mean temperature of a
planetary mantle makes a big change in the rate of heat transport. Tozer reasoned that
the internal temperature of a convecting planet adjusts itself to carry all internally gen-
erated heat to the surface, but that the necessary temperature variations are relatively
small. He then concluded that the viscosity of all convecting planetary mantles is virtu-
ally identical: 1021 Pa-s. Although the logic behind this assertion is somewhat elusive,
detailed modeling of convection in planetary mantles does exhibit strong self-regulation
and the models show only one or two orders of magnitude variation in the mean viscosity
(Davies, 1980; Schubert, 1975), either from one planet to another or over the course of
planetary evolution.
Convecting planets. Provisionally accepting Tozer’s prescription for viscosity, Table 4.7
summarizes the major features of mantle convection and the tectonic framework for the
terrestrial planets and Earth’s moon. Not unexpectedly Earth, the largest planet, convects
the most vigorously, has the highest plume velocities, and its lithosphere is subject to the
largest stresses. Perhaps this is the reason that Earth, apparently alone of the terrestrial
planets, possesses plate tectonics, although the presence of a weakening agent, water, in
its lithosphere may play a larger role (water is not taken into account in the estimates
presented here). Plume velocities in Table 4.7 predicted by Equation (4.25) are about an
order of magnitude lower than average plate velocities. This is because the crude convec-
tion model described in this section does not consider plate tectonics explicitly. The vel-
ocities of the plates are largely determined by a balance between the negative buoyancy
of the subducted lithosphere and drag along the transform plate boundaries and are, thus,
not correctly computed from simple convection models. Venus is very similar to Earth,

EBSCOhost - printed on 11/3/2023 9:54 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
4.4 Rates of tectonic deformation 127

but its elastic lithosphere is even thinner because of its high surface temperature: If Venus
possesses plates, they are individually much smaller than the Earth’s, and there is little
evidence for subduction recycling. On Venus, volcanic resurfacing seems to be the rule.
The interiors of Mercury, Mars, and the Moon are imprisoned within thick lithospheres,
convective stresses are small and internal strain rates are low. These are one-plate plan-
ets whose tectonic framework is dominated by global stress systems that are apparently
insensitive to internal convective processes.
An important discovery that derived from the parameterized convection models in the
1970s is that the Earth’s internal heat generation and present heat loss through the surface
are not in balance. Previous to this time it was assumed that because of the rapid turnover
of its internal convective cells, the Earth’s primordial heat had long since been dissipated.
What was not realized is that the interiors of these cells exchange heat with the rising and
falling plumes by conduction, for which the time constant is very long. Careful thermal
modeling shows that about half of the Earth’s present heat flow comes from heat buried at
the time the planet formed (Schubert et al., 2001). In the case of less vigorously convect-
ing planets this time constant is even longer, so that it is doubtful that any of the terrestrial
planets’ heat flow today reflects the rate of internal heat generation from radioactive decay.
This fact makes simple computations such as those reported in Table 4.7 somewhat dubi-
ous: Each planetary body must be treated on a case-by-case basis, with careful attention to
the particulars of each planet and its history.

4.4 Rates of tectonic deformation

The tempo of planetary tectonic movements varies over the entire spectrum, from the near-
instantaneous response to large asteroid impacts to very slow movements that are per-
ceptible only when integrated over the age of the Solar System. Rates are important for
understanding the mechanisms that form tectonic features: Stresses due to creep and vis-
cosity are functions of the strain rate. The Maxwell time plays a dominant role in rate con-
siderations. For example, on the timescale of an impact, earthquake, or even the monthly
lunar tide, the Earth’s mantle acts like a massive elastic body with a shear modulus larger
than steel. However, on the much longer timescales associated with internal convection, the
Earth’s mantle is utterly strengthless and flows like warm honey under the small stresses
caused by thermal expansion and contraction. The mantle’s Maxwell time of about a cen-
tury defines the watershed between these two contrasting types of behavior.
The slowest rates of tectonic deformation are those related to thermal conduction in bod-
ies of planetary dimensions. The cooling and contraction timescale for an object the size of
our Moon is measured in tens of billions of years. Even the larger asteroids, like Ceres or
Vesta, are still gradually losing their primordial heat. Tectonic deformation of such objects
is very slow and strains are small: Unless other processes, such as impacts, intervene, the
surfaces of such objects should reflect their most ancient histories.
Early in Solar System history, however, radiogenic heating of planetary precursors was
intense. Short-lived radioactive species, especially 26Al and 60Fe, were active. Planetesimals

EBSCOhost - printed on 11/3/2023 9:54 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
128 Tectonics

as small as 30 km in diameter, if formed early enough, could have melted and differenti-
ated. We know little of such objects at present, aside from a small number of differentiated
meteorites, but continued exploration of objects in the asteroid and Kuiper belts may reveal
a few survivors whose tectonic and volcanic evolution was both short and intense.
The rates of tectonic processes caused by tidal heating vary enormously. Because tidal
dissipation depends upon the inverse sixth power of the distance from the primary, tidal
heating may vary from dominance, as for Jupiter’s moon Io, to unimportance, as for
Jupiter’s Callisto, even within a single satellite system. Strain rates vary similarly. Because
of volcanic resurfacing, Io’s crust is continually sinking (and being regenerated) at the rate
of about 1.5 cm/yr, leading to strain rates in the thin elastic crust Δa/a ~ 3 x 10–16 s–1. On
the other hand, Mercury lost its primordial spin by solar tidal friction over a period of per-
haps 300 Myr, during which time its original equatorial bulge of, say, 20 km, relaxed away,
implying a strain rate of about 10–18 s–1.
Strain rates in planets large enough to support internal convection are controlled by the
speed of convective overturn and, when convection is vigorous, by the impingement of ris-
ing and sinking plumes on the surface. In the Earth and Venus this occurs at velocities of a
fraction of a centimeter per year and the associated strain rates are around 10–15 s–1. Earth’s
plate tectonic regime implies much higher strain rates, up to a maximum near 10–13 s–1 near
plate boundaries. Similar strain rates may have affected the crust of Venus during its lat-
est episode of global resurfacing. Internal strain rates in Mars and Mercury are both much
smaller and only dubiously manifested on their surfaces: On these bodies the dominant tec-
tonic features seem to be related to surface loading by either impact or volcanic processes.

4.5 Flexures and folds

4.5.1 Compression: folding of rocks

The distortion of layered rocks into spectacular drape-like folds was one of the earliest
observations in the history of geology. Even the earliest observers, such as James Hutton
(1726–1797), recognized that folding is caused by compression. Although Hutton had no
clear idea of the origin or even magnitude of the forces necessary, folding clearly indicates
shortening of the rock layers and was quickly attributed to a cooling, and therefore shrink-
ing, Earth by a series of “dynamical geologists” that extended from Lord Kelvin through
Harold Jeffreys. Mapping of the Alps and Appalachian Mountains revealed that their rocks
are thrown into enormous trains of folds whose wavelengths often reach tens of kilometers
(Figure 4.3). The analogy between these mountain-scale folds and the more humble folds
that develop in crumpled piles of paper or fabric is irresistible: It led to many elaborate
experiments to reproduce mountain structure in the laboratory. USGS geologist Bailey
Willis, for example, devoted years to re-creating the Appalachian Mountains’ structure in a
large wooden box filled with layers of carpeting and tar (Willis, 1893).
Visual analogies, however, are often poor guides to mechanics, and folding provides a
particularly poignant illustration of this fact. The mathematical theory of folding by elastic

EBSCOhost - printed on 11/3/2023 9:54 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
4.5 Flexures and folds 129

(a)

(b)

Figure 4.3 Folded mountains on Earth and Venus. (a) Shuttle synthetic aperture radar image of the
area near Sunbury, PA, USA, in the Appalachian folded Valley and Ridge Province. The frame is
30.5 km wide and 38 km high. North is to the upper right. The large river is the Susquehanna and the
tributary is the West Branch River. NASA image PIA01306. (b) Northern portion of Ovda Regio near
the equator of Venus. This image shows compressional east–west ridges cut by orthogonal extensional
fractures. Note that in these radar backscatter images created by Magellan synthetic aperture radar,
dark is smooth (at 12.6 cm wavelength) and bright is rough. The frame is 300 km wide and 225 km
high, approximately 10 times larger than the image in (a). NASA image PIA00202.

instability is briefly outlined in Box 4.1, and illustrated in Figure 4.4. Equation (B4.1.7)
gives the minimum stress to initiate buckling in a floating elastic plate of thickness t. The
minimum stress grows as the square root of the plate thickness. Inserting values of the elastic
constants and density suitable for laboratory materials, one finds good agreement with the
results of small-scale experiments, such as those of Bailey Willis, or the archetypical wrink-
ling skin of a drying apple. However, as soon as these results are scaled up to the dimensions
of the Earth, trouble arises. Harold Jeffreys, through many editions of his famous book, The
Earth, scoffed at the idea that elastic instability had anything to do with mountain ranges. He
turned Equation (B4.1.7) around and asked the question: “What is the maximum thickness
of an elastic layer that can buckle before the stress reaches the fracture limit?” His answer
is surprising: 12 m is the maximum thickness of a granite layer that can buckle into folds
by elastic instability before the stress exceeds its crushing strength! If instead one inserts
values of the oceanic lithosphere’s flexural rigidity deduced directly from seamount loading,
D = 6 x 1022 N-m, corresponding to a 16 km thick elastic lithosphere, one finds that it would
buckle into folds with a wavelength of about 240 km, but that the minimum buckling stress
is 5.3 GPa – about 50 times larger than the crushing strength of basalt!
The theory of elastic folding, thus, offers a classic case of appealing small-scale model
experiments whose extrapolation to the planetary scale runs into insuperable conflict
with a well-tested mechanical analysis. Many efforts have been made to modify the

EBSCOhost - printed on 11/3/2023 9:54 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
130 Tectonics

Box 4.1 Elastic and viscous buckling theory

Leonhard Euler (1707–1783) initiated the theory of elastic buckling in a treatise on elastic
flexure in 1743. However, the theory was not applied to geological problems until the dawn
of the twentieth century. The theory begins with the equation for the deformation of a floating
elastic plate, discussed in Box 3.3, when no external load is applied, q(x) = 0, but in the
presence of a compressive load N (per unit width of the plate) in the plane of the plate:

∂4w ∂2 w
D + N + ρa g w = 0. (B4.1.1)
∂x 4 ∂x 2

The buckling instability essentially involves the amplification of a small initial deformation
to the point where it dominates the appearance of the plate (see Figure 4.4). It is assumed
that any real planetary surface possesses some initial topography at all wavelengths (in the
sense that Fourier analysis of even random irregularities contains non-zero components at all
possible wavelengths). The vertical deflection of the plate, w, is thus divided into an initial
deflection, w0, and a component wb representing the additional deflection due to the applied
horizontal force N, w = w0 + wb. Assuming that the plate is in stress-free equilibrium when wb
is absent (that is, w0 is a solution to (B4.1.1) when N = 0), substitution of w into (B4.1.1) yields
an equation for wb:

∂ 4 wb ∂ 2 ( w0 + wb )
D 4
+N + ρa g wb = 0. (B4.1.2)
∂x ∂x 2

Assuming that the Fourier component of the initial deflection w0 is equal to a0 sin k x,
where k is the wavenumber (equal to 2π/λ, where λ is the wavelength), the solution to (B4.1.2)
is found by supposing that wb is of form C sin k x, substituting, then solving the algebraic
equation for the unknown C. The overall result of this process is that:

 Dk + ρa g 
4
w = (a0 + C )sin kx =  4 2  a0 sin kx. (B4.1.3)
 Dk − Nk + ρa g 

The main thing to notice about this solution is that the “amplification factor” in brackets
becomes infinite if the denominator vanishes. This occurs when the compressive force N
attains the special value:

ρa g
N b = Dk 2 + . (B4.1.4)
k2

The buckling force Nb depends on the wavenumber k. Because the first term in Equation
(B4.1.4) grows as k2 for large values of k and the second term grows as 1/k2 for small values of
k, there is some intermediate value of k at which the amplification factor grows toward infinity
for a minimal applied compressive load. Setting the derivative of Nb with respect to k equal to
zero, it is easy to show that this unique wavenumber is:
1/ 4
 ρ g 2
kb =  a  = . (B4.1.5)
 D  α

EBSCOhost - printed on 11/3/2023 9:54 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
4.5 Flexures and folds 131

Box 4.1 (cont.)

where α is the flexural parameter defined in Chapter 3. The buckling wavelength that
corresponds to this wavenumber is given by λb = 2 π α , so that the characteristic wavelength
for buckling is directly proportional to the flexural parameter (by a factor of “order one” equal
to 4.44).
Of course, the deformation at this unique “buckling wavelength” is not actually infinite,
because we have used a flexure theory that is only valid for deformations that are small
compared to the plate thickness. Nevertheless, numerical studies of buckling that incorporate
realistic deformation models agree with this simple model that, when the buckling force is
exceeded, large-amplitude sinusoidal deformation pops up rapidly at just the wavenumber
computed from Equation (B4.1.5).
The minimum force needed to initiate buckling, Nmin, is computed by inserting the
wavenumber (B4.1.5) in Equation (B4.1.4):

4D
N min = = 2 ρa g D . (B4.1.6)
α2

The corresponding stress, σmin, is obtained by dividing the minimum buckling force by the
lithosphere thickness t. Expanding the definition of the flexural rigidity D of a plate:

N min ρa g E t
σ min = = (B4.1.7)
t (
3 1 −ν 2 )
where ν is Poisson’s ratio and E is Young’s elastic modulus.
Buckling of compressed layers is not confined to floating elastic plates. The sinusoidal
deformation of layers that we call folding arises from an antagonism between the elastic forces
tending to relieve in-plane compression by lateral displacement and restoring forces tending to
limit lateral motion. In Equation (B4.1.1) the restoring force is the weight of the rising folds,
expressed by the third term on the left, ρagw. However, this is not the only possible restoring
force.
Another buckling scenario is a stiff elastic layer embedded in a surrounding medium of
lower elastic modulus. In this case the restoring force in (B4.1.1) is replaced by an elastic
reaction force, E0kw/(1−ν02), where E0 is Young’s modulus of the soft layer and ν0 its Poisson’s
ratio. Note the explicit appearance of the wavenumber k in this expression. After an analysis
similar to that described above, the buckling wavelength is computed to be:
1/ 3
 E (1 − ν 02 ) 
λb = 2π t  2 
. (B4.1.8)
 6 E0 (1 − ν ) 

The most important result of this calculation is that the buckling wavelength is directly
proportional to the layer thickness. This prediction agrees well with many observations of folds
in rock layers, where the wavelength of the fold is clearly proportional to the layer thickness:
Thin layers are buckled into a large number of short-wavelength folds, while thick layers are
distorted in broad, open folds. An assembly of data on folds and layer thickness shows a good
correlation over an astonishing range of wavelengths, ranging from centimeters to 30 km

EBSCOhost - printed on 11/3/2023 9:54 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
132 Tectonics

Box 4.1 (cont.)

(Currie et al., 1962). Over this range the wavelength seems to be approximately 27 times the
layer thickness. The reason for this particular numerical value is presently obscure.
The stress required to initiate buckling in the layer in this case is given by a rather
complicated formula:
2/3
3 D1 / 3  E0 
σ min =  2 
. (B4.1.9)
t  2 (1 − ν 0 ) 

A very similar analysis can be performed for compression of layers of differing viscosity.
Although the mechanics of elastic and viscous flow differs profoundly, elastic strain can often
be simply replaced by the viscous strain rate in the mathematical equations and the same
analysis carried through. This works well for compression of layers of differing viscosity.
The result of such an analysis is an equation for the viscous buckling wavelength that closely
resembles (B4.1.8):
1/ 3
 η 
λb = 2π t   . (B4.1.10)
 6η0 

As for elastic buckling, the wavelength is directly proportional to the layer thickness.
The major difference between elastic and viscous buckling is that for viscous layers there is
no minimum buckling stress, thus avoiding the problem (discussed in the text) that elastic
buckling often requires stresses far exceeding the fracture stress. Another difference is that,
although the “buckling” wavelength corresponds to the fastest growing instability, the growth
rate is not infinite but instead is limited by the viscosity of the layer and its surrounding
medium. For more details see the books by Johnson et al., cited below.
Elastic and viscous buckling are the subjects of several monographs. The interested reader
will find discussions of elastic buckling applied to folds in terrestrial rocks in Johnson (1970).
Viscous buckling is discussed at length by Johnson and Fletcher (1994) and Pollard and
Fletcher (2005). Ramsay (1967) offers a comprehensive description of folded rocks on the
Earth along with a clear general discussion of the theory of buckling.

conclusions of the mechanical analysis. One of the most plausible is to recognize that
layered rocks are not simple monolithic masses but, like reams of paper or piles of car-
peting, are composed of much thinner layers that can slide over one another. Because the
flexural rigidity D of a layer is a function of the thickness of the layer cubed, if the layer
is divided into n sublayers each of thickness t/n, and if the layers flex independently of
one another, the flexural rigidity of the entire pile of n layers is proportional to n (t/n)3, so
that splitting a layer into n sublayers decreases its flexural rigidity by 1/n2. The buckling
stress, Equation (B4.1.7), depends on the square root of D, so that dividing the layer into
n sublayers decreases the buckling stress by 1/n, and decreases the buckling wavelength
by 1/ n . Of course, this explanation depends on the layers’ ability to slip freely over
one another, and friction between layers greatly diminishes their independent motion.

EBSCOhost - printed on 11/3/2023 9:54 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
4.5 Flexures and folds 133

N ⇒ E,ν w
⇒N
ρa

λb

Figure 4.4 Buckling of a floating elastic plate under a horizontal load N. The deflection of the
midplate neutral sheet (dashed line) is w, the wavelength of the buckled plate is λb, the Young’s
modulus of the plate is E, its Poisson’s ratio is ν, the density of the plate is ρl, and that of the
underlying asthenosphere is ρa. The surface acceleration of gravity is g.

Other efforts to preserve the major results of buckling theory appeal to alternation of soft
and stiff layers or to buckling of viscous or even viscoelastic layers. It seems that buck-
ling theory does apply in many cases to small-scale folds in rocks, and many monographs
expound buckling theory for that reason, finding good qualitative and even quantitative
agreement between small-scale observations of crumpled rocks and this theory.
Buckling theory has been applied to explain undulations observed on the surface of
Europa (Prockter and Pappalardo, 2000), in this case appealing to the special rheology of
ice to argue for a very thin elastic layer of small elastic modulus and thus a low buckling
stress. Seismic reflection images of the Indian plate on the Earth reveal east–west ridges
that strongly resemble compressional buckles consistent with the stress state in the plate
(Weissel et al., 1980). Although the geometry of these ridges suggests elastic buckling,
the required stresses are again excessive, around 2.4 GPa, greatly exceeding the estimated
crushing strength of rock. There is no evidence for thin layering in this oceanic plate.
However, it has been suggested that the pervasive fracturing of the plate might lower its
flexural rigidity by the required amount (Wallace and Melosh, 1994).

4.5.2 Folding vs. faulting: fault-bend folds

Folded mountain belts are apparent on both the Earth and Venus, as shown in Figure 4.3.
Even the Moon and Mars possess smaller-scale folds in the form of mare ridges. If these
are not the result of elastic buckling, then how do they form? Geologist John Suppe,
who made a detailed study of the Pine Mountain structure in the southern Appalachians,
discovered an answer to this question (Suppe, 1983). Suppe found that the large-scale
folds in this area, and elsewhere in the Appalachians, developed as the layered rocks
were pushed up and over sloping steps created by compressional thrust faults that cut
through the rock layers. This model requires that the rock layers glide freely along nearly
horizontal detachment faults between the rock layers until they meet a sloping step, up
which they are forced to move by regional compressional stresses. The rocks then bend

EBSCOhost - printed on 11/3/2023 9:54 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
134 Tectonics

Figure 4.5 Formation of a fault-bend fold. The top frame shows a layered stack of rocks that are cut
by a thrust fault dipping at about 30° from the horizontal. The fault becomes horizontal, following
weaker beds, both above and below the dipping fault. The middle frame shows the evolution of the
fault as the layered rocks are thrust up and over the fault ramp. Bending takes place in narrowly
defined hinge zones, here represented as perfect planes (dashed lines). The lowest frame shows the
further evolution of the compressed stack of rocks, indicating the formation of a ridge on the surface
that overlies horizontal rocks below it. Concept after Suppe (1983).

to accommodate the distortion, as shown in Figure 4.5. Although this model does a good
job explaining the geometry of the folds, it only shifts the stress problem from one of
elastic buckling to one of explaining how rocks can glide apparently effortlessly along
detachment faults. Nevertheless, the fault-bend fold model gives an excellent description
of folds in a variety of circumstances (Woodward et al., 1989), and is widely used in oil
exploration. The stress problem may be partly resolved, at least on Earth, by the presence
of pressurized fluids (water or oil) in the compressed rock mass, but this explanation can-
not apply to Venus or the Moon.

EBSCOhost - printed on 11/3/2023 9:54 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
4.5 Flexures and folds 135

Brittle

Ductile

Figure 4.6 Schematic of mare-ridge formation on Mars. The steep scarp of these mare ridges occurs
along a jump in surface elevation that reveals the presence of a deep-seated thrust fault. The width
of the gently sloping surface above the dipping fault is about two times the depth of the 30° dipping
fault. After Mueller and Golombek (2004).

The fundamental idea of fault-bend folding is consistent with the appearance of mare-
ridge folds on Mars (Mueller and Golombek, 2004). As shown in Figure 4.6, the Martian
ridges separate areas of different surface elevation, strongly suggesting that the surface
folding is a manifestation of deep-seated detachment faulting in the Martian crust.

4.5.3 Extension: boudinage or necking instability

Whereas crumpling due to compression is a matter of daily experience, the formation of
wavelike thickness variations in a stretched layer is much less familiar. Nevertheless, geol-
ogists have long recognized sinusoidal variations in the thickness of stretched rock layers.
Because these pinch-and-swell variations resemble strings of sausages on the face of rock
outcrops, they are commonly called boudins, after the French word for sausage. In three
dimensions they do not resemble sausages at all, as they form in parallel waves on the sur-
face of the stretched layers. They are easily distinguished from folds because the thickness
of a folded layer remains nearly constant through the fold, whereas boudins vary greatly in
thickness (Figure 4.7), to the extent that individual boudins may become detached, at which
point the stretched layer comes to resemble a collection of parallel, elliptical cylinders.
Boudinage is unfamiliar because it requires special rheological conditions to form. A
Newtonian viscous or elastic layer does not spontaneously break into sinusoidal waves as
it is stretched: It merely becomes thinner. Mathematical analysis of stretching reveals the
fundamental reason: Boudins form only in materials in which the stress–strain relation
is non-linear. Sometimes called necking or pinch-and-swell instability, it occurs only in
materials that flow more readily when subjected to larger stresses or strains. As a uniform
layer is stretched, local regions that may be just a bit thinner than average concentrate stress
and strain. If that concentration lowers the resistance to further strain, the layer becomes

EBSCOhost - printed on 11/3/2023 9:54 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
136 Tectonics

⇒ ⇒

Figure 4.7 Schematic illustration of boudinage or pinch-and-swell deformation of a floating viscous

or plastic lithosphere that is undergoing extension. Boudinage requires a non-linear relation between
stress and strain and so is somewhat unfamiliar in daily experience.

still thinner than average and the process runs away, eventually producing regularly spaced
variations in layer thickness.
Obscure as this process might seem, it is now believed to account for one of the most
widespread terrain types on Ganymede (Figure 4.8), as well as the periodically spaced
mountain ranges of the Earth’s basin and range provinces (Fletcher and Hallet, 1983).
Ganymede’s grooved terrain exhibits a variety of dominant wavelengths, ranging from
about 2 to 15 km. The prominent grooves, with amplitudes up to 500 m and lengths ranging
from hundreds to thousands of kilometers, occur in regional sets that crosscut one another
and evidently record a long history of extension (Bland and Showman, 2007; Collins et al.,
1998). Because there is little evidence for compression or subduction on Ganymede, they
strongly suggest an era in which the satellite expanded, perhaps due to internal rearrange-
ments or tidal heating.

4.5.4 Gravitational instability: diapirs and intrusions

In addition to the horizontal forces of compression or extension, vertical forces arising
from density instabilities may warp planetary lithospheres. In actively convecting planets
these could be generated by warm plumes of material rising from deep within the mantle.
Even within a planetary crust, basal melting caused by intrusion of hot melts can mobil-
ize rocks, which then rise toward the surface because of their lower density. On the Earth,
this process was first recognized in mushroom-shaped structures produced by the rise of
low-density salt buried under accumulating thicknesses of overlying sediment (Figure 4.9).
Swedish geologist Hans Ramberg made this process the prime focus of his research career
(Ramberg, 1967). Through mathematical analysis and extensive experimentation with cen-
trifuged models, he elucidated the forms taken by gravitationally unstable configurations
and correlated them with structures observed in the Earth’s crust.
The slow upward movement of viscous masses of low-density rock material (whether it is
hot rock, salt, warm ice or even mud) often disrupts and pierces through overlying rock lay-
ers. Such structures are called diapirs and the process by which they are emplaced is called
diapiric intrusion. Because petroleum is often associated with salt diapirs, particularly in

EBSCOhost - printed on 11/3/2023 9:54 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
4.5 Flexures and folds 137

Figure 4.8 Grooved terrain in Nippur Sulcus on Ganymede. Groove spacing is 5–10 km. This
Galileo image is approximately 79 km wide by 57 km high, and centered at 51° latitude and 204°
longitude. North is toward the top. The largest crater in the image is 12 km in diameter. NASA image
PIA01086.

Incipient Growing Mature

Figure 4.9 The initiation and growth of a viscous layer of density lower than that of the overlying
material. An incipient instability forms, triggered by small accidental variations in layer thickness.
This layer begins to thicken, bulging upward while the adjacent low-density material flows toward
and into the rising column. Eventually this rising plume impinges on the surface. It then spreads
laterally, warping the overlying surface upward and often fracturing it. The base of the low-density
column is usually surrounded by a depressed moat, called a rim syncline.

the Middle East and the Gulf Coast of the United States, they are well studied and their
structures at depth are well understood. As diapirs rise, adjacent rocks are broken and tilted
upward and away from the core of the rising mass. Where they reach the surface, overlying
rocks are uplifted and stretched. Radial extensional faults often mark their location, as does
a local minimum in the acceleration of gravity. Where multiple diapirs reach the surface,

EBSCOhost - printed on 11/3/2023 9:54 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
138 Tectonics

Figure 4.10 Selu Corona on Venus measures 350 km in diameter. It is an archetypical corona marked
by both radial and concentric fractures, probably created as hot plumes from Venus’ deeper mantle
rose and uplifted the surface. The plumes released melt as they rose, feeding lava flows on the surface
and perhaps injecting lava beneath the surface. Portion of NASA Magellan image C145S110.

their horizontal spacing offers clues to both the thickness of the buried source layer and the
depth of the overlying rocks.
Planetary examples of diapiric surface modifications include coronae on Venus,
which are crudely circular volcano-tectonic structures that range from about 100 to
more than 1000 km in diameter (Figure 4.10). Venusian coronae are interpreted as the
product of rising masses of hot mantle material that first approached the surface, bowed
it upwards and thus generated an initial radial fracture system (Squyres et al., 1992).
The hot underlying material then spread out underneath the surface, generating melts
and feeding lava flows in addition to subsurface intrusions. The region eventually cooled
and subsided, producing a flexural trough and the associated concentric faults surround-
ing the surface load.
On a much smaller scale, the similarly named coronae on Uranus’ tiny moon Miranda
(480 km in diameter) are also believed to be the surface manifestations of large diapirs
(Figure 4.11) that pushed toward the surface early in Miranda’s history. The major diffe-
rence is that upwarping of the surface on this small body created concentric, not radial,
extensional faults. This is almost certainly due to the large curvature of Miranda’s litho-
sphere and the flip in tectonic style that results from the curvature of the elastic shell (Janes
and Melosh, 1990).

EBSCOhost - printed on 11/3/2023 9:54 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
4.6 Fractures and faults 139

Figure 4.11 Mosaic of Uranus’ satellite Miranda from Voyager 2. Miranda is 480 km in diameter.
This image shows the striped and ridged terrain composing three large coronae, Elsinore at the top,
Inverness to the right of center, and Arden at the bottom. The ridges are believed to represent normal
faults, and the terrain is created as plumes rose from Miranda’s interior to just under the surface.
NASA image PIA01490.

4.6 Fractures and faults

4.6.1 Why faults? Localization

A casual glance at nearly any steep hillside or roadside outcrop reveals that nearly every rock
mass is broken by fractures. Cracks in rocks form at all scales, from tiny microcracks along
grain boundaries to enormous fractures that extend over much of a planet’s circumference.
When cracks cut through a rock mass without sensibly displacing the rocks on either side,
they are called joints, whereas when the rocks are displaced relative to one another, the dis-
placement is called a fault. It is common experience that an overstressed solid breaks along
almost invisibly thin surfaces, leaving the adjacent material largely unchanged: Anyone
who has dropped a glass or plate onto a hard surface can verify this. Rocks underlying the
surfaces of planets are no different. Because this occurrence is so common, few people stop
to ask themselves, why do fractures occur in such narrow zones?

EBSCOhost - printed on 11/3/2023 9:54 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
140 Tectonics

Field geologists have a wider experience of rock fracture. In many places on the Earth
uplift, tilting, and erosion have exposed expanses of rock that once extended deep below
the surface. In these fortunate places one can sometimes find an ancient fault and trace it
from the surface down into the former depths. One finds that the fault, while knife-sharp
near the surface, gradually widened and blurred as it went deeper. At great depths the fault
zone may have widened to the point of invisibility.
The transition between narrow fractures and broad zones of deformation is called the
“brittle–ductile” transition. Brittle behavior implies cracks and narrow zones of deform-
ation. The sudden loss of strength upon fracture may also cause “stick-slip” motion that
may be related to earthquakes. Crockery and glasses, as well as rocks at the surface of our
planet, all fracture in the brittle domain. Nevertheless, increasing the pressure or tempera-
ture in materials like rock can push them into the ductile flow domain. Ductile behavior
is more characteristic of some metals and plastics. Ductile flow is not the same thing as
pseudoviscous creep: Ductile materials retain considerable strength and their strain is not
a function of time. Only at still higher temperatures do rock materials begin to flow under
small stresses.
It is evident that materials such as rock can be either brittle or ductile, depending on
external conditions of temperature or pressure. Figure 4.12 illustrates this transition for a
particularly well-studied rock, Westerly granite (Stesky et al., 1974), and shows how the
observed pressure and temperature dependence of the brittle–ductile transition translates
into mechanical behavior as a function of depth.
But we still have not explained what really underlies the formation of narrow fractures,
the “brittle” behavior so familiar from daily experience. The basic reason for brittle frac-
ture is simple, although a full understanding has taxed the mathematical ingenuity of a
generation of material scientists. The key is to apply the concept of positive feedback.
Positive feedback occurs when the response to some disturbance is enhanced by the dis-
turbance itself, resulting in an overall response that quickly “runs away.” In the case of
brittle fracture, a small region of a stressed, heterogeneous solid may suffer more strain
than adjacent areas. If this extra increment of strain weakens the region, it leads to a fur-
ther increase of strain, and a strain runaway begins. Deformation quickly concentrates in
the already-deformed region and a narrow zone of macroscopic displacement – a joint or
fault – develops. This self-reinforcing response to stress is called strain softening and the
process is called localization. Localization is typical of all brittle materials. Simply stated,
a macroscopic crack forms when an initial increment of failure locally weakens the solid
and causes more failure to concentrate around it, weakening it still further.
The above description of localization is oversimplified: Decades of experimental and
mathematical analysis have considerably refined our understanding of when and how local-
ization occurs. Readers interested in this fascinating topic must probe the original literature
(for example, Hobbs et al., 1990; Rice, 1976). Nevertheless, this idea provides the rationale
for the observed complexities of rock fracture, and offers a qualitative explanation of why
rock failure transitions from brittle to ductile as the pressure and temperature rise.

EBSCOhost - printed on 11/3/2023 9:54 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
4.6 Fractures and faults 141

(a)

0.5 Ductile Pseudoviscous

creep
P, GPa
Brittle

0 600 900
T, K

(b)

Extension
fractures
Schizosphere
Shear Lithosphere
depth

faults

Ductile
flow Tm
~
~ 2
Pseudoviscous Asthenosphere
creep

Figure 4.12 The brittle–ductile transition in rock. Panel (a) shows the three domains of mechanical
behavior as a function of pressure and temperature. Rock is brittle, subject to sudden failure in narrow
fracture zones at low pressures and temperatures. At higher pressures and temperatures it undergoes
a transition to ductile fracture, in which it still retains considerable long-term strength, but failure is
gradual and distributed over wide zones. Only at temperatures exceeding about half of the melting
temperature does rock begin to flow like a highly viscous fluid with no long-term strength. This plot
is constructed on the basis of data on Westerly granite (Stesky et al., 1974). Panel (b) applies these
domains of rock fracture to the crust of the Earth or planets whose surface conditions begin in the
brittle domain. The outermost shell fails in an extensional regime, where fractures are parallel to the
most compressive stress. At higher pressures this transitions to the shear failure regime, in which
fractures form at an acute angle to the most compressive stress (refer to Figure 4.14 for more on this).
Both regions are in the brittle fracture domain termed the schizosphere by C. Scholz (1990). Still
deeper is the ductile flow domain. All of these domains that possess long-term strength are termed the
elastic lithosphere. Still deeper, rising temperatures permit rock to flow under arbitrarily low stresses
by pseudoviscous creep. This region is the asthenosphere.

4.6.2 Joints, joint networks, and lineaments

Joints are ubiquitous cracks in rocks across which no perceptible offset occurs. They fre-
quently form in long parallel sets, periodically spaced at intervals that range from centime-
ters to hundreds of meters. Joints are found in all consolidated rock types, ranging from soft
shales to hard igneous rocks. Special types that define hexagonal columns, called columnar
joints, result from shrinkage of cooling basalt or volcanic tuff layers. Joints are clearly
tensile fractures: Detailed surface features such as hackle marks and plumose patterns
record the propagation of fast-running tensile cracks. Nevertheless, the origin of joints is

EBSCOhost - printed on 11/3/2023 9:54 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
142 Tectonics

not clearly understood, although the mechanics of many special cases has been worked out.
Many circumstances conspire to produce tensile stresses: Cooling near the surface, uplift
of deeply buried rock masses (with attendant cooling), or surface exposure by erosion can
all produce tensile cracks in various orientations (a good description of joints can be found
in Chapter 6 of Suppe, 1985).
Although it may seem that the formation of tensile cracks must be strictly limited to
shallow depths, in the Earth, Mars, and Titan the presence of pressurized fluids in the
crust (water, natural gas or, in Titan’s case, liquid methane), may extend the tensile regime
to many kilometers in depth. Moreover, experiments on the crushing of unconfined rock
specimens indicates that tensile fractures, in which a crack forms in the plane of the com-
pressional force, may occur in the regime where the average of the three principal stresses
is slightly compressive.
Griffith crack theory. The mechanics of tensile failure is now well understood (Gordon,
2006; Lawn and Wilshaw, 1975). A young aeronautical engineer, A. A. Griffith (1893–
1963), was the first to comprehend the relation between cracks and strength. Griffith
supposed that all materials contain initial flaws, tiny cracks of various lengths l. When a
tensile stress acts perpendicular to the plane of the crack, the crack may grow longer if the
energy cost of lengthening the crack is less than the strain energy released by its extension.
Equating these two terms, he found that if the surface energy of the crack is γ J/m2, then
the crack grows when the tensile stress exceeds:

Eγ
σt = (4.28)
l

where E is Young’s modulus. Tensile cracks extend in their own planes and, because of the
1/ l dependence in Equation (4.28), once started they continue to extend until either the
stress is relieved or the crack reaches some impassible barrier.
Because tensile failure depends on the presence of initial flaws, the “strength” of a
rock mass in tension depends on the history of the rock and its population of initial flaws.
Moreover, since large rock masses are more likely to contain an especially vulnerable flaw
than small rock samples, the tensile strength of a rock unit depends on its size. There is no
necessary relation describing the dependence of strength on size, but many empirical stud-
ies suggest that tensile strength depends roughly on the inverse square root of the linear
dimension of a rock.
On a planetary scale rocks are nearly strengthless in tension: When tensile forces develop,
the rocks generally yield and joints, thus, form readily. The great lengths attained by many
joints are due to the ability of tensile cracks to grow in their own plane: Once begun, a ten-
sile crack just gets longer until the applied stress is relieved.
Because joints are weaker than the surrounding rock and offer ingress to corrosive flu-
ids on planets with a hydrosphere, they are frequently expressed in the landscape. Joints
often strongly control the course of rivers and the directions of valleys on the Earth and
perhaps on Mars and Titan. Maps of the density, direction and regional extent of joints
on the Earth are an important part of every geotechnical project, whether for building

EBSCOhost - printed on 11/3/2023 9:54 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
4.6 Fractures and faults 143

roads or evaluating sites for nuclear power plants. Joint directions are used to infer stress
orientations in the lithosphere of the Earth and planets (Engelder, 1993), although care
must be given to the time at which the joints formed. The orientations of impact crater
walls and the head scarps of landslides are often visibly controlled by joint directions in
the pre-existing rocks.
Lineament analysis. Joint analysis has long been an industry on the Earth and it is now
frequently extended to the other planets. Because joints themselves are small-scale fea-
tures, their presence and direction is usually inferred from their effects on the landscape.
This is especially true in planetary science, where high-resolution surface investigations
are currently rather limited. The art of inferring joints (including small-offset faults) from
landscape patterns is called “lineament analysis,” because the fractures that form major
joint sets frequently occur in linear arrays. It is easy to pick out apparently straight features
and alignments on planetary images, a technique at which the human visual processing
system seems particularly adept. However, because of this very ability, great caution needs
to be applied when pursuing this type of analysis.
Non-existent linear features have been reported on the surfaces of planets since Percival
Lowell believed he saw artificial canals on Mars in 1895. More recently, the Apollo 15
astronauts reported layering in the face of Mount Hadley, 4.5 km from their landing site
and took photographs of the “layers” (Figure 4.13a). Subsequent analysis showed that the
apparent “layers” were all aligned with the direction of the Sun. As shown in Figure 4.13b,
experiments on illuminating model hills with random surface textures demonstrate the illu-
sion of layering (Wolfe and Bailey, 1972). These investigations indicate that lineament
analysts must be conscious of such illumination effects (Howard and Larsen, 1972), and
must avoid working near the limits of resolution, where the eye has a tendency to “connect
the dots” (which was probably what fooled Lowell).

4.6.3 Faults: Anderson’s theory of faulting

Faults on planetary surfaces betray their presence by sudden changes of elevation (scarps
and troughs), offset surface features, or linear alignments. Miners and geologists first
recognized them as abrupt breaks or shifts in rock layers. By definition, a fault is a nar-
row surface along which the rocks have moved relative to one another. Although faults
have been recognized and described for several centuries, it was not until the late 1940s
that experimenters, among them the ubiquitous David Griggs (Griggs and Handin, 1960),
began to clarify the mechanics of how they form. Even now, the process of earthquake
faulting, especially on large active faults, is shrouded in deep mystery, but progress has
been made.
Griggs and similar experimenters crushed centimeter-sized rock specimens to failure
under conditions of pressure and temperature approximating those in the Earth. They found
that when the stresses in all directions are compressive, failure occurs by sliding along pla-
nar surfaces inclined at an acute angle, typically about 30°, to the direction of the maximum
compressive load (Figure 4.14). No tensile stresses occur under these conditions, and the

EBSCOhost - printed on 11/3/2023 9:54 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
144 Tectonics

(a)

(b)

Figure 4.13 Panel (a) is an Apollo 15 photograph taken from the lunar surface. It shows Mount
Hadley 4.5 km away from the landing site, on which the astronauts reported prominent sloping
layers. Top portion of NASA photograph AS15–90–12208. Panel (b) is a photograph of a pile of
structureless cement powder, 15 cm high, taken under oblique illuminations similar to that at the
Apollo 15 landing site, indicating how solar illumination can often produce the appearance of layered
structure when none is present. From Figure 7 of Wolfe and Bailey (1972).

simple Griffith crack theory does not apply. The maximum shear stress in the specimen
occurs on planes tilted at 45° to the direction of the applied load, so there is more to this
failure mode than simple shearing.
For many years these results have been rationalized by a graphical stress representation
invented by Otto Mohr (1835–1918). Mohr’s circles enable one to correlate the failure
angle with the shape of the yield envelope by resolving the stresses into normal and tangen-
tial forces acting across the eventual failure plane. Although this type of analysis, which is
expounded in nearly every text and monograph on the subject of rock failure, is applicable
to frictional sliding on a pre-existing fracture, it begs the question of how the inclined frac-
ture surface develops in the first place.
Very recent work applies the theory of sheared microcracks (Paterson and Wong, 2005).
The emerging picture is far from simple: Unlike cracks in tension, sheared cracks do not
propagate in their own plane and thus cannot simply lengthen to create a fracture surface.
Instead, sheared microcracks sprout secondary “wing” cracks that interact with neighboring

EBSCOhost - printed on 11/3/2023 9:54 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
4.6 Fractures and faults 145

Extension Shear Shear Ductile

fracture fracture zone
σ1
σ1
σ1
σ1

σ3 σ3 σ3 σ3

< 1% 1 - 5% 2 - 8% > 10%

τ τ τ τ

ε ε ε ε

Figure 4.14 The effect of increasing pressure on the style of rock fracture. At very low pressures
extensional cracks form parallel to the direction of compression. As pressure increases the failure
surface becomes inclined to the direction of maximum compression at an acute angle, often close to
30°. Increasing pressure results in broader shear zones, until at very high pressures deformation is
distributed throughout the failing rock specimen and the fracture is described as ductile. The upper
row of figures schematically illustrates the failure mode and the lower row plots the relation between
shear stress τ and strain ε. The arrows indicate sudden failure. After Griggs and Handin (1960).

cracks to eventually link up and form the observed inclined failure plane. This explains the
generally irregular fracture plane that forms at low strains. Once a fault is formed and
sliding progresses along the fault, the surface is ground smooth and characteristic “slicken-
side” grooves develop in the direction of motion.
Although such detailed microscopic theories are greatly advancing our understanding
of how faults work, these theories are fortunately not necessary for a basic correlation
between fault type and stress orientation. E. M. Anderson (1877–1960) published a short
but influential book on faults and stress directions (Anderson, 1951) that applied the prem-
ise that most faults are produced by shear. He recognized three basic types of fault and
related each type to the stress field that produces it.
Anderson began by resolving the general stress tensor into its principal stresses and
directions. He observed that, because the surface of a planet (supposed to be essentially
horizontal) is free of shear stresses, one of these principal stresses must be perpendicular to
the surface. There are then three possible kinds of fault, depending on whether that stress
is the maximum, intermediate or minimum principal stress. Accepting the empirical obser-
vation that shear faults form at an acute angle to the maximum compressive direction, he
supposes that the fault plane also contains the intermediate principal stress (Figure 4.15,
left panels). There is an ambiguity, however, because there are two planes that satisfy this
criterion. These are called conjugate planes and either one is a potential fault. Which one
is more important is not determined by the theory: It is supposed that boundary conditions

EBSCOhost - printed on 11/3/2023 9:54 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
146 Tectonics

σ3

σ2
σ1
THRUST

σ1

σ2
σ3

NORMAL

σ2
σ3

σ1

STRIKE
SLIP

Figure 4.15 The relation between shear fractures and large-scale faults according to the theory of
Anderson (1951). One of the three principal stresses must be perpendicular to the free surface, which
results in the separation of possible fracture directions into thrust faults (top), normal faults (middle),
or strike-slip faults (bottom). Right panel after Figure 11–22 in Hartmann (1972).

or geometric constraints finally determine the plane actually selected. Experimentally, in

the initial stages of crushing both planes are often activated, producing X-shaped fracture
patterns.
Normal faults. Anderson’s theory predicts that so-called “normal” faults arise when the
maximum principal compressive stress is vertical, strike-slip faults appear when it is inter-
mediate, and thrust faults occur when it is minimum. Normal faults typically dip steeply, at
an angle of about 60° with respect to the vertical (See Box 4.2). The name “normal” fault
comes from the English Coal Measures, where the most common faults are of this type.
The names thrust and strike-slip should be self-explanatory (Figure 4.15, right panels).
Thrust faults. Thrust faults arise when the maximum compressive stress is horizontal
and the minimum stress is vertical. The dip angle of thrust faults is typically shallow, close
to 30° (See Box 4.2). In a fault of this type a wedge of rock is thrust up and over a sloping
ramp of underlying rock, uplifting the surface on the upthrust side and slightly depressing
it on the opposite side of the fault. This type of faulting commonly produces asymmetric
scarps across which the surface elevation suddenly changes. In plan, thrust faults are typ-
ically lobate. The most famous planetary thrust faults are the lobate scarps of Mercury
(Figure 4.16), which may be several kilometers high and wind hundreds of kilometers

EBSCOhost - printed on 11/3/2023 9:54 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
4.6 Fractures and faults 147

Box 4.2 Dip angle of Anderson faults

Normal faults and thrust faults are two of the most common types observed on planetary
surfaces. Their appearances in images are strongly affected by the angle at which the faults
dip below the surface. Thrust faults dip at shallow angles, typically about 30° with respect to
the surface, and frequently present an irregular, wavy trace where they cut the surface. Normal
faults dip more steeply, at angles near 60°, and their trace is typically straighter because the
steep dip masks small variations in the depth of the fault surface.
The difference in the dip of thrust and normal faults is a simple geometric consequence of
how shear and normal stresses are localized on the fault surface. Although the main text of
this book has avoided dealing with the details of stress components, this box will show how
this profound difference in fault dip comes about for either compressional horizontal stresses
(thrust faults) or extensional stresses (normal faults).
The derivation begins with Coulomb’s frictional law for sliding on a rock surface, Equation
(3.20):
σs = ± ff σn (B4.2.1)

where σs is the shear stress on the potential plane of sliding, ff is the coefficient of friction and
σn is the “normal stress,” or pressure perpendicular to the potential sliding plane. The alternate
+ and – sign indicates that sliding may occur either up the plane or down the plane when this
shear stress is exceeded.
These stresses on the potential sliding surface, which we suppose dips at angle θ with
respect to the horizontal (Figure B4.2.1), must now be related to the vertical stress, σV, and
horizontal stress, σH, acting below the solid surface. The full derivation of this relation is

Figure B4.2.1 Schematic illustration of the stresses on a fault plane dipping at an angle θ with
respect to a horizontal plane. Stresses σn and σs are stresses normal (perpendicular) to the fault
plane and acting across it in shear, respectively. The vertical principal stress σV is nearly equal
to the weight of the overlying rock, while the horizontal principal stress σH varies with tectonic
processes until failure occurs in either compression or extension.

EBSCOhost - printed on 11/3/2023 9:54 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
148 Tectonics

Box 4.2 (cont.)

presented in the book by Turcotte and Schubert (2002, Section 8.4). We simply quote the result
here:

1
σn = (σ + σ V ) − 21 (σ H − σ V ) cos 2θ .
2 H
(B4.2.2)

1
σs = (σ − σ V ) sin 2θ .
2 H
(B4.2.3)

The vertical stress σV is nearly equal to the weight of the rock overlying the fault plane,
σV  ρgz, where ρ is the density of the rock, g is the surface acceleration of gravity, and z
is the depth below the surface. Tectonic forces in addition to the confining pressure of the
surrounding rocks generate the horizontal stress σH, which we treat as a variable in this
computation. We, thus, insert the expressions for the normal (B4.2.2) and shear stresses
(B4.2.3) into the failure Equation (B4.2.1) and solve for the horizontal stress, with the result
that, at failure,

 sin 2θ ± f f (1 + cos 2θ ) 
σH =  σV . (B4.2.4)
 sin 2θ  f f (1 − cos 2θ ) 

To find the dip θ of the plane most susceptible to failure, minimize the expression in square
brackets with respect to this angle by differentiating it with respect to θ and then setting the
derivative equal to zero. The result is:
cot 2θ± = ± ff. (B4.2.5)

This can be simplified by defining the angle of internal friction φ f = arctan ff and using
trigonometric identities to obtain:

π φf
θ± =  . (B4.2.6)
4 2

The angle of internal friction is typically about 30° for most rocks. We can then deduce that
for compressional stresses (upper sign) θ+  30°, whereas for extensional stresses (lower sign)
θ−  60°, in good agreement with observations.

across the surface, cutting craters in their path. The compressive deformation is also indi-
cated by the distortion and shortening of initially circular craters cut by the fault. The
prevalence of thrust fault scarps on Mercury indicates a dominantly compressional stress
regime in Mercury’s crust at the time that they formed, perhaps as a result of cooling and
shrinking of the planet.
Strike-slip faults. Strike-slip faults can be easily recognized in planetary images because
they produce horizontal offsets of surface features. On the Earth, strike-slip faults grow
to enormous lengths because they form one of the three major types of plate boundary.

EBSCOhost - printed on 11/3/2023 9:54 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
4.6 Fractures and faults 149

Figure 4.16 Discovery scarp on Mercury is a thrust fault roughly 350 km long that reaches a
maximum height of 3 km. It transects two craters of 35 (top) and 55 km (middle) diameter. Note that
these craters are shortened in the direction perpendicular to the scarp, as expected for a compressional
thrust fault. NASA Mariner 10 image, PIA02446.

The San Andreas Fault in California and the Alpine Fault in New Zealand are examples.
Strike-slip motion is not so evident on planets that lack plate tectonics, but careful mapping
usually reveals some displacement of this type. Venus and Europa both exhibit strike-slip
faults with a few kilometers of offset, ranging up to more than 40 km on Europa (Hoppa et
al., 1999). Global planetary grids, discussed below, may be networks of conjugate strike-
slip faults with undetectably small offsets.
This simple scheme captures the appearance of many faults on the Earth and other plan-
ets and so is in wide use. Lest the reader think that this scheme is inevitable, a strange
and little-documented episode in the history of American structural geology should serve
as a warning. In the 1920s the first American textbook on structural geology by C. K.
Leith (1923) expounded a geometrical theory of rock failure. Based on distortions of the
strain ellipsoid, this theory predicted that failure occurs on planes making an obtuse angle
to the direction of maximum compression. Structural geologists who accepted this idea
were evidently unable to make any sense of field relations and an entire generation of
American geologists grew to distrust any relationship between faults and stress directions.
Their European counterparts, who did not accept this theory, were more fortunate and the
literature of the time shows them more confident in their stress determinations. This error
was not corrected until the 1940s when a new structural geology textbook by C. M. Nevin
(1942) finally got things right. Of course, this confusion did not prevent the young David
Griggs (1935) from weighing in on the right side of the argument.
Grabens. There is more to faulting, however, than Anderson’s simple theory can describe.
Normal faults have a strong tendency to form together in conjugate pairs, dropping a long,

EBSCOhost - printed on 11/3/2023 9:54 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
150 Tectonics

Figure 4.17 Dozens of parallel grabens form a corduroy-like terrain on parts of Mars. These grabens
north-east of Tharsis at 25° N and 101° W in Ceraunius Fossae are spaced approximately 6–10 km
apart and can be traced up to 300 km along their length. Image is 135 x150 km. NASA Viking frame
39B59.

flat-floored wedge down relative to its surroundings. Such fault-bounded troughs are known
as grabens, from the German word for ditch, and often form in parallel and numerous sets
(Figure 4.17). The reason for this common association is probably due to the fact that
it minimizes distortional energy in the surrounding rocks (Melosh and Williams, 1989).
Telescopic observers were the first to describe grabens on the Moon. Given names like
“straight rilles” and “arcuate rilles,” they form flat-bottomed troughs that stretch hundreds
of kilometers across the surface of the Moon. In a now-classic study George McGill (1971),
and later Matt Golombek (1979), recognized that the increases in the apparent width of
these rilles as they cut through ridges reflect the dips of the bounding faults (Figure 4.18)
and they showed that the normal faults bounding these lunar grabens dip at angles identical
with those of fault-bounded troughs on the Earth. Referring to Figure 4.18, if the apparent

EBSCOhost - printed on 11/3/2023 9:54 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
4.6 Fractures and faults 151

W+
∆W

∆H

ANGLE OF
DIP (α)

Figure 4.18 Schematic illustration of the interaction of a graben with a topographic ridge. Because
the sides of the graben slope at a constant dip angle, when the graben cuts the ridge the apparent
width of the graben expands from W to W+ΔW. If the height of the ridge ΔH is known, the dip angle
can be computed from Equation (4.29). After Figure 1 of McGill (1971).

width of the graben expands from W to W+ΔW and the height of the ridge is ΔH, then the
dip angle α is given by:

 ∆H 
α = tan −1  . (4.29)
 ∆W / 2 

Grabens are found on virtually every planet and satellite in the Solar System and are
one of the most recognizable indications of extensional strain. The floor depth and spacing
of grabens makes it easy to compute the amount of extension quantitatively and, thus, to
estimate the extensional strain, presuming that the terrain between grabens behaves as es-
sentially rigid blocks. Grabens or vertical open cracks also form locally over the tips of ver-
tical dikes that do not quite reach the free surface, and so are common features in volcanic
terrains. Many of this type have been identified on Venus and perhaps Mars (Ernst et al.,
2001). Distinctive lines of nearly rimless conical drainage pits sometimes mark incipient
grabens where extension may create gaping vertical “dilational” faults. Pit chains of this
kind occur on Mars (Ferrill et al., 2004) and maybe explain the distinctive crater chains
on Phobos (Horstman and Melosh, 1989). The spacing between these pits, which require a
loose regolith overlying a more competent substrate, is nearly equal to the thickness of the
regolith layer, and, thus, provides an estimate of the thickness of the loose debris.
A much larger type of graben creates large rift zones on the Earth, such as the East
African rift valleys. These grabens apparently originated with a single normal fault that

EBSCOhost - printed on 11/3/2023 9:54 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
152 Tectonics

cut the entire lithosphere. Flexure then produces extensional strain a distance from the first
fault roughly equal to the flexural parameter and, when the stress is large enough, a second
normal fault forms. Heiskanen and Vening-Meinesz (1958, Part 10D) first described flex-
ural rift zones of this type.
Other combinations of stress produce distinctive structures: Simultaneous compression
and shear create “transpression” structures (known as “flower structure” in the petroleum
geology literature), while extension and shear create “transtension.” Shearing of soft mater-
ial between more rigid blocks produces “Riedel shears,” while shearing of stiff material
between soft layers generates “bookshelf” kinematics, so called because of its similarity
to the motions between a laterally sheared row of upright books. Bends in faults produce
characteristic associations of secondary faults. It would take an entire book to describe
all of the known variations and varieties of faults. Fortunately, there are several, of which
Mandl (1988) is one of the more comprehensive.
Mare ridges. Mare ridges, first observed telescopically on the Moon, where they are
also called “wrinkle ridges,” are attributed to compressional tectonics. These distinctive
features form ridges up to 1 km high, several kilometers wide and hundreds of kilometers
long (Figure 4.19a). The narrow crenulated ridge often lies near the summit of a much
broader upwarp. On the Moon mare ridges are confined to the mare lava plains. They have
now been observed on lava plains on all of the terrestrial planets, with the apparent excep-
tion of the Earth. Even on the Earth, an unusual type of ridge found on the lava plains of the
Columbia Plateau in the Yakima Fold Belt may provide local examples of this otherwise
ubiquitous tectonic feature. The internal structure of lunar mare ridges was probed by the
Apollo 17 electromagnetic sounder experiment, but those images did not lead to a clear
understanding of the nature of these ridges. However, they resemble small-scale (meters
to tens of meters) compressional structures formed in soil during the 1968 Meckering,
Australia, earthquake (Bolt, 1970) and so have been generally attributed to compression.
Analysis of Martian mare ridges using the MOLA laser altimeter revealed offsets in the
elevation of plains on opposite sides of the ridges, strongly implying a kind of incipient
fault-bend folding over deep-seated detachment faults (Mueller and Golombek, 2004). See
also Figure 4.6. Earlier studies of lunar mare ridges also indicated elevation changes across
the ridge. Nevertheless, the en-echelon structure of the mare ridges continues to suggest a
component of strike-slip motion (Tjia, 1970) in the formation of at least the lunar ridges
(Figure 4.19b). At the present time, the nature of mare ridges is not fully understood.
Detachment faults. One of the principal types of fault not described by Anderson’s
theory is the detachment fault, also called a low-angle thrust fault. First mapped in the
Alps, detachment faults are characteristic of compressive mountain belts. As described
above, they are the foundation of the theory of fault-bend folding. Detachment faults
elude Anderson’s fault classification because they are nearly horizontal, parallel to the
free surface. None of the faults shown in Figure 4.15 have this orientation. Either a major
reorientation of the principal stress axes must take place between the free surface and
the fault plane (Melosh, 1990b), or the detachment surface must be a plane of extreme
weakness.

EBSCOhost - printed on 11/3/2023 9:54 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
4.6 Fractures and faults 153

(a) (b)

Figure 4.19 Panel (a) illustrates a prominent mare ridge in Mare Serenitatis on the Moon. North is
toward the top and the frame is 3.5 km across. NASA Apollo 17 image AS17 2313 (P). Panel (b)
is a schematic drawing of the “flower structure” that develops along faults that are simultaneously
compressed and sheared (transpression). After Figure 9 of Lowell (1972).

The first geologists to map detachment faults could scarcely believe their eyes: Many
require displacements of kilometer-thick plates of rock over horizontal distances up to 100
km. It was quickly realized that such faults must be exceedingly weak: If normal frictional
forces act across the fault plane, the overlying plates of rock would crumble long before
they could be pushed along the detachment surface. In some cases fluid materials, such as
salt or evaporites, underlie the detachment and it is plausible that the overlying plate moved
on this viscous substrate. However, in many others the detachment cuts through hard rocks
that cannot be reasonably treated as viscous.
Strength paradox. The apparent low strength of detachment faults has long been attrib-
uted to high fluid pressures in the deforming rocks, beginning with a seminal paper by
Hubbert and Rubey (1959). The empirically determined yield strength of rock was dis-
cussed in Section 3.4.2 and is well described by equations similar to (3.23) and (3.24). A
widely used low-pressure limit of this type of equation that explicitly shows the effect of
fluid pore pressure pf is:
|σs| = Y0 + fF (p − pf) (4.30)
where Y0 is the cohesion, fF the “coefficient of internal friction” and p is the total over-
burden pressure. It is clear from (4.30) that if the pore pressure becomes comparable to the
overburden pressure the strength of rock may become very small: On a pre-existing fault,

EBSCOhost - printed on 11/3/2023 9:54 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
154 Tectonics

for which Y0 is already zero, the strength may vanish. High water pressures are indeed
found in many terrestrial subduction zones where thrust faulting is active. However, there
are some areas (such as the Heart Mountain detachment in Wyoming, USA) where a case
for high fluid pressure is difficult to make. And if Venusian fold belts or lunar mare ridges
are related to detachment faults at depth, it is hard to make a case for subsurface fluids of
any kind.
Strike-slip faults have their own strength problems. Since the late 1960s it has been
suspected that the San Andreas Fault in California is sliding with an apparent coefficient of
friction at least a factor of 10 smaller than expected from rock friction experiments (Zoback
et al., 1987). Although this is not the place to air the ongoing controversy about the strength
of active faults, it is important to indicate that a full understanding of tectonic processes
has not yet been achieved, even on the well-studied and relatively accessible Earth. This
situation, however, adds extra interest and urgency to comparative tectonic studies of the
planets. Because other planets offer examples of tectonic processes in vastly different set-
tings of gravity, pressure, and the abundances of subsurface fluids, comparative planetary
studies of tectonics may be of great importance in resolving the fundamental problem of
what determines the strength of faults.

4.7 Tectonic associations

4.7.1 Planetary grid systems

Geologic mapping of the Moon began well before the spaceflight era. Mare ridges, normal
faults, and grabens were all recognized before about 1960. In addition, several “selenolo-
gists,” a group that included mining geologists familiar with lineament analysis, proposed
a global system of faults that became known as the “lunar grid” (Baldwin, 1963). Deduced
from straight segments of crater walls (Figure 4.20a), bona fide faults, and, in some cases,
confused by troughs created by impact ejecta from the Imbrium basin, the lunar grid
appears to delineate an ancient tectonic fabric buried deep within the lunar crust. Although
the status of this feature is still somewhat contentious, the prominent NW–NE peaks in
both the number and lengths of linear features (Figure 4.20b), with an acute E–W opening
angle, strongly suggest a network of small-offset conjugate strike-slip faults that formed
very early in the Moon’s history (Strom, 1964).
The possible reality and significance of global grid systems got a major boost in 1974,
when Mariner 10’s images of Mercury also revealed a Moon-like planetary grid. In the
case of Mercury, it was quickly suggested that the fault pattern is consistent with strike-
slip faulting expected from a relaxing rotational bulge, perhaps a remnant of a time early
in Solar System history when Mercury spun faster than its current 59-day period (Melosh
and Dzurisin, 1978). In addition, Mercury is covered with a possibly north–south oriented
system of lobate thrust fault scarps that indicate planetary contraction, perhaps also influ-
enced by the despinning stress field. Mars and Venus are so complex tectonically that no
global grid system has been found.

EBSCOhost - printed on 11/3/2023 9:54 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
4.7 Tectonic associations 155

(a) (b) N

W E
50

N
1

W E
2

W E
50

Figure 4.20 Panel (a) is an image returned by the Ranger 9 B-camera 18.5 minutes before it impacted
the Moon on March 24, 1965. The large, flat-floored hexagonal crater in the center is Ptolemaeus, 164
km in diameter. Its straight wall segments are probably controlled by pre-existing joints in the Moon’s
crust. Crater Alphonsus (108 km diameter) is at the lower left and Albategnius (114 km diameter)
is at the right. (NASA Ranger 9, B001). Panel (b) shows “rose diagrams” plotting the frequency of
occurrence of lineaments with different orientations on the Moon. Top shows lineaments in Region 1
of the index map in the center, bottom is lineaments in Region 2. Orientations are plotted only in the
upper half circle because a full circle is redundant. After Strom (1964).

In addition to a decrease in spin rate, tidal distortion (plus early despinning?) and
reorientation of the lithosphere with respect to the rotation axis may generate global stress
patterns. Such a system has been suggested for Enceladus (Matsuyama and Nimmo, 2008)
and evidence for other global systems is currently being sought on other moons. The major
requirements for the production of a coherent global tectonic pattern are a one-plate planet
and a globally coherent source of stress.

4.7.2 Flexural domes and basins

Roughly circular upwarps and downwarps produce distinctive tectonic associations.
Kilometer-scale domelike upwarps over salt domes or igneous intrusions (laccoliths)
were the first to be recognized. As shown in Figure 4.21, stretching over an uplifted rock
layer produces brittle fracture that results in gaping tensile fractures on a very small scale,
or more usually normal faults and grabens. As uplift begins, failure is most likely along
spoke-like radial faults that are later connected by more irregular circular normal faults.
A good planetary example of this tectonic pattern is seen on the floor of the lunar crater
Humboldt.

EBSCOhost - printed on 11/3/2023 9:54 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
156 Tectonics

D
U

0 3 km

U
D UD

U
D D
U
U
D
U D U
D
UD

U D
U D
D U

U
D

U U
D U D
D D D
U U
D U
D U

Figure 4.21 The pattern of faults and contours in strata uplifted and domed by the Hawkins salt dome
in Texas. Drawn on the top of the Woodbine sand horizon, the outermost contour (light line) is 1400
m below the surface and the inner contour is 1300 m below the surface. The dark lines are faults,
where U stands for up and D for down, giving the relative displacement across the fault. Redrawn
from Suppe (1985), Figure 8–20.

On a much larger scale, the entire Tharsis region on Mars is underlain by a broad rise
up to 9 km high and approximately 5000 km broad. Created by a combination of intrusive
and extrusive volcanism, the rise is traversed by numerous approximately radial exten-
sional faults or dikes that transition into a compressional aureole more than 2000 km away
from the center of the uplift (Banerdt et al., 1992). The prominent radial grabens cutting
the oldest terrain near the crest of the Tharsis Rise strongly suggest extensional doming
and led some of the first workers to propose that the rise originated as a broad, isostatically
supported uplift (Wise et al., 1979). However, further analysis and gravity data reveals a
more complex picture: Beyond the dome itself, most tectonic features indicate deformation
of the lithosphere by loads that produced a circumferential girdle of compressional tectonic
features. The compressional terrain coincides with a topographic low that is best explained

EBSCOhost - printed on 11/3/2023 9:54 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
4.7 Tectonic associations 157

as a thick-lithosphere flexural response to the load of Tharsis (Phillips et al., 2001). The
load itself is attested by a large (500 mGal) positive free-air gravity anomaly. The only
simple way to reconcile these disparate observations is to suppose that most of the eleva-
tion of the Tharsis dome is due to magma intruded into the crust at depths below most of
the grabens. Horizontal intruded magma bodies, therefore, loaded the lithosphere while the
ancient surface above the intrusions was uplifted and stretched contemporaneously with
the intrusion.
Mascon loads provide a cleaner example of tectonics created by subsidence. Large
impact basins on the Moon’s nearside were flooded by several kilometers of basalt,
the last outpourings ending about 1000 Myr after basin formation. This age diffe-
rence provided ample time for a coherent lithosphere to form before the lava loads
were imposed. The heavy basalt loads visibly depressed the underlying terrain and
fractured the surface rocks into a characteristic tectonic pattern. Radial and concentric
mare ridges occupy the interiors of the mascon basins, while the basins themselves
are surrounded by concentric sets of grabens. Theoretical computations of the stress
pattern expected at the surface of a flat, floating elastic plate subjected to mascon-like
loads showed, as observed, an inner zone of compressional faulting surrounded by an
annulus of concentric normal faults (Melosh, 1978). However, the theory indicates that
these two zones are separated by an unobserved annulus of strike-slip faults (Figure
4.22). The absence of the expected strike-slip faults became so troublesome that it was
known as the “strike-slip fault paradox” until more detailed modeling finally showed
that on the Moon the lithospheric shell’s curvature, among other things, suppresses
strike-slip faulting (Freed et al., 2001). The diameter of the transition between com-
pressional and extensional tectonics, in relation to the diameter of the load (measured
from gravity anomalies) serves to determine the thickness of the lithosphere and has
been widely used for this purpose, on the Moon and elsewhere (Comer et al., 1979).
The Moon’s elastic lithosphere varied between 25 and 75 km in thickness during the
time of the mare basalt emplacement.

4.7.3 Stress interactions: refraction of grabens by loads

Some of the most striking tectonic patterns reflect interference between different stress
systems. Figure 4.23a illustrates the broad loop created by extensional grabens on Mars,
as they seem to sidestep the Alba Patera volcanic center on their otherwise gently curving
course from the south to north (Cailleau et al., 2005). Similar patterns are observed on
Venus (Cyr and Melosh, 1993), where grabens approaching corona centers either avoid
the center of the corona load, shown in Figure 4.23b, or approach the extensional stress
system of a nascent corona or “nova” dome in an “arachnid”-style corona. These patterns
can be understood as the superposition of two stress systems: A broad regional system of
extension that creates grabens overprints an existing stress field due to flexure of the litho-
sphere beneath an existing load. The sum of these two stress systems produces a tectonic
response that clearly indicates its multiple origin. Interference patterns of this kind can be

EBSCOhost - printed on 11/3/2023 9:54 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
158 Tectonics

Concentric
normal Strike Radial
faults slip thrust
(Graben) faults faults
(CN) (SS) (RT)

σφφ
σrr

Figure 4.22 Pattern of stresses and faults around a circular mascon-like load on a flat floating elastic
plate. The inner region is characterized by compressional tectonics, predicted to be radial thrust
faults in this approximation. This region is surrounded by an annulus of strike-slip faults where the
circumferential stress is compressive while the radial stress is extensional. This is, in turn, surrounded
by an outer annulus of concentric normal faults, where both horizontal stresses are extensional, but the
radial stress is more extensional than the circumferential stress. Volcanic dikes reaching the surface,
represented here by concentric arcs of small circles, occur wherever the radial stress is extensional.
On the Moon, the annulus of strike-slip faults is conspicuously absent, a fact explained mainly by the
curvature of the Moon’s lithospheric shell. After Figure 2 of Melosh (1978).

quantitatively useful for estimating the sizes of stress acting in the lithosphere if the size
of one of the stress systems is known (Turtle and Melosh, 1997). The stresses created by
loads can often be computed if the gravity anomalies associated with the load are known
and the thickness of the lithosphere estimated. When this is possible, the angles of deflec-
tion of tectonic structures then determine the size of the other stress system. In this way
the stresses acting in a planetary lithospheric shell may be mapped both in direction and
absolute magnitude.

4.7.4 Io’s sinking lithosphere

Like the legendary Sisyphus, fated to eternally roll a rock to the top of a hill only to have
it roll back again, the crust of Jupiter’s moon Io is eternally renewed by volcanism, only to
sink once again. Io’s internal heat comes from powerful tidal dissipation, its mantle flexing
every 1.8 days as the eccentricity of its orbit is maintained by resonances with Jupiter’s

EBSCOhost - printed on 11/3/2023 9:54 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
4.7 Tectonic associations 159

(a) (b)

Figure 4.23 Panel (a) Mosaic of Alba Patera on Mars showing the divergence of the trends of graben
around the volcanic load. NASA Viking Orbiter frames 783A11 to 16. Panel (b) Similar pattern
around 85 km diameter Kvasha Patera on Venus: Portion of NASA Magellan image FL-09S069
centered on 9.5° S, 69° E. North is to the right to facilitate comparison with panel (a).

other large moons. Io’s heat seems to keep its mantle in a semi-molten state, from which
the excess heat is advected by magmas that flow onto Io’s surface. The resurfacing rate is
estimated to be 1.5 cm per year on average. Io seems to function as a kind of planetary lava
lamp, in which the crust is renewed at the surface even as deeper portions are simultan-
eously heated and recycled back onto the surface.
The tectonic consequences of all this recycling now seem inevitable, but were so sur-
prising that it took many years before they were fully appreciated. Schenk and Bulmer
(1998) first attributed Io’s mesa-like mountain masses to thrust faulting and realized that
the sinking crust must lead to compressive stresses. More detailed exploration of this idea
(McKinnon et al., 2001) showed that, while some of the compressive stress is generated by
the decrease in radius of the sinking crustal layers, by far the largest portion is produced
by the heating and expansion of the layers as they sink, warm, and expand. Io’s crust is
thus in compression, which leads to thrust faulting and tilted mountain blocks that rise up
to 20 km above its sulfur-coated surface. The visible mountains do not form any apparent
global pattern, although an anticorrelation between mountain location and volcanic centers
makes some sense from the viewpoint of lithosphere thickness. If Io did not possess a sink-
ing lithosphere that advects heat downward, its lithosphere would be much thinner than it
actually is and the observed high mountains could not be supported.

EBSCOhost - printed on 11/3/2023 9:54 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
160 Tectonics

Figure 4.24 Plate tectonics, Earth style. Tectonic plates are created at oceanic spreading centers,
travel horizontally across the surface of the Earth, thickening and cooling as they go, then plunge
back into the mantle in subduction zones where they are partly reabsorbed. Plates may also slide past
one another along strike-slip faults, of which the San Andreas Fault is the most famous example.

The lesson we learn from the bizarre-looking moon Io is that, although established tec-
tonic principles seem to apply to all of the terrestrial planets and moons, we can still be
surprised and initially baffled by the circumstances in which they occur.

4.7.5 Terrestrial plate tectonics

The Earth possesses a unique tectonic style now known as plate tectonics (Figure 24).
Whether or not this tectonic style is an inevitable consequence of Earth’s highly vigorous
convective interior is presently unresolved. Venus, Earth’s “twin” planet, certainly does
not operate in the same manner as the Earth, but perhaps this is a consequence of Venus’
high surface temperature, which may render the existence of large, effectively rigid plates
impossible. Or, on the contrary, perhaps the presence of water on the Earth makes plate
boundaries so weak that large expanses of the lithosphere can move as rigid blocks until
they meet another plate, where they deform with little resistance.
Whatever the ultimate cause of plate tectonics, the surface of the Earth can best be
described as consisting of a dozen or so large lithospheric plates that glide over the surface
of the planet, interacting mainly at their edges (Turcotte and Schubert, 2002). There are two
basic types of lithosphere composing the plates: Oceanic lithosphere is created at diver-
gent, extensional mid-ocean ridges. It scrapes past adjacent plates at strike-slip transform
plate boundaries and is consumed in compressional subduction zones where the oceanic
lithosphere sinks back into the mantle and is eventually recycled by mantle convection.
Continental lithosphere is chemically different from oceanic lithosphere. It is topped by
more silica-rich, less dense and thus more buoyant rocks, which often ride over a deep keel
of depleted upper mantle material. Because of its buoyancy, continental lithosphere cannot
be subducted, at least not very deeply, and so preserves a record of events much older than

EBSCOhost - printed on 11/3/2023 9:54 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
Further reading 161

the oceanic lithosphere (whose oldest portions date back only about 180 Myr). Because
of its high silica content, the deeper portions of the continental crust are more fluid than
oceanic lithosphere and may, therefore, deform more readily.
How subduction starts is still controversial: Both types of lithosphere resist small
downward deflections. Even oceanic lithosphere probably needs to be forced downward
into the mantle by some amount before it becomes unstable. Once started, however, the
negative buoyancy of subducted oceanic lithosphere drives the plate tectonic engine.
Extensional stresses are transmitted from subduction zones throughout the plates as they
are pulled inexorably into the oceanic trenches. Plate velocity correlates closely with the
negative buoyancy of attached subduction zones. Increasingly sophisticated computer
simulations of terrestrial convection patterns suggest that plate tectonics develops only
when the plate boundaries are extremely weak: Stresses of only a few tens of megapas-
cals are typically required for the lithosphere of a planet to organize into independently
moving plates.
Plates exhibit the gamut of possible tectonic styles: Extensional faulting at spreading
centers where oceanic lithosphere is born, strike-slip transform faulting as they move past
other plates, and compressional thrust faulting where they plunge back into the mantle in
subduction zones. In addition, folds and detachments often develop where continental crust
is compressed, and nearly flat-lying normal faults may form where it is extended. Chains
of volcanoes develop on the overriding limb of subduction zones, where deep melting is
enhanced by the infusion of subducted water and carbonate. Mountain belts arise where
continental lithosphere meets continental lithosphere in collapsing subduction zones. Rift
zones tear continental lithosphere apart when a new spreading center develops in the midst
of previously intact continents.
Earth and its complex tectonic system is certainly the best understood of all the planets
in the Solar System. However, even terrestrial tectonics is not fully understood and remains
the subject of vigorous ongoing research. Outstanding questions focus on the strength of
the lithosphere and its role in plate tectonic processes. The deformation modes of rock
materials at very low strain rates, and how creep is affected by volatiles such as water still
engage the attention of large numbers of Earth scientists. Nevertheless, careful studies of
the tectonics of other planets is certainly not premature, because tectonic activity in very
different settings than those on Earth illuminates these processes under conditions not ac-
cessible on our own planet. Another puzzle in its own right is that, for reasons that are not
presently understood, compression is almost completely absent on the icy satellites, while
extension dominates their tectonics.

Further reading
The prescient writings of James Hutton on Earth’s tectonic and sedimentary cycles are so
obscurely written that few even of his contemporaries read them. Fortunately, his friend
John Playfair presented them in a clear, appealing style that can be read with profit even

EBSCOhost - printed on 11/3/2023 9:54 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
162 Tectonics

today (Playfair, 1964). The fundamental reference on geodynamics and tectonics for this
chapter, as for the last, is the famous textbook by Turcotte and Schubert (2002). A good,
more intuitive introduction to mantle convection is provided by G. Davies (1999), which
concentrates on understanding the problem using simple geometries, while the massive
tome by Schubert et al. (2001) covers all aspects of the convection problem. A classic intro-
duction to fracture mechanics of brittle materials at a high mathematical level is provided
by Lawn and Wilshaw (1975). A clear and appealing discussion of the buckling theory of
folding and boudinage, among other topics, is in Arvid Johnson’s book (Johnson, 1970).
The best and most comprehensible introduction to structural geology and tectonics without
excessive dumbing-down is Suppe (1985). Suppe’s book also contains a nice exposition of
the fault-bend model of large-scale folding, which he largely created. A recent summary of
tectonics from a planetary perspective has just appeared, with individual chapters by many
authors (Watters and Schultz, 2010).

Exercises

4.1 Hot and cold asteroids

The steady-state heat conduction equation for a spherically symmetric planet is:

dT
q(r ) = − k
dr
where q(r) is the heat flux passing through a spherical shell of radius r in units of J/m2-s,
T is the temperature in K and k is the thermal conductivity (about 2 W/m-K for “typical”
rock). Use this equation to show that the internal temperature of a solid, spherical, asteroid
with uniform heat production H W/m3 is:

H 2 2
T (r ) = TS + (a − r )
6k
where TS is the asteroid’s surface temperature and a is its radius. If Hρ = 5.23 x 10–12 W/kg
(be careful of units!!), the average value for carbonaceous chondrites, compute the present-
day difference between the surface and central temperatures of:

Asteroid Name Radius (km)

1 Ceres 512
3 Juno 134
4 Vesta 263
243 Ida ~15 (not actually very spherical!)
433 Eros ~5 (not actually very spherical!)

EBSCOhost - printed on 11/3/2023 9:54 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
Exercises 163

4.2 The towering inferno

Images returned by the Galileo orbiter reveal that Io hosts some of the tallest mountains
in the Solar System: The highest peaks rise some 17 km above the surrounding plains.
Although Io’s surface is topped with a colorful frosting of sulfur compounds, the mountains
are believed to be composed of silicates, as are the extremely hot lavas that are extruded
at the Ionian paterae. Supposing that Io has a basaltic crust (density about 3000 kg/m3) of
unknown thickness overlying a peridotitic mantle (density about 3300 kg/m3), what is the
minimum depth of an Airy-isostasy root that could underlie the mountains?
The average heat flux on Io lies between 2 and 2.5 W/m2. Most of this is lost by volcan-
ism at the paterae, so that the conductive heat flux through the balance of the crust may be
much smaller. Suppose that it is as low as the Earth’s, 80 mW/m2. Further supposing that
the thermal conductivity of Io’s crust is typical of that of silicates, k = 2 W/K-m, what is the
temperature at the base of this root? It may be useful to know that the surface temperature
of Io is about 100 K.
What does this tell you about the support of high topography on Io?

4.3 The fuzzy lithosphere

Assume that the temperature in the upper mantle of the Earth is 0°C at the surface and rises
linearly to 1200°C at 100 km depth (roughly the source depth of basaltic rocks that erupt at
this temperature). Still deeper the temperature, controlled by convection, is approximately
constant.
Use the wet olivine creep law of Chopra and Paterson (1984) for Anita Bay dunite,
ε˙Y(s−1) = 104.0 σ 3.4 (MPa) exp (−444kJ mol−1 / RT)
with R = 8.314 J/K, to estimate the Maxwell time as a function of depth for a typical
stress σ ≈ 30 MPa (produced by loads ca. 1 km thick). The upper mantle shear modulus
μ = 65 GPa.
How thick is the “lithosphere” for loads applied for: 1 Myr, 10 Myr, 100 Myr, and 1 Gyr?
How sensitive is this thickness to the assumed stress?
The bottom line: Is the floating elastic plate approximation any good for loads of this
duration?
Beware of unit conversions!

4.4 Hot, flexed lithospheres

Jupiter’s satellite Europa is squeezed and flexed by tides every 3.55 days as it orbits the
giant planet. Stresses in its icy surface shell vary sinusoidally with an amplitude of about
105 Pa. Using the Maxwell viscoelastic model, compute the strain as a function of time in a
cubic meter of the ice shell, assuming a shear modulus of 1010 Pa and an effective viscosity
(at 105 Pa) of 1013 Pa-s. How does the power dissipated depend on frequency? (Remember

EBSCOhost - printed on 11/3/2023 9:54 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
164 Tectonics

that this is stress times strain rate – you might want to integrate this over some standard
time interval if it varies throughout the tidal cycle!) Under present Europan conditions,
estimate the surface heat flux from the flexing ice shell if it is 10 km thick.
For this exercise it is enough to consider only one stress component. The energy dis-
sipated per cycle per unit volume is then:
∆E = ∫ εxzσ xz dt.
cycle

4.5 A primary Europan concern

As you learned in the previous homework problem, the ice shell of Europa is continually flexed
by tides over the 3.55 day orbital cycle. Stresses in the shell oscillate sinusoidally with ampli-
tudes of about 105 Pa. In Problem 4.4 we assumed that the viscous part of the deformation
(the only portion that dissipates heat) is the same as the steady-state creep rate. However, in
real life, the creep curve has two major parts (shown schematically in Figure 4.25) in addition
to the elastic deformation. In particular, the primary, or transient, portion of the creep curve
has been ignored in all publications thus far on the dissipation of tidal heat in Europa. Your
task is to qualitatively discuss how primary creep may contribute to the budget of heat dissi-
pation due to tidal flexing. Try to couch your analysis in terms of the Maxwell time associated
with steady-state creep, τM = (elastic strain)/(steady creep strain rate) = η/μ, and that due to
primary creep relaxation, τR, versus the period P of tidal flexing. Remember that the rate of
heat dissipation (power per unit volume) due to stress and strain variations is given by:

1
W= ∑ ∫ ε (t ) σ ij (t ) dt
P i , j cycle ij

where εij is the strain tensor and σij is the stress tensor.
Strain

steady creep strain = elastic strain

primary creep

elastic

0 τR
Time
τM

Figure 4.25

4.6 Radiant Earth

Within hours of the impact of a Mars-sized body that struck the ancient Earth, internal read-
justments in the Earth and fallback of debris from the projectile distributed the projectile’s

EBSCOhost - printed on 11/3/2023 9:54 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
Exercises 165

initial kinetic energy into heat that raised the mean temperature of the entire mantle of the
planet to about 3000 K, well beyond the melting point of olivine (which is about 2000 K to
sufficient accuracy for our purposes). Not surprisingly, the liquid silicate mantle began to
cool from the top, which induced vigorous convection throughout its 3000 km depth. The
surface temperature of this magma ocean rose to about 900 K, limited largely by the rate
that heat could be radiated from the surface.
a) Using your knowledge of thermal convection, estimate the Rayleigh number of the
convecting magma ocean, then compute the surface heat flux and the thickness of the
thermal boundary layer. Does it seem likely that chilled surface material will form a
crust that may inhibit convection? Compare the convected heat flux to the rate at which
radiation can carry it away from the surface.
b) Neglecting, for now, the temperature dependence of viscosity (which is much smaller
for a liquid silicate than for crystalline silicates), estimate how long the magma ocean
took to cool to the melting point of olivine. If you were to visit the early Earth sometime
during its 100 Myr long history of accretion, how likely is it that you would find it in a
totally molten state?
Some numerical data you may need: Thermal expansion coefficient of melt: 2 × 10–5 K–1,
Thermal diffusivity of melt: 10–6 m2/s, Thermal conductivity: 2 W/(mK), heat capacity: 3.5
× 106 J/m3K, kinematic viscosity (=η/ρ): 3 m2/s, Stefan–Boltzmann constant, σ = 5.67 ×
10–8W/(m2 K4).

4.7 Planetary warps

Consider a disk-shaped region of planetary lithosphere of diameter L and thickness H
(Figure 4.26). Suppose it is bowed up in the center by an amount U. A central plane, the
neutral sheet (dashed in Figure 4.27), is unchanged in length as upbowing occurs. The top
surface is stretched as shown.
a) Show that the increase in circumference of a circle originally located half way between
the disk’s center and its rim, using purely geometrical considerations, is given by
δ = HU/L. From this result, deduce the circumferential strain εc = Δl/l, where l is the cir-
cumference. Compute the elastic stress σ = Eεc produced by this upwarping if Young’s
modulus, E = 6 x 1010 Pa, typical of basalt, L = 500 km, and U = 10 km. Compare this
to the “strength” of rock.
b) Suppose this strain is taken up by three radial graben that form as upwarping pro-
ceeds. Each graben is composed of two normal faults, dipping 60° toward one another
(Fig. 4.28). If the upwarped region is 500 km in diameter and is uplifted 10 km, how
deep will the graben be? Do you think they could be detected from spacecraft? What if
more than three graben form?
c) Suppose the disk is downwarped by the same amount it was upwarped in (b). Suppose
now that three radial thrust faults form, each dipping at 30°. How high will the thrust
scarps become if they accommodate all circumferential strain?

EBSCOhost - printed on 11/3/2023 9:54 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
166 Tectonics

Figure 4.26

L
4 +δ
Neutral
sheet

U H

Figure 4.27

60°

⇒ ⇒
Figure 4.28

EBSCOhost - printed on 11/3/2023 9:54 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
Exercises 167

NB: This problem is oversimplified. If you were to attack these problems professionally,
you would use elastic-plastic flexure theory to evaluate stresses and strains more exactly.
The curvature of planetary surfaces also cannot be neglected for broad upwarps. However,
the geometrical approach in this problem works reasonably well. For more information see
Turcotte’s review (Turcotte, 1979) or Turcotte and Schubert (2002).

4.8 Boiled cantaloupe lands on Triton

A large percentage of Triton’s surface is covered by “cantaloupe terrain,” a peculiar land-
form dominated by irregular round-to-oval spots 30–50 km in diameter with dark rims that
stand 300–800 m higher than their centers. Although this terrain was initially supposed to
arise from viscously relaxed impact craters, the rather uniform size distribution suggests
an internal origin. It has been proposed that the pattern was created by the convection of a
near-surface ice layer that overturned in response to a sudden heating event (perhaps tidal
heating when Triton was captured by Neptune). The surface terrain would, thus, be analo-
gous to the patterns preserved in a congealed pot of boiled oatmeal. The composition of the
upper crust of Triton is unknown, but probably consists of a mixture of H2O, CO2, or NH3
ices. The radius of Triton is 1350 km and its mean density is 2070 kg/m3, which gives a
surface acceleration of gravity of 0.78 m/s2.
Using your knowledge of convection, estimate:
a) The depth of the ice layer beneath the cantaloupe terrain.
b) Because there is no evidence of melt formation in the cantaloupe terrain, the bottom of
the ice layer could at most have reached the melting point of water ice. Triton’s surface
temperature is about 45 K, so compute the maximum temperature difference ΔT that
could have developed across the ice layer. Use this number to estimate the effective
viscosity of the ice layer for convection to begin. In the light of your knowledge of
the rheology of ice and other similar substances, does this seem reasonable? Note that
for water ice the coefficient of thermal expansion α = 3 × 10–6 K–1, thermal diffusivity
κ = 3 × 10–6 m2/s.
c) Finally, speculate on whether the convection hypothesis is consistent with the observa-
tion that the centers of the spots are lower than the rims.

4.9 Io’s crunched crust

Estimates of the volcanic resurfacing rate on Io indicate that new surface material is added
at a global average rate of 1.5 cm/yr. As new material is added the lithosphere sinks, warmed
by heat conducted from the interior, until it softens and deviatoric stresses relax at a depth
of about 25 km.
(a) Estimate the percentage change in area (area strain) of a lithospheric shell formed at
the surface, at Io’s average radius of 1815 km, as it sinks to the softening depth of
25 km. How is the area strain related to the linear strain of a great circle inscribed on
the surface of the sinking crust?

EBSCOhost - printed on 11/3/2023 9:54 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
168 Tectonics

(b) Compute the stress in the plate just before it softens and begins to flow. Compare this to
the crushing strength of granite, about 0.1 GPa. The relation between horizontal com-
pressive stress σ and the breadth L of a plate uniformly compressed in its own plane by
an amount ΔL is:

E ∆L E
σ =− = ε
(1 − ν ) L (1 − ν )

where E is the Young’s modulus of the crust, about 65 GPa, ν is its Poisson’s ratio, about
0.25, and ε is the horizontal strain.
(c) Compare the compressive strain due to sinking, to the strain caused by thermal expan-
sion as the lithospheric shell warms from Io’s surface temperature of 100 K to ½ of the
melting point of olivine, Tm = 2200 K (presumed to approximate the geologic softening
temperature). Which is more important, compression due to sinking or compression
due to heating? The volume expansion coefficient αV of olivine is 2.5 × 10–5 K–1 and
the linear thermal strain is αV ΔT/3.
Reference: McKinnon et al. (2001).

4.10 Venusian ridge belts

Box 4.1 is required for this problem!
Magellan radar images, confirming hints from the much lower resolution Venera 16 and
17 images, have clearly revealed the presence of fold-like ridges on many parts of the
planet. These ridges are hundreds of kilometers long, spaced 15–20 km apart and attain
heights of hundreds of meters to several kilometers. Using the theory of buckling of float-
ing elastic plates:
(a) Compute the flexural rigidity and thickness of a single plate that buckles to produce
λ = 15–20 km folds (gvenus= 8.6 m/s2, ρc ≅ 2.7 gm/cm3). Also compute the buckling stress.
(b) If the maximum stress supportable by the surface rocks is 0.1 GPa horizontal com-
pression, what is the mean layer thickness required to obtain the wavelength observed?
(Beware! This is not so simple as it seems. You will have to solve coupled algebraic
equations to answer this question, which requires you to consider multiple layers.)
After you have completed this exercise, you may be interested to consult Solomon and
Head (1984) and Brown and Grimm (1997) for a more detailed look at this problem.

EBSCOhost - printed on 11/3/2023 9:54 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
5
Volcanism

Burning mountains and volcanoes are only so many spiracles serving

for the discharge of the subterranean fire … And where there happens
to be such a structure on conformation of the interior parts of the Earth,
that the fire may pass freely and without impediment from the caverns
therein, it assembles unto these spiracles, and then readily and easily gets
out from time to time …
Bernhard Varenius, 1672, quoted in Sigurdsson (1999), p. 148

5.1 Melting and magmatism

The German geographer Varenius (1622–1650) was one of the first to suggest that vol-
canic activity is ultimately caused by the escape of hot melted rock from the interior of our
planet. Written at a time when most geologists believed that the Earth’s interior is filled
with molten rock, the source of the melt was not problematic: Any break or fracture would
allow molten rock to leak out to the surface, just as puncturing the skin of an animal allows
blood to flow out. However, with the study of solid earth tides and the advent of seismology
at the end of the nineteenth century, it became plain that the bulk of the Earth is solid and
the origin of magma became less obvious.
At the present time we believe that melted rock is a secondary manifestation of the ther-
mal regime of our planet and that heat transport by magma is of slight importance com-
pared to thermal conduction and lithospheric recycling, at least on the Earth. Volcanism
and its subsurface accompaniment, igneous intrusion, is, nevertheless, an important pro-
cess affecting the surface of the terrestrial planets, to the extent that almost no planetary
surface seems to have escaped its effects.
Once-molten material from their interiors diversifies the surfaces of many, if not most,
planets and satellites. Basalt has even erupted onto the surfaces of large asteroids such as
Vesta, although this took place at a time when now-extinct heat-producing elements were
active. In the outer Solar System the surfaces of icy satellites exhibit flows of congealed
Copyright 2011. Cambridge University Press.

water-rich melts reflecting low-temperature “cryovolcanism.” Although the materials

differ, the morphology of all these flows is similar, a manifestation of similar physical
processes.

169

EBSCO Publishing : eBook Academic Collection (EBSCOhost) - printed on 11/3/2023 9:55 AM via YDUQS
AN: 399390 ; H. Jay Melosh.; Planetary Surface Processes
Account: s4526441.main.estacio
170 Volcanism

Table 5.1 Heat budget of the Earth

Source Heat flow (1012 W) Percent of total

Oceanic plate recycling, excluding crust formation 23.1 55

Intra-crustal radiogenic heat (mostly continental) 6.6 16
Conduction through continental plates 5.0 12
Oceanic crust formationa 3.1 7.4
Conduction through oceanic plates 2.9 6.9
Intra-continental advection (erosion, orogeny, magmatism) 1.1 2.6
Volcanic centers 0.2 0.5
Totals 42. 100.

Data from Sclater et al. (1980). Estimated accuracy of each entry ca. 10%.
a
Assumes oceanic crust is generated at 18 km3/yr, which advects 1.81 MJ/kg.

5.1.1 Why is planetary volcanism so common?

One might suppose that volcanism acts as a kind of relief valve for pent-up heat in the
interior of large Solar System bodies. Although there may be some truth to this idea (the
planetary volcanic resurfacing event that affected Venus some 700 Myr ago is often sup-
posed to mark an episode of high subsurface temperatures), a detailed inventory of the
Earth’s heat flow shows that volcanic heat transport accounts only for a small fraction of
the total heat lost from our own planet’s interior (Table 5.1), which is otherwise dominated
by plate recycling (62% of the total) and lithospheric conduction (37%, including heat
transported by oceanic crust formation). The amount of heat transported by volcanism, qvol,
can be estimated from the volume rate of eruption, QE, by:

qvol = ρ(cPΔT + ΔHf)QE (5.1)

where ρ is the density of the magma brought to the surface, cP its heat capacity, ΔT the
temperature difference between the erupted magma and the surface, and ΔHf is the en-
thalpy of fusion of the solid magma. The volume rate of eruption is often estimated from
the volume of magma observed on the surface and the duration of an eruption, allowing
estimates of the heat flux from volcanism alone. When compared to the heat flux conducted
through the lithosphere, the volcanic flux is usually found to be small, except in the case of
Io, whose heat transport does seem to be dominated by the eruption of magma (probably
ultramafic silicate magma, accompanied by large amounts of the volatiles sulfur and SO2).
We do not have good estimates of the eruption rate on Venus during its short volcanic resur-
facing event, but there, too, volcanic heat transport may have dominated its planetary heat
flow. Venus should serve as a warning against too-simple classifications of planetary heat
transfer: Different processes may dominate at different times in a planet’s history.
Classifications are, nevertheless, useful, if one keeps their approximate nature in mind.
A common and appealing classification of planetary heat transfer is the triangular ternary

EBSCOhost - printed on 11/3/2023 9:55 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
5.1 Melting and magmatism 171

diagram (Solomon and Head, 1982), borrowed from igneous petrology and illustrated in
Figure 5.1a. Dividing planetary heat transfer into lithospheric conduction, plate recyc-
ling and volcanism, whose sum must equal 100%, one plots small bodies and one-plate
planets close to the conduction vertex, the Earth near the plate-recycling vertex and vol-
canic bodies like Io close to the volcanic vertex. An alternative classification diagram
is shown in Figure 5.1b in which heat transport is plotted versus mean mantle tempera-
ture, normalized by the melting temperature. On this one-dimensional diagram we also
see the three processes, conduction, convection, and volcanism, but now laid out on a
line reflecting the role of increasing internal heat generation. Volcanism, however, is
not considered a unique mode of heat transport, but as an accessory phenomenon that
can occur in any of the three regimes, although it becomes more important as the mean
temperature rises.
The transition between pure conduction and convection occurs where the temperature
of the mantle rises to about half its melting temperature, when viscous creep becomes
important and the mantle’s viscosity drops to about 1021 Pa-s. Because of the strong tem-
perature dependence of solid-state creep, discussed in the last chapter, this regime tends
to be self-regulating and can accommodate a large range of heat transport. However, once
large-scale melting occurs the viscosity drops very rapidly to 103 Pa-s or even less, and the
rate of heat transport, proportional to the inverse cube root of viscosity – see Equations
(4.19) and (4.26) – increases million-fold. Heat transport rates higher than those sustained
by such magma oceans are possible, but here we enter the realm of giant planet or even
stellar interiors for which the concept of a planetary surface disappears, so we defer this
topic to other texts.
Volcanism occurs on planets dominated by each of the major modes of heat transport.
This is partly due to the physical nature of melts: no matter where they are produced, melts,
once formed, are typically more mobile and less dense than their parent materials and so
they may rise to the surface, leaving their parents behind. However, the principal reason for
volcanism seems to be the highly variable susceptibility of planetary materials to melting.
The ease with which materials melt varies with both position in the planet and composition
of the material.
Melting in a convecting planet is most likely to occur just below the conductive thermal
boundary layer. The boundary layer itself is, of course, the coldest part of the convecting
system and temperatures rise linearly with increasing depth (refer back to Figure 4.1).
This steep rise ends at the base of the thermal boundary layer where the temperature gradi-
ent becomes approximately adiabatic (Box 5.1). If the melting temperature were constant,
melting would begin much deeper still, at the bottom of the convecting region. However,
pressure increases the melting point, so the location where melting begins is established
by a competition between the rate at which the temperature rises and the rate at which the
melting point increases.
Clausius–Clapeyron equation. É. Clapeyron (1799–1864) in 1834 first deduced that the
pressure derivative of the melting point of a substance is proportional to its latent heat of

EBSCOhost - printed on 11/3/2023 9:55 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
172 Volcanism

Lithosphere conduction
a)

Moon, Mercury

Mars

Europa
Earth

Plate recycling Volcanism

b)
Volcanism

Solid-state Magma ocean

Conduction convection convection

Tm
Temperature

Tm
2

Mars
Moon Mercury Earth Europa Io

Log (heat transport)

Figure 5.1 Mode of heat transport as a function of process. (a) Location of various planets and moons
with respect to the three major heat transport processes of heat conduction through the lithosphere,
plate recycling, and volcanism. (b) Dependence of mean mantle temperature and the transport process
on net heat transport. As the amount of heat transport increases, the mode of transport switches from
conduction to solid-state convection to liquid-state magma ocean convection. Surface volcanism
may occur at any stage of this progression, although it is more likely to occur at net heat transport
increases.

EBSCOhost - printed on 11/3/2023 9:55 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
5.1 Melting and magmatism 173

Box 5.1 The adiabatic gradient

The pressure experienced by materials in a planetary interior frequently changes as they move
about. Because of the slow conduction of heat into large masses of material, this motion
commonly takes place with little or no heat exchange to the surroundings. Nevertheless,
compression or expansion of the material does change its internal energy and, therefore, its
temperature varies as it ascends or descends. Such temperature changes are called adiabatic
(equivalently, constant entropy or isentropic, as long as the process is also reversible).
Adiabatic temperature changes play an important role in both convection and volcanism. The
characteristic adiabatic gradient is a crucial concept in describing these processes.
The adiabatic gradient can be derived from fundamental thermodynamics in various ways.
One of the simplest is to note that the entropy, S, is a state variable that is typically a function
of only the pressure P and temperature T (in more special circumstances it may also depend
on the composition, especially the latent heat during melting, the magnetic field, or other
variables that affect the internal energy of the moving material), expressed as S(P,T). Then
standard multivariable calculus tells us that a change in the entropy can be expressed in terms
of changes in temperature and pressure as:

∂S ∂S
dS = dT + dP. (B5.1.1)
∂T ∂P

Recall the definitions of the heat capacity at constant pressure, cP, and volume thermal
expansion coefficient αV:

∂S
cP ≡ T
∂T
1 ∂V ∂V (B5.1.2)
αV ≡ =ρ
V ∂T ∂T

where ρ is the density of the material. To these equations add the thermodynamic Maxwell
relation between derivatives:

∂S ∂V (B5.1.3)
=− .
∂P ∂T

Then, inserting (B5.1.3) into (B5.1.1) and using the definitions (B5.1.2), we obtain an
expression for the entropy change in terms of material properties:

cP α
dS = dT − V dP. (B5.1.4)
T ρ

The adiabatic (reversible, no heat exchange) condition assures that dS = 0, so that the
temperature changes with pressure according to:

dT αV T (B5.1.5)
= .
dP adiabatic ρ cP

EBSCOhost - printed on 11/3/2023 9:55 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
174 Volcanism

Box 5.1 (cont.)

Finally, applying the hydrostatic relation between pressure and depth z and local
gravitational acceleration g, dP = ρgdz, we obtain the standard expression:

dT α V gT
= . (B5.1.6)
dz adiabatic cP

For the Earth’s mantle, typical values of these constants are αV = 3 × 10–5 K–1, T = 1600
K, cP = 1000 J/(kg K) and g = 10 m/s2, resulting in a gradient of about 0.5 K/km. This is
far smaller than the conductive temperature gradient near the Earth’s surface, about 30 K/
km. Nevertheless, across the entire 3000 km thickness of the Earth’s mantle, if the gradient
were independent of depth, then a temperature increase of about 1500 K would be implied,
exclusive of the temperature jumps across the hot and cold conductive boundary layers at
the bottom and top of the mantle. The adiabatic gradients in the other planets and satellites
are smaller than that of the Earth, but the gradient, nevertheless, plays an important role in
planetary-scale bodies, especially in relation to the gradient in the melting temperature of
planetary materials.

melting. Stated in modern terms as the Clausius–Clapeyron equation, the pressure deriva-
tive of the melting temperature Tm is:

dTm ∆Vm
= (5.2)
dP ∆Sm

where ΔVm = Vliq–Vsolid is the volume change upon melting and ΔSm = Sliq–Ssolid is the entropy
change upon melting, often expressed as the latent heat L divided by the melt temperature,
ΔSm=L/Tm. The volume change upon melting is typically about 10% of the specific volume
of a substance and ΔSm is typically a few times the gas constant, based on Boltzmann’s rela-
tion S = Rln W, where R is the gas constant and W the probability of a given state (Pauling,
1988, p. 387). Table 5.2 lists the slope of the melting curve for several pure substances of
geologic interest.
Decompression melting. Comparing the slopes of these melting curves with the adiabatic
gradient, about 16 K/GPa for silicates, makes it clear that the adiabatic gradient is generally
less steep than the melting curve of common minerals, at least at low pressures. We will see
later that this must be modified at high pressures where mineral phase transformations take
place, but at least near the surface of silicate planets it is clear that melting is most likely to
occur just below the lithosphere, while at greater depths the planet may remain solid.
When hot, deep-seated solid material rises toward the surface by solid-state creep, its
temperature gradually approaches the melting curve and, if the melting curve is reached,
magma forms. This is one of the principal causes of melting in the Earth and probably the
other silicate planets. It is called pressure-release or decompression melting. The other
principal cause of melting, at least in the Earth, is the reduction of the melting point of sili-
cates through the addition of volatiles, principally water. This is known as flux melting.

EBSCOhost - printed on 11/3/2023 9:55 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
5.1 Melting and magmatism 175

Table 5.2 Dependence of melting point upon temperature for various minerals

Volume change Entropy

Volume change as a fraction of change upon Slope of melting
upon melting, molar volume, melting, ΔSm curve at 1 bar,
Mineral ΔVma (cm3/mol) ΔVm/V (J/mol-K) dTm/dP (K/GPa)

Ice Ih (0 to 0.2075 GPa) -1.634 -0.083 22.0 -74.3

Ice VI (0.6 to 2.2 GPa) 1.65 0.120 16.2 102.0
Quartz 1.96 0.086 5.53 355.0
Forsterite 3.4 0.072 70.0 48.0
Fayalite 4.6 0.094 60.9 75.0
Pyrope 8.9 0.077 162.0 55.0
Enstatite 5.3 0.157 41.1 128.0

Silicate data from Poirier (1991), water data from Eisenberg and Kauzmann (1969).
a
ΔVm adjusted to agree with melting curve slope.

The peculiar melting behavior of water in icy bodies complicates this picture: Increasing
pressure decreases the melting point of ice to a minimum of 251 K at a pressure of 0.208 GPa,
after which the melting point increases again at an average rate of about 55 K/GPa. This per-
mits the existence of stable subsurface liquid oceans in bodies such as Europa and possibly
others in the outer Solar System, but makes it difficult for pure water to reach the surface.
Although the effect of pressure on the melting curve is of great importance in planetary
interiors, the effect of composition, and of mixed compositions in particular, is probably
even greater. It is not possible to understand volcanism on the Earth and other silicate
planets without understanding the melting of heterogeneous mixtures of different minerals.
Even on the icy satellites the melting behavior of mixtures of ices probably dominates their
cryovolcanic behavior.

5.1.2 Melting real planets

Planetary composition: rocky planets. The composition of planetary bodies is ultimately
determined by the mix of chemical elements that they inherited when they condensed from
an interstellar cloud of gas and dust. Most of this material is too volatile to condense into
rocky or icy planets: 98% of the mass of the Solar System is H and He, augmented by about
0.2% of “permanent” noble gases such as Ne and Ar. Of the remaining 1.8% of the mass,
about 3/4 comprises “ices” and only 1/4 forms the high-temperature rocky material from
which the terrestrial planets are built. Table 5.3 lists oxides of the major elements present
in the “rocky” component of Solar System material. More than 90% of this rocky material
is composed of only four elements: O, Fe, Si, and Mg.
The listing of elements as oxides in Table 5.3 is purely conventional: These elements
are actually found in more or less complex minerals that are combinations of the simple

EBSCOhost - printed on 11/3/2023 9:55 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
176 Volcanism

Table 5.3 Cosmic abundances of metal oxides

Metal in oxide Abundance

combination by mass (%)

FeO 38.6
SiO2 30.6
MgO 21.7
Al2O3 2.2
CaO 2.1
Na2O 1.9
All others 2.9

Data from Table VI.2 of Lewis (1995).

oxides. The four most abundant elements typically produce a mixture of the minerals
olivine (Mg2SiO4 or Fe2SiO4) and pyroxene (MgSiO3 or FeSiO3) plus metallic Fe. Doubly
charged iron and magnesium ions are rather similar in size and readily substitute for one
another in the crystal lattices of olivine and pyroxene, which, thus, commonly occur as
solid-solution mixtures of both elements. These minerals, along with their high-pressure
equivalents, comprise the bulk of the Earth and other terrestrial planets, moons, and aster-
oids. Geologists are more familiar with rocks that contain a higher proportion of the less
abundant elements Al, Ca, Na, and others. These elements generally form minerals of lower
density than olivine and pyroxene and so have become concentrated in the surface crusts
of differentiated planets.
Planetary composition: icy bodies. “Icy” materials, listed in Table 5.4, are more volatile
than those forming the terrestrial planets and are, thus, mostly confined to the outer Solar
System. The format in Table 5.4 is also conventional, listing the elements O, C, N, and S as
chemical species that condense from a slowly cooling gas of average solar composition that
is dominated by H. Water ice is by far the most abundant species, but carbon may occur as
methane, as listed here, or as the more oxidized CO or CO2. In comets, carbon seems to be
present mainly as CO and CO2 ices, often loosely bound in water as clathrates, as well as in
more complex hydrocarbon “tars.” Nitrogen may occur as N2 rather than as ammonia, and
sulfur may form compounds with other elements.
The main lesson from this brief discussion of planetary composition is that all plan-
ets and satellites are heterogeneous mixtures of a variety of different chemical species.
Although the proportions may vary depending upon their location in the Solar System and
the chemical and physical accidents of their assembly, planets are anything but pure chem-
ical species. This fact is the root cause of volcanic phenomena, and it requires a careful
inquiry into the complex process of melting.
Melting rocks. Our most common experience with the melting of a solid is also one of
the most misleading: Everyone is familiar with the conversion of solid ice into liquid water
as heat is added. Probably everyone also knows that this takes place at a fixed temperature,

EBSCOhost - printed on 11/3/2023 9:55 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
5.1 Melting and magmatism 177

Table 5.4 Cosmic abundances of ices

Chemical species Abundance by mass (%)

Water, H2O (6.4% bound to silicates) 53.8

Methane, CH4 33.0
Ammonia, NH3 11.0
Sulfur, S 2.2

Data from Table VI.2 of Lewis (1995).

0°C at 1 bar, and that this temperature remains constant until all of the ice is converted
into water. This behavior is typical of pure and nearly pure materials, such as fresh water.
However, planetary materials are usually far from pure and so the melting behavior of het-
erogeneous mixtures of materials is most relevant in a planetary context.
Studies of the melting and reactions of complex mixtures of materials is as old as the
“science” of alchemy, but a clear understanding of its basic principles only emerged in
1875 when Yale physicist J. W. Gibbs (1839–1903) published his masterwork “On the
Equilibrium of Heterogeneous Substances.” Austerely written and published only in the
Transactions of the Connecticut Academy, it took many years for the scientific commu-
nity to absorb the principles that he set forth. In addition to mechanical work and internal
energy, Gibbs associated energy, which he called the “chemical potential,” with each dif-
ferent chemical species and phase in a mixture of materials. Using the tendency of the
entropy of an isolated system to increase, he defined the conditions under which reac-
tions and phase changes occur in thermodynamic equilibrium and showed how to perform
quantitative computations of the abundances of each species in a chemical system, once
the chemical potential of each reactant is known. Much of the research in petrology and
physical chemistry over the subsequent century has centered about measuring these chem-
ical potentials (now known as the Gibbs’ free energy) of a wide variety of substances at
different temperatures and pressures. The detailed application of these methods to the
melting of ices and minerals is discussed in standard texts, such as that of McSween et
al. (2003).
Melting of solid solutions. For a basic understanding of volcanic processes, it is enough
to recognize that there are two fundamental types of melting in mixed systems. The first
occurs in systems in which the different components dissolve in one another in both the
liquid and the solid phase. Illustrated in Figure 5.2a for a mixture of iron and magnesium
olivine at atmospheric pressure, the pure end member forsterite (Mg2SiO4) melts at a single
temperature of 2163 K, while fayalite (Fe2SiO4) melts at the lower temperature of 1478 K.
Mixtures of the two components, however, do not have a single melting temperature but
melt over an interval that may be larger than 200 K, depending on the mixing ratio. The pair
of curves connecting the pure endpoints on Figure 5.2a indicates this melting range. The
lower curve, the solidus, marks the temperature at which the first melt appears at the given
composition. The upper curve, the liquidus, marks the temperature at which the last solid

EBSCOhost - printed on 11/3/2023 9:55 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
178 Volcanism

a) 2163

li q
uid liquid
us
1900

Temperature, K
liquid +
so crystals
li d
us
1700

solid
1500 1478

0 20 40 60 80 100
Mg 2SiO4 Wt % Fa Fe 2 SiO4
Forsterite Fayalite

300
b) 273

liquid
Temperature, K

ice + liquid eutectic

194
clathrate
+ liquid
175
ice + clathrate
160
0 0.1 0.2 0.3 0.4 0.5
H2 O Mole fraction NH3 NH 3• H 2 O
Ice Clathrate

Figure 5.2 Melting relations in heterogeneous mixtures of substances. (a) Melting of a mixture of
two materials that dissolve in one another, forming a solid solution. This example is for the pair of
silicate minerals forsterite and fayalite. Note that melting of the solution occurs over a considerable
interval of temperature. (b) Melting of a mixture of materials that do not form a solid solution but
crystallize as distinct phases. This example is for water ice and water–ammonia clathrate, simplified
after Figure 1 of Durham et al. (1993). Melting also takes place over a range of temperatures except
at a single point, the eutectic.

crystal disappears. In between, a mixture of crystals and liquid is present. The crystals and
melt do not, however, have the same composition: The crystals are always richer in magne-
sium (the higher melting point end member) than the melt. The equilibrium compositions
of crystals and melt at any given temperature can be read off the diagram at the intersection

EBSCOhost - printed on 11/3/2023 9:55 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
5.1 Melting and magmatism 179

of a horizontal line drawn at the given temperature and initial composition on the plot with
the liquidus and solidus lines.
Eutectic melts. The second type of melting behavior is illustrated in Figure 5.2b for a
simplified version of the melting of a mixture of pure water and water–ammonia clathrate,
NH3·H2O. Water and ammonia clathrate do not mix in arbitrary proportions, but form nearly
pure compounds that separately melt at 273 K and 194 K, respectively. When crystals of
the two species are mechanically mixed together, however, they react to form a liquid solu-
tion at about 175 K, less than the melting temperature of either pure end member. This first
melt, called the eutectic composition, is composed of about 0.36 NH3 by mole fraction. As
the temperature continues to rise, what happens depends upon the mixture of crystals. The
crystals always remain pure, but the melt composition changes as more crystals melt and
mix into the liquid. If the overall composition is richer in ice than the eutectic melt, then all
of the ammonia clathrate reacts with the water to form the liquid and only ice crystals are
present in contact with the melt, whose composition is indicated by the line to the left of the
eutectic point. As the temperature continues to rise, more and more of the water reacts with
the melt until an upper temperature is reached at which all of the ice crystals disappear, also
indicated by the line to the left of the eutectic point. If the original mixture of crystals is
richer in ammonia clathrate than the eutectic composition, then all of the water crystals dis-
appear at the eutectic temperature and a mixture of liquid plus ammonia clathrate crystals
persists until the temperature reaches the line to the right of the eutectic point.
Complex melting. Real materials may exhibit either type of behavior in different ranges
of composition due to partial solubility of one phase within another or more complex rela-
tions due to thermal decomposition of one phase before it finally melts. Despite the com-
plexity of such behavior, it is all governed by Gibbs’ rules and, with sufficient experimental
data, the melting relations of any mixture of materials can be understood. The number of
components that must be added to understand real rocks, however, is distressingly large.
Ternary and quaternary diagrams have been devised to represent the mixtures of three or
four components (Ehlers, 1987), but in the Earth as well as the other planets, many more
than four elements, in addition to volatiles such as water and CO2, are present and the
situation commonly exceeds the ability of any graphical method to illustrate the outcome.
At the present time one of the frontiers in this field is gathering all of the data that has
been collected by several generations of petrologists and physical chemists into computer
programs that use Gibbs’s thermodynamics and various models of mixing to predict the
outcome of any natural melting event.
Pyrolite and basalt. A simple generalization, however, is possible for the terrestrial
planets. As described above, most of their mass is composed of a mixture of olivine and
pyroxene. Adding to this the “second tier” elements such as Al, Ca, Na, and K in the
form of either feldspar (at low pressure) or garnet (at high pressure), geochemist A. E.
Ringwood (1930–1993) concocted a hypothetical material he called “pyrolite” that forms a
fair approximation of the bulk composition of any rocky body in our Solar System. When
this material melts the first liquid appears at a eutectic temperature of about 1500 K and
has a composition generally known as “basaltic.” Basalt is a name applied to a suite of

EBSCOhost - printed on 11/3/2023 9:55 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
180 Volcanism

d ark-colored rocks rich in Fe, Mg, and Ca, among others, with about 50% of SiO2. It is
the typical rock produced on Earth at mid-ocean ridges, forms the dark mare of the Moon,
and underlies extensive plains on Venus and Mars. Eucrite meteorites, believed to originate
from the large asteroid Vesta, are basaltic in composition. Basalt is the quintessential vol-
canic melt from rocky bodies in the Solar System. Even on the Earth’s continents, where
more SiO2-rich volcanic rocks are common, volcanologists have been known to describe
volcanic activity as “basically basalt” because basalts from the Earth’s mantle are now
believed to provide most of the heat for even silica-rich volcanism.
Role of pressure in melting. Pressure affects the melting behavior of rock both by chan-
ging the properties (volume, entropy) of a given mineral phase and by changing the sta-
ble phases of the minerals themselves. As olivine is compressed it undergoes a series of
transformations to denser phases, assuming the structure of the mineral spinel beginning
at about 15 GPa (depending on composition and temperature; there are actually two spinel-
like phases), then transforming to a still denser perovskite phase at 23 GPa. These phase
changes are clearly seen seismically as wave velocity jumps in the Earth’s interior. The
phase changes are reflected in the melting curve, as shown in Figure 5.3a, which illustrates
the behavior of the solidus and liquidus of the olivine–pyroxene rock known as peridotite
(equivalent to basalt-depleted pyrolite, and taken to represent the Earth’s mantle) as a func-
tion of pressure up to 25 GPa (equivalent to that at a depth of 700 km in the Earth or the
core-mantle boundary in Mars). It is clear that the steep melting gradients computed for
individual minerals in Table 5.2 cannot be extrapolated to great depths, although the aver-
age melting temperature does continue to rise as pressure increases. Figure 5.3b expands
the low-pressure region to show several adiabats along with the solidus and liquidus. Note
the low slope of the adiabats in comparison with the solidus, as well as the change in slope
of the adiabats when melt is present. A hot plume rising along one of the adiabats begins
to melt when the adiabat intersects the solidus and continues melting as it rises further,
although in reality the melt separates from the solid when more than a few percent of liquid
is present, causing a compositional change in the residual material that must be modeled in
more detail than is possible from Figure 5.3b.
Flux melting. The properties of silicate magmas are strongly controlled by small quan-
tities of volatile species, particularly water, which has a high affinity for the silica molecule.
The dramatic effect of water in lowering the melting point of rock of basaltic composition
is illustrated in Figure 5.4, where water saturation is seen to lower the melting point by
as much as 500 K at a pressure of a few gigapascals. This strong dependence of melting
temperature on water content leads to the possibility of melting caused by addition of
volatiles – flux melting. The name comes from the practice of adding “fluxes” like lime-
stone to iron smelters to produce a low melting point slag. On Earth, fluxing by water is
particularly important in subduction zones, where hydrated minerals formed in the oceanic
crust sink into the mantle. As these minerals heat up, they lose much of their water, which
then invades the overlying crustal wedge and lowers the melting point of these rocks. The
result is the massive volcanism associated with the overriding plate in subduction zones. As
a measure of the importance of water fluxing, note that the average eruption temperature

EBSCOhost - printed on 11/3/2023 9:55 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
5.1 Melting and magmatism 181

(a) 3500

3000
Temperature, K

2500
Liquidus

2000 Solidus

1500

1000
0 5 10 15 20 25 30
Pressure, GPa

(b) 2400

Liquidus
2200
X = 0.8 X = 0.6
Solidus
2000 X = 0.4
Temperature, K

X = 0.2 1500 C

1800 1400 C

1300 C
1600
1200 C

1400

1200
0 1 2 3 4 5 6 7
Pressure, GPa

Figure 5.3 Phase diagram of peridotite representative of the Earth’s mantle. (a) The solidus
and liquidus temperatures of peridotite to a pressure of 25 GPa. The inflections in these curves
are due to phase transformations of the constituent minerals as the pressure is increased. After
Ito and Takahashi (1987). (b) Detail of the peridotite phase curve (heavy solid lines) to 7 GPa
showing adiabats for various temperatures (light solid lines; temperatures indicated in centigrade
to facilitate comparison with petrologic data) and light dashed lines showing different degrees of
partial melting. Data from Ito and Takahashi (1987), computational method after McKenzie and
Bickle (1988).

EBSCOhost - printed on 11/3/2023 9:55 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
182 Volcanism

100
3

Depth in Earth, km
Pressure, GPa

2
with
excess 60
water

1 40

wet 20
dry

0
273 1000 2000
Temperature, K

Figure 5.4 The effect of water on the melting temperature for magma of gabbro composition. The
shaded regions indicate the interval between the solidus and liquidus. Note that the gabbro–eclogite
transition takes place just below the solidus temperature of dry gabbro. The presence of excess water
in the magma dramatically lowers the melting temperature at high pressure, by as much as 500 K for
pressures near 1 GPa. After Figure 6–12 of Wyllie (1971).

of silica-rich island arc magmas is only about 1200 K, whereas the eruption temperature
of basaltic lava is typically about 1500 K. Because silica-rich rocks are more susceptible
to the fluxing effects of water than silica-poor rocks, the overall effect of flux melting in
subduction zones is to enhance the abundance of silica in melts produced in this environ-
ment. The high-silica granitic rocks of Earth’s continents may thus be a direct consequence
of subduction plus water. It is presently unknown whether this mechanism also plays a
role on planets that lack plate recycling, but this makes the discovery of silica-rich gran-
itic rocks on other terrestrial planets a question of great interest. To date, there is no clear
evidence for such rocks on Mars, in spite of a brief flurry of excitement over the possible
discovery of an andesitic composition rock by the Pathfinder mission – since retracted by
the discovery team.
Carbon dioxide also plays a large role in magmas as its presence reduces the solubility
of water – water and CO2 are the dominant gases released in volcanic eruptions. On the
Moon, where water is scarce and the lunar mantle is more reducing, CO seems to have
been the major gas released during eruptions of basaltic magma. The important role of
these volatiles in driving explosive volcanic eruptions will be discussed below in more
detail.
Cryovolcanism. Cryovolcanism on the icy satellites presents a still-unsolved problem.
We do not yet have samples of the material that flowed out on their surfaces, so the

EBSCOhost - printed on 11/3/2023 9:55 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
5.1 Melting and magmatism 183

exact composition of the cold “lavas” on these bodies is still conjectural, except that
spectral reflectance studies reveal an abundance of water ice. But pure water has a rela-
tively high melting temperature for these cold worlds and, moreover, it is denser than the
icy crusts through which these “cryomagmas” apparently ascend, making it difficult to
understand how pure liquid water could reach the surface. So either the crusts are mixed
with some denser phase (silicate dust or maybe CO2 ice), or the liquid water is impure,
mixed perhaps with ammonia or bubbles of some more volatile phase that lower its
average density.

5.1.3 Physical properties of magma

The mechanics of volcanic eruptions and flows are largely dependent upon the physical
properties of magma as it separates from its source rock, rises from depth, and spills out
onto a planetary surface. Eruptions and flows generally occur so rapidly that chemical equi-
librium is not attained and so it is physics, not equilibrium chemistry, which governs the
final stages of magma evolution. The most important property in eruptions is the viscosity
of the melt, which is a strong function of temperature, composition, pressure, and crystal
content.
The viscosity of silicate magmas depends strongly upon their silica content. The small
tetravalent silicon ion bonds strongly with four oxygen ions to form very stable tetrahedra.
In silica-poor minerals, such as olivine, the tetrahedra are isolated and their charge is bal-
anced by adjacent metal ions. However, as the silica content increases the tetrahedra share
corners, forming linear chains in pyroxenes, then sheets and ultimately create a space-fill-
ing lattice in pure SiO2. Silica tetrahedra can, thus, form long polymers in silica-rich melts
and, similar to the polymers in carbon-based compounds, the viscosity increases rapidly as
the length and abundance of the polymer chains increase. At the same temperature, granitic
melts with SiO2 abundances in the range of 65–70% typically have viscosities about 105
times larger than basaltic melts with SiO2 near 50%. Moreover, as temperature increases
and breaks up the polymers, the viscosity of silica-rich melts changes much faster than that
of basaltic melt. These relations are shown in Figure 5.5.
Water and silicate magma. The tendency of silica tetrahedra to polymerize also explains
the strong dependence of the viscosity of silica-rich magmas on water content. Water does
not dissolve in a silicate melt as the triatomic molecule H2O. Instead, one of its hydrogen
atoms bonds with an oxygen ion in the melt to form a pair of OH– radicals that bond with
the silica tetrahedra. In the process, as shown in Figure 5.6, silica polymers are broken into
shorter pieces and the viscosity consequently drops. This effect is important only in very
silica-rich melts: The viscosity of basaltic melts is hardly affected by dissolved water, as
shown in Table 5.5, which compares the viscosity of wet and dry silica magmas at typical
terrestrial eruption temperatures. Thus, water both lowers the melting temperature of silica-
rich melts, such as granites, and decreases their viscosity. Granitic melts, thus, typically
arrive at the surface with high water contents and low temperatures, whereas basaltic mag-
mas are hot and relatively dry. This circumstance has major implications for the processes

EBSCOhost - printed on 11/3/2023 9:55 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
184 Volcanism

Log viscosity, Pa s
1

5
2

3
4

0
800 1000 1800
Temperature, K

Figure 5.5 The viscosity of silicate melts depends strongly upon both temperature and silica content.
The four curves are shown for dry but increasingly silica-poor rocks: (1) rhyolite, (2) andesite, (3)
tholeiitic basalt, (4) alkali basalt. After Figure 2–4 of Williams and McBirney (1979).

O O O

Si Si Si
O O O O
O O H O

Add water: O
H

O O O

Si Si Si
O O O O O
O O H O
H
Cleaved polymer

Figure 5.6 Water has a strong affinity for silica polymers. When a water molecule reacts with a long
silica polymer chain in the upper half of this figure, it easily breaks the polymer into pieces, while
inserting OH groups into the resulting silicate chains. Simple counting of O–H and Si–O bonds shows
the same number before and after the insertion, so this process is almost energy-neutral. Breakdown
of the silica polymers lowers the viscosity of silicate magma.

EBSCOhost - printed on 11/3/2023 9:55 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
5.1 Melting and magmatism 185

Table 5.5 Effects of water on the viscosities of silicate melts

Approximate Viscosity with

SiO2 content Temperature Viscosity dry dissolved H2O
Melt composition (Wt %) (K) (Pa s) (Pa s)

Granite 72 1058 1011 104 (5% H2O)

Andesite 60 1423 103 102.5 (4% H2O)
Tholeiitic basalt 50 1423 102.2 102 (4% H2O)
Olivine basalt 48 1523 101 101 (4% H2O)

Data from Williams and McBirney (1979).

that occur in an eruption and lend silica-rich magmas a dangerous tendency to explode due
to exsolution of their water.
Viscosity and crystal content. Magmas are typically not entirely liquid. Upon eruption,
most magmas contain a heterogeneous assemblage of liquid, crystals, and bubbles of gas.
The presence of both crystals and bubbles strongly affects the viscosity and flow properties
of the mixture. If the magma contains more than about 55% by volume of solids it may not
be able to flow at all: The solid crystals interlock with one another and the flow of the bulk
magma is controlled by solid-state creep of the crystal framework rather than the liquid
matrix. For crystal contents φ up to about 30% by volume, the Einstein–Roscoe formula
gives the average viscosity of the liquid mass η in terms of the viscosity η0 of the liquid
alone (McBirney and Murase, 1984):

η0
η= . (5.3)
(1 − φ )2.5
The viscosity of natural magmas varies with crystal content even more than this equation
suggests, because the crystals that form in a cooling magma are typically less rich in silica
than the magma itself. Thus, as crystallization proceeds the melt becomes progressively
enriched in silica and, because of polymerization, the viscosity rises dramatically. Because
of its importance in terrestrial volcanology, the rheology of silicate melts has received a
great deal of attention. Unfortunately, the same cannot yet be said of the water-rich melts
present in the icy satellites of the outer Solar System, for which only a few rheological
measurements exist.
Bingham rheology. In addition to increasing the viscosity of the melt, the presence of
a dense mass of crystals alters the flow properties of magma still more profoundly. In
1919 the chemist E. C. Bingham (1878–1945) discovered the peculiar behavior of a dense
emulsion of solids while trying to measure the viscosity of paint (Bingham and Green,
1919). Paint, like magma, is a dispersion of small solid particles in a liquid, although in
the case of paint it is usually compounded from a ceramic powder like titanium oxide
mixed into organic oil. Bingham was trying to measure its viscosity by forcing paint under

EBSCOhost - printed on 11/3/2023 9:55 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
186 Volcanism

pressure through narrow capillary tubes and using a well-established equation for the flow
of a viscous fluid through a tube to determine its viscosity. Much to his surprise, he dis-
covered that the paint would not flow at all until the pressure reached some finite thresh-
old, after which the rate of flow depended linearly on the pressure, as he had expected.
Bingham quickly realized that he had made an important discovery and could, for the first
time, explain why wet paint does not immediately flow off a vertical wall: Paint has a
finite yield stress, now called the “Bingham yield stress,” that must be exceeded before it
can flow. The thickness of a layer of fresh paint is proportional to this yield strength, and
does not depend on the viscosity. Previous to Bingham’s work, viscosity alone was used to
determine the quality of paint until the American Society for Testing Materials compared
240 samples of paint at its Arlington, VA, laboratory. In this test many samples, prepared
to have the same viscosity but, unbeknownst to the testing staff, having different yield
stresses, ran off the boards of a fence and left gaping, unsightly, bare spots. Following his
success in explaining this fiasco, Bingham went on to a distinguished career during which
he coined the term “rheology,” introduced the “poise” as a unit of viscosity, and founded
the Society of Rheology.
Although it might seem that the rheology of paint has little in common with volcanic
phenomena, it has been abundantly shown that lava is also a Bingham material and that
the Bingham yield stress is a crucial parameter for computing the length and thickness
of lava flows and domes. Indeed, almost any dense mixture of solid and liquid is likely
to behave as a Bingham material: Even kitchen staples such as mashed potatoes (surely
you have noticed that mashed potatoes can only be piled so high, after which the pile
collapses – their Bingham yield stress has been exceeded!), apple sauce, and pudding are
properly described as Bingham materials, as are basaltic magma, mudflows, and rock
glaciers. Although the Bingham yield stress is thus a central parameter in many applica-
tions, it unfortunately cannot be computed from first principles for nearly any mixture.
There are literally hundreds of empirical equations relating the Bingham yield stress
to solid volume fraction, particle shape, size and liquid composition, but the ability of
these formulas to predict the yield stress of previously unmeasured materials is prac-
tically nil. The reason for this failure is that the Bingham stress depends on the surface
energy of contact between the solids in the mixture. This surface energy depends on so
many presently unknown factors that prediction is nearly impossible: All current work
is empirical.
The relation between shear strain rate and shear stress for a Bingham material is
given by:
ε = 0 for σ < YB
ε = (σ − YB ) / ηB for σ ≥ YB (5.4)

where YB is the Bingham yield stress and ηB is the Bingham viscosity. This equation will
be used below to describe the behavior of lava flows on a planetary surface. First, how-
ever, we need to understand how magma gets up to the surface of a planet in the first
place.

EBSCOhost - printed on 11/3/2023 9:55 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
5.1 Melting and magmatism 187

5.1.4 Segregation and ascent of magma

When solid material from deep within a planetary body begins to melt, small pockets of
melt first form at high-energy locations such as grain boundary intersections and where dif-
ferent crystals can react to produce eutectic liquids. At first, these tiny melt pockets have no
tendency to join together and remain trapped in the rock. At this stage an often-overlooked
phenomenon controls the fate of these small particles of melt. If the surface contact ener-
gies of the melt and crystals surrounding them permits the melt to wet the crystal faces and
run along the grain boundaries, melt will begin to accumulate into larger volumes. On the
other hand, if the contact angle between the melt and the solid crystals is greater than about
60°, the melt beads up and much larger volumes of melt must form before the melt can
separate from its parental rock (Watson, 1982). In both silicate rocks and water–ammonia
mixtures the contact angle is small and melt readily percolates out of the matrix. However,
some combinations of materials, such as liquid iron and silicate, have larger contact angles
and percolation is strongly inhibited.
Magma percolation flow. In silicate rocks, when the melt fraction exceeds a few percent,
the melt begins to percolate along grain boundaries and flows out of the source rock. The
process of “Rayleigh distillation” then controls the chemical evolution of the melt, in which
the continuously extracted melt carries away the elements that enter the liquid and, thus, the
composition of the source material gradually changes. The contrasting process of “batch
melting” occurs when the melt remains in chemical communication with the parent rock.
If the matrix through which the melt percolates can be treated as a rigid structure, the
flow of the melt is described by the Darcy equation (which was originally devised to de-
scribe the percolation of water through porous rock: See Section 10.2.2). In one dimension,
this equation relates the volume discharge of fluid (magma in this case) per unit area Q to
the gradient of pressure driving the flow and the viscosity η of the liquid,

k dP
Q=− . (5.5)
η dz

The permeability k has dimensions of (length)2 and depends upon the size and spacing
of the pores through which the magma percolates (see Turcotte and Schubert, 2002 for
more on permeability and how to calculate it). The most uncertain part of this equation
is the permeability, but this equation has, nevertheless, often been used to estimate the
length of time necessary to, say, differentiate a basaltic crust on a heated asteroid or to
produce enough magma to feed an observed surface flow. In this case the vertical pressure
gradient is equal to the difference in density between the liquid and solid matrix times the
gravitational acceleration, dP/dz  Δρ g. Taking the permeability very roughly to equal
the square of the grain size d, the timescale for magma to flow out of a layer of thickness
h is given by:

φηh
t percolate = . (5.6)
d 2 ∆ρ g

EBSCOhost - printed on 11/3/2023 9:55 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
188 Volcanism

For an asteroid like Vesta, the magma percolation time is only about 60 yr for a 125 km
thick mantle (half of Vesta’s radius), assuming a density difference of 300 kg/m3, a grain
size of 1 mm, total melt fraction of 10%, and a basaltic magma viscosity of 10 Pa–s. This
timescale is very short compared to the thermal heating timescale, and indicates that the
rate-determining step in Vesta crust formation is not melt percolation but the rate at which
its mantle heats up. This seems to be the case in many circumstances: In general, melt per-
colation is so fast that melt leaves its parental rock as fast as it is formed.
Diapirs vs. dikes. Studies of depleted source rocks on the Earth suggest that the simple
rigid percolation model is quite inadequate. It appears that melt in hot rocks, especially if
they are deforming, quickly collects into pockets and veins that are much larger than the
grain size. These melt rivulets join to form larger veins that drain the mass of source rock
more efficiently than uniform percolation. As the magma accumulates in ever larger bodies,
the difference in density between the melt and matrix becomes more important and buoy-
ant bodies of melt may begin to slowly rise through the high-viscosity source rock. Many
book illustrations depict magma, especially highly viscous silica-rich magma, in this stage
as rising in mushroom-shaped diapirs, similar to those depicted in Figure 4.9. However,
petrologists are currently in doubt about the validity of this picture.
A low-density fluid, such as magma, enclosed in a higher density but deformable matrix,
has two means of ascending through the matrix. One is a diapir, discussed above. The other
is a dike, a vertical fluid-filled crack that pierces directly through the matrix and permits
much more rapid ascent of the fluid. Whereas diapirs exploit the viscous property of a
fluid, developing when a low-density fluid displaces an overlying higher density viscous
fluid, dikes exploit the elastic property of the enclosing material. Hot rocks, however, ex-
hibit both viscosity and elasticity, depending on the timescale: They are best described as
Maxwell solids, as described in Section 3.4.3. Whether the rise of magma is governed by
the viscous or elastic response of the surrounding rocks depends on timescale, and, thus,
on the ratio between the viscosity of the magma and that of the host rocks. The precise
conditions for the dominance of one process or the other are still somewhat uncertain: It
is presently an area of active research (Rubin, 1993). However, near the surface it seems
clear that most basaltic magmas ascend via dikes. This is also true for at least some gran-
itic magmas (Petford, 1996), so the following discussion focuses on the mechanics of dike
ascent. An older, and now-discredited model of volcanic eruption is discussed in Box 5.2.
Whereas this “standpipe” model has some apparent successes, in the light of the discussion
below it cannot possibly be correct, but it is still of interest because the elastic dike model
cannot, as yet, reproduce the major success of the old, impossible model!
Dikes differ from simple cracks for two major reasons: They are filled with a viscous
liquid and, because of gravity, the pressure that the fluid exerts on the walls of the vertical
crack differs greatly between its top and bottom. Much of our present understanding of
dikes rests on the summer research of material scientist Johannes Weertman. Weertman,
who has made fundamental contributions to the study of dislocations and creep in solids, is
also a highly regarded glaciologist. He may have become interested in dikes while watch-
ing a stream flowing over the surface of a glacier disappear into a crevasse and wondered

EBSCOhost - printed on 11/3/2023 9:55 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
5.1 Melting and magmatism 189

Box 5.2 The standpipe model of magma ascent

A striking observation made by pilots flying along the volcanic range of the Andes or the
Aleutian volcanic chain is that the summits of all the volcanoes seem to be at nearly the same
elevation (Ben-Avraham and Nur, 1980). This holds even though the bases of the volcanic
constructs often differ greatly in elevation: Extruded lavas build their ultimate cone up to
an apparently fixed elevation. A good planetary example of the same relation is the summit
elevations of the four major volcanic centers on Mars. The summit calderas of the three major
volcanoes on the Tharsis Rise – Arsia Mons, Pavonis Mons, and Ascraeus Mons – all rise to
an elevation of almost 20 km above the Martian datum, each about 12 km above their bases
near the summit of the Tharsis Rise (Table B5.2.1). Olympus Mons, however, rising from the
lowlands of Amazonis Planitia far to the west of the Tharsis Rise, also rises to about 20 km,
even though its base lies 1 km below the Martian datum.
The accordance of these summits was stunningly discovered near the beginning of the
Mariner 9 orbital mission. Arriving at Mars during the height of a planetary dust storm in 1972,
the first images showed only a uniform haze of red dust. However, as the dust began to settle,
four dark spots – the tops of the major volcanoes – began to emerge almost simultaneously.
Planetary geologist Hal Masursky, seeing these spots, immediately proposed that they were the
summits of giant volcanoes, much to the astonishment of the planetary geologic community.
Their simultaneous appearance out of the haze was an indication of the similarity of their
summit elevations.
Even though the concordances of height of most volcanoes are approximate, their
consistency does not seem accidental. This concordance has been “explained” for many years
by a model that might be called the “standpipe” model, because it posits that the magma rises
up a rigid open channel (pipe) and settles at the level of hydrostatic equilibrium.
Illustrated in Figure B5.2.1, the model supposes that the magma rises from its source depth
through a rigid lithosphere of density ρl. The pressure in the magma, of density ρm, is equal
to the pressure of the lithosphere in the magma source region at the base of the lithosphere.
Because the magma is less dense than the lithosphere, the column of magma must be taller
than the thickness of the lithosphere, of thickness t, by an amount h, equal to the height of the
volcano’s summit above the planetary datum, independent of the actual height of the volcano’s
base. Balancing pressures,
(t + h) ρmg = t ρl g. (B5.2.1)

Table B5.2.1 Elevations of the major Martian volcanoes

Edifice height Base elevation Summit elevation

Volcano (km) (km) (km)

Olympus Mons 22 –1 21.229

Ascraeus Mons 15 3 18.225
Pavonis Mons 10 4 14.058
Arsia Mons 12 6 17.761

Data from USGS (2003).

EBSCOhost - printed on 11/3/2023 9:55 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
190 Volcanism

Box 5.2 (cont.)

t ρm ρ ρm
l

magma source

Figure B5.2.1 The standpipe model of volcanic summit heights. According to this model, the
summit heights of volcanoes are all the same, independent of the elevation of the top of the crust
(here represented as supported by Pratt isostasy, although Airy isostatic support gives the same
result) because the magma originates at a common depth and connects to the surface through a
rigid standpipe.

The gravitational acceleration at the planet’s surface cancels out of this equation and we
obtain a linear relation between the summit height and lithosphere thickness:

 ρ − ρm 
h=t l . (B5.2.2)
 ρm 

This model neatly explains the concordant summits of chains of volcanoes, supposing only
that the magma originates at the same depth. It has been widely used to estimate lithosphere
thickness or source depth and gives very reasonable results. On the other hand, it cannot
possibly be correct, because the lithosphere thickness derived, typically of the order of 100
km, is so large that the assumption of a rigid open channel makes no sense: The difference in
pressure between the magma and the walls along the channel is so large that the walls would
deform elastically, expanding near the top and closing off the channel near the bottom, as
described in the text.
The standpipe model is a beautiful example of a simple, clear model that neatly explains
the data, but cannot possibly be right. Unfortunately, the more detailed, “correct” model of
dike intrusion cannot easily explain the observational fact of accordant summits or the heights
themselves. Obviously, this is a problem that needs more work.

whether the stream would ever be able to fill the crevasse. These ruminations led to a funda-
mental paper (Weertman, 1971) that elucidated the role of gravity and elasticity of the wall
rock in determining the shape of a vertical dike (from a mechanical point of view, a dike is
just a water-filled crevasse turned upside-down). One of his most important discoveries is
that dikes cannot be arbitrarily deep: They are strictly limited in their vertical extent by the

EBSCOhost - printed on 11/3/2023 9:55 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
5.1 Melting and magmatism 191

pressure

ρm ρr push
open
d
e
w p
Lc
t
h
pinch
closed
rock

Figure 5.7 Pressure as a function of depth in a vertical dike. Because the magma in the dike is less
dense than the surrounding rock, the pressure in the dike falls less rapidly (here shown in highly
exaggerated form as a vertical line in the right half of the figure). If the average pressure in the dike
and the surrounding rock are equal, the pressure in the dike exceeds that of the enclosing rock at the
top of the dike, whereas the pressure in the rock exceeds that in the dike at its bottom. The magma
thus pushes the crack open at its head, while it is squeezed closed at its tail, forcing its way upward
through the rock. The actual length of the crack Lc is a function of the elastic properties of the
surrounding rock as well as the pressure drop in the crack and is computed in the text.

elastic deformation of the surrounding matrix. This fact implies that the quantity of magma
that can rise in a dike is quantized into a fixed volume.
Dike mechanics. The full analysis of the length and ascent velocity of a fluid-filled crack
is complex (Rubin, 1995), but some simple order-of-magnitude estimates can illustrate the
main outlines of the theory. As magma rises into a vertical, slowly moving crack, the low
density of the magma compared to the wall rock means that the pressure in the crack falls
less rapidly than in the surrounding denser rock (Figure 5.7). If the magma and rock are
at the same pressure in the source region, the pressure at the head of the crack will, thus,
exceed that in the adjacent rock and the top of the crack will tend to balloon outward. This
extra stress at the crack tip, if large enough, may rip apart the rock ahead of the crack and
permit the mass of magma to ascend. However, as the tip of the crack balloons, the tail of
the crack grows narrower because of the elastic reaction of the surrounding medium (this
is similar to the bulge that forms adjacent to a loaded area on an elastic half space – or the
bulge next to a person who sits down on a springy sofa). As the crack lengthens the tail
nearly pinches off, although the fluid in the crack prevents it from closing completely. At
this stage the rising crack can accept no more magma and it continues to ascend toward the
surface as an independent pod of hot magma.
A simple relation between the length L and width w of such a crack is obtained by equat-
ing the average stress generated by the crack in the elastic medium to the pressure drop
between the head and the tail of the crack. The elastic stress is computed from the strain,
ε = w/L, times Young’s modulus E: σ = Ew/L, according to the definition (3.9). The pres-
sure drop through the crack is just (ρl – ρm)gL = ΔρgL. Equating these stresses, we obtain a

EBSCOhost - printed on 11/3/2023 9:55 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
192 Volcanism

relation between the critical crack length and width that is nearly identical to the relation
derived from the detailed theory of crack-tip dynamics (Rubin, 1995):

Ew
Lc = . (5.7)
∆ρ g

For parameters appropriate for the Earth, this tells us that a 1 m wide vertical dike would
have a height of about 2 km, in reasonably good accord with geologic observations. This
result, however, does not give us an independent way of estimating the crack width and
length. Weertman solved this difficulty by balancing the elastic stress of the crack against a
regional extensional stress T, which he considered necessary to permit the crack to ascend.
Although dikes do generally ascend perpendicular to regional extensional stress, experi-
ments on the injection of dyed liquids into gelatin matrices suggest that extension is not
an essential factor in dike ascent. What has been neglected so far is the flow of the fluid
included in the dike. Magma-filled dikes ascend at rates limited by the viscosity of the fluid
and the width of the dike. A magma-filled dike cannot ascend so fast that the pressure drop
in the viscous magma exceeds the pressure gradient in a static crack (otherwise the pressure
gradient would reverse and the magma would decelerate), so the pressure drop due to the
fluid flow provides another equation to determine w in terms of the rate at which magma
flows in the dike. Using the definition of viscosity, Equation (3.12), it is easy to show that
the mean velocity ν ̅ of a viscous fluid flowing through a channel of width w is, up to factors
of order 2, given by:

w 2 dP w 2
v≈ = ∆ρ g. (5.8)
2η dz 2η

The volume discharge of a planar dike Qd per unit length is equal to ν ̅w. Solving for w
in terms of the discharge, and inserting it into Equation (5.7), one can show that the length
of a dike carrying a fixed discharge of magma Qd is given by:

E1 / 2 (2Qdη )1 / 6
Lc = . (5.9)
( ∆ρ g )2 / 3

The most notable feature of this equation is its dependence on 1/g. This has the sur-
prising and important implication that the volume of a magma “quantum” ascending from a
magma source, which is roughly equal to Lc2w, actually increases as the gravitational accel-
eration decreases. This makes a good deal of sense: In order to break through to the surface
of a small body, the magma’s buoyancy force must overcome the resistance of the elastic
medium through which it ascends. On a low-gravity body this means that the volume of
magma must increase. On Earth these magma quanta are relatively small, a few 100 m3,
and may account for the almost regular pulsing activity seen in erupting volcanoes: Each
pulse represents the arrival of a new package of magma traveling up a dike connecting the
surface with the magma reservoir below.

EBSCOhost - printed on 11/3/2023 9:55 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
5.1 Melting and magmatism 193

flank summit
eruption reservoir
–10
0
10 crust level of neutral buoyancy
20 mantle
magma-filled
lithosphere
depth, km

dikes ascend
40
asthenosphere
melt concentration,
60 porous flow

80
partial melting
100
0 20 40 60 80 100 120 140 160 180 200 220
distance, km

Figure 5.8 Ascent and eruption of magma beneath the Hawaiian volcanoes. Magma originates in
the hot mantle below the volcanoes, concentrates into pods and pockets, and finally ascends to the
surface in dikes. As it reaches the level of neutral buoyancy it stalls, collecting in magma chambers
and moving laterally to emerge in flank eruptions. Summit eruptions occur when the driving pressure
from new material becomes high enough to push magma from the level of neutral buoyancy up to the
summit of the edifice. Figure simplified after Tilling and Dvorak (1993).

Eruption volume and gravity. This analysis shows that volcanic eruptions should become
both more voluminous and more rare as the size of the body decreases. Lunar volcanic erup-
tions should be much larger and more catastrophic than terrestrial eruptions, a deduction
that seems to agree well with the large volume of lunar lava flows. Going to still smaller
bodies, many of the small icy satellites seem to have been volcanically resurfaced only
once in their history, by an eruption so large that it covered nearly their entire surface.
Level of neutral buoyancy. The behavior of magma near the surface depends largely
upon the density contrast between the magma and the surrounding rocks. Although the
height to which magma can ascend depends upon the driving pressure at depth, this factor
often seems to be eclipsed by the level at which magma achieves neutral buoyancy: That
is, where the density of the magma equals that of the surrounding rock. First expressed by
G. K. Gilbert in his famous monograph on the Geology of the Henry Mountains (Gilbert,
1880), this concept has found abundant support from detailed studies of the Hawaiian vol-
canoes (Ryan, 1987). Magma beneath Kilauea caldera rises until it encounters rocks of
similar density, then spreads out laterally beneath the surface in the form of vertical dikes
(Figure 5.8). The depth to the upper portion of these dikes fluctuates as the magmatic pres-
sure fluctuates. When this pressure becomes especially high the top of the dike reaches the
surface and magma pours out in a fissure eruption.
As magma-filled dikes approach the surface dissolved volatiles may come out of solu-
tion and form a pocket of gas that leads the liquid toward the surface, as discussed in more

EBSCOhost - printed on 11/3/2023 9:55 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
194 Volcanism

detail in the next section. When this occurs the average density of the fluid filling the dike
decreases and it may be possible for the dike to rise higher than the buoyancy of the liquid
magma itself might suggest. Interactions between the liquid, gas and surrounding solid
rocks then become complex and their consequences have yet to be fully understood.
Intrusion vs. extrusion. Magma reaching its level of neutral buoyancy has no further
tendency to ascend and may become “stalled” underground, creating magma chambers
and reservoirs. It may also spread out in the form of either vertical dikes or horizontal sills,
depending on whether the local minimum principal stress is horizontal (dikes) or vertical
(sills). When sills extend to a critical size that is controlled by the elastic properties of the
overlying rock, they may bodily lift the overlying rocks and create a turtle-shaped intrusive
mass named a “laccolith” by Gilbert. Laccoliths on Earth may reach several kilometers in
diameter and uplift the overlying rocks by up to 1 km.
An important statistic for volcanism on any planet is the ratio between the volume of
magma extruded on the surface and that intruded below the surface. Estimates for the Earth
suggest that far more is injected below ground than ever reaches the surface, perhaps by as
much as a factor of 40. The oceanic crust typically consists of about 0.5 km of extrusive
pillow basalts that overlie a total of about 5–10 km of vertical dikes and plutonic gabbro
(the intrusive equivalent of basalt), giving a ratio of intrusion to extrusion of 10:1 to 20:1.
In continental rifts, basaltic lava has more difficulty reaching the surface through the low-
density continental crust and this ratio may be still larger: A large fraction of the magma
of the well-studied Central Atlantic Magmatic Province that formed as Africa rifted away
from North America appears to be intrusive.

5.2 Mechanics of eruption and volcanic constructs

5.2.1 Central versus fissure eruptions

A common observation is that volcanic materials on the surface of a planet may either pile
up in a heap, recognized as a volcanic center or mountain, or they may spread over a broad,
low-lying area. In the latter case the source of the volcanic flow is often obscure, but when
it can be located it frequently turns out to be a long fissure from which lava poured over a
relatively short interval to feed the surface flows. Because such feeder fissures tend to cover
themselves, they can be difficult to find without detailed topographic maps supported by
high-resolution images.
What determines the pattern of central vs. plains volcanism is unclear: Every planet
appears to possess both volcanic mountains and extensive volcanic plains, although the
proportion of central to fissure eruptions varies greatly from one planet to another, or even
from one location to another on a single planet. On Earth, we recognize volcanic centers
that create steep-sided volcanoes such as Japan’s Mount Fuji, and low domes such as the
islands of Hawaii. Earth also possesses broad volcanic plains such as the US’s Columbia
Plateau or India’s Deccan Plateau. Venus similarly exhibits large central volcanoes, such as
Sif Mons, that contrast with the broad volcanic plains that underlie most of the planet’s sur-
face. Even the Moon, which seems to lack steep-sided volcanic mountains in favor of broad

EBSCOhost - printed on 11/3/2023 9:55 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
5.2 Mechanics of eruption and volcanic constructs 195

mare basalt plains, possesses a volcanic center on the Aristarchus plateau. Recent images
of Mercury from the MESSENGER spacecraft revealed a volcanic center southwest of the
Caloris Basin, which contrasts with the otherwise planet-wide volcanic plains.
Mantle plumes and hot spots. Regional concentrations in the intensity of volcanic
activity are often related to the activity of hot, buoyant plumes that rise through the planet’s
mantle (see Figure 4.9), carrying heat from depth as part of the normal convective heat
engine that cools the planet’s interior (Ernst and Buchan, 2003). As a new teardrop-shaped
plume head nears the surface and its pressure drops, its temperature may cross the solidus.
The melts that, thus, form quickly separate and rise further, either intruding the crustal
rocks just below the surface or erupting onto the surface. Plumes eventually spread out
in the mantle beneath the crust, generating melts at a lower rate than the initial spurt. Hot
mantle material may continue to rise for a long time along the warm trail (the “plume tail”)
left in the wake of the original plume head. This extended flow of hot material becomes the
source of a long-lived volcanic center. Earth’s Hawaiian island chain is believed to origin-
ate as the Pacific plate drifts over such a long-lived source of magma rising from deep in
the mantle. The Tharsis Rise on Mars may similarly be located over a long-lived plume in
the Martian mantle. In the case of Mars, however, there are no moving tectonic plates and
the large volume of Tharsis is attributed to the long-term accumulation of volcanic melts at
one location. The evolution of Venusian coronae is likewise linked to the activity of plumes
rising from its deeper mantle (Squyres et al., 1992).
Although plumes can explain regional concentrations of volcanic activity, they do not
determine whether the eruption will be either of the central or fissure type: On Earth,
plume sources are invoked both for volcanic centers such as Hawaii or Yellowstone and
for plateau basalts such as the Deccan or Siberian flows. Does magma composition play a
role in determining whether activity is centralized or diffused into regional fissure systems?
What about the persistence of the heat source – do long-lived sources of magma (plume
tails) favor central activity, while short, hot pulses (plume heads) favor fissure-fed plains?
These questions are presently unresolved.
The Earth is unique among the terrestrial planets in its possession of plate tectonics.
Subduction-zone magmas are highly enriched in silica as well as volatiles from subducted
oceanic plates bearing sediments and hydrated minerals. Subduction zones also tend to per-
sist for geologic periods. The result is long, linear chains of central volcanoes such as the
Andes or Aleutians. Hot, plume-generated, basaltic magmas may also intrude the base of
silica-rich continental crust, melting the overlying rocks and creating local pockets of silica-
rich melts that underlie volcanic centers such as Yellowstone,which are not related to sub-
duction zones.

5.2.2 Physics of quiescent versus explosive eruptions

Volcanic eruptions can, in principle, be simple outpourings of liquid magma onto the sur-
face of a planet. In practice, however, they usually involve a complex mixture of solid,
liquid, and gas components whose behavior upon reaching the surface is anything but
simple. One of the most consistent observations of terrestrial volcanic eruptions is that

EBSCOhost - printed on 11/3/2023 9:55 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
196 Volcanism

0.6

0.5
B G A

0.4

Pressure, GPa
0.3

0.2

0.1

0 2 4 6 8 10
Weight % H2O

Figure 5.9 The solubility of water in magmas of different composition at a common temperature of
1373 K: B is basalt, G granite, and A is andesite magma. Decreasing the pressure greatly decreases
the solubility for all compositions. Although all three materials can dissolve similar amounts of water
at a given temperature and pressure, the eruption temperature of these three kinds of magma varies
greatly on Earth. After Figure 2–13 in Williams and McBirney (1979).

they progress from gas-rich initial phases into the eruption of successively gas-depleted
magmas, although the course of any one eruption may involve a complex alternation
of gas-rich to gas-depleted pulses (Williams and McBirney, 1979). This progression is
readily explained by the separation of gas from liquid magma as it rises toward the sur-
face: The upper tip of rising dikes or the near-surface zones of masses of magma become
enriched in gas that is vented as the dike or magma column breaches the surface. Gas-
depleted liquid magma (often mixed with crystals) then follows.
The tendency for dissolved gases to separate from their parent magmas near the surface
is a simple consequence of the pressure and temperature dependence of gas solubility. Gas
solubility in a liquid generally increases with increasing pressure. This is especially true
for water in silicate melts because of the strong affinity of water for silica. Measurements
of the solubility of water in magma as a function of pressure, Figure 5.9, show that deep
in planetary crusts silica-rich magmas may hold up to 10% water by weight, but at surface
pressures this drops by almost two orders of magnitude. This tendency for depressurized
liquids to exsolve gas is familiar to anyone who has quickly opened a sealed container of a
carbonated drink: Upon opening, the pressure suddenly drops and bubbles of carbon diox-
ide gas appear throughout the liquid.

EBSCOhost - printed on 11/3/2023 9:55 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
5.2 Mechanics of eruption and volcanic constructs 197

The consequences of gas exsolution near the surface depend strongly upon the viscosity
of the magma and the rate at which the pressure drops. Fluid magmas, such as basalt, tend
to erupt relatively quietly. Their viscosities are low (Figure 5.5) so that bubbles of exsolved
water or carbon dioxide readily escape the magma, collecting in large pockets of gas at
the tip of rising dikes. The presence of such gas pockets in moving dikes contributes to the
seismically observed “harmonic tremor” that often precedes Hawaiian eruptions. When
such a dike breaches the surface, the first material to erupt is mostly gas, driving the dra-
matic (but localized) fire-fountain activity that ushers in the main flow of magma onto the
surface. Fire-fountaining may be renewed during a prolonged eruption as new dikes arrive
to discharge their own gas pockets, then add their magma to the overall flow. Basaltic erup-
tions on the sea floor may not possess a gas-rich phase because the pressure beneath 4 km
of seawater (about 0.04 GPa) is too large for much gas to exsolve. Such deep-sea eruptions
are much more quiescent than eruptions onto the Earth’s surface. Venus’ surface pressure
may likewise be large enough to suppress intense exsolution of volatiles and, thus, preclude
explosive eruptive activity (presuming, of course, that Venus’ interior possesses Earth-like
quantities of water or carbon dioxide).
Silica-rich magmas, such as andesites or rhyolites, have such high viscosities that vola-
tiles have great difficulty separating from them. As pressures drop in rising magmas of this
type, the volatiles form bubbles that remain trapped in the viscous melt. This fact, along
with the strong pressure dependence of solubility, leads to a dangerous tendency for silica-
rich volcanoes to catastrophically “explode.” Note that, although the word “explosion” is
commonly used to described catastrophic volcanic eruptions, these events are not actually
explosions in the sense that a rapid conversion of solid to gas results in a sudden increase
of pressure: In volcanic eruptions the pressure always decreases. This behavior strongly
differentiates volcanic eruptions from impacts, in which large pressure increases do occur.
The shocked minerals that characterize impacts have never been reliably associated with
volcanic eruptions. The presence of such minerals, thus, serves to discriminate the two
types of event.
Catastrophic, silica-rich eruptions can proceed from either fissures or central volcanoes,
as demonstrated by deposits in Earth’s geologic record. However, no fissure eruptions have
been observed during recorded history, so the following discussion will focus mainly on
the eruptions of central volcanoes.
Silica-rich magmas are seldom observed reaching the surface directly. Instead, they
accumulate for some time beneath the surface, cooling by conduction and interaction with
surrounding groundwater and thus partially crystallizing as they lose their initial heat. Fluid
pressure in the magma increases as more of the melt solidifies into crystals because fluids
are excluded from the regular crystal lattice. Eventually this fluid pressure is released,
either as a gas-rich eruption or in response to some unrelated event, such as the landslide
that preceded the 1980 eruption of Mount St. Helens and suddenly uncapped its magma
chamber. A sudden pressure release, from whatever cause, initiates a rapid chain reaction
that we recognize as an explosive eruption.

EBSCOhost - printed on 11/3/2023 9:55 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
198 Volcanism

cold overlying
rocks

vesicular pumice volcanic

lava ash

Figure 5.10 The upper row illustrates the sequence of events that occurs during an explosive volcanic
eruption. At the beginning, pressure builds up below a strong cap sealing the volcanic vent. Bubbles
exsolve from the magma as it cools, which create vesicular lava when it erupts on the surface (lower
row). In the middle panel, the increasing pressure removes the cap overlying the vent and the sudden
pressure release causes more gas to exsolve. The gases expand explosively, accelerating volcanic
debris and magma to high speed. Frothy magma erupted at this stage is called pumice. In the right
panel, so much material has been removed from the vent that it collapses, sealing in the magma
below. The expanding gases disrupt frothy magma into small glass shards that mix with varying
amounts of ambient air and produce volcanic “ash.”

Any sudden pressure release causes vapor to exsolve from the liquid magma. In a highly
viscous, silica-rich magma the bubbles formed by this vapor cannot easily escape from the
magma. Near the surface, where pressure is low, the bubbles grow large and the volume of
the magma suddenly increases by a large amount. This bubbly froth spills out of the magma
chamber onto the surface, continuing to expand as the pressure drops. It expands upward
as well as outward and may accelerate to velocities of hundreds of meters per second. The
magma may be completely dispersed by this large expansion, forming an emulsion of gas
and magma fragments that is commonly called “volcanic ash,” even though no actual com-
bustion takes place. The vapor cools rapidly as it expands, chilling the magma fragments,
which often form tiny glass shards whose shapes are recognizably portions of the walls of
former liquid bubbles. In more fluid magmas the bubble walls may have time to reform
into spherical liquid droplets, as is observed for the silica-poor glassy products of lunar
fire fountains.
The ultimate fate of such erupting gas/liquid emulsions depends strongly upon the ratio
of gas to liquid (glass). Gas-poor magmas erupt as bubbly liquids in which the bubbles are
widely separated from one another (see Figure 5.10, lower row). These magmas chill to

EBSCOhost - printed on 11/3/2023 9:55 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
5.2 Mechanics of eruption and volcanic constructs 199

form lavas containing small, often roughly spherical cavities known as vesicles. This ma-
terial is thus called vesicular lava. Even lunar basalts contain vesicles whose gas phase is
now believed to have been carbon monoxide, CO. Magmas containing more gas cool to
form a rock whose bubbles are nearly in contact, giving it an average density that may be
less than that of water. Lavas of this type are called pumice. Volcanic ash results when the
bubbles coalesce and the emulsion’s volume is dominated by gas.
Gas-rich emulsions may expand at high speed. The ultimate velocity of such an expand-
ing mixture is determined by the thermodynamic properties of the gas, in particular by its
molecular weight and initial temperature. The expanding emulsion often incorporates other
material from the vent walls, clots of gas-depleted magma or even cold rocks. These acci-
dental inclusions are accelerated with the gas and may be thrown substantial distances from
the vent. Such fragments are called volcanic bombs and pose a major hazard to volcanolo-
gists trying to approach the site of an active eruption. The impact of large volcanic bombs
sometimes forms craters a few to ten meters in diameter many kilometers from the vent.
Although volcanic debris may be thus accelerated to high speed, it is unlikely to exceed
escape velocity of even moon-sized bodies. Box 5.3 explains how the maximum ejection
speed is determined. Note that the size of the volcano itself is not a factor in determining
the ejection speed – only the eruption temperature and nature of the gas are important.
As a gas-rich emulsion of hot gas and melt fragments expands above the surface it may
reach substantial heights before falling back. Its initial velocity at the surface, vej, gives the
maximum height that this mixture can reach ballistically. Equating its kinetic and gravita-
tional potential energies, this height is given by vej2 / 2g, where g is the surface acceleration
of gravity (Figure 5.11a). On an airless body the mixture of gas and fragments may rise to
considerable heights – the Prometheus plume on Io rises about 75 km above its surface –
then spreads laterally, driven by the entrained gas, into a wide umbrella that rains back onto
the surface over a broad area (Figure 5.11b).
When an atmosphere is present, the erupting emulsion may incorporate some of the sur-
rounding atmosphere. The average density of this mixture may become lower than that of
the ambient atmosphere as the incorporated gas is heated. In this case the erupted material
rises still further as a buoyant plume (Figure 5.11c). On the Earth, such buoyant erup-
tion columns commonly rise into the stratosphere, tens of kilometers above the surface.
The incorporated glassy particles then drift with the local winds and rain out later, at a
rate depending on their size. Fine volcanic ash may, thus, spread globally over the Earth,
although most of the coarser volcanic ash falls closer to the vent. Many terrestrial volcanic
ash deposits can be recognized thousands of kilometers from their sources.
Even when an eruption plume does not become buoyant, it spreads rapidly away from
the vent. The emulsion of gas and glassy particles is then denser than the surrounding
atmosphere, but the mass is still fluidized by the gas phase, much as a dry snow avalanche
is fluidized by air incorporated with the snow particles. It, thus, spreads as a density cur-
rent, often overrunning topographic obstacles near the vent. Such hot, mobile density cur-
rents are known as pyroclastic flows. They can be devastating to human life and buildings

EBSCOhost - printed on 11/3/2023 9:55 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
200 Volcanism

Box 5.3 A speed limit for volcanic ejecta

When magma rises to the surface and dissolved gases come out of solution, the hot, high-
pressure gases expand and lower the pressure. As the gas expands it accelerates both itself
and any entrained solid or liquid droplets to high velocity. However, thermodynamics imposes
a strict limit on the maximum velocity to which this material can expand. This maximum is
mainly a function of the initial temperature and composition of the gas phase.
The gas itself attains the highest velocity: Any burden of entrained material lowers the
ultimate velocity of the mixture. We thus focus on computing the maximum expansion velocity
of the gas alone, with the understanding that any admixture of solid or liquid material slows the
final velocity in proportion to the square root of its additional mass loading. We assume that,
for the time period under consideration, the flow is approximately steady, with a reservoir of
hot, high-pressure gas at pressure P1 expanding though some complex but energy-conserving
process to a final pressure P2.
Referring to Figure B5.3.1, consider two successive times during the steady expansion,
tA and tB. At time tA there is a mass m of gas on the left with a total energy equal to the
sum of its specific internal energy e1 and kinetic energy 1/2 u12 per unit mass times m:
 1 
m  e1 + u12  + (energy of hatched region ) . At time tB the same mass of gas has emerged
 2 
on the right (this is the steady-flow approximation) and the total energy is now given by:
 1 
m  e2 + u22  + (energy of hatched region ) . Furthermore, in steady flow the energy in the
 2 
hatched region has not changed. Between the times tA and tB, the work done by the pressure P1
on the mass on the left is P1ΔV1, where the volume change ΔV1 = m/ρ1, where ρ1 is the density

P1 U1 U2 P2 tA

P1 U1 U2 P2 tB

Figure B5.3.1 Schematic illustration of volcanic gases expanding steadily through a complex
vent, represented by a hatched block. So long as the expansion does not add or subtract energy,
the details of the expansion do not matter. The high-pressure gas on the left at time tA expands so
that its increase in kinetic energy per unit mass at tB is equal to its decrease in enthalpy per unit
mass, as described in the text.

EBSCOhost - printed on 11/3/2023 9:55 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
5.2 Mechanics of eruption and volcanic constructs 201

Box 5.3 (cont.)

of the fluid on the left. Similarly, the work done by P2 on the right is P2ΔV2 = mP2/ρ2. Equating
the sum of the energy and the work done at tA and tB then gives:

 1  P  1  P
m  e1 + u12  + m 1 = m  e2 + u22  + m 2 (B5.3.1)
 2  ρ1  2  ρ2

where the energy of the hatched region cancels out, because it does not change from tA to tB:
This is the crucial assumption that no energy is added or lost during the expansion. We can
cancel the common factor m and note that the definition of the specific enthalpy h is:

P
h≡ e+ (B5.3.2)
ρ

so Equation (B5.3.1) becomes:

1 2 1
h1 + u1 = h2 + u22 (B5.3.3)
2 2

or, collecting like terms:

u22 − u12 = 2(h1 − h2 ). (B5.3.4)

If the initial velocity u1 is either zero or much smaller than u2, as is usually the case in a
volcanic eruption, we can neglect it and set u2 equal to the maximum expansion velocity umax,
which from Equation (B5.3.4) we find:

umax = 2(h1 − h2 ). (B5.3.5)

For a perfect gas at temperature T the specific enthalpy is:

h = cpT (B5.3.6)

where cP is the specific heat at constant pressure (SI units are J/kg-K) and T is the temperature
in K. Note that, although the molar heat capacity for many substances is similar, the heat
capacity per unit mass depends strongly on the molecular weight of the material, so that low-
molecular-weight gases typically have a much higher specific heat than high-molecular-weight
gases and consequently expand faster. Hydrogen-powered volcanoes thus eject material much
faster than volcanoes erupting water vapor or CO2.
If h2 is much less than h1 because of the strong cooling during adiabatic expansion, then we
can write simply:

umax  2 cP Tmagma (B5.3.7)

where Tmagma is the pre-eruption temperature of the magmatic gases. This equation provides
a convenient and easily evaluated estimate of the maximum expansion velocity possible in
volcanic eruptions. It gives a good estimate of the maximum observed velocity of volcanic
bombs from terrestrial volcanoes. Some typical results are given in Table B5.3.1.

EBSCOhost - printed on 11/3/2023 9:55 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
202 Volcanism

Box 5.3 (cont.)

Table B5.3.1 Maximum expansion velocity in volcanic eruptions, Equation (B5.3.7)

Working Heat capacity, cP Magmatic temperature umax

gas (kJ/kg-K) (K) (km/s)

H2O 2.0 1473 2.4

CO 1.02 1473 1.7
CO2 0.84 1473 1.6
H2 14.3 1473 6.5

It is possible to incorporate modifications of this simple formula for the presence of solids
or liquids loading the gas, but there are no simple formulas because the results are sensitive to
the details of how heat is exchanged between the gases and solid. As a start, one notes that the
solids or liquids make only a negligible contribution to the pressure. Defining s = msolids/mgas,
we can write for the density ρ =(1+s)ρgas, so using the perfect gas law the enthalpy of the gas
itself is given by:

 s 
hgas =  cP − R T (B5.3.8)
 1 + s 

where R is the gas constant. Values of s greater than zero clearly lower the enthalpy of the
gas, but this neglects heat transfer between the entrained solids and liquids and gas during
expansion. Further treatment requires modeling beyond the level of this book.

tens to hundreds of kilometers from a volcanic vent. The phenomenology of such flows is
both complex and of great interest from many points of view. For more information the
reader is referred to the recent monograph of Branney and Kokelaar (2002).
The ultimate deposits of explosive volcanic eruptions range from welded tuffs, which are
deposited from density currents that are so hot that the glass particles weld together once
they come to rest, to airfall tuffs in which the chilled, glassy magma particles fall relatively
gently to the surface through either air or water. Less is known of deposits on airless bod-
ies such as the moon. There, the chilled droplets of magma must be emplaced ballistically,
with less ability to flow far from their original vent. An important characteristic of all such
deposits is that they tend to blanket pre-existing terrain and, unlike water-laid sediments,
may be deposited with initial slopes that follow the terrain rather than lying in initially
horizontal beds.
Back at the vent from which the emulsion of gas and liquid magma erupted, changes
occur as more and more material is ejected. The pressure on the deeper-lying magma is
relieved as material from the top of the magma chamber is erupted (Figure 5.10). Volatiles
dissolved in this deeper-seated magma then exsolve, creating more bubbles and increasing

EBSCOhost - printed on 11/3/2023 9:55 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
5.2 Mechanics of eruption and volcanic constructs 203

(a)

2
pyroclastic flow h = vej
2g

welded tuff vej

(b)

~75 km

vej

(c) buoyant plume

vbuoyant
hstability
ash fall

air entrainment

vej coarse ash fall tuff fine

Figure 5.11 Panel (a) illustrates the ejection of an emulsion of magma and volcanic gas from a vent
at high speed. The plume rises ballistically to a height h determined by the conversion of its kinetic
energy to gravitational energy and then falls back. As it descends, it forms a hot density current
known as a pyroclastic flow. When this material finally vents its gas and settles into a deposit on the
surface it creates an ignimbrite or welded tuff. In Panel (b) the eruption occurs on an airless body.
In this case the plume rises so high above the surface that it cools (by thermal radiation) before it
descends to the surface. It spreads out as it rises, forming an umbrella-shaped plume such as those on
Io. Panel (c) illustrates an eruption on a body with a dense atmosphere, such as the Earth or Venus.
In this case the plume incorporates atmospheric gases and becomes buoyant, rising until its density
equals that of the surrounding atmosphere at hstability, after which it drifts laterally, responding to local
winds. As it drifts, cooled volcanic ash falls out of the cloud and is deposited as loose volcanic tuff
on the surface. The largest particles fall out first, the finer material later, producing a deposit in which
the particle size grades laterally from coarse near the vent to fine farther away.

EBSCOhost - printed on 11/3/2023 9:55 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
204 Volcanism

the magma’s volume. This material then follows the first-erupted material out of the vent,
further expanding and accelerating as it reaches surface pressures. This initiates a chain
reaction in which progressively deeper material expands and flows out onto the surface at
progressively higher velocities. This chain reaction ceases only when rock surrounding the
magma chamber loses its lateral support and collapses into the void left by the magma that
has now erupted onto the surface. Overlying rocks, if not initially removed by the violence
of the eruption, also subside to fill the void. The final result is a depression, a volcanic cal-
dera, surrounded and perhaps partially buried by the magma that has so recently left its ori-
ginal location. The size of the depression reflects the size of the original magma chamber
and its included volume of unstable magma. The vents producing small eruptions, such as
that of the 1980 eruption of Mount St. Helens, may not totally collapse because the sur-
rounding rock is strong enough to support the evacuated cavity. Much larger magma bod-
ies, however, leave caldera depressions that reach dimensions of hundreds of kilometers.
Such collapse calderas are observed on most volcanically active bodies, including Venus,
Mars, and Io as well as the Earth.

5.2.3 Volcanic surface features

Although nearly every volcanic eruption involves gas and liquid phases, the relative con-
tributions of these two major phases vary widely from one eruption to another. At one
extreme, some eruptions are driven almost entirely by gas, in which little or no magma
may appear at the surface. At the other extreme, enormous volumes of liquid magma pour
rapidly onto the surface with little or no accompanying gas. The morphology of the final
deposits depends strongly upon the relative volumes of gas and liquid.
Maar craters. The structures most likely to be confused with impact craters are produced
by gas-dominated eruptions. Eruptive vents are often nearly circular, are surrounded by a
raised rim of ejecta, and are depressed below the pre-eruption surface, similar to impact
craters. Such vents are called maar craters or tuff rings. On Earth, they frequently contain
lakes (hence the name maar, which derives from the German word for a small lake). They
form as volcanically heated gases breach the surface and eject a gas-fluidized mass of
rock debris that frequently contains no magma. The gas may either be deep-seated, rising
from depths of a few hundred kilometers in the case of diamond-bearing kimberlite pipes,
or near-surface water that is explosively vaporized by contact with hot, rising magma. In
either case the gas/debris expulsion velocity may be very high, several hundred meters per
second, and the vent is often surrounded by a bedded deposit containing debris from great
depth displaying bedforms that indicate high-velocity radial outflows of gravity currents
initiated by the eruption. The volume of this ejecta (the tuff ring surrounding the vent) is
often much smaller than the volume of the crater itself, indicating withdrawal or subsid-
ence of the magma following the eruptive phase. Craters of this type on Earth include many
examples in the Franconian Lake District, Germany, MacDougal and Sykes craters, among
others, in the Pinacate Mountains of Sonora, Mexico, and Ubehebe crater in California’s
Death Valley. The latter crater is often described as a phreatic explosion crater because the

EBSCOhost - printed on 11/3/2023 9:55 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
5.2 Mechanics of eruption and volcanic constructs 205

source of the water in this case appears to be near-surface groundwater. Maar craters are
typically only a few kilometers in diameter.
It is difficult to recognize maar-type volcanic craters on other planets, largely because
of their close resemblance to impact craters. In this case one often must appeal to the
departure of volcanic vents from perfect circularity, their tendency to form linear clus-
ters over subsurface dikes, and their small-volume rim deposits compared to impact cra-
ters. However, in some cases there is no ambiguity because gas-rich eruptions have been
caught in the act. The prominent SO2-rich plumes discovered on Io, the water geysers on
Enceladus, and nitrogen geysers on Triton are all spectacular examples. Cometary jets and
outbursts may be further examples of gas-fluidized eruptions on the surfaces of small extra-
terrestrial bodies.
Cinder cones. Cinder cones are steep-sided conical mounds of pumice and volcanic
bombs that build up near the sources of more magma-rich volcanic flows. Driven by gas
bubbling out of silica-poor magmas, these landforms often dot the traces of dikes feed-
ing large flows, or stand in isolation over the sources of smaller flows. They may be built,
destroyed, and rebuilt many times during any given eruption on Earth. They form because
of the relatively small range to which pumice and volcanic bombs can be ejected from the
vent and so pile up until landslides transport loose debris farther away from the vent. They
are typically only a few kilometers across and 1 km high, with sides standing near the angle
of repose for loose rock debris, around 30° from the horizontal. Cinder cones have been
identified on Venus and Mars as well as Earth. On the Moon and other low-gravity bod-
ies, the range of volcanic debris may be so large that cinder cones are much more spread
out and thus do not achieve steep sides (McGetchin and Head, 1973). In this case cinder
cones grade insensibly into pyroclastic deposits and the name cinder cone might not be
appropriate.
Shield volcanoes. Silica-poor magmas, when they persistently erupt from a single
center, produce low, broad volcanic mountains known as shield volcanoes. This name
comes from the profile of an ancient Greek soldier’s shield, placed concave-side down on
a flat surface. The gases dissolved in silica-poor magmas readily escape near the vent and
eruptions are characterized mainly by outpouring lava accompanied by only minor fire-
fountaining. These eruptions are not explosive and mainly feed low-viscosity flows that
carry liquid lava away from the vent. Shield volcanoes may spread hundreds of kilometers
horizontally while achieving elevations of a few to a few tens of kilometers. The largest
known example in the Solar System is Olympus Mons on Mars, with a basal diameter of
550 km and an elevation of 21 km above its base. Its summit is crowned by a complex
caldera depression about 80 km across and 3 km deep. The slopes of shield volcanoes are
low, much less than the angle of repose, and they are built up from many thousands of
individual lava flows that either proceed from a summit caldera or are erupted from dikes
breaching the flanks of the mound. The shield often contains a central magma chamber
where rising magma accumulates before eruption. Eruption events that partially drain
the magma chamber produce collapse calderas near the summit that alternately fill and
reform during the life of the volcano. Shield volcanoes have been identified on Venus

EBSCOhost - printed on 11/3/2023 9:55 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
206 Volcanism

Figure 5.12 An extensive field of shield volcanoes on Venus. This field contains approximately 200
small volcanoes ranging from 2 to 12 km in diameter, many of which possess summit calderas.
These are identified as shield volcanoes, although some cinder cones may occur among them. NASA
Magellan image, left portion of PIA00465. Located at 110°E and 64° N. North is to the top.

(Figure 5.12), Earth (the Hawaiian volcanoes are the iconic examples), Mars, and perhaps
on Mercury. Shield-like constructs also form the Moon’s Aristarchus plateau and Io’s
Prometheus patera.
Long, fissure eruptions produce broad, nearly level volcanic plains that will be discussed
in more detail in the next section. Near the fissure, exsolution of small quantities of gas
may build cinder cones or, if less gas is present, small, typically elongated spatter ramparts
around the dike where magma wells out of the ground.
Composite cones. Silica-rich magmas are highly viscous and volcanic material tends
to build up around their eruptive centers because viscous flows are typically thick and
short. As described above, their eruptions tend to be much more violent than silica-poor
magmas because this type of magma is likely to be internally ruptured by bubble forma-
tion. Volcanic ash is created abundantly by such eruptions and pyroclastic flows spread

EBSCOhost - printed on 11/3/2023 9:55 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
5.2 Mechanics of eruption and volcanic constructs 207

it far and wide, blanketing pre-existing terrain with smooth-surfaced deposits that may
reach hundreds of meters in thickness. This type of magma is more likely than any other
to form recognizable volcanic mountains. These constructs are often very steep-sided and
are frequent sites of catastrophic rock avalanches that limit the size and extent of silica-rich
volcanic mountains. Called “composite cones,” these piles of volcanic materials consist of
alternating lava flows and volcanic ash deposits. All known examples occur on Earth and
are the direct product of plate tectonics, their parent magmas being created by flux melt-
ing of silica-rich rocks in subduction zones, although it has recently been suggested that
Ascraeus Mons on Mars might be a composite cone.
Laccoliths. In addition to lavas that reach the surface, a few volcanic features are created
by lava that is intruded beneath the surface. Horizontal sills may extend tens of kilometers
from their source, but are difficult to recognize because they uplift the surface only a few
meters to perhaps 100 m over a broad area. However, if the overburden above a sill is thin
enough, it may become elastically unstable, flexing upward into a broad dome known as
a laccolith. First recognized and named by G. K. Gilbert in the Henry Mountains of Utah
(Gilbert, 1880), laccoliths have been found in many locations on Earth and the mechanics
of their formation has been analyzed in detail (Johnson, 1970). Updoming results in charac-
teristic radial fractures that may themselves be the source of surface lava flows. Laccoliths
are typically a few kilometers to tens of kilometersin diameter. It has been suggested that
several features on Mars are laccoliths, but definitive proof of a laccolithic structure is dif-
ficult on the basis of remote sensing alone.
Calderas. Calderas are one of the few negative-relief volcanic surface features. When
large volumes of magma ascend over the same route from a persistent source, the liquid
magma may accumulate beneath the surface at the level of neutral buoyancy and grow into
a compact, long-lived, hot mass by displacing or partially assimilating the pre-existing cold
rock. Called a magma chamber, slow cooling of the magma results in gradual crystalliza-
tion and compositional changes. Rocks formed from large magma masses that slowly cool
below the surface are called plutonic rocks and are characterized by large crystal sizes,
typically millimeters to centimeters.
Eruptions may rapidly deplete the volume of liquid magma in a magma chamber. When
this occurs the sudden loss of volume undermines the overlying rocks, which then col-
lapse into the space formerly occupied by the magma, forming a depression. Such volcanic
depressions are called calderas and their size reflects the diameter of the underlying magma
chamber. Calderas on the terrestrial planets range from a few kilometers up to 100 km in
diameter and range from a few 100 m in depth to many kilometers. They form at the sites
of both silica-poor and silica-rich volcanic eruptions.
Not all volcanic centers display calderas: If the rate of magmatic replenishment to the
chamber can keep up with the eruption rate, wholesale foundering of the surface into the
magma chamber does not occur and a caldera never forms. Nevertheless, erupted magma
piles up on the surface and a large volcanic edifice may be constructed from the products
of many small eruptive events.

EBSCOhost - printed on 11/3/2023 9:55 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
208 Volcanism

5.3 Lava flows, domes, and plateaus

Lava flows form when predominantly liquid magma reaches the surface and flows away
from the vent or fissure, eventually to cool into a solid surface deposit. Magmas are not
simple liquids: They almost universally contain bubbles formed by exsolved volatiles and
crystals created during cooling before eruption. Rapidly erupted magmas may also be
loaded with solid xenoliths, inclusions of rock from either a deep source region or the walls
of the rock surrounding the magma’s route of ascent. For this reason they exhibit complex
rheologies, which are reflected by their deposits.

5.3.1 Lava flow morphology

Magma that erupts onto a planetary surface is called lava. When this hot, complex liquid
flows out onto the surface of a planet it suddenly enters a much cooler environment. The
response of the lava’s surface to this rapid cooling creates a variety of textures that depend
strongly on the magma composition and cooling rate. The interior of the flow cools more
slowly and behaves very differently from the lava at its surface.
Highly viscous, silica-rich lavas form short, stubby flows that usually contain a large
proportion of glass upon cooling. Differential thermal stresses fracture the surface of the
hot magma into glassy fragments that bury and then insulate the interior of the flow. Slow,
viscous flows usually exhibit a wavy, strongly textured surface on the scale of tens to hun-
dreds of meters that is later cut by deep cooling cracks.
Pillow lava. More fluid lavas develop surface textures that are described as being of
four general types. Submarine flows cool very quickly upon contact with water. The glassy
surface is splintered into fine glassy fragments (called palagonite), while the mass of the
magma collects into lava-filled sacks that often detach from the flow front and pile up in
front of it, forming structures known as pillows. Individual pillows range from a few tens
of centimeters to meters in diameter and their presence in a cooled lava flow is diagnostic
of an underwater eruption. Subaerial flows cool more slowly and produce surface textures
denoted pahoehoe, aa, or block types.
The peculiar surface textures of silica-based lavas derive in part from the tendency of silica
tetrahedra to polymerize. At high temperatures (between about 800°C and 1070°C) highly
polymerized silica glass behaves like rubber (a high-polymer substance based on carbon).
Capable of sustaining large strains without breaking, the rubbery surface of cooling silicate
lavas can inflate with liquid lava like a balloon or, under compression, collapse into folds
resembling drapery on a scale of tens of centimeters. Incautious Hawaiian volcanologists
once impressed onlookers by jumping up and down on fresh lava flows, which responded like
a giant waterbed (this practice is now forbidden due to several unfortunate accidents).
Aa lava. Subaerial lava that is depleted of volatiles or moving rapidly develops a clink-
ery, fractal surface known by the Hawaiian name aa. Its surface disaggregates into porous,
decimeter-sized fragments as the flow moves. Glassy spines pull out of the hot, separating
fragments like the sugar-rich spines of pulled taffy. Aa surfaces are almost impossible to

EBSCOhost - printed on 11/3/2023 9:55 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
5.3 Lava flows, domes, and plateaus 209

walk across and tend to shred shoes or boots. The fronts of aa flows are typically a few to
tens of meters high and advance like a caterpillar tractor tread, with cooled chunks of aa
falling from the top of the steep-fronted flow as hot lava in its interior overruns the previ-
ously fallen debris. A section of a cooled aa flow, thus, shows both a clinkery basal zone
and a rubbly top, separated by a more massive zone of vesicular lava. Although the surfaces
of aa flows are unforgettable for those who have experienced them, they are not volumin-
ous on the Earth’s surface.
Block lava. Block lavas, as their name suggests, are composed of meter-scale polyhedral
blocks of lava. Formed from more viscous lava than either aa or pahoehoe, on Earth they
form thick, short flows and seem to advance in a mode similar to aa. They are also not very
voluminous on Earth.
Pahoehoe lava. Pahoehoe is, by far, the most abundant type of lava on Earth and probably
on the other terrestrial planets. It forms both small flows and the giant sheets that blanket
Earth’s most extensive flood basalt provinces. Its name is derived from a Hawaiian word
that indicates “smooth going.” The surface of fresh pahoehoe is smooth, with a thin glassy
rind that weathers rapidly on Earth. Its surface is marked by broad billows and swales and
is locally puckered into drapery-like folds. Pahoehoe advances by a unique mechanism
that has only recently been understood. Called “inflation,” this mode of motion permits it
to travel large distances with only modest eruption rates (Self et al., 1998). Inflation was
discovered during observations of active pahoehoe lava flows in Hawaii (Hon et al., 1994).
Flows first advance as a thin sheet that rapidly covers the terrain. The upper surface of the
sheet cools rapidly, forming an insulating blanket under which hot lava continues to flow.
On a timescale of hours, hot, fluid lava intrudes beneath the chilled surface, uplifting it
and creating a nearly planar surface. Deep cracks form near the margins of inflated flows
as the upper surface is lifted above the original base and slabs of chilled crust tilt away
from the main mass of the flow. Uninflated areas form irregular depressions in the over-
all, nearly uniform, lava surface. Individual pahoehoe flows achieve thicknesses of tens of
meters in flood basalt provinces and may continue to inflate and spread for more than a
year. Over time, the lava feeding an inflated flow organizes into distinct streams, or “tubes,”
beneath the thickening crust. The flow velocity in an individual tube thus exceeds the rate of
advance of the lava flow as a whole. When the supply of fresh lava from the vent declines,
these tubes may drain and remain open as lava caves, or in the case of a thinner cover, the
empty lava tubes may collapse to leave sinuous channels on the flow surface. The surface
of large pahoehoe flows is locally marked by pits where inflation failed to occur, mounds
called “tumuli” where lava locally broke through the surface when lava channels became
blocked and small mounds accumulated, and “rootless cones” where the lava flowed over
wet ground, creating small phreatic explosions that threw up blocky mounds (often called
“hornitos”) or larger rimmed craters up to a few hundred meters in diameter.
Columnar jointing. As lava flows cool and crystallize the lava shrinks and cracks open
throughout the solidified flow. Because lava flows cool from outside inward, cracks initiate
on their surfaces and propagate toward their interiors. Uniform contraction of a thin surface
layer typically produces a polygonal fracture pattern, familiar from the surface of drying mud

EBSCOhost - printed on 11/3/2023 9:55 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
210 Volcanism

puddles. As cooling cracks propagate into the flow perpendicular to the cooling surface, they
elongate, creating a pattern of polygonal columns called “columnar jointing.” Columnar
joints are often prominent at the eroded edges of lava flows, forming spectacular outcrops at
Devils Postpile in California and the Giant’s Causeway in Ireland. Columnar joints form in
both silica-poor and silica-rich lava flows as well as pyroclastic deposits. Columns are typ-
ically tens of centimeters to meters in width. They have recently been observed on Mars, in
lava flows outcropping on the rims of the great canyons. The formation of columnar joints is
a universal process that occurs for any cooling, shrinking mass and can be analyzed in terms
of fundamental principles (Goehring et al., 2009).

5.3.2 The mechanics of lava flows

Spurred by the hazards posed by advancing lava flows, volcanologists have made many
efforts to compute the length and width of lava flows from their basic properties. Planetary
scientists, observing lava flows on other planets, have inverted these efforts to compute the
properties of lava from the morphology of the final flow. Although early efforts treated lava
as a Newtonian fluid, it is now agreed that lava behaves as a Bingham fluid and modern
flow models are based on this rheology. The Bingham model provides a number of simple
relations for the dimensions of a lava flow (Moore et al., 1978). One of the goals of planet-
ary volcanology is to relate estimates of the physical properties of the lava to its compos-
ition and, thus, learn about the chemistry of planetary crusts and mantles without having to
directly sample them.
The thickness H of an extensive sheet of lava resting on a uniform slope standing at angle
α to the horizontal is given in terms of the Bingham yield stress as simply:

YB
H= (5.10)
ρ g sin α

where g is the acceleration of gravity and ρ is the density of the lava. This formula strictly
applies only to an infinitely wide sheet of lava and is derived by setting the shear stress at
the base of the lava flow equal to the Bingham yield stress (Figure 5.13a). If the thickness
and other parameters of a lava flow can be estimated, this equation can be inverted to deter-
mine the Bingham yield stress. This model makes the uncomfortable prediction that a lava
flow on a level surface (α = 0) is infinitely thick. A more sophisticated model, still applying
the Bingham rheology, examines the force equilibrium of a lava sheet of variable thickness
resting on a level surface (Figure 5.13b). If y(x) is the thickness of the lava flow and x is the
distance to the edge of the flow, by balancing the horizontal thrust of the lava toward its edge
at x, given by ½ ρgy(x)2, by the basal force at the yield stress, xYB, for a thin slice of the flow,
an equation for the thickness of the lava flow at any distance from its edge is found:

2 xYB
y( x ) = . (5.11)
ρg

EBSCOhost - printed on 11/3/2023 9:55 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
5.3 Lava flows, domes, and plateaus 211

(a)

α
} semi-rigid plug,
τ < YB

} fluid, τ > YB

(b)
crystal / liquid mush
Pressure
ρgy y(x)

YB
Bingham stress

Figure 5.13 Panel (a) shows the shear stress acting on the base of an infinitely wide sheet of lava
resting on a surface sloping at a uniform angle α. The shear stress acting on its base is ρg H sin a. The
Bingham rheology implies that, so long as the shear stress is less than the Bingham yield stress YB, the
lava does not deform. At the higher shear stresses found beneath the uppermost semi-rigid plug, the lava
flows like a viscous fluid. Panel (b) shows the stresses near the edge of a lava flow, where the horizontal
thrust of the pressure in the lava flow to the left is balanced against the shear resistance of the Bingham
material to the right. The profile of the lava flow margin is a parabola, as described in the text.

This is the equation of a parabola and, indeed, measurements of the shape of the edges
of many lava flows give approximately parabolic profiles from which the Bingham yield
stress can be computed. This equation is often applied to estimate the Bingham yield stress
from the width of a lava flow. So long as the cross-profile of the flow is crudely parabolic
(not flat-topped: If the flow has a flat top, this indicates that the lava flow thickness is con-
trolled by the slope, through Equation (5.10)), then the width of the flow W = 2x and its
centerline thickness H determines the Bingham yield stress:

ρ g H2
YB = . (5.12)
W
Bingham yield stress. Equation (5.11) makes the prediction that the thickness of a lava
flow on a level surface increases as its width increases. This might be a sensible prediction,
but planetary surfaces are seldom level over very long distances. A better model results
from combining Equations (5.10) and (5.11) to model the edge of a wide lava flow on a

EBSCOhost - printed on 11/3/2023 9:55 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
212 Volcanism

semi-rigid plug

flowing

stagnant

surface
velocity
distance
WL

Figure 5.14 Cross section of a lava flow and the leveed channel it leaves behind. The upper section
shows the profile of the lava flow at its most active. The chilled bottom is frozen to the bed, but the
hotter lava above it flows as a viscous fluid so long as the shear stress exceeds the Bingham yield
stress. In low-stress regions the material moves as a semi-rigid plug, similar to the infinite flow in
Figure 5.13a. The surface velocity of the flow is shown in the middle panel. The lower section shows
the levees that remain after the lava drains away. The width and height of the levees can be related to
the Bingham yield stress.

gently sloping surface. In this case, the thickness of the center of a wide lava flow is given
by Equation (5.10) while its edges are described by (5.11). The combined equations also
give a nice description of the formation of the levees sometimes observed flanking lava
flows. So long as lava is erupting at a high rate, the profile of the lava flow bulges up in the
middle. However, as the eruption rate declines, the lava runs out of the channel, leaving the
material in the cooler levees behind (see Figure 5.14). The levee width WL and height HL
are easily measured on images of the flow so, applying Equation (5.11), the Bingham yield
stress can be computed from these observables:

ρ g H L2
YB = . (5.13)
2 WL

An alternative formula can be derived by combining this equation with (5.10) to elimin-
ate the flow thickness HL:

YB = 2 ρgWLsin2 α. (5.14)

The Bingham yield stress measured in this way suffers from the problem that the yield
stress applies to the stagnant levees, not the hotter active portion of the flow, but it does give
an order-of-magnitude estimate useful for comparisons between lava flows.

EBSCOhost - printed on 11/3/2023 9:55 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
5.3 Lava flows, domes, and plateaus 213

Lava viscosity. Once the Bingham yield stress is exceeded, a Bingham material flows as
a Newtonian viscous fluid. The Newtonian rheology has also been widely used to estimate
the viscosity η of actively flowing lava by measuring the surface velocity Usurf and estimat-
ing the flow depth H. In the case of laminar flow, the velocity profile is parabolic and the
surface velocity is:

ρ g H 2 sin α
Usurf = . (5.15)
2η

This velocity is approximately related to the total lava discharge QE (volume/unit time)
of the channel of width Wc by:

2
QE = U H Wc . (5.16)
3 surf
Similar equations can be derived for turbulent flow, but because most known lava flows
seem to have moved in the laminar regime, with low Reynolds number, they are not tabu-
lated here.
Equations (5.15) and (5.16) are the basis of a widely used viscosity estimate (Nichols,
1939) called the Jeffreys’ equation (after a 1925 paper by Harold Jeffreys):

ρ g H 3 Wc sin α
η= (5.17)
nQE

where n is a numerical parameter equal to 3 for broad flows and 4 for narrow flows. This
equation assumes that lava is a Newtonian fluid and so its predictions for planetary lava
flows are open to question.
For a Bingham fluid, the flow depth H should properly be the depth of the fluid exclud-
ing the semi-rigid plug of lava on top of the flow (the flowing portion must be subject to a
shear stress exceeding the Bingham yield stress, and so requires some overburden before
this stress is reached) and the overall discharge should include the volume of this plug. In
practice these distinctions are usually ignored, as there is no simple way of separating them
from image data. Thus, planetary estimates of the viscosity and Bingham yield stress are
only rough estimates, not precise measurements.
Lava effusion rate. The effusion rate of lava from a feeder vent provides information
about how magma is transported beneath the surface of a planet and is, thus, important for
comparing volcanism on different bodies. It also plays a role in determining the maximum
length of a lava flow. The total volume VL of a lava flow can be estimated from its area A
and thickness H: VL = A H. If the duration of the eruption te is known, then the average
effusion rate is simply QE = VL / te = AH/te. However, the duration of past eruptions cannot
be measured directly and so other ways of estimating eruption rates must be found. One
current approach is to use a dimensionless measure of heat transport, the Grätz number,
Gz, which is the ratio between the heat advected in a flow to the heat conducted. The Grätz
number for a lava flow is defined as:

EBSCOhost - printed on 11/3/2023 9:55 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
214 Volcanism

QE H
Gz = (5.18)
κA

where κ is the thermal diffusivity of the lava composing the flow. It has been observed
that this number is typically about 300 for basaltic lava flowing in channels (Gregg and
Fink, 1999). Using this number, Equation (5.18) can readily be inverted to determine the
effusion rate. However, tube-fed pahoehoe flows probably cool much more slowly and this
may greatly overestimate the effusion rate for such flows. Unfortunately, no systematic
estimates yet exist for the Grätz number of pahoehoe flows, although one estimate of the
cooling time of the crust (Hon et al., 1994) suggests that it may be of order 10. Another
estimate from the 1300 km2 Roza flow of the Columbia River Basalt Province gives
Gz ~ 32. Eruption rates for inflated flows based on the Grätz number for channel flows may
thus be about 10 times too large.
An estimate of the effusion rate for a single lava flow on a surface of average slope α and
Bingham yield stress YB can be obtained by combining Equations (5.10) and (5.18):

κ AGz
QE = ρ g sin α . (5.19)
YB

Equations of this sort can be used to make models of lava flow formation if the mater-
ial parameters of the flow can be estimated. Rearrangement of this equation indicates that
the effusion rate is probably the major factor in controlling the final area of a lava flow.
Table 5.6 collects a number of estimates of the parameters describing lava flows on the ter-
restrial planets and moons.
Lava composition and rheology. It is widely believed that the Bingham yield stress
is correlated with silica content, as is the viscosity. Data compilations suggest that yield
stresses in the range of 100 Pa indicate basalt with silica contents of about 40 wt%,
whereas values near 105 Pa indicate andesitic lavas with silica contents around 65 wt%.
However, this relationship does not seem to be monotonic at higher silica contents, as
rhyolites also have yield stresses in the range of 105 Pa (Moore et al., 1978). It is clear
from Table 5.6 that the properties of silicate lavas are generally similar on all of the ter-
restrial planets. Silica-rich lavas tend to have higher Bingham strengths and viscosities
than silica-poor lavas. The major exception is Io, where very high eruption temperatures
produce weak, highly fluid lavas. Eruption rates are highly variable and it seems that the
eruption rate, more than any other factor, is the primary determinant of the size of an indi-
vidual lava flow.

5.3.3 Lava domes, channels, and plateaus

Lava flows are highly complex landforms whose surface features are too varied to treat in
a chapter of this length, so the interested reader who wants to go further is urged to consult
some of the specialized works listed at the end of this chapter. However, a few features that
are prominent in images of planetary surfaces deserve a brief mention here.

EBSCOhost - printed on 11/3/2023 9:55 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
5.3 Lava flows, domes, and plateaus 215

Table 5.6 Rheologic properties of lava flows in the Solar System

Bingham yield strength Viscosity Effusion rate

Location (type) (Pa) (Pa-s) (m3/s)

Earth
Mauna Loa, HI (basalt) (3.5–72) × 102 1.4 × 102 –5.6 × 106 417–556
Makaopuhi, HI (basalt) (8–70) × 103 (7–45) × 102 –
Mount Etna, Italy (basalt) 9.4 × 103 9.4 × 103 0.3–0.5
Sabancaya, Peru 5 × 104 – 1.6 × 106 7.3 × 109 – 1.6 × 1–13
(andesite) 1013
Mono Craters, CA (1.2–3) × 105 – –
(rhyolite)
Moon
Mare Imbrium (1.5–4.2) × 102 – –
Gruithuisen domes (7.7–14) × 104 (3.2–14) × 108 5.5–120
Venus
Artemis festoons (4.1–13) × 104 7 × 106 – 7.3 × 109 (2.5–10) × 103
Atalanta Festoon 1.2 × 105 2.3 × 109 950
Mars
Arsia Mons (2.5–3.9) × 103 9.7 × 105 (5.6–43) × 103
Ascraeus Mons (3.3–83) × 103 (2.1–640) × 103 18–60
Tharsis plainsa (1.2–2.4) × 102 (8–58) × 102 (2–25) × 102
Io
Shield volcanoesb (1–10) × 101 103–105 ~3000
Ariel cryovolcanism
Flowsc (6.7–37) × 103 (9–45) × 1014 –

Data from Hiesinger et al. (2007), except:

a
Hauber et al. (2010)
b
Schenk et al. (2004)
c
Melosh and Janes (1989)

Pancake domes. Figure 5.15 illustrates a cluster of “pancake domes” on Venus. These
domes resemble silica-rich rhyolite domes on the Earth (an example is the Mono Craters
in California), although the 25 km-wide Venusian domes are larger than most terrestrial
occurrences. These constructs are believed to form from the extrusion of highly viscous
lava that flows only a short distance from an underlying vent. When lava of this type
erupts onto steep slopes the lava oozes downhill to freeze into thick, stubby, elongated
lobes.
Sinuous rilles. First named by German astronomer Johann Schröter (1745–1816) in
1787, sinuous rilles immediately caught the attention of some of the first observers of the
Moon, who thought that they had discovered river valleys. They often head in circular to
irregular pits and tend to decrease in width as they meander downslope in wide curves
(Figure 5.16). Typically several kilometers wide and hundreds of meters deep, they may

EBSCOhost - printed on 11/3/2023 9:55 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
216 Volcanism

Figure 5.15 Pancake domes on Venus. These seven domes in eastern Alpha Regio are each about 25
km in diameter and 750 m high. They are interpreted as extrusions of very viscous lava from a central
source. NASA Magellan radar image PIA00215. North is up.

Figure 5.16 Apollo 15 image of the Aristarchus plateau of the Moon, showing flooded craters
and sinuous rilles of this highly volcanic region. The flooded crater in the foreground is Prinz,
46 km in diameter. Aristarchus crater, 40 km diameter, is in the right background. NASA image
A15_m_2606.

EBSCOhost - printed on 11/3/2023 9:55 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
5.3 Lava flows, domes, and plateaus 217

stretch hundreds of kilometers in length. Some appear to stop for a short distance and then
begin again as if whatever flowed through them went underground for a while. Sinuous
rilles occasionally degenerate downslope into a line of rimless pit craters, again suggesting
collapse into an underlying cavity. A small rille on the floor of the much larger Schröter’s
Rille seems to wind enigmatically in and out of its side walls. They are invariably associ-
ated with lava plains. The Apollo 15 astronauts famously landed on the margin of Hadley
Rille, but were unable to demonstrate its origin other than to show that the rocks outcrop-
ping on its rim are basalts.
Sinuous rilles are observed on the Earth, Mars, and Venus as well as the Moon. All of
these characteristics suggest that sinuous rilles represent channels through which lava once
flowed. However, much controversy still centers about how completely they were filled,
when they were active, and whether they are collapsed lava tubes or were once open to
the sky. Some of them cut deep into the surrounding surface, suggesting that the flow-
ing lava somehow deepened its channel. Much discussion has focused on the process of
thermal erosion, by which the hot, flowing lava heats and softens the underlying rocks,
melting them away and, thus, slowly deepening the channel. On the other hand, flowing
lava is quite capable of quarrying away jointed blocks in its bed through the hydrodynamic
“plucking” process by which water carves channels into bedrock. While thermal erosion
is theoretically capable of deepening channels, its action has never been unambiguously
demonstrated on the Earth, whereas there are good Hawaiian examples of plucking.
Lava plateaus. Lava plateaus are stacks of many individual lava flows. When a rising,
hot, mantle plume arrives near a planet’s surface it may deliver both heat and differentiated
melts to the base of the crust for a long period of time. In this case magma spills out onto
the surface in many, perhaps hundreds to thousands, of individual flows. So long as magma
can continue to penetrate the lava flows already congealed on the surface, more material
builds up, creating a thick pile of individual flows with a combined volume that may rival
the volume of the crust itself.
On the Moon, lava began to erupt through the thin nearside crust about 500 Myr after
the large basins were themselves created by impacts. Because of the long time interval
between basin formation and the lava flows, there does not seem to be any genetic connec-
tion: The impacts did not initiate the volcanism. Instead, the impacts created topographic-
ally low basins and thin crust through which the lava extruded. Although individual lunar
flows are hundreds of meters thick, owing to the low lunar gravity, it required the accumu-
lation of many individual flows to build up the multi-kilometer thick piles of lava that we
now recognize as the lunar mare.
Similar stacks of individual lava flows from long-continued volcanism created the tens
of kilometers thick Tharsis plateau on Mars. Thinner lava plains cover much of the rest of
the surface of Mars, with individual thin flows extending thousands of kilometers. Most of
the surface of Venus was covered by extensive lava flows around 700 Myr ago, obliterating
most of whatever surface Venus originally possessed. A single lava channel, Baltis Vallis,
on Venus stretches 6800 kilometers across its surface. The Earth possesses dozens of broad,
large igneous provinces that have erupted hundreds of thousands of cubic kilometers of

EBSCOhost - printed on 11/3/2023 9:55 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
218 Volcanism

basalt throughout the geologic history of our planet, each eruptive episode lasting only a
few million years. Recent images returned from the MESSENGER spacecraft show that
Mercury’s extensive intercrater plains are volcanic, again indicating the importance of
repeated volcanic eruptions.
Io is the volcanic body par excellence, whose crust seems to be entirely created by
repeated volcanic flows, which renew its surface at the average rate of 1.5 cm/yr. Although
its visible surface is largely coated with volatile sulfur and SO2, its overall density and the
high temperatures of its lavas measured by the Galileo probe indicate that its volcanism is
predominantly silicic, perhaps dominated by ultramafic lavas.
Less is known of the water-mediated cryovolcanism on the icy satellites, but it is clear
that flowing liquids have covered many of the icy satellites’ surfaces. Volcanism has been
active on nearly all of the solid bodies in the Solar System, so understanding this process is
a vital part of the study of planetary surfaces in general.

Further reading
The history of investigation of volcanoes and magma on the Earth is engagingly told in the
history by Sigurdsson (1999). The book by Williams and McBirney (1979) provides excel-
lent quantitative coverage of the basic ideas of volcanic phenomena, but is now becoming
somewhat dated. A more up-to-date reference with similar coverage is Schmincke (2003),
but the best and most readable reference, although not as quantitative as Williams and
McBirney is Francis and Oppenheimer (2003). A good general reference to the chemistry
of rocks and melting, as well as much more, is McSween et al. (2003). Io is the volcanic
moon par excellence and an entire book is now devoted to it Davies (2007).

Exercises
Note: As for all problems of this kind, your best guide to a correct answer is to make sure
that the dimensional units of all results are correct!

5.1 Squeezing magma sponges

Derive Equation (5.6) for the timescale over which magma percolates out of a magma-
saturated layer of thickness h by equating the volume discharge per unit area Q times the
percolation time tpercolate to the volume of melt φ h in the layer.
Use this equation to estimate the timescale over which the 100 km thick asthenosphere
underlying oceanic plates on Earth would lose 10% of its melt if the grain size in the mantle
is about 1 mm and the viscosity of hot basaltic melt is 104 Pa-s.
Now shift to the outer Solar System and estimate the rate at which water “magma” seg-
regated from the body of Uranus’ 1160 km diameter satellite Ariel (Voyager imaged lava
flows on its surface). Ariel’s mean density is 1660 kg/m3 and the viscosity of water at its

EBSCOhost - printed on 11/3/2023 9:55 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
Exercises 219

melting point is about 1.5 ×10–3 Pa-s. What does this tell you about the ability of planets to
retain melt in their interiors?

5.2 Feeding Ariel’s volcanoes

If pure liquid water ascends from the interior of Ariel by means of dikes of width about 1 cm,
use Equation (5.8) to estimate the mean velocity of the water in the dike. Next, use Equation
(5.7) to estimate the vertical depth Lc of the dike as it rises toward the surface. Finally,
assuming that the horizontal extent of a dike segment is approximately the same as its ver-
tical depth, compute the volume of water carried by this dike. Combining the mean velocity
and area of the dike as it breaches the surface, estimate the eruption rate of water onto the
surface. At this rate, how long would it take to build up a cryovolcano 10 km in diameter and
1 km high? How many dikes must discharge their contents to build the volcano?

5.3 Plumbing Enceladus’ geysers

The NASA–ESA Cassini Mission discovered that water-dominated geysers erupt continu-
ously from the south pole of Enceladus, a Saturnian satellite about 500 km in diameter.
The geysers spew out of four long fissures called “tiger stripes” whose hottest portions
extend about 50 km horizontally. The CISR Infrared Spectrometer observed an excess
heat flow from the entire region of about 6 GW. If the liquid water freezes and cools to
the ambient temperature of 69 K, use Equation (5.1) to estimate the volume eruption rate,
per unit length of the fissures, of water necessary to supply this heat flux. Supposing that
the geysers are fed by pure liquid water, apply Equation (5.8) (plus a little creative think-
ing) to estimate the width of the dikes feeding water to the surface and the mean velocity
of the water moving up the dike. Finally, use either Equation (5.7) or (5.9) to estimate the
depth of the fissures feeding the eruption. If these eruptions are fed by “quanta” of water
trapped in rising dikes of these dimensions, what is the duration of a single eruptive pulse
as an individual dike breaches the surface? Note that there are multiple valid ways to get
the correct answer.
Useful data: The latent heat of freezing for water is 334 kJ/kg and its heat capacity is 4.2
kJ/kg-K. The viscosity of water is about 1.5 × 10–3 Pa-s. Young’s modulus of Enceladus’ ice
crust is about 1010 Pa. The density contrast between the erupting fluid and the surrounding
ice is almost completely unknown. A crude estimate is to suppose that it is about 10% of
the density of ice. The surface acceleration of gravity on Enceladus is 0.11 m/s2.

5.4 Volcanic bombs in orbit: a natural answer to StarWars

Ronnie, an inquisitive sixth-grade visitor to the planetarium, wants to know if big, noisy
volcanoes on Earth can eject rocks into space. Johnnie, another budding intellect, thinks that
lunar volcanoes eject tektites. Use your knowledge of the thermodynamics of expansion to
compute the maximum expansion velocity v∞ = 2h , where h is the specific enthalpy of

EBSCOhost - printed on 11/3/2023 9:55 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
220 Volcanism

the expanding gas for the volcanic gases CO, H2O, and H2, at a typical eruption tempera-
ture of 1200°C. Compare these velocities to the escape velocities of the Earth, Mars, and
the Moon.
If water vapor is the primary volcanic gas on Mars, how high an eruption temperature
would be required for Martian volcanoes to eject volcanic bombs from the planet? How
hot a volcano is needed for it to eject material from Earth? Do you think that Ronnie’s and
Johnnie’s ideas make any sense?
Useful data: The specific enthalpy h of a gas is approximately h= cPT, where T is the
absolute temperature and cP is the specific heat at constant pressure, cP = 7/2 R, where R
is the gas constant, 8.317 J/mol-K.

5.5 Go with the flow

(a) Steep flow fronts are observed at the edges of broad, extensive lava flows on the lunar
Mare. These lava flows are considerably thicker than terrestrial lava flows, often reach-
ing 100 m in height. The average slope of one such flow is about 0.5°. Use Equation
(5.10) to compute the Bingham yield stress of this lava flow and compare it to terres-
trial lava flows. Is lunar magma substantially stronger than terrestrial magma?
(b) Viking images of Olympus Mons on Mars reveal numerous leveed lava flows running
down its flanks, which slope at about 7° to the horizontal. The widths of the levees are
easily measured from orbit. One large flow is observed to possess levees about 1 km
wide. Estimate the Bingham yield stress of this flow and compare it to that measured
on terrestrial lava flows. What can you deduce (if anything) about the magma that cre-
ated this flow? This leveed channel is observed to have fed a small lava flow on the
flanks of Olympus Mons that expanded to about 8 km wide and ran 50 km down the
slope. Estimate the effusion rate of this flow. What additional information would you
need to estimate the viscosity of the lava? Can you envisage obtaining this information
from orbit?

5.6 Big volcanoes on little planets

Use the theory of lava flow lengths derived in Section 5.3.2 to relate the radius of a volcanic
edifice, L, with a height H to the eruption rate QE. Assume that central eruptions last long
enough that the length of each flow that builds up the edifice is limited by its solidification
time. That is, if L is of order QE te /h, where te is the duration of the flow and h is its thick-
ness, then te is of order h2/k, where the thermal diffusivity of rock, k, is about 10–6 m2/s.
Assume that the thickness of the flow is given by the Bingham yield stress YB,
h = YB/(ρg sin α)

where α is the mean angle of the volcano’s slope, tan α = H/L. Derive an expression
relating volcano radius (that is, maximum lava flow length, L) to eruption rate QE and

EBSCOhost - printed on 11/3/2023 9:55 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
Exercises 221

height H. Note that you will have to make some approximations to account for the fact
that the volcano is circular in plan while the estimates above for L and h are on a per unit
width basis.
Use this relation to derive an eruption rate and mean flow thickness for Olympus Mons
on Mars, a 400 km diameter central volcano made predominantly of basaltic lava, where YB
= 103 Pa. The average surface slope of this 24 km high volcanic edifice is about 7°. Compare
the derived eruption rate to the typical eruption rate of terrestrial basaltic volcanoes, ca.
3 × 107 m3/day. What does this mean?
Note: There is no single “right answer” to this problem, which requires you to make a
number of “reasonable” approximations. This problem is a thinking exercise in how simple
theories are concocted.

EBSCOhost - printed on 11/3/2023 9:55 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
6
Impact cratering

The dominant surface features of the Moon are approximately circular

depressions, which may be designated by the general term craters …

Solution of the origin of the lunar craters is fundamental to the unravel-
ing of the history of the Moon and may shed much light on the history of
the terrestrial planets as well.
E. M. Shoemaker (1962)

Impact craters are the dominant landform on the surface of the Moon, Mercury, and
many satellites of the giant planets in the outer Solar System. The southern hemisphere
of Mars is heavily affected by impact cratering. From a planetary perspective, the rarity
or absence of impact craters on a planet’s surface is the exceptional state, one that needs
further explanation, such as on the Earth, Io, or Europa. The process of impact cratering
has touched every aspect of planetary evolution, from planetary accretion out of dust or
planetesimals, to the course of biological evolution.
The importance of impact cratering has been recognized only recently. E. M. Shoemaker
(1928–1997), a geologist, was one of the first to recognize the importance of this process
and a major contributor to its elucidation. A few older geologists still resist the notion that
important changes in the Earth’s structure and history are the consequences of extraterres-
trial impact events. The decades of lunar and planetary exploration since 1970 have, how-
ever, brought a new perspective into view, one in which it is clear that high-velocity impacts
have, at one time or another, affected nearly every atom that is part of our planetary system.
Impact cratering is crucially important for the accumulation of the planets in the first place
and has played major roles from the formation of the most ancient planetary landscapes
to the creation and maintenance of the modern regolith of airless bodies. In an important
sense, impact cratering is the most fundamental geologic process in the Solar System.

6.1 History of impact crater studies

Craters were discovered in 1610 when Galileo pointed his first crude telescope at the Moon.
Copyright 2011. Cambridge University Press.

Galileo recognized the raised rims and central peaks of these features, but described them
only as circular “spots” on the Moon. Although Galileo himself did not record an opinion
on how they formed, astronomers argued about their origin for the next three centuries.

222

EBSCO Publishing : eBook Academic Collection (EBSCOhost) - printed on 11/3/2023 9:56 AM via YDUQS
AN: 399390 ; H. Jay Melosh.; Planetary Surface Processes
Account: s4526441.main.estacio
6.2 Impact crater morphology 223

Astronomer J. H. Schröter first used the word “crater” in a non-genetic sense in 1791. Until
the 1930s most astronomers believed the Moon’s craters were giant extinct volcanoes: the
impact hypothesis, proposed sporadically over the centuries, did not gain a foothold until
improving knowledge of impact physics showed that even a moderately oblique high-speed
impact produces a circular crater, consistent with the observed circularity of nearly all of
the Moon’s craters. Even so, many astronomers clung to the volcanic theory until high-
resolution imagery and in situ investigation of the Apollo program in the early 1970s firmly
settled the issue in favor of an impact origin for nearly every lunar crater. In the current
era spacecraft have initiated the remote study of impact craters on other planets, beginning
with Mariner 4’s unexpected discovery of craters on Mars on July 15, 1965. Since then
craters have been found on almost every other solid body in the Solar System.
Meteor Crater, Arizona, was the first terrestrial structure shown unambiguously to be of
impact origin. D. M. Barringer (1860–1929) investigated this 1 km diameter crater and its
associated meteoritic iron in detail from 1906 until his death in 1929. After Barringer’s work
a large number of small impact structures resembling Meteor Crater have been found. Impact
structures larger than about 5 km in diameter were first described as “cryptovolcanic” because
they showed signs of violent upheaval but were not associated with the eruption of volcanic
materials. J. D. Boon and C. C. Albritton in 1937 proposed that these structures were really
caused by impacts, although final proof had to wait until the 1960s when the shock-metamor-
phic minerals coesite and stishovite proved that the Ries Kessel in Germany and subsequently
many other cryptovolcanic structures are the result of large meteor impacts.
Finally, theoretical and experimental work on the mechanics of cratering began during
World War II and was extensively developed in later years. This work was spurred partly
by the need to understand the craters produced by nuclear weapons and partly by the fear
that the “meteoroid hazard” to space vehicles would be a major barrier to space explor-
ation. Computer studies of impact craters were begun in the early 1960s. A vigorous and
highly successful experimental program to study the physics of impact was initiated by D.
E. Gault (1923–1999) at NASA’s Ames facility in 1965.
These three traditional areas of astronomical crater studies, geological investigation of
terrestrial craters, and the physics of cratering have blended together in the post-Apollo era.
Traditional boundaries have become blurred as extraterrestrial craters are subjected to direct
geologic investigation, the Earth’s surface is scanned for craters using satellite images, and
increasingly powerful computers are used to simulate the formation of both terrestrial and
planetary craters on all size scales. The recent proposals that the Moon was created by the
impact of a Mars-sized protoplanet with the proto-Earth 4.5 Gyr ago and that the Cretaceous
era was ended by the impact of a 15 km diameter asteroid or comet indicate that the study of
impact craters is far from exhausted and that new results may be expected in the future.

6.2 Impact crater morphology

Fresh impact craters can be grossly characterized as “circular rimmed depressions.”
Although this description can be applied to all craters, independent of size, the detailed

EBSCOhost - printed on 11/3/2023 9:56 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
224 Impact cratering

form of craters varies with size, substrate material, planet, and age. Craters have been
observed over a range of sizes varying from 0.1 μm (microcraters first observed on lunar
rocks brought back by the Apollo astronauts) to the more than 2000 km diameter Hellas
Basin on Mars. Within this range a common progression of morphologic features with
increasing size has been established, although exceptions and special cases are common.

6.2.1 Simple craters

The classic type of crater is the elegant bowl-shaped form known as a “simple crater”
(Figure 6.1a). This type of crater is common at sizes less than about 15 km diameter on the
Moon and 3 to about 6 km on the Earth, depending on the substrate rock type. The inter-
ior of a simple crater has a smoothly sloping parabolic profile and its rim-to-floor depth is
about 1/5 of its rim-to-rim diameter. The sharp-crested rim stands about 4% of the crater
diameter above the surrounding plain, which is blanketed with a mixture of ejecta and deb-
ris scoured from the pre-existing surface for a distance of about one crater diameter from
the rim. The thickness of the ejecta falls off as roughly the inverse cube of distance from
the rim. The surface of the ejecta blanket is characteristically hummocky, with mounds and
hollows alternating in no discernible pattern. Fields of small secondary craters and bright
rays of highly pulverized ejecta that extend many crater diameters away from the pri-
mary may surround particularly fresh simple craters. Meteor Crater, Arizona, is a slightly
eroded representative of this class of relatively small craters. The floor of simple craters is
underlain by a lens of broken rock, “breccia,” which slid down the inner walls of the crater
shortly following excavation. This breccia typically includes representatives from all the
formations intersected by the crater and may contain horizons of melted or highly shocked
rock. The thickness of this breccia lens is typically 1/2 to 1/3 of the rim-to-floor depth.

6.2.2 Complex craters

Lunar craters larger than about 20 km diameter and terrestrial craters larger than about 3
km have terraced walls, central peaks, and at larger sizes may have flat interior floors or
internal rings instead of central peaks. These craters are believed to have formed by the
collapse of an initially bowl-shaped “transient crater,” and because of this these more com-
plicated structures are known as “complex craters” (Figure 6.1b). The transition between
simple and complex craters has now been observed on the Moon, Mars, Mercury, and the
Earth, as well as on some of the icy satellites in the outer Solar System. In general the tran-
sition diameter scales as g–1, where g is the acceleration of gravity at the planet’s surface,
although the constant in the scaling rule is not the same for icy and rocky bodies. This is
consistent with the idea that complex craters form by collapse, with icy bodies having
only about 1/3 the strength of rocky ones. The floors of complex craters are covered by
melted and highly shocked debris, and melt pools are sometimes seen in depressions in the
surrounding ejecta blanket. The surfaces of the terrace blocks tilt outward into the crater
walls, and melt pools are also common in the depressions thus formed. The most notable

EBSCOhost - printed on 11/3/2023 9:56 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
6.2 Impact crater morphology 225

(a) (b)

Figure 6.1 Impact crater morphology as a function of increasing size. (a) Simple crater: 2.5 km
diameter crater Linné on the Moon (Apollo 15 Panometric Photo strip 9353). (b) Complex crater with
central peak: 102 km diameter crater Theophilus on the Moon (Apollo 16 Hasselblad photo 0692).
(c) Complex crater with internal ring: Mercurian craters Strindberg (165 km diameter) to the lower
right and Ahmad Baba (115 km) to the upper left (Mariner 10 FDS 150, rectified). (d) Multiring
basin: 620 km diameter (of most prominent ring) Orientale basin on the Moon (LROC WAC mosaic.
Full width of image PIA13225 is 1350 km. NASA/GSFC/ASU).

structural feature of complex craters is the uplift beneath their centers. The central peaks
contain material that is pushed upward from the deepest levels excavated by the crater.
Study of terrestrial craters has shown that the amount of structural uplift hsu is related to the
final crater diameter D by:

EBSCOhost - printed on 11/3/2023 9:56 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
226 Impact cratering

hsu = 0.06 D1.1 (6.1)

where all distances are in kilometers. The diameter of the central peak complex is roughly
22% of the final rim-to-rim crater diameter in craters on all the terrestrial planets.
Complex craters are shallower than simple craters of equal size and their depth increases
slowly with increasing crater diameter. On the Moon, the depth of complex craters increases
from about 3 km to only 6 km while crater diameters range from 20 to 400 km. Rim height
also increases rather slowly with increasing diameter because much of the original rim slides
into the crater bowl as the wall collapses. Complex craters are thus considerably larger than
the transient crater from which they form: estimates suggest that the crater diameter may
increase as much as 60% during collapse. A useful scaling relation suggests that the rim-to-
rim diameter of a complex crater is related to the transient crater Dt diameter by:

Dt1.13
D = 1.17 (6.2)
Ds0−.13c

where Ds-c is the diameter at the simple to complex transition, about 3.2 km on the Earth
and 15 km on the Moon.
As crater size increases, the central peaks characteristic of smaller complex craters give
way to a ring of mountains (Figure 6.1c). This transition takes place at about 140 km diam-
eter on the Moon and about 20 km diameter on the Earth, again following a g–1 rule. Known
as “peak-ring craters,” the central ring is generally about 0.5 of the rim-to-rim diameter of
the crater on all the terrestrial planets.
The ejecta blankets of complex craters are generally similar to those of simple craters,
although radial troughs and ridges replace the “hummocky” texture characteristic of simple
craters as size increases. Fresh complex craters also have well-developed fields of sec-
ondary craters, including frequent clusters and “herringbone” chains of closely associated,
irregular, secondary craters. Very fresh craters, such as Copernicus and Tycho on the Moon,
have far-flung bright ray systems.

6.2.3 Multiring basins

The very largest impact structures are characterized by multiple concentric circular scarps,
and are, hence, known as “multiring basins.” The most famous such structure is the 930 km
diameter Orientale basin on the Moon (Figure 6.1d), which has at least four nearly com-
plete rings of inward-facing scarps. Although opinion on the origin of the rings still varies,
most investigators feel that the scarps represent circular faults that slipped shortly after
the crater was excavated. There is little doubt that multiring basins are caused by impacts:
most of them have recognizable ejecta blankets characterized by a radial ridge-and-trough
pattern. The ring diameter ratios are often tantalizingly close to multiples of 2 , although
no one has yet suggested a convincing reason for this relationship.
In contrast to the simple/complex and central peak/internal ring transitions discussed
above, the transition from complex craters to multiring basins is not a simple function of

EBSCOhost - printed on 11/3/2023 9:56 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
6.2 Impact crater morphology 227

Figure 6.2 The Valhalla basin on Callisto. The original impact was within the central bright patch,
which is 300 km in diameter and may represent ejecta from a still smaller (now unrecognizable)
crater. This central zone is surrounded by an annulus of sinuous ridges, which in turn is surrounded
by an annulus of trough-like grabens, which can be recognized up to 2000 km from the basin center.
Voyager 1 mosaic PIA02277. NASA/JPL.

g–1. Although multiring basins are common on the Moon, where the smallest has a diameter
of 410 km, none at all has been recognized on Mercury, with its two times larger gravity,
even though the largest crater, Caloris Basin, is 1540 km in diameter. The situation on Mars
has been confused by erosion, but it is difficult to make a case that even the 1200 km diam-
eter Argyre Basin is a multiring structure. A very different type of multiring basin is found
on Jupiter’s satellite Callisto, where the 4000 km diameter Valhalla basin (Figure 6.2) has
dozens of closely spaced rings that appear to face outward from the basin center. Another
satellite of Jupiter, Ganymede, has both Valhalla-type and Orientale-type multiring struc-
tures. Since gravity evidently does not play a simple role in the complex crater/multiring
basin transition, some other factor, such as the internal structure of the planet, may have to
be invoked to explain the occurrence of multiring basins. The formation of such basins is
currently a topic of active research.

EBSCOhost - printed on 11/3/2023 9:56 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
228 Impact cratering

Figure 6.3 Distinctive asymmetrical ejecta surrounds a 370 m diameter crater on the lunar mare near
the Linné crater 25° N, 17° W. This pattern is typical of ejecta surrounding a crater formed by an
impact at an angle between 20° and 45° from the horizontal. The crater itself is still circular, but there
is an uprange wedge in which very little ejecta is present. Portion of Apollo 15 Panometric image
AS15-P-9337.

6.2.4 Aberrant crater types

On any planetary surface a few craters can always be found that do not fit the simple
size–morphology relation described above. These are generally believed to be the result of
unusual conditions of formation in either the impacting body or the planet struck. Circular
craters with asymmetric ejecta blankets (Figure 6.3) or elliptical craters with “butterfly-
wing” ejecta patterns are the result of very low impact angles. Although moderately oblique
impacts yield circular craters, at impact angles less than about 6° from the horizontal the
final crater becomes elongated in the direction of flight. Small, apparently concentric, cra-
ters or craters with central dimples or mounds on their floors are the result of impacts into
a weak layer underlain by a stronger one. The ejecta blankets of some Martian craters
show petal-like flow lobes that are believed to indicate the presence of liquid water in the

EBSCOhost - printed on 11/3/2023 9:56 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
6.3 Cratering mechanics 229

excavated material. Craters on Ganymede and Callisto have central pits at a diameter where
internal rings would be expected on other bodies. The explanation for these pits is still
unknown. In spite of these complications, however, the simple size–morphology relation
described above provides a simple organizing principle into which most impact craters can
be grouped.

6.2.5 Degraded crater morphology

All crater morphologies are observed in either “fresh,” pristine landforms or as erosionally
degraded forms. The dominant degradational process determines the detailed changes in
crater morphology. Other impact craters are often the exclusive agent of degradation on
airless bodies like the Moon (although burial by floods of lava can be locally important
there as well). This form of erosion fuels a variety of surface creep (see Section 8.1 for
more detail). Sharp terrain features such as crater rims are rounded and battered out of
line by smaller impacts, crater bowls are gradually filled and slopes become gentler. At the
extreme limit, craters may fade into invisibility as their place is occupied by large numbers
of overlapping craters. Moon mappers have established “degradation classes” for lunar cra-
ters that range from fresh to nearly invisible and depend upon the initial crater size. Used in
conjunction with crater density data, the numbers of craters in different degradation classes
can be used to infer the age and cratering history of a given site for different populations
of impactors.
On ancient Mars, fluvial processes dissected impact craters by gullying and channel for-
mation. Old craters there were filled with sediment and lava. Wind-blown sand and dust fill
small craters on Mars today and erode their rims into crenulated yardang ridges.
Erosion on Earth is so active that craters are among the most rare landforms. Fluvial
deposition fills in closed depressions, such as crater bowls, and fluvial erosion gullies rims
and quarries away ejecta blankets. Many of the craters that are fortunately preserved were
once completely buried, preserving them, and are only now being exhumed: The Ries cra-
ter in Germany is an example of this fortunate circumstance. Differential erosion of the
various rock units etches out the present morphology of the crater to create its modern
landscape.
The varieties of degraded crater morphologies are as diverse as the different agencies of
erosion or deposition. Recognition of degraded crater forms must, thus, take the behavior
of each process into account as observers attempt to reconstruct the original structure of
an impact crater.

6.3 Cratering mechanics

The impact of an object moving at many kilometers per second with the surface of a planet
initiates an orderly sequence of events that eventually produces an impact crater. Although
this is really a continuous process, it is convenient to break it up into distinct stages that
are each dominated by different physical processes. This division clarifies the description

EBSCOhost - printed on 11/3/2023 9:56 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
230 Impact cratering

of the overall cratering process, but it should not be forgotten that the different stages
really grade into one another and that a perfectly clean separation is not possible. The most
commonly used division of the impact cratering process is into contact and compression,
excavation, and modification.

6.3.1 Contact and compression

Contact and compression is the briefest of the three stages, lasting only a few times longer
than the time required for the impacting object (referred to hereafter as the “projectile”) to
traverse its own diameter, tcc ≈ L / vi, where tcc is the duration of contact and compression, L
is the projectile diameter, and vi is the impact velocity. During this stage the projectile first
contacts the planet’s surface (hereafter, “target”) and transfers its energy and momentum
to the underlying rocks. The specific kinetic energy (energy per unit mass, ½ vi2) possessed
by a projectile traveling at even a few kilometers per second is surprisingly large. A. C.
Gifford, in 1924, first realized that the energy per unit mass of a body traveling at 3 km/s
is comparable to that of TNT. Gifford proposed the “impact-explosion analogy,” which
draws a close parallel between a high-speed impact and an explosion. During contact and
compression the projectile plunges into the target, generating strong shock waves as the
material of both objects is compressed. The strength of these shock waves can be computed
from the Hugoniot equations, first derived by P. H. Hugoniot in his 1887 thesis, that relate
quantities in front of the shock (subscript 0) to quantities behind the shock (no subscript):

ρ (U − u p ) = ρ0 U
P − P0 = ρ0 u p U
( P + P0 )  1 1 
E − E0 =  ρ − ρ  . (6.3)
2 0

In these equations P is pressure, ρ is density, up is particle velocity behind the shock (the
unshocked material is assumed to be at rest), U is the shock velocity, and E is energy per
unit mass. These three equations are equivalent to the conservation of mass, momentum,
and energy, respectively, across the shock front. They hold for all materials, but do not pro-
vide enough information to specify the outcome of an impact by themselves. The Hugoniot
equations must be supplemented by a fourth equation, the equation of state, that relates
the pressure to the density and internal energy in each material, P = P(ρ, E). Alternatively,
a relation between shock velocity and particle velocity may be specified, U= U(up). Since
this relation is frequently linear, it often provides the most convenient equation of state in
impact processes. Thus,

U = c + Sup (6.4)
where c and S are empirical constants (c is the bulk sound speed and S is a dimension-
less slope). Table 6.1 lists the measured values of c and S for a variety of materials. These

EBSCOhost - printed on 11/3/2023 9:56 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
6.3 Cratering mechanics 231

Table 6.1. Linear shock-particle velocity equation of state parameters

Material ρ0 (kg/m3) c (km/s) S

Aluminum 2750 5.30 1.37

Basalt 2860 2.6 1.62
Calcite (carbonate) 2670 3.80 1.42
Coconino sandstone 2000 1.5 1.43
Diabase 3000 4.48 1.19
Dry sand 1600 1.7 1.31
Granite 2630 3.68 1.24
Iron 7680 3.80 1.58
Permafrost (water saturated) 1960 2.51 1.29
Serpentinite 2800 2.73 1.76
Water (25°C) 998 2.393 1.333
Water Ice (–15°C) 915 1.317 1.526

Data from Melosh (1989).

equations can be used to compute the maximum pressure, particle velocity, shock velocity,
etc. in an impact.
Planar impact approximation. A rough estimate of the parameters describing the highest
pressure portion of the contact and compression stage is obtained from the planar impact
approximation (sometimes called the impedance matching solution), which is valid so long
as the lateral dimensions of the projectile are small compared with the distance the shock
has propagated. This approximation is, thus, valid through most of the contact and com-
pression stage. A simultaneous solution to the Hugoniot jump equations is obtained in both
the target and projectile by noting that, at the interface between the two, both the particle
velocity and pressure must be the same in both bodies. Unfortunately, there is no simple
formula for this approximation. The simplest expression is for the particle velocity in the
target, ut (the particle velocity in the projectile is vi-ut by the velocity matching condition),
which is the solution of a simple quadratic equation:

− B + B2 − 4 AC
ut = (6.5)
2A
where A, B, and C are defined as:

A = ρ 0 t St − ρ 0 p S p
B = ρ0 t ct + ρ0 p c p + 2 ρ0 p S p vi .
C = − ρ0 p vi (c p + S p vi ) (6.6)

The subscripts p and t refer to the projectile and target respectively. The above equation
can be used in conjunction with the Hugoniot equations and equation of state to obtain

EBSCOhost - printed on 11/3/2023 9:56 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
232 Impact cratering

(a) (b)
0 50 km

t = 0.33 T t = 0.99 T

60 33
180 100
540 540 170
420 420 33
300 300 170
170 100 100
180 170
60 230
230 300
0
170 10
0 50 km
33
(c) (d)
0 50 km 0 50 km
t = 3.03 T t = 6.11 T

7 7
20 20

20
33 8 10
33
50 10 17 7
60 2

Figure 6.4 The first three frames (a to c) illustrate the evolution of shock waves in the contact
and compression stages of the vertical impact of a 46.4 km diameter iron projectile on a gabbroic
anorthosite target at 15 km/s. The last frame (d) is a very early phase of the excavation stage. Pressure
contours are labeled in GPa, and the times given are in multiples of the time that the projectile takes
to pass through its own diameter, about 3 s in this case. Note the change in length scale from frame
to frame.

any other quantities of interest. Thus, the pressure behind the shock is given by P = ρot ut
(ct+St ut). The pressures in both the target and projectile are the same by construction of
the solution.
As the projectile plunges into the target, shock waves propagate both into the projectile,
compressing and slowing it, and into the target, compressing and accelerating it downward
and outward (Figure 6.4). At the interface between target and projectile the material of each
body moves at the same velocity. This equals 1/2 the impact velocity if they are composed
of the same materials (note that in the above equation, A = 0 in this case, but the numer-
ator also vanishes and the right-hand side of the equation approaches –C/B, which equals
vi / 2). The shock wave in the projectile eventually reaches its back (or top) surface. When
this happens, the pressure is released as the surface of the compressed projectile expands

EBSCOhost - printed on 11/3/2023 9:56 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
6.3 Cratering mechanics 233

upward and a wave of pressure relief propagates back downward toward the projectile–
target interface. The contact and compression stage is considered to end when this relief
wave reaches the projectile–target interface. At this time the projectile has been compressed
to high pressure, often reaching hundreds of gigapascals, and upon decompression it may
be in the liquid or gaseous state due to heat deposited in it during the irreversible com-
pression process. The projectile generally carries off 50% or less of the total initial energy
if the density and compressibility of the projectile and target material do not differ too
much, while the balance of the energy moves into the target. This energy will eventually be
expended in opening the crater as well as heating the target. The projectile–target interface
at the end of contact and compression is generally less than one projectile diameter L below
the original surface.
Contact and compression are accompanied by the formation of very high-velocity “jets”
of highly shocked material. These jets form where strongly compressed material is close to
a free surface, for example near the circle where a spherical projectile contacts a planar tar-
get. The jet velocity depends on the angle between the converging surface of the projectile
and target, but may exceed the impact velocity by factors as great as 5. Jetting was initially
regarded as a spectacular but not quantitatively important phenomenon in early impact
experiments, where the incandescent streaks of jetted material only amounted to about
10% of the projectile’s mass in vertical impacts. However, recent work on oblique impacts
indicates that in this case jetting is much more important and that the entire projectile may
participate in a downrange stream of debris that carries much of the original energy and
momentum. Oblique impacts are still not well understood and more work needs to be done
to clarify the role of jetting early in this process.

6.3.2 Excavation
During the excavation stage the shock wave created during contact and compression
expands and eventually weakens into an elastic wave, while the crater itself is opened by
the much slower “excavation flow.” The duration of this stage is roughly given by the period
of a gravity wave with wavelength equal to the crater diameter D, equal to tEX ~ (D/g)1/2, for
craters whose excavation is dominated by gravity g (this includes craters larger than a few
kilometers in diameter, even when excavated in hard rock). Thus, Meteor Crater, Arizona,
was excavated in about 10 s, while the 1000 km diameter Imbrium Basin on the Moon took
about 13 minutes to open. Shock wave expansion and crater excavation, while intimately
linked, occur at very different rates and may be usefully considered separately.
The high pressure attained during contact and compression is almost uniform over a
volume roughly comparable to the initial dimensions of the projectile, a region called the
“isobaric core.” However, as the shock wave expands away from the impact site the shock
pressure declines as the initial impact energy spreads over an increasingly large volume of
rock and loses energy to heating the target. The pressure P in the shock wave as a function
of distance r from the impact site is given roughly by:

EBSCOhost - printed on 11/3/2023 9:56 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
234 Impact cratering

Figure 6.5 Shatter cones from the Spider Structure, Western Australia, formed in mid-Proterozoic
orthoquartzite. This cone-in-cone fracture is characteristic of shattering by impact-generated shock
waves. The scale bar is 15 cm long (courtesy of George Williams).

n
 a
P = P0   (6.7)
r

where a ( = L/2) is the radius of the projectile, P0 is the pressure established during con-
tact and compression, and the power n is between 2 and 4, depending on the strength
of the shock wave (n is larger at higher pressures – a value n = 3 is a good general
average).
Shock metamorphism. The shock wave, with a release wave immediately following,
quickly attains the shape of a hemisphere expanding through the target rocks. The high-
shock pressures are confined to the surface of the hemisphere: the interior has already
decompressed. The shock wave moves very quickly, as fast or faster than the speed of
sound, between about 6 and 10 km/s in most rocks. As rocks in the target are overrun by
the shock waves, then released to low pressures, mineralogical changes take place in the
component minerals. At the highest pressures the rocks may melt or even vaporize upon
release. As the shock wave weakens high-pressure minerals such as coesite or stishovite
arise from quartz in the target rocks, diamonds may be produced from graphite, or maske-
lynite from plagioclase. Somewhat lower pressures cause pervasive fracturing and “planar
elements” in individual crystals. Still lower pressures create a characteristic cone-in-cone
fracture called “shatter cones” (Figure 6.5),which are readily recognized in the vicinity of
impact structures. Indeed, many terrestrial impact structures were first recognized from
the occurrence of shatter cones. Table 6.2 lists a number of well-established shock meta-
morphic changes and the pressures at which they occur.

EBSCOhost - printed on 11/3/2023 9:56 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
6.3 Cratering mechanics 235

Table 6.2. Petrographic shock indicators

Material Indicator Pressure (GPa)

Tonalite (igneous rock) Shatter cones 2–6

Quartz Planar elements and fractures 5–35
Stishovite 15–40
Coesite 30–50
Melting 50–65(?)
Plagioclase Planar elements 13–30
Maskelynite 30–45
Melting 45–65(?)
Olivine Planar elements and fractures 5–45
Ringwoodite 45
Recrystallization 45(?)–65(?)
Melting >70
Clinopyroxene Mechanical twinning 5–40(?)
Majorite 13.5
Planar elements 30(?)–45
Melting 45(?)–65(?)
Graphite Cubic diamond 13
Hexagonal diamond 70–140

Data from Melosh (1989).

Spallation. The expanding shock wave encounters a special condition near the free sur-
face. The pressure at the surface must be zero at all times. Nevertheless, a short distance
below the surface the pressure is essentially equal to P, defined above. This situation results
in a thin layer of surface rocks being thrown upward at very high velocity (the theoret-
ical maximum velocity approaches the impact speed vi). Since the surface rocks are not
compressed to high pressure, this results in the ejection of a small quantity of unshocked
or lightly shocked rocks at speeds that may exceed the target planet’s escape velocity.
Although the total quantity of material ejected by this “spall” mechanism is probably only
1–3% of the total mass excavated from the crater, it is particularly important scientifically
as this is probably the origin of the recently discovered meteorites from the Moon, and of
the SNC (shergottite, nakhlite, and chassignite) meteorites, which are widely believed to
have been ejected from Mars.
Seismic shaking. The weakening shock wave eventually degrades into elastic waves.
These elastic waves are similar in many respects to the seismic waves produced by an
earthquake, although impact-generated waves contain less of the destructive shear-wave
energy than earthquake waves. The seismic waves produced by a large impact may have
significant effects on the target planet, creating jumbled terrains at the antipode of the
impact site if they are focused by internal planetary structures, such as a low-velocity core.

EBSCOhost - printed on 11/3/2023 9:56 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
236 Impact cratering

Meteoroid
impact

Interference A
zone
Excavation
flow
Shock

A′
Particle velocity

Pressure
I/r2 I/r2

A A′ A A′
Excavation Shock Shock
flow

Figure 6.6 Illustration of the expanding shock wave and excavation flow following a meteorite impact.
The contours in the upper part of the figure represent the pressure at some particular time after the
impact. The region of high-shock pressure is isolated or “detached” on an expanding hemispherical
shell. The lower graphs show profiles of particle velocity and pressure along the section AA′. The
dashed lines on these graphs show the particle velocity and pressure some time later than those shown
by the solid lines, and the solid curves connecting the peaks are portions of the “envelopes” of peak
particle velocity and peak pressure.

This effect has been observed opposite Caloris Basinon Mercury and opposite Imbrium
and Orientale on the Moon. The equivalent Richter magnitude M caused by an impact of
energy E (= 1/2 mpvi2) is given approximately by:
M = 0.67 log10 E – 5.87. (6.8)
Excavation mechanics. Target material engulfed by the shock wave is released a short
time later. Upon release the material has a velocity that is only about 1/5 of the particle vel-
ocity in the shock wave. This “residual velocity” is due to thermodynamic irreversibility in
the shock compression. It is this velocity field that eventually excavates the crater (Figure
6.6). The excavation velocity field has a characteristic downward-outward-then upward
pattern that moves target material out of the crater, ejecting it at angles close to 45° at the
rim. The streamlines of this flow cut across the contours of maximum shock pressure, so
that material ejected at any time may contain material with a wide range of shock levels
(Figure 6.7). Nevertheless, the early, fast ejecta generally contain a higher proportion of
highly shocked material than the later, slower ejecta. Throughout its growth the crater is
lined with highly shocked, often melted, target material.

EBSCOhost - printed on 11/3/2023 9:56 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
6.3 Cratering mechanics 237

CL Ejecta curtain

Slower
Vapor Fast ejecta ejecta

Melt

Maximum pressure contours

Figure 6.7 Geometry of the excavation flow field that develops behind the rapidly expanding shock
front, which has moved beyond the boundaries of this illustration. The lines with arrows indicate
streamtubes along which material flows downward and outward from the crater. The streamtubes cut
across the contours of maximum shock pressure, showing that material ejected at any given range
from the impact is shocked to a variety of different maximum pressures. When material flowing
through a streamtube crosses the initial surface it forms part of the ejecta curtain. Ejecta emerging
near the impact site travel at high speed, whereas ejecta emerging at larger distances travel at slower
velocities.

Inside the growing crater, vaporized projectile and target may expand rapidly out of the
crater, forming a vapor plume that, if massive enough, may blow aside any surrounding
atmosphere and accelerate to high speed. In the impacts of sufficiently large and fast pro-
jectiles some of this vapor plume material may even reach escape velocity and leave the
planet, incidentally also removing some of the planet’s atmosphere. Such “impact erosion”
may have played a role in the early history of the Martian atmosphere. Even in smaller
impacts the vapor plume may temporarily blow aside the atmosphere, opening the way for
widespread ballistic dispersal of melt droplets (tektites) above the atmosphere and perhaps
permitting the formation of lunar-like ejection blankets even on planets with dense atmos-
pheres, as has been observed on the Soviet Venera 15/16 images of Venus.
Crater growth rate. The growing crater is at first hemispherical in shape. Its depth
H(t) and diameter D(t) both grow approximately as t0.4, where t is time after the impact.
Hemispherical growth ceases after a time of about (2Ht /g)1/2, where Ht is the final depth
of the transient crater. At this time the crater depth stops increasing (it may even begin
to decrease as collapse begins), but its diameter continues to increase. The crater shape,
thus, becomes a shallow bowl, finally attaining a diameter roughly three to four times its
depth. At this stage, before collapse modifies it, the crater is known as a “transient” cra-
ter. Even simple craters experience some collapse (which produces the breccia lens), so

EBSCOhost - printed on 11/3/2023 9:56 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
238 Impact cratering

Impact
site
Spalled
Vaporized
Melt
Ejected Ejected Hexc
Hat
Displaced
Transient
crater

Figure 6.8 Provenance of material expelled from an impact crater. Vaporized material expands
outward in a vapor plume. Of the remaining material, some is ejected and some is displaced out of
the crater and deforms the adjacent rocks, uplifting the surface near the rim and downwarping rocks
beneath the crater floor. The ejected material is excavated from a maximum depth Hexc that is only
about one-third of the transient crater depth or one-tenth of the transient crater diameter. The dashed
lines show the profile of the transient crater.

that the transient crater is always a brief intermediate stage in geological crater formation.
However, since most laboratory craters are “frozen” transient craters, much of our know-
ledge about crater dimensions refers to the transient stage only, and must be modified for
application to geological craters.
Maximum depth of excavation. Laboratory, field, and computer studies of impact craters
have all confirmed that only material lying above about 1/3 of the transient crater depth (or
about 1/10 of the diameter) is thrown out of the crater. Material deeper than this is simply
pushed downward into the target, where its volume is accommodated by deformation of
the surrounding rocks (Figure 6. 8). Thus, in sharp contrast to ejecta from volcanic craters,
material in the ejecta blankets of impact craters does not sample the full depth of rock inter-
sected by the crater, a surprising fact that has led many geologists astray in their estimation
of the nature of the ejected debris.
The form of the transient crater produced during the excavation stage may be affected by
such factors as obliquity of the impact (although the impact angle must be less than about
6° for a noticeably elliptical crater to form at impact velocities in excess of about 4 km/s),
the presence of a water table or layers of different strength, rock structure, joints, or initial
topography in the target. Each of these factors produces its own characteristic changes in
the simple bowl-shaped transient crater form.

6.3.3 Modification
Shortly after the excavation flow has opened the transient crater and the ejecta has been
launched onto ballistic trajectories, a major change takes place in the motion of debris
within and beneath the crater. Instead of flowing upward and away from the crater center,
the debris comes to a momentary halt, then begins to move downward and back toward
the center whence it came. This collapse is generally attributed to gravity, although elastic
rebound of the underlying, compressed rock layers may also play a role. The effects of

EBSCOhost - printed on 11/3/2023 9:56 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
6.3 Cratering mechanics 239

collapse range from mere debris sliding and drainback in small craters to wholesale alter-
ation of the form of larger craters in which the floors rise, central peaks appear, and the rims
sink down into wide zones of stepped terraces. Great mountain rings or wide central pits
may appear in still larger craters (Figure 6.9).
These different forms of crater collapse begin almost immediately after formation of
the transient crater. The timescale of collapse is similar to that of excavation, occupying
an interval of a few times (D/g)1/2. Crater collapse and modification, thus, take place on
timescales very much shorter than most geologic processes. The crater resulting from this
collapse is then subject to the normal geologic processes of gradation, isostatic adjustment,
infilling by lavas, etc. on geologic timescales. Such processes may eventually result in the
obscuration or even total obliteration of the crater.
The effects of collapse depend on the size of the crater. For transient craters smaller than
about 15 km diameter on the Moon, or about 3 km on the Earth, modification entails only
collapse of the relatively steep rim of the crater onto its floor. The resulting “simple crater”
(see Figure 6.1a) is a shallow bowl-shaped depression with a rim-to-rim diameter D about
five times its depth below the rim H. In fresh craters the inner rim stands near the angle
of repose, about 30°. Drilling in terrestrial craters (Figure 6.10) shows that the crater floor
is underlain by a lens of broken rock (mixed breccia) derived from all of the rock units
intersected by the crater. The thickness of this breccia lens is typically 1/2 the depth of the
crater H. Volume conservation suggests that this collapse increases the original diameter
of the crater by about 15%. The breccia lens often includes layers and lenses of highly
shocked material mixed with much less-shocked country rock. A small volume of shocked
or melted rock is often found at the bottom of the breccia lens.
Complex craters (Figure 6.1b,c) collapse more spectacularly. Walls slump, the floor is
stratigraphically uplifted, central peaks or peak rings rise in the center, and the floor is
overlain by a thick layer of highly shocked impact melt (Figure 6.11). The detailed mech-
anism of collapse is still not fully understood because straightforward use of standard rock
mechanics models does not predict the type of collapse observed (see Box 8.1). The current
best description of complex crater collapse utilizes a phenomenological strength model in
which the material around the crater is approximated as a Bingham fluid, a material that
responds elastically up to differential stresses of about 3 MPa, independent of overburden
pressure, and then flows as a viscous fluid with viscosity of the order of 1 GPa-s at larger
stresses. In a large collapsing crater the walls slump along discrete faults, forming terraces
whose widths are controlled by the Bingham strength, and the floor rises, controlled by the
viscosity, until the differential stresses fall below the 3 MPa strength limit. A central peak
may rise, and then collapse again in large craters, forming the observed internal ring (or
rings). Figure 6.9 illustrates this process schematically. The rock in the vicinity of a large
impact may display such an unusual flow law because of the locally strong shaking driven
by the large amount of seismic energy deposited by the impact.
The mechanics of the collapse that produces multiring basins (Figure 6.1d) is even
less well understood. Figure 6.12 illustrates the structure of the Orientale basin on the
Moon with a highly vertically exaggerated cross section derived from both geological and

EBSCOhost - printed on 11/3/2023 9:56 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
Excavation and
Excavation and
beginning of uplift
beginning of uplift

Central uplift
Central uplift and overshoots stability
rim collapse

Final crater Peak ring crater

EBSCOhost - printed on 11/3/2023 9:56 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
(a) (b)

Figure 6.9 Illustration of the formation of complex craters of either (a) central peak morphology or (b) peak ring morphology. Uplift of the crater floor
begins even before the rim is fully formed. As the floor rises further, rim collapse creates a wreath of terraces surrounding the crater. In smaller craters
the central uplift “freezes” to form a central peak. In larger craters the central peak collapses and creates a peak ring before motion ceases.
6.3 Cratering mechanics 241

Original plane

Sediments

Present
surface

Brecciated & Breccias

fractured gneiss
Highly mixed, partly melt
Shocked gneiss Melt Mixed with glassy melt fragments

0 0.5 km Slightly mixed, weakly shocked

Slightly mixed, generally unshocked

Basal breccia

Figure 6.10 Geologic cross section of the 3.4 km diameter Brent Crater in Ontario, Canada. Although
the rim has been eroded away, Brent is a typical, simple crater that forms in crystalline rocks. A small
melt pool occurs at the bottom of the breccia lens and more highly shocked rocks occur near its top.

Terrace zone Terrace zone

Incipient
central
Crater rim Inner ring Inner ring Crater rim
uplift
Suevite

km
3
2
1
0
0 1 2 3 km

Figure 6.11 Geologic cross section of the 22 km diameter Ries Crater in Germany. Drilling and
geophysical data suggest that this is a peak ring crater. Its central basin is filled with suevite, a
mixture of highly shocked and melted rocks and cold clasts.

geophysical data. Note the ring scarps are interpreted as inward-dipping faults above a pro-
nounced mantle uplift beneath the basin’s center. One idea that is currently gaining ground
is that the ring scarps are normal faults that develop as the crust surrounding a large crater
is pulled inward by the flow of underlying viscous mantle material toward the crater cavity
(Figure 6.13). An important aspect of this flow is that it must be confined in a low-viscosity
channel by more viscous material below, otherwise the flow simply uplifts the crater floor,
and radial faults instead of ring scarps are the result. Special structural conditions are, thus,
needed in the planet for multiring basins to form on its surface, so that a g–1 dependence for
the transition from complex craters to multiring basins is not expected (or observed). This

EBSCOhost - printed on 11/3/2023 9:56 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
242 Impact cratering

Box 6.1 Maxwell’s Z model of crater excavation

The details of the excavation flow can be determined only by experiment or by elaborate
numerical computations. Even such numerical work may have difficulty in correctly computing
the final dimensions of the transient crater. However, in 1973 D. Maxwell and K. Seifert
proposed a simple analytical model of excavation flow (Maxwell, 1977). This model gives
a useful kinematic, although not dynamic, description of the cratering flow field. Like all
approximate models, it should not be used to determine fine details.
Maxwell and Seifert noted that in explosion cratering computations the radial component
of the excavation flow velocity ur usually falls as a simple inverse power of distance r from the
explosive charge.

α (t )
ur = (B6.1.1)
rZ
where α(t) is a function of time and describes the strength of the flow, while Z is a
dimensionless power.
The incompressibility of the excavation flow, ∇ • u = 0, requires that the angular component
of the flow velocity uθ in polar coordinates (r, θ), is:

sin θ
uθ = ( Z − 2) ur . (B6.1.2)
1 + cosθ

The geometry of the velocity field defined by this model, seen in Figure 6B.1, is remarkably
similar to that computed in both explosion and impact cratering events Z  3. The equation of
streamlines in polar coordinates is:
1
r = r0 (1 − cosθ ) Z − 2 (B6.1.3)

where r0 is a constant that is different for each streamline. It is equal to the radius at which the
streamline emerges from the surface (θ = 90°). Taking r0 = Dat/2, the radius of the transient
crater, the maximum depth of excavation Hexc is:
1− Z
Dat
H exc = ( Z − 2)( Z − 1) Z −2 . (B6.1.4)
2
For Z = 3 the maximum depth of excavation Hexc = Dat/8, or about one-third of the final
transient crater depth.
The total mass ejected from a crater described by the Z model, Mej, is a fraction of the total
mass displaced from the transient crater Me:

Z −2
M ej = M. (B6.1.5)
Z −1 e
The Z model also predicts that the vertical and horizontal velocity components uV and uH of
the ejecta launched at a distance s along the surface from the impact point are:

uV = α / s Z
uH = ( Z − 2) uV (B6.1.6)

EBSCOhost - printed on 11/3/2023 9:56 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
6.3 Cratering mechanics 243

Box 6.1 (cont.)

or that the angle of ejection is φ = tan−1 (Z − 2), equal to 45° for Z = 3.
The Z-model presented thus far is a kinematical model useful for describing the form of
the excavation flow. Maxwell and Seifert attempted to give it more dynamical content by
computing the function α(t) in Equation (B6.1.1). This function gives the strength of the flow
at any particular time. Its value is different for each streamtube in the flow. It is estimated by
using energy conservation in each of these streamtubes, neglecting interactions with adjacent
tubes. Thus, the sum of the kinetic, gravitational, and distortional energies is found in each
streamtube at some initial time. The total energy in each streamtube is conserved as the flow
progresses. However, the kinetic energy declines at the expense of the gravitational and
distortional energies, so that the net flow velocity declines. Unfortunately, this aspect of the
Z model has not worked out well in practice: The actual course of the excavation flow is best
determined through detailed dynamical models.

θ
r

45° 63.4°

z=2 z=3 z=4

Figure B6.1

The Z-model described here can be (and has been) improved and extended in several ways.
One of the most straightforward is to move the source of the flow, r = 0, from the surface to
some depth below the surface, taking into account the depth of the effective center of the shock
wave (Croft, 1980). Other workers have attempted to refine Maxwell and Siefert’s methods of
estimating energies in the streamtubes. The Z-model, however, is fundamentally limited by its
neglect of interactions between the streamtubes. For this reason, it can never become an exact
description of the cratering flow, however accurately the dynamics within a single streamtube
is represented.
In spite of all its faults, the Z-model gives a reasonably accurate representation of the gross
geometric features of the cratering flow and can even be used to predict some first-order
dynamical properties. It has the unfortunate feature of not being a truly dynamical model,
so that further refinements are not necessarily closer approximations to the full dynamical
equations of motion. Nevertheless, the excellent properties of this model are probably still far
from being fully exploited.

EBSCOhost - printed on 11/3/2023 9:56 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
244 Impact cratering

1 - Inner basin ring

2 - Inner rock
3 - Rook
4 - Cordillera
5 - Questionable ring

0 200 400 600 800 km

S 4 5? N
1 2 3
10

0
–10
–20 Crust
Depth, km

density = 2900 kg/m3

–30

–40
–50

–60 Mantle

–70 Density = 3400 kg/m3

–80

Prebasin rocks Hevelius Rock Crackled Mare

(highly brecciated formation material floor material basalt
toward basin (ejecta) (slumps?) (impact melt?)
center)

Figure 6.12 Geologic and geophysical structure of the Orientale basin on the Moon, one of the
freshest and best-studied multiring basins. A dense mantle uplift underlies the center of the basin.
The crustal thinning above the uplift is due to the ejection of about 40 km of crustal material from the
crater that formed the basin. The great ring scarps shown in cross-section formed during collapse of
the crater. Note the 10X vertical exaggeration necessary to show the ring scarps.

theory is capable of explaining both the lunar-type and Valhalla-type multiring basins as
expressions of different lithosphere thicknesses.

6.4 Ejecta deposits

A deposit of debris ejected from the crater interior surrounds essentially all impact craters.
The only exceptions are craters on steeply sloping surfaces or on satellites with too little grav-
ity to retain the ejecta or too much porosity to produce it. This ejecta deposit is thickest at the
crater rim and thins with increasing distance outwards. Where this deposit is recognizably
continuous near the crater it is called an “ejecta blanket.” Ejecta beyond the edge of the con-
tinuous deposit are thin and patchy. Secondary craters occur in this zone and beyond it. Figure
6.14 shows the ejecta blanket of the 30-km diameter lunar crater Timocharis. The pre-existing

EBSCOhost - printed on 11/3/2023 9:56 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
6.4 Ejecta deposits 245

(a)

Elastic-plastic
half-space

(b)
Elastic

Fluid

Figure 6.13 The ring tectonic theory of multiring basin formation: (a) shows the formation of a normal
complex crater on a planet with uniform rheology; (b) shows the inward-directed flow in a more fluid
asthenosphere underlying a lithosphere of thickness comparable to the crater depth and the resulting
scarps; (c) shows a Valhalla-type basin developing around a crater formed in a very thin lithosphere.

terrain within about one crater radius of the rim is buried and mostly obliterated by the con-
tinuous ejecta blanket. Light patches show where thinner deposits overlie the mare surface.
The thickness of the ejecta deposit varies greatly with direction away from the rim.
Azimuthal thickness variations, at the same radius, can be as large as ten to one. The ejecta
is concentrated into rays that often are observed to form at angles of about 30° to one
another. However, it is also true that, at a given azimuth, the thickness falls rapidly and sys-
tematically with distance away from the crater center. Compilation of many data sets from
both impact and explosion craters shows that the thickness δ(r) as a function of radius r
away from the crater center is given by an approximate inverse cube relation:
−3± 0.5
r
δ (r ) = f ( R)   (6.9)
 R

where R is the radius of the crater rim. Integration of this relation indicates that most
of the mass of the ejecta is located near the crater rim. According to the approximate
“Schröter’s rule,” the volume of the ejecta is approximately equal to the volume of the crater
bowl. Although this rule seems to make a great deal of sense (however, it does ignore the
increase in volume of the ejected material due to fragmentation), it is unverifiable in practice
because the original ground surface can seldom be located with adequate precision.

EBSCOhost - printed on 11/3/2023 9:56 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
246 Impact cratering

Figure 6.14 Near-vertical view of the 30 km diameter lunar crater Timocharis. The ejecta deposits
are dunelike near the crater rim but grade into a subradial facies beyond about 2 R from the crater’s
center. Secondary craters occur at greater distances. A pattern of bright material surrounding the
crater indicates the presence of ejecta too thin to greatly modify the pre-existing terrain at this scale.
Apollo 15 photo AS15–1005.

The rims of many fresh craters are littered with blocks of rock ejected from beneath
the crater. Compilations of data show that block size generally decreases as a function
of distance from the rim. The maximum size of block observed on the rim of a crater
is related to the size of the crater itself and an empirical relation that holds over a wide
range of crater sizes relates the mass of the largest ejected fragment, mf, to the total mass
ejected from the crater, Me. If both masses are expressed in kilograms, this relation is:

m f = 0.8 M e0.8 . (6.10)

6.4.1 Ballistic sedimentation

The ejecta deposited around an impact crater on an airless planet are emplaced ballistically,
that is, ejecta are thrown from the crater with some initial velocity, follow a nearly para-
bolic trajectory above the planet’s surface, then fall back to the surface, striking with the

EBSCOhost - printed on 11/3/2023 9:56 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
6.4 Ejecta deposits 247

T
7
1.5T
6

Altitude, units of crater radius

2T
5

4 2.5T

0
1 3 4 5 6 7 8 9 10
CE
Range, units of crater radius

Figure 6.15 The debris ejected from an impact crater follows ballistic trajectories from its launch
position within the final crater (the rim of the final crater is located at a range equal to one crater
radius). The innermost ejecta are launched first and travel fastest, following the steepest trajectories
shown in the figure. Ejecta originating farther from the center are launched later and move more
slowly, falling nearer the crater rim. Because of the relationship between position, time, and velocity
of ejection, the debris forms an inverted cone that sweeps outward across the target. This debris
curtain is shown at four separate times during its flight, at 1, 1.5, 2, and 2.5 Tf, where Tf is the crater
formation time, (D/g)1/2. Coarser, less-shocked debris travels near the base of the curtain, whereas
the fast, highly shocked ejecta fraction tends to travel near the top. The three lower figures show
details of the pre-existing ground surface when the ejecta curtain arrives. As the range increases the
ejecta strike with progressively larger velocities, incorporating larger amounts of surface material and
imparting a larger net horizontal velocity.

same velocity as on ejection. Some interaction may occur between ejecta fragments in the
denser parts of the ejecta curtain, but the general motion is dominated by ballistics alone.
The ground surface around the crater is profoundly affected by the ejecta as it lands, and its
interaction with this falling debris determines the character of the ejecta deposits of large
craters.
The debris ejected from an impact crater travels together in the form of an “ejecta
curtain.” Although each fragment follows a parabolic trajectory, the times and velocities
of ejection from the crater are organized so that most of the debris lies on the surface of
an expanding inverted cone. Figure 6.15 illustrates the parabolic trajectories of a number

EBSCOhost - printed on 11/3/2023 9:56 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
248 Impact cratering

of fragments from an impact crater. The positions of these fragments at several times are
indicated. At any one time the debris lies on the surface of a cone that makes an angle
of about 45° with the ground surface. This cone sweeps rapidly outward from the crater
rim. Debris from the curtain strikes the ground at its base, impacting first near the crater
rim, then at greater distances as time progresses. The size of the ejecta fragments near the
base of the ejecta curtain is expected to be larger than the fragments higher in the curtain,
and the proportion of highly shocked fragments and glass increases with height in the
curtain.
Ballistically emplaced debris falling near the crater rim strikes with a low velocity
because it travels only a short distance. At the rim itself this velocity is so low that rock
units may retain some coherence and produce an overturned flap with inverted stratigraphy.
Eugene Shoemaker first recognized such overturned beds at Meteor Crater, Arizona. At
greater distances from the crater rim, the debris strikes with a higher velocity. When this
velocity is large enough, surface material is eroded and mixes with the debris. The falling
ejecta also possess a radially outward velocity component. Although the vertical velocity
is cancelled when the debris strikes the surface, the mixture of debris and surface material
retains its outward momentum (see the lower inserts in Figure 6.15). This mixture moves
rapidly outward as a ground-hugging flow of rock debris, similar in many ways to the
flow of a large dry-rock avalanche. Depositional features such as dunes, ridges, and radial
troughs indicative of high-speed flow may result from this motion. The deposit itself con-
sists of an intimate mixture of primary crater ejecta and of secondary material scoured
from the pre-existing ground surface.

6.4.2 Fluidized ejecta blankets

The ejecta blankets of impact craters on Mars are dramatically different from those on the
Moon or Mercury. Martian craters smaller than about 5 km in diameter closely resemble
their counterparts on the Moon. However, craters between 5 and 15 km in diameter have
a single ejecta sheet that extends about one crater radius from the rim and ends in a low
concentric ridge or outward-facing escarpment. These are called “rampart craters,” from
the continuous ridge surrounding the ejecta deposit. The ejecta of most craters larger than
15 km in diameter are divided into petal-like lobes that extend two or more radii from the
rim (Figure 6.16), approximately twice as far as the continuous ejecta deposits of lunar or
Mercurian craters. A few large craters with lunar-type ejecta blankets are known, but they
are rare. Many Martian craters have abnormally large central peaks and other internal col-
lapse structures compared with lunar or Mercurian craters, also suggesting the presence
of some fluidizing agent peculiar to Mars.
The fluidized ejecta blankets of Martian craters appear to have been emplaced as
thin, ground-hugging flows. When impeded by topographic obstacles that could not
be overridden, the flows are deflected and either spread out elsewhere or pond against
the obstacle. Ejecta lobes fail to overtop low hills and mesas that are only a few times

EBSCOhost - printed on 11/3/2023 9:56 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
6.4 Ejecta deposits 249

Figure 6.16 The 19 km diameter crater Yuty on Mars is surrounded by thin, petaloid flow lobes
that extend approximately twice as far from the crater as the continuous ejecta deposits of lunar or
Mercurian craters. Viking Orbiter frame 3A07.

higher than the flow thickness itself, suggesting that the lobe could not have traveled as
dispersed clouds of the base-surge type, nor were they emplaced by ballistic sedimen-
tation, because the ejecta curtain should have fallen on the topographic obstacle from
above.
The peculiar form of Martian ejecta blankets is generally attributed to the presence of
liquid water in the substrate. Ejected along with subsurface material, liquid water mixed
into the ejecta would greatly enhance the mobility of the debris, converting the dry, frag-
mental ejecta flows characteristic of lunar craters to fluid debris flows similar to terres-
trial mudflows. Nevertheless, not all Mars researchers agree with this interpretation and
an alternative viewpoint attributes at least part of the mobility of these flows to interaction
with the thin atmosphere of Mars.

EBSCOhost - printed on 11/3/2023 9:56 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
250 Impact cratering

Figure 6.17 Night-time IR thermal image of the 6.9 km diameter Martian crater Gratteri, located
at 17.8° S, 202° E. The dark streaks are created by secondary impact craters that extend up to 500
km from the crater center. In images of this type, dark regions are cold and emit little IR radiation
because they have low thermal inertia, indicating that the streaks are composed of fine-grained
material compared to their surroundings. The overall image measures 545 x 533 km across. THEMIS
image courtesy of Phil Christensen. NASA/JPL/ASU. See also color plate section.

6.4.3 Secondary craters

Numerous secondary impact craters, variously occurring either singly or in loops, clusters
and lines, surround large impact craters. Figure 6.17 shows the secondary crater field around
the crater Gratteri on Mars, as revealed by the THEMIS thermal mapper. Recognizable sec-
ondary craters extend from just beyond the continuous ejecta blanket out to distances of up
to thousands of kilometers from their source crater. Close to the primary crater, secondary
craters are produced by relatively low-velocity impacts and are, thus, irregular in shape,
shallow, and obviously clustered, and are often separated by V-shaped dunes known as the
“herringbone pattern.” Farther from the primary impact, velocities are larger and secondary
craters are more dispersed, which makes them difficult to discriminate from small primary
craters.
An important controversy is presently raging about the importance of secondary craters
in masking the primary flux. If a majority of the small craters (less than a few 100 m diam-
eter) on a planet’s surface are secondary, then ages assigned to cratered surfaces based on the

EBSCOhost - printed on 11/3/2023 9:56 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
6.5 Scaling of crater dimensions 251

assumption that the craters are primary will be too great. On the other hand, many experienced
crater counters claim that they can exclude secondary craters because they are clustered, a
claim that is disputed by other experts. At the moment there is no consensus on this problem.
The maximum size of secondary craters is approximately 4% of the primary diameter
on the Moon, Mars, and elsewhere. However, on Mercury obvious secondary craters are
apparently several times larger. This observation is at odds with the larger impact velocity
on Mercury, which is expected to result in smaller ejected fragments, not larger ones. Is
the crust of Mercury somehow stronger than that of the Moon or Mars? At the moment, the
solution to this conundrum is unknown.

6.4.4 Oblique impact

Although high-velocity impact craters are circular down to very low angles of approach,
the pattern of the ejecta may betray impact obliquity at angles as large as 45°. The first sign
of an oblique impact is an asymmetric, but still bilaterally symmetrical, ejecta blanket. The
ejecta in the uprange direction are thinner and less extensive than those in the downrange
direction at impact angles near 45°. At impact angles near 30° an uprange wedge free
of ejecta develops, an example of which is shown in Figure 6.3. As the angle decreases
still farther, to 10°, ejecta-free regions appear in both downrange and uprange directions,
although bright streaks may extend downrange in very fresh craters. At such low angles
the crater itself becomes elliptical, with its long axis parallel to the flight direction of the
projectile. At these highly oblique angles the projectile essentially plows a furrow into the
target surface, throwing ejecta out to both sides to form a “butterfly-wing” pattern.

6.5 Scaling of crater dimensions

One of the most frequently asked questions about an impact crater is, “How big was the
meteorite that made the crater?” Like many simple questions this has no simple answer. It
should be obvious that the crater size depends upon the meteorite’s speed, size, and angle
of entry. It also depends on such factors as the meteorite’s composition, the material and
composition of the target, surface gravity, presence or absence of an atmosphere, etc. The
question of the original size of the meteorite is usually unanswerable, because the speed
and angle of impact are seldom known. The inverse question, of how large a crater will be
produced by a given-sized meteorite with known speed and incidence angle is in principle
much simpler to answer. However, even this prediction is uncertain because there is no
observational or experimental data on the formative conditions of impact craters larger than
a few tens of meters in diameter, while the impact structures of geologic interest range up
to 1000 km in diameter. The traditional escape from this difficulty is to extrapolate beyond
experimental knowledge by means of scaling laws.
C. W. Lampson, who studied the craters produced by TNT explosions of different sizes,
introduced the first scaling law in 1950. Lampson found that the craters were similar to

EBSCOhost - printed on 11/3/2023 9:56 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
252 Impact cratering

one another if all dimensions (depth, diameter, depth of charge placement) were divided
by the cube root of the explosive energy W. Thus, if the diameter D of a crater produced
by an explosive energy W is wanted, it can be computed from the diameter D0 of a crater
produced by a smaller explosive energy W0 using the proportion:
1/ 3
D W (6.11)
= .
D0  W0 

An exactly similar proportion may be written for the crater depth, H. This means that
the ratio of depth to diameter, H/D, is independent of yield, a prediction that agrees rea-
sonably well with observation. In more recent work on large explosions the exponent 1/3
in this equation has been modified to 1/3.4 to account for the effects of gravity on crater
formation.
Although impacts and explosions have many similarities, a number of factors make them
difficult to compare in detail. Thus, explosion craters are very sensitive to the charge’s
depth of burial. Although this quantity is well defined for explosions, there is no simple
analog for impact craters. Similarly, the angle of impact has no analog for explosions.
Nevertheless, energy-based scaling laws were very popular in the older impact literature,
perhaps partly because nothing better existed, and many empirical schemes were devised
to adapt the well-established explosion scaling laws to impacts.

6.5.1 Crater diameter scaling

This situation has changed rapidly in the last few decades, however, thanks to more impact
cratering experiments specifically designed to test scaling laws. It has been shown that
the great expansion of the crater during excavation tends to decouple the parameters
describing the final crater from the parameters describing the projectile. If these sets of
parameters are related by a single, dimensional “coupling parameter” (as seems to be the
case), then it can be shown that crater parameters and projectile parameters are related by
power-law scaling expressions with constant coefficients and exponents. Although this is
a somewhat complex and rapidly changing subject, the best current scaling relation for
impact craters forming in competent rock (low-porosity) targets whose growth is limited
by gravity rather than target strength (i.e. all craters larger than a few kilometers in diam-
eter) is given by:
1/ 3
 ρp 
Dtc = 1.161   L0.78 vi0.44 g −0.22 sin1 / 3 θ (6.12)
 ρt 

where Dtc is the diameter of the transient crater at the level of the original ground surface,
ρp and ρt are densities of the projectile and target, respectively, g is surface gravity, L is
projectile diameter, vi is impact velocity and θ is the angle of impact from the horizontal.
All quantities are in SI units.

EBSCOhost - printed on 11/3/2023 9:56 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
6.5 Scaling of crater dimensions 253

Dat = 10 km
Vaporized
Melted

Dat = 100 km
Melted Vaporized

Dat = 1000 km
Vaporized
Melted

Figure 6.18 The different scaling laws for crater diameter and melt or vapor volume imply that as the
crater diameter increases, the volume of melted or vaporized material may approach the volume of
the crater itself. This figure is constructed for impacts at 35 km/s on the Earth.

The transient crater depth, Htc, appears to be a constant times the diameter Dtc. Although
a few investigations have reported a weak velocity dependence for this ratio, the experi-
mental situation is not yet clear.

6.5.2 Impact melt mass

The amount of material melted or vaporized by an impact is a strong function of the impact
velocity itself. The melt mass depends principally upon the velocity and mass of the pro-
jectile, but does not depend upon the gravitational acceleration of the target body. Very little
melt is produced until a threshold velocity of about 12 km/s is reached. Once this threshold
is exceeded, the mass of melt Mm is given in terms of the mass of the projectile Mp as:

Mm v2
= 0.25 i sin θ , vi ≥ 12 km/s (6.13)
Mp εm

where εm is the specific internal energy of the Rankine–Hugoniot state from which isen-
tropic decompression ends at the 1 bar point on the liquidus. It is equal to 5.2 MJ/kg for
granite, which can be taken to be representative of crustal rocks.
Although the mass of melt does not depend on the gravitational acceleration of the planet,
the crater size does, through Equation (6.12). The volume of the melt relative to the volume

EBSCOhost - printed on 11/3/2023 9:56 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
254 Impact cratering

of the crater is relatively small for small craters, but as the crater size increases it becomes
a progressively larger fraction of the total crater volume, as shown in Figure 6.18. At some
sufficiently large diameter (about 1000 km on the Earth), the volume of the melt equals the
volume of the crater itself and substantial changes in the morphology of the crater can be
expected, although these changes are not well understood at the present time.

6.6 Atmospheric interactions

As fast-moving meteoroids enter the atmosphere of a planet, they are slowed by friction
with the atmospheric gases and compressed by the deceleration. Small meteoroids are
often vaporized by frictional heating and never reach the surface of the planet. Larger
meteoroids are decelerated to terminal velocity and fall relatively gently to the surface of
the planet. The diameter of a meteorite that loses 90% of its initial velocity in the atmos-
phere is typically about 1 m for Earth and 60 m for Venus. However, this assumes that the
projectile reaches the surface intact, whereas in fact aerodynamic stresses may crush all
but the strongest meteorites. Once fractured, the fragments of an incoming meteorite travel
slightly separate paths and strike the surface some distance apart from one another. This
phenomenon gives rise to the widely observed strewn fields of meteorites or craters on the
Earth, which average roughly a kilometer or two in diameter. On Venus, clusters of small
craters attributed to atmospheric breakup are spread over areas roughly 20 km in diameter.
Rather surprisingly, clusters of small craters are also observed on Mars, where the spread of
small craters in a cluster averages only a few tens of meters to hundreds of meters across.
The aerodynamic crushing stress experienced by an incoming meteoroid is of the order
of the stagnation pressure, given by
Pstagnation ≈ ρ av2 (6.14)
where ρa is the density of the atmosphere through which the meteoroid is traveling and v
is the relative velocity of the meteoroid and atmosphere. Evidently, even the thin Martian
atmosphere is enough to fracture and partially disperse weak incoming meteoroids.
It can be shown that the dispersion of a cluster of fragments is a maximum when breakup
occurs at twice the atmospheric scale height, Hs. In this case the expected dispersion ΔY is
given by

2 Hs ρa
∆Y ≈ (6.15)
sin θ ρp

where θ is the angle of entry of the meteoroid with respect to the horizontal and ρp is its
density. Because the scale heights of the atmospheres of the Earth, Venus, and Mars are all
similar, the dispersion of clusters of fragments is expected to be roughly a factor of ten dif-
ferent among these bodies, increasing from Mars to Earth to Venus, as observed.
The atmospheric blast wave and thermal radiation produced by an entering meteorite
may also affect the surface: The 1908 explosion at Tunguska River, Siberia, was probably

EBSCOhost - printed on 11/3/2023 9:56 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
6.7 Cratered landscapes 255

Figure 6.19 Oblique view of a heavily cratered landscape on the Moon. This area is to the northeast
of crater Tsiolkovskiy on the Moon’s farside. The large crater near the center is about 75 km in
diameter, but craters as small as a few tens of meters in diameter can be discerned in the foreground.
Apollo 17 photo 155–23702 (H).

produced by the entry and dispersion of a 30 to 50 m diameter stony meteorite that leveled
and scorched about 2000 km2 of meter-diameter trees. Radar-dark “splotches” up to 50 km
in diameter on the surface of Venus are attributed to pulverization of surface rocks by strong
blast waves from meteorites that were fragmented and dispersed in the dense atmosphere.

6.7 Cratered landscapes

Impact craters have been treated as individual entities in the preceding sections. However,
as spacecraft images abundantly illustrate, the surfaces of most planets and satellites are

EBSCOhost - printed on 11/3/2023 9:56 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
256 Impact cratering

scarred by vast numbers of impact craters that range in size from the limit of resolution to a
substantial fraction of the planet’s or satellite’s radius. In some places impact craters are the
dominant landform: little of the observed topography can be ascribed to any other process
(see Figure 6.19). Craters on such a surface exhibit degrees of preservation ranging from
fresh craters with crisp rims and bright rays to heavily battered or buried craters that may
only betray their presence by a broken rim segment or a ragged ring of peaks.
The present crater population on surfaces such as the lunar highlands or the more lightly
cratered lunar mare is the outcome of a long history of impact cratering events. Analysis
of the existing crater population in conjunction with some assumptions about the rate of
crater formation may reveal a great deal about the geologic history of a surface. A typical
population is composed of craters with a wide range of sizes, some of which are relatively
fresh, with sharp rims, extensive rays, and crisp fields of small secondary craters, whereas
others are progressively more degraded. On parts of such airless bodies as the Moon, the
principal agent of degradation is other impacts, producing a surface that appears to be
crowded with craters. In other regions volcanism has created plains that are more or less
sparsely cratered. Elsewhere in the Solar System the activities of wind, water, or tectonic
processes such as subduction erase craters within a short time of their formation, leading to
landscapes like the Earth’s, where impact craters are among the rarest of landforms, or like
that of Venus, where the low abundance of craters may be due to an ancient era of resur-
facing that obliterated most pre-existing craters.
Study of crater populations is, thus, a powerful tool for geologic investigation of the sur-
faces of other planets and satellites. If the flux and size distribution of the impacting bodies
were known, studies of crater populations could yield absolute ages of the surface and
some of its features. Although the original flux is often unknown, relative ages can usually
be obtained. Before geologic inferences can be drawn from crater populations, however,
we must have an effective means of describing and comparing them. Unfortunately, a large
number of descriptions have evolved over the years as each group of scientists studying
a particular problem created their own specialized means of presenting population data,
making it difficult to compare the results of different groups. In this section I adopt the
major recommendations of a NASA panel convened in 1978 (NASA, 1978) to standardize
the presentation of crater population data.

6.7.1 Description of crater populations

The first step in an investigation of the crater population on a given surface is to select
an area that is believed to have had a homogeneous geologic history. It would make little
sense, for example, to combine the crater population of a sparsely cratered lava plain with
that of a densely cratered upland. Once such an area is selected, the craters that lie within
it are counted. Most crater population studies include all recognizable craters, regardless
of their state of degradation. Where a large enough population exists, more specialized
studies may be performed in which the numbers of fresh craters, slightly degraded craters,
degraded craters, etc., are counted separately. Although these studies leave some room for

EBSCOhost - printed on 11/3/2023 9:56 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
6.7 Cratered landscapes 257

interpretation as to what a “recognizable” or “fresh” crater is, intercomparison of results

between different groups of crater counters has generally shown good agreement.
Craters occur in a wide variety of sizes, so that the principal information about a cra-
ter population is the number of craters per unit area as a function of crater diameter. It is
presumed that impact cratering is a random process and that there is no significance to the
particular location of craters within the selected area, so that only data on the number and
diameter of craters is kept.
Incremental distribution. Numerous ways of representing the number of craters as a
function of diameter have been developed. One very simple method is to list the number
N of craters per unit area with diameters between two limits, say between Da and Db. The
problem with this method is that the resulting number of craters depends upon the interval
ΔD = Db – Da, and different crater counters may choose different intervals. Furthermore, if
the interval ΔD is fixed at, say, 1 km, this might be convenient for craters with diameters
between 5 and 20 km, but would be too large for craters with diameters less than 1 km and
too small for craters larger than 100 km. A simple way to overcome this problem is to let
the interval depend on crater size. Thus, the number of craters may be tabulated between
D and 2D, where the intervals increase in octaves. Actually, this binning has been found
to be too coarse in practice, so that most such incremental size-frequency distributions
use an interval of D to 2D . The incremental distribution still suffers, however, from the
arbitrary choice of a starting diameter D. It is now recommended that the bins be chosen so
that one bin boundary is at D = 1 km.
Cumulative distribution. Although the incremental size-frequency distribution could be
successful if the same bin sizes and boundaries are universally adopted, it lacks funda-
mental simplicity. Another distribution has long been used that is independent of bin size:
this is the cumulative size-frequency distribution. In this distribution the number Ncum of
craters per unit area with diameters greater than or equal to a given diameter D is tabulated.
Not only is the resulting distribution Ncum(D) independent of bin size, but any desired incre-
mental distribution can be easily generated from it, since the number of craters N per unit
area in the interval between Da and Db is simply:
N(Da, Db) = Ncum(Da) − Ncum(Db) (6.16)
where Db> Da. N(Da,Db) is necessarily positive or zero by the definition of the cumula-
tive number distribution. The only disadvantage of the cumulative distribution is that the
cumulative number of craters at some given diameter depends upon the number of craters
at all larger diameters. Although this is rarely a major problem, cumulative distributions in
limited-diameter intervals (often controlled by the size of the region being analyzed) have
to be adjusted in overall value to join with the distributions from other diameter ranges. The
slope of the cumulative plot is not, of course, affected by such adjustments.
It has been found in practice that the cumulative number distribution closely approxi-
mates a power function of diameter:
Ncum = c D−b (6.17)

EBSCOhost - printed on 11/3/2023 9:56 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
258 Impact cratering

Figure 6.20 A population of craters with a slope of b = 2 is structured such that for each crater of
diameter D, there are four times as many craters of diameter D/2 and 16 times as many as for diameter
D/4. When mixed together, the population looks very much like that seen on the heavily cratered
areas of the Moon and other planets. Note that the total area in each size class of this distribution is
the same, a characteristic that may explain why such distributions are so common when formed by
either coagulation or impact fragmentation – processes that both depend on surface area.

where b = 1.8 for post-mare craters on the Moon between 4 and several hundred kilometers in
diameter (and, within the limited data, also seems to hold for impact craters on the Earth).
It is intriguing that the power b is close to 2. A power 2 in Equation (6.17) is a kind of
“magic” number because when b = 2 the coefficient c is dimensionless (remember Ncum
is number per unit area). There is no fundamental length or size scale in a crater popula-
tion with this power law, so that such a population looks the same at all resolutions (see
Figure 6.20 for an illustration of such a distribution). It is impossible to tell from a photo-
graph of such a cratered landscape whether the scale of the photograph is 100 km or 1 m
(of course, other clues than crater population alone may give a hint about the actual scale).
A population of craters described by a power near 2 might arise either from a simple for-
mation process in which there is truly no fundamental length scale or from a series of inde-
pendent processes that are so complex and chaotic that no one scale dominates.
Because the cumulative number distribution of Equation (6.17) falls as a power of D, it is
conventional to graph such distributions on a log-log plot on which a power law is a straight
line with slope equal to –b (Figure 6.21). Unfortunately, cumulative number distributions plot-
ted in this way have a tendency to look all the same, apparently differing only in the absolute
number density of craters. Although this may be adequate or even desirable for some purposes,

EBSCOhost - printed on 11/3/2023 9:56 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
(a)
10–4

G
al

eo
lc
10–6

m
ra

et
ert

ric
eq

sa
ui

tu
lib

ra
riu

iot
m
Cumulative number ≥ D, #/m2

n
10–8

=
7%
NS
10–10

10–12

10–14
300 m

4 km

10–16 1
10 102 103 104 105 106
Crater diameter D, m
(b)
101
Geometric saturation
3.12

100

Small crater equilibrium, R = 0.22

10–1
R

10–2
300 m

4 km

10–3
101 102 103 104 105 106
Crater diameter, m

Figure 6.21 The post-mare crater population on the Moon in both a cumulative plot (a) and R-plot (b)
format. The population shows three distinct segments. At small diameters (D < 300 m) the population
is in equilibrium and the cumulative number is proportional to D–2. R is constant for this population.
At intermediate diameters (300 m < D < 4 km) the cumulative number is proportional to D–3.4,
whereas at larger diameters (D > 4 km) it falls as D–1.8. The curves are dashed at the largest diameters
(D > 200 km) because no craters of this size have yet formed on the mare. The changes in slope of
the crater population are especially evident in the R-plot format. The dashed line labeled “geometric
saturation” is an upper limit to the crater density on any surface. After Melosh (1989, Figure 10.2).

EBSCOhost - printed on 11/3/2023 9:56 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
260 Impact cratering

such a mode of presentation obscures slight but significant differences in crater populations.
Since these differences can be useful in deciphering the source of the crater population or other
geologic processes that acted on it, another type of plot is in common use.
The R plot. This type of plot exploits the close approach of b to 2 in Equation (6.17) by
graphing essentially the ratio between the actual crater distribution and a distribution with
slope –2. A crater population in which the actual slope is −2 would, thus, plot as a hori-
zontal line. The conventional plot of this type is called an R plot (R stands for “Relative”).
It is based on an incremental distribution with 2 intervals between diameter bins. Note
that the slope of this type of incremental plot is the same as the cumulative distribution of
Equation (6.17), although the coefficient is different. It is easy to show that if b is constant
over the interval D to 2D , then the incremental number density is:

N ( D, 2 D) = c (1 − 2 − b / 2 ) D − b . (6.18)

The definition of the R plot includes several numerical factors. In terms of the cumula-
tive number distribution it is given by:
23 / 4
R (D ) ≡ D 2  N cum ( D) − N cum ( 2 D) . (6.19)
2 −1 

Figure 6.21 compares the cumulative and R plots for pose-mare craters on the Moon.
A useful interpretation of R is to note that, up to a factor of 3.65, R is equal to the fraction
fc(D) of the total area covered by craters in the diameter interval D to 2D :
R(D) = 3.65 fc(D) (6.20)
where both R and fc are dimensionless numbers. With this interpretation it is easy to see
that in a crater population with b = 2, for which R and fc are constant, craters in every size
interval occupy the same fraction of the total area. If b < 2, as it is for post-mare craters on
the moon, R(D) increases as D increases so that large craters occupy a larger fraction of
the surface than small craters. If b > 2, small craters occupy a larger fraction of the surface
than do large ones.
The production population. The impact of the primary meteoroid flux on a planetary
surface results in some definite rate of crater production as a function of diameter. As
the surface, initially taken to be craterless, ages, more and more craters accumulate on it.
The integral of the crater-production rate over the age of the surface is a special, theoret-
ical, crater population called the production population. The production population is the
size-frequency distribution of all the craters, excluding secondary craters, that have ever
formed since craters began to accumulate on the surface. The population is theoretical in
the sense that it neglects all crater-obliteration processes and is, hence, formally unobserv-
able, although the crater population on lightly cratered surfaces may approach the produc-
tion population closely enough for practical applications.
The production population is a useful concept for the study of the evolution of crater pop-
ulations. Such studies usually begin with an assumed or inferred production population and

EBSCOhost - printed on 11/3/2023 9:56 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
6.7 Cratered landscapes 261

on
ucti
od
Geometric saturation Pr

Craters/unit area
Ns

Equilibrium Observed
Neq

teq
Time

Figure 6.22 Evolution of a crater population in which all craters are the same size. The graph shows
how the crater density increases as a function of time. Although the production population rises
linearly with time, the number of craters that can be counted on the surface eventually reaches a
limit well below the geometric saturation limit. Once the population has reached equilibrium, each
additional crater obliterates, on average, one old crater.

then postulate a model of crater obliteration that will, it is hoped, result in a predicted crater
population that matches the observed population. The crater-obliteration model is a function
of the process being modeled. In the next section we shall consider only the process of crater
obliteration by other craters. As we shall see, this can be a surprisingly complex process.

6.7.2 Evolution of crater populations

Given the fact that craters accumulate randomly on the surface of a planet at a rate that is,
on average, constant, it should be possible to begin with a knowledge of the rate at which
craters form as a function of their size and then predict the crater population at any future
time. Furthermore, it should be possible, within certain limits, to invert an evolved crater
population and deduce both how long it has been accumulating (that is, the age of the sur-
face) and the rate of crater formation as a function of size.
Simple as these propositions may seem, there has been great difficulty in actually imple-
menting them. The main problem is the interaction between craters of different sizes: the
formation of a single large crater on a surface may obliterate many smaller craters, while it
takes many small craters to batter a large crater beyond recognition. It has taken many years
to fully understand the effects of this interaction on crater populations, and some aspects of
it are not completely understood today.
The conceptually simplest population is one in which all craters are the same diameter.
Although no natural examples of such a population are known, the study of its evolution
introduces several important concepts. Moreover, there are crater populations, such as that
on Mimas, in which many of the craters fall within a relatively narrow size range and that
may, therefore, be approximated by a population of craters all of the same size.

EBSCOhost - printed on 11/3/2023 9:56 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
262 Impact cratering

The equilibrium population. The evolution of this crater population is illustrated in

Figure 6.22 as a function of time. The plot shows that the observed crater population and
the production population initially increase at the same rate. However, as the density of
craters on the surface increases a few older craters are either overlapped by new ones or are
buried by their ejecta. As this process continues, some older craters are completely oblit-
erated by younger craters and the observed crater density falls below the production line.
Eventually, the crater density becomes so high that each new crater that forms obliterates,
on average, one older crater. At this stage the crater population has reached equilibrium: no
further increase in crater density is possible, although new craters continue to form and the
production population increases steadily in number.
The attainment of equilibrium places severe constraints on attempts to date planetary
surfaces by crater counting. Up until equilibrium is attained at time teq in Figure 6.22 the
crater density increases with the age of the surface, so that knowledge of the crater produc-
tion rate permits computation of an absolute age from the crater density. Even if the pro-
duction rate is not known, the relative ages of two surfaces may be obtained by comparing
their crater densities. However, once equilibrium is attained the crater density becomes
constant and only a lower limit on the age can be obtained. The relative ages of two sur-
faces in equilibrium are also completely unconstrained, since their crater densities are iden-
tical even though the surfaces may be widely different in age.
Geometric saturation. A useful concept introduced by Don Gault in 1981 (Project, 1981)
is that of geometric saturation. The idea is to define a crater density that serves as an upper
limit to the number of craters that can possibly be recognized on a heavily cratered surface.
For a population of craters all of the same size this limit is simply the number of craters
per unit area in a hexagonally closest packed configuration, neglecting any possibility of
obliteration by overlapping ejecta blankets. In this case, Ns = 1.15 D−2. The definition of a
limiting crater density when craters are of different sizes is less objective. Gault proposed
the limit:
Ncs = 1.54 D−2. (6.21)
This crater density corresponds to R = 3.12 or a fractional area coverage fc = 0.85.
These limits are shown in Figure 6.21: It is clear that lunar crater populations fall well
short of this limiting density, although the mare surface is apparently in equilibrium for
crater diameters less than about 300 m. The crater density on even the most heavily cra-
tered surfaces seldom reaches more than about 3–5% of the geometric saturation limit,
Equation (6.21).

6.8 Dating planetary surfaces with impact craters

It has long been recognized that the number of impact craters per unit area could date
planetary surfaces. Shoemaker et al. (1963), who wrote the classic paper on this subject,
realized that relative dates on the same planet could always be attained, but absolute dates
require knowledge of the flux of impacting bodies.

EBSCOhost - printed on 11/3/2023 9:56 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
6.8 Dating planetary surfaces with impact craters 263

Time t1 Time t2

Pro
Pro
Ob
se

duc
duc
rv
Log cumulative frequency
ed

tion
tion
G Ob
eo se Ge
m rv om
et ed
ric et
sa ric
Eq tu sa
uil ra tu
ibr tio ra
ium n tion
Eq
uil
ibr
ium

Deq (t1) Deq (t2)

Log diameter

Figure 6.23 Evolution of a crater population with slope b > 2. The production population exceeds
the equilibrium line at small crater diameters. Small craters are, thus, in equilibrium up to some
diameter Deq, above which the observed population follows the production population. The left panel
illustrates the population at a relatively early time tl and the right panel shows how the population has
changed at a later time t2. The equilibrium diameter Deq increases as a function of time, although this
increase is generally not linear.

A full understanding of how to interpret crater populations as a function of size has been
long in coming. The attainment of equilibrium by crater populations divides into two dis-
tinct cases (and a trivial intermediate). The first, and simplest, case occurs when the pro-
duction population has a slope b steeper than 2. This case was studied by Gault (1970) and
offers the fewest conceptual difficulties. Unfortunately, this distribution is only appropriate
for small (≤300 m) craters on the lunar mare (see Figure 6.21), although at the time Gault
performed his analysis it was believed to be valid for all crater sizes. The size-frequency
distribution of larger lunar craters follows a power law with a slope b = l.8. This distribu-
tion fits the second case of an evolving population with slope b smaller than 2, and will be
treated shortly.

6.8.1 b > 2 population evolution

Figure 6.23 illustrates the evolution of a production population with slope b steeper than
2. The left frame depicts the population at an early time t1 and the right frame is at a later
time t2. Because the production population is steeper than the geometric saturation line,
mutual obliteration must occur for sufficiently small craters (craters smaller than Deq(t1)
in Figure 6.23) no matter how early the time, unless, of course, the population has had
so little time to evolve that the statistics of small numbers of craters begins to play a role.
Because the actual crater density cannot reach the geometric saturation limit, the observ-
able crater density reaches equilibrium somewhere below this line. The difference in the

EBSCOhost - printed on 11/3/2023 9:56 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
264 Impact cratering

Figure 6.24 A laboratory-scale demonstration of the concept of crater equilibrium. The photographs
are of a box 2.5 m square filled 30 cm deep with quartz sand. The sand is topped with 2 cm of
carborundum powder to provide a color contrast. Six sizes of projectile were fired into the box at
random locations, simulating a production population with a slope index b = 3.3, similar to that of
small craters on the Moon. Time increases from the upper left horizontally to lower right. Equilibrium
is attained about halfway through the simulation: Although individual surface details vary from
frame to frame, the crater population in the later frames remains the same. From Gault (1970); photo
courtesy of R. Greeley.

number of craters between the projected production population and the observed popu-
lation is equal to the number of craters obliterated by later impacts. It seems intuitively
reasonable that the equilibrium population should follow a line parallel to, but below, the
geometric saturation distribution. Gault showed empirically in small-scale impact experi-
ments (Figure 6.24) that this is the case, and subsequent work has confirmed this result
both analytically (Soderblom, 1970) and by Monte Carlo computer simulations (Woronow,
1977). In Gault’s experiments the crater equilibrium density depends on the slope b of
the production population, with steeper slopes giving a lower equilibrium crater density.
Similar results were also obtained from the theoretical studies.
The observed population at any one time in Figure 6.23 is, thus, composed of two
branches. Small craters follow an equilibrium line with slope b = 2. Larger craters follow
the steeper production population curve. The inflection point between these two curves is
at diameter Deq(t1) where the production curve crosses the equilibrium line. The right-hand

EBSCOhost - printed on 11/3/2023 9:56 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
6.8 Dating planetary surfaces with impact craters 265

Log cumulative frequency/unit area

Ge
o
me
tric
sa
Pro

tu
du

ra
ctio

tio
n

n
Ob
se
rve
d

1
Study area

Log diameter

Figure 6.25 Evolution of a crater population with slope b < 2. The production population exceeds the
geometric saturation line at large diameters. Under these circumstances the evolution of the population
is dominated by large impacts that allow a number of small craters to accumulate before wiping the
surface clean. The observed population line is always parallel to the production population, but may
lie considerably below it. The production population line is dashed at the large-crater end because
there must be at least one crater within the study area for the line to be meaningful. The observed
population may approach geometric saturation if the entire study area is the site of one large impact,
but the population will be dominated by fluctuations because of the statistics of small numbers.

frame in Figure 6.23 shows the crater density at a later time. The crater population is quali-
tatively similar to that at the earlier time, except that the transition diameter Deq(t2) is larger.
If the rate at which craters accumulate is constant (that is, c in Equation (6.17) is a linear
function of time), it is easy to show that the transition diameter Deq(t) grows as (time)1/(b–2).
Conversely, if the crater production rate, slope b, and the present Deq are known, the age of
the surface can be computed.

6.8.2 b < 2 population evolution

Figure 6.25 illustrates the evolution of a crater population with slope b less than 2. Note that
in the unlikely event that b exactly equals 2, the observed crater density simply maintains
the slope of 2 until the density reaches equilibrium (which occurs simultaneously at all
diameters) after which the observed crater density remains constant and the slope remains
2. When b is less than 2 the situation is more complex. In this case the production curve
exceeds the geometric saturation line at the large-diameter end of the scale. Gault’s (1970)
model could not deal with this situation and it took nearly 15 years before the implications
of this large-diameter crossing were understood, in spite of the developing knowledge that
b ~ 1.8 for craters more than about 4 km in diameter on the lunar mare. The situation was

EBSCOhost - printed on 11/3/2023 9:56 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
266 Impact cratering

resolved by C. R. Chapman (Chapman and McKinnon, 1986) who performed a Monte

Carlo simulation of crater population evolution that included a wider range of diameters
than had previously been possible.
Chapman realized that, first of all, the large-diameter end of the cratering curve is domi-
nated by the statistics of small numbers. Even though the production curve apparently
crosses the geometric saturation line at all times, enough time must pass after the formation
of the initial uncratered surface that at least one large crater has formed on it. The crossing
of the two curves makes little sense unless the cumulative number of craters larger than Ds
is at least 1 in the finite area under study. Since the probability of a small impact is larger
than the probability of a large impact for any b > 0, a population of small craters initially
develops that follows the production curve closely. Eventually, however, a large impact
occurs. With b < 2, there is a high probability that this large crater obliterates all or a signifi-
cant fraction of the study area, wiping out nearly all previous smaller craters. The number
of observed craters thus suddenly drops below the production curve. As time passes, small
craters again accumulate on the surface. The slope of this new population is equal to that of
the production curve but, as shown in Figure 6.25, the cumulative number of such craters
is smaller than the production population. The number of small craters continues to grow
until the next large impact wipes the slate clean once more.
Under these circumstances there is no “equilibrium” population, although the observed
crater density is always well below the geometric saturation limit. The observed crater
density fluctuates widely and irregularly, controlled by the large, rare catastrophic-impact
events. In spite of these wide variations in density, the slope of the observed population at
any given time is roughly equal to that of the production population. The crater densities
on a surface of this type are spatially patchy, being low at the sites of recent large impacts
and high in areas that have not been struck for a long time by one of the large impacts.
Dating such a surface is nearly impossible after the first large impact, unless some area can
be found that has escaped all large impacts. In practice, all that can be determined from
the crater density on such a surface is the date since the last large impact that affected the
particular study region.

6.8.3 Leading/trailing asymmetry

An important assumption in the relative dating of planetary surfaces is that the cratering
rate is uniform over the entire body. Although this is true to a high degree of approximation
for most bodies in the Solar System, it may be badly violated for synchronously locked
planets and moons. Our Moon and the Galilean satellites of Jupiter circulate about their
primary with the same face always leading. Just as a car driving into a rainstorm encounters
more raindrops on its front windscreen then on the back window, the impact flux on syn-
chronous satellites is higher on the leading side than the trailing side, leading to an asym-
metry in cratering rate that may be as large as a factor of 20 or more, depending on the
cratering population (Zahnle et al., 2001). The degree of asymmetry depends on the orbits
of the impactors. The asymmetry is largest for impactors in orbit about the same primary

EBSCOhost - printed on 11/3/2023 9:56 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
6.9 Impact cratering and planetary evolution 267

or moving at a low relative velocity. In this case the impact velocity on the leading hemi-
sphere is the sum of the orbital velocities of the satellite and impactor, while the velocity on
the trailing hemisphere is the difference, so that not only is the flux of impactors different
between the leading and trailing hemispheres, but the different impact velocities mean that
the same size of impactor yields a different size of final crater. The leading/trailing asym-
metry is smaller for heliocentric populations of impactors: Comets, for example, produce
much more uniform crater populations than co-orbiting objects on synchronous satellites.
Neptune’s moon Triton has a large leading/trailing crater asymmetry, but the Galilean
satellites of Jupiter show much less crater asymmetry than expected on theoretical grounds.
Some process must, therefore, occasionally reorient their surfaces relative to Jupiter: Either
an exchange of their primary-facing and opposite hemispheres due to exchange of their A
and B moments of inertia by internal or external (cratering?) mass transfers, or perhaps the
momentum impulse of a large cratering event.

6.9 Impact cratering and planetary evolution

Over the last few decades it has become increasingly clear that impact cratering has played
a major role in the formation and subsequent history of the planets and their satellites.
Aside from their scientific interest, impact craters have also achieved a modest economic
importance as it has become recognized that the fabulously rich Sudbury nickel deposit
in Ontario, Canada, is a tectonically distorted 140-km diameter impact crater. Similarly,
the Vredefort structure in South Africa is a large, old, impact crater. Oil production has
been achieved from a number of buried impact craters, such as the 10-km diameter Red
Wing Creek crater in the Williston Basin and the 3.2-km diameter Newporte structure in
North Dakota. On a more homely level the 60-km diameter ring-shaped depression in the
Manicouagan crater in eastern Quebec is currently used as a reservoir supplying water to
New York City.

6.9.1 Planetary accretion

Modern theories of planetary origin suggest that the planets and the Sun formed simul-
taneously 4.6 × 109 yr ago from a dusty, hydrogen-rich nebula. Nebular condensation and
hydrodynamic interactions were probably only capable of producing ca. 10-km diameter
“planetesimals” that accreted into planetary-scale objects by means of collisions. The time-
scale for accretion of the inner planets by mutual collisions is currently believed to be
between a few tens and one hundred million years. Initially rather gentle, these collisions
became more violent as the random velocities of the smaller planetesimals were increased
during close approaches to the larger bodies. The mean random velocity of a swarm of
planetesimals is comparable to the escape velocity of the largest object, so as the growing
planetary embryos reached lunar size, collisions began to occur at several kilometers per
second. At such speeds impacts among the smaller objects were disruptive, whereas the
larger objects had sufficient gravitational binding energy to accrete most of the material

EBSCOhost - printed on 11/3/2023 9:56 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
268 Impact cratering

(a) (b)

Ncum = C D–2 A = C’

log Ncum

log A
slope = 0
slope = –2

log D log D

(c)

M = C” D
log M

slope = 1

log D

Figure 6.26 The numerical foundation of impact catastrophism. The size-frequency distribution of
a typical crater population with slope b = 2 is shown in panel (a). In any such distribution there are
many more small craters than large ones. Panel (b) shows the area of craters as a function of diameter.
This distribution is flat: The total area of small craters and large craters is equal (refer back to Figure
6.20). Panel (c) shows, however, that the mass is concentrated at the large-size scale, so that despite
the overwhelming numerical superiority of small craters, the few large ones are the most important
in terms of either mass or energy delivered.

that struck them. Infalling planetesimals bring not only mass, but also heat to the growing
planets. In the past it was believed that the temperature inside a growing planet increased
in a regular way from near zero at the center to large values at the outside, reflecting the
increase in collision velocity as the planet became more massive. However, it now seems
probable that the size distribution of the planetesimal population was more evenly graded
between large and small objects, and that each growing planetary embryo was subjected
to many collisions with bodies comparable in size to the embryo itself. Such catastrophic
collisions would deposit heat deep within the core of the impacted body, wiping out any
regular law for temperature increase with increasing radius and making the thermal evolu-
tion of growing planets rather stochastic.

6.9.2 Impact catastrophism

Although small craters (and the impactors that created them) are overwhelmingly more
abundant than large ones in terms of numbers, the characteristic b ≈ 2 slope of most such

EBSCOhost - printed on 11/3/2023 9:56 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
6.9 Impact cratering and planetary evolution 269

distributions in the Solar System implies that most of the mass actually resides at the large
end of the size spectrum (see Figure 6.26). This circumstance gives rise to the idea of
impact catastrophism: Although small impacts are relatively common, all of the small
impacts combined do less geologic work than a few large impacts. This is the formal def-
inition of a catastrophic process and it is a good description of the way that impacts affect
the geological evolution of the planets. It also implies that stochastic processes cannot be
neglected where impacts play a role, except in situations where b is substantially larger
than 2, such as for the Shoemaker–Gault theory of regolith evolution on the Moon.
Note that the usual statistical mechanical idea of energy equipartition does not play a
role in the Solar System: The dynamics of the Solar System are such that the time to relax
to a state of thermal equilibrium is far, far longer than the age of the Solar System, so that
the biggest, most massive objects also carry the most energy.

6.9.3 Origin of the Moon

The origin of the Moon is now attributed to a collision between the proto-Earth and a Mars-
size protoplanet near the end of accretion 4.5 × 109 yr ago. This theory has supplanted the
three classic theories of lunar origin (capture, fission, and co-accretion) because only the
giant impact theory provides a simple explanation for the Moon’s chemistry, as revealed
in the lunar rocks returned by Apollo. One view of this process is that a grazing collision
vaporized a large quantity of the proto-Earth’s mantle, along with a comparable quantity of
the projectile. While most of the mass of the projectile merged with the Earth (incidentally
strongly heating the Earth: If the Earth was not molten before this impact it almost certainly
was afterward), one or two lunar masses of vapor condensed into dust in stable Keplerian
orbits about the Earth and then later accumulated together to form the Moon.

6.9.4 Late Heavy Bombardment

Sometime after the Moon formed, and before about 3.8 × 109 yr ago, the inner planets and
their satellites were subjected to the “Late Heavy Bombardment,” an era during which the
impact fluxes were orders of magnitude larger than at present. The crater scars of this period
are preserved in the lunar highlands and the most ancient terrains of Mars and Mercury. A
fit to the lunar crater densities using age data from Apollo samples gives a cumulative crater
density through geologic time of:
Ncum (D > 4 km) = 2.68 × 10−5 [T + 4.57 × 10−7 (eλT − 1)] (6.22)
where Ncum(D > 4 km) is the cumulative crater density (craters/km2) of craters larger than
4 km diameter, T is the age of the surface in Gyr (T=0 is the present) and λ = 4.53 Gyr–1.
The current cratering rate on the moon is about 2.7 × 10–14 craters with D > 4 km/km2/yr.
On the Earth the cratering rate has been estimated to be about 1.8 × 10–15 craters with
D > 22.6 km/km2/yr, which is comparable to the lunar flux taking into account the dif-
ferent minimum sizes, since the cumulative number of craters Ncum(D) ~ D–1.8. There is

EBSCOhost - printed on 11/3/2023 9:56 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
270 Impact cratering

currently much debate about these cratering rates, which might be uncertain by as much
as a factor of 2.
Although Equation (6.22) describes an impact flux that decreases monotonically with
time, there is currently much debate about the possible reality of an era of “heavy bombard-
ment” between about 4.2 and 3.9 Gyr ago in which the cratering rate reached a local peak
(Strom et al., 2005). It is now supposed that eccentric asteroids from the main asteroid belt
caused this peak in flux. These asteroids were mobilized as destabilizing orbital resonances
swept through the asteroid belt when Jupiter and Saturn underwent an episode of planetary
migration. The heavily cratered surfaces of the Moon and terrestrial planets are supposed
to have formed at this time. The intense flux obliterated any evidence of an earlier surface
on these bodies.
The cratering rates on Mars and Venus are believed to be comparable to that on the Earth
and Moon; however, the exact fraction of the Earth/Moon rate is presently uncertain and
a subject of controversial discussion. The exact rates will probably remain unknown until
radiometric dates on cratered surfaces are determined by means of sample returns from
these bodies. Cratering rates in the outer Solar System are even more uncertain and con-
troversial: Most of the craters on the satellites of Jupiter and Saturn are probably formed
by comets, whose flux is very uncertain at the small sizes represented by most observed
craters (Dones et al., 2009).
The high cratering rates in the past indicate that the ancient Earth must have been heavily
scarred by large impacts. Based on the lunar record it is estimated that more than 100 impact
craters with diameters greater than 1000 km should have formed on the Earth. Although
little evidence of these early craters has yet been found, it is gratifying to note the recent
discovery of thick impact ejecta deposits in 3.2 to 3.5 Gyr Archean greenstone belts in both
South Africa and Western Australia. Since rocks have recently been found dating back to
4.2 Gyr, well into the era of heavy bombardment, it is to be hoped that more evidence for
early large craters will be eventually discovered. Heavy bombardment also seems to have
overlapped the origin of life on Earth. It is possible that impacts may have had an influence
on the origin of life, although whether they suppressed it by creating global climatic catas-
trophes (up to evaporation of part or all of the seas by large impacts), or facilitated it by
bringing in needed organic precursor molecules, is unclear at present. The relation between
impacts and the origin of life is currently an area of vigorous speculation.

6.9.5 Impact-induced volcanism?

The idea that large impacts can induce major volcanic eruptions is one of the recurring
themes in the older geologic literature. This idea probably derives from the observation
that all of the large impact basins on the Moon’s nearside are flooded with basalt. However,
radiometric dates on Apollo samples made it clear that the lava infillings of the lunar basins
are nearly 1 Gyr younger than the basins themselves. Furthermore, the farside lunar basins
generally lack any lava infilling at all. The nearside basins are apparently flooded merely
because they were topographic lows in a region of thin crust at the time that mare basalts

EBSCOhost - printed on 11/3/2023 9:56 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
Further reading 271

were produced in the Moon’s upper mantle. Simple estimates of the pressure release caused
by stratigraphic uplift beneath large impact craters make it clear that pressure release melt-
ing cannot be important in impacts unless the underlying mantle is near the melting point
before the impact (Ivanov and Melosh, 2003). Thus, it is probably safe to say that, to date,
there is no firm evidence that impacts can induce volcanic activity. Impact craters may
create fractures along which pre-existing magma may escape, but themselves are probably
not capable of producing much melt. Nevertheless, there were massive igneous intrusions
associated with both the Sudbury and Vredefort structures whose genesis is sometimes
attributed to the impact, although in this case they may have been triggered by the uplift of
hot lower crust. Further study of these issues is needed.

6.9.6 Biological extinctions

The most recent major impact event on Earth was the collision between the Earth and a 10
to 15 km diameter asteroid 65 Myr ago that ended the Cretaceous era and caused the most
massive biological extinction in recent geologic history. Evidence for this impact has been
gathered from many sites over the last decade, and is now incontrovertible (Schulte et al.,
2010). First detected as an enrichment of the siderophile element iridium in the ca. 3 mm
thick K/Pg (Cretaceous/Paleogene) boundary layer in Gubbio, Italy, the iridium signature
has now been found in more than 100 locations worldwide, in both marine and terrestrial
deposits. Accompanying this iridium are other siderophile elements in chondritic ratios,
shocked quartz grains, coesite, stishovite and small (100–500 μm) spherules resembling
microtektites. All these point to the occurrence of a major impact at the K/Pg boundary.
The impact crater is located beneath about 1 km of sedimentary cover on the Yucatan
Peninsula of Mexico. Known as the Chicxulub crater, it is about 170 km in diameter and is
presently the subject of intensive study.

Further reading
A general survey of all aspects of impact cratering from which several sections of this
chapter were abstracted was published by Melosh (1989). This book is presently out of
print and often difficult to obtain, but many university libraries possess a copy. A more
popular but generally clear and accurate description of terrestrial impact craters can be
found in Mark (1987). A good description of the three largest craters on Earth can be found
in Grieve and Therriault (2000). Although deeply eroded or otherwise obscured, these cra-
ters provide the “ground truth” about impacts that images of extraterrestrial craters cannot
give. The surface of our Moon is dominated by impact craters and so the account of its
geology by Wilhelms (1987) is necessarily an account of the process of impact cratering.
Don Wilhelms also wrote an insightful and very readable historical account of lunar explor-
ation that emphasizes the growing appreciation of impact cratering though the course of the
Apollo missions (Wilhelms, 1989). The importance and mechanics of impact crater col-
lapse and the simple-to-complex transition is treated by Melosh and Ivanov (1999), while

EBSCOhost - printed on 11/3/2023 9:56 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
272 Impact cratering

the special features of oblique impacts are the subject of a review by Pierazzo and Melosh
(2000). The best current summary of the scaling of impact crater ejecta was stimulated by
analysis of the ejecta from the Deep Impact Mission cratering event (Richardson et al.,
2007). A fine review of the details of crater counting and the application of crater counts
to the dating of surfaces appears in a rather unlikely-sounding book, which was the subject
of a collaborative project called the Basaltic Volcanism Study Project (Project, 1981). The
near-field effects of an impact on the Earth are reported by an on-line computer program,
whose basic algorithms are described in detail by Collins et al. (2005).

Exercises

6.1 Crater dimensions

a) Suppose that the transient crater that collapsed to form the Imbrium Basin on the Moon
was Dtc = 800 km in diameter. Assuming that it struck vertically at vi = 22 km/s, what
was the diameter of the asteroid that produced Imbrium?
Use the revised Schmidt–Holsapple scaling law (this form implicitly assumes that the
target and projectile have the same density):

 g
0.277

d = 0.671 D  
1.28
. (MKS units)
v  tc 2

b) If this object struck at 45° to the vertical, what is the period of the Moon’s rotation
thus imparted, assuming the Moon was free in space (unlikely, but easy to analyze)?
Compare this to the Moon’s current rotation rate. What can you conclude about the abil-
ity of large impacts to unlock the Moon’s present synchronous rotation state?

6.2 Crater collapse

a) The empirical relation for the thickness T of an impact crater’s ejecta blanket indi-
cates that the average thickness declines as (Rc /r)3, where Rc is the crater radius and

1 r
r3 Tc

h
T(r)

Figure 6.27

EBSCOhost - printed on 11/3/2023 9:56 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
Exercises 273

More exact

Simple geometry
- OK for this problem

Figure 6.28

r is distance from the crater’s center. Assuming that the interior shape of a simple
(or transient) crater is described by a paraboloid of revolution with diameter/depth
ratio Dc /h ≈ 5 (use the rim-to-rim diameter Dc and depth h below the original ground
surface), and assuming that the volume of the ejecta blanket and crater bowl are
equal, derive an equation for the ejecta blanket thickness Tc at the rim. (Hint: It is
permissible at this level of accuracy to treat the inner edge of the ejecta blanket as if
it were a vertical cliff. An exact result, however, can be obtained with a bit of effort,
and anyone who can legitimately get the result Tc=h/5 should get extra credit for this
assignment!)
b) Now suppose the crater collapses to a uniform, constant depth hf (3 km for the Moon)
over its entire interior. Again using volume conservation, how much does the rim-to-rim
diameter increase for a 100 km (final diameter) crater?

6.3 Swedish rock rain

In addition to the 5 proven and 33 possible impact crater scars that adorn the country of
Sweden, Lilljequist and Henkel (1996) proposed the existence of a truly world-class cra-
ter, the Uppland structure, which is supposed to be 320 km in diameter. It encompasses
the area around Stockholm up to Uppsala. Using Shoemaker’s estimated cratering rate on
Earth, Ncum (D > 22.6 km) = 1.8 x 10–15 craters/km2/yr, and the post-mare size exponent
b = 1.8 (Ncum ~ D–1.8), compute the largest crater likely to have formed in Sweden (area ~
450 000 km2) since the formation of its surface rocks (1600 to 2300 Myr; use 2000 Myr as
a reasonable average).
How does this compare to the size of the putative Uppland structure? Put another way,
what is the probability of the formation of a crater as large as Uppland in the past 2000

EBSCOhost - printed on 11/3/2023 9:56 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
274 Impact cratering

Myr in Sweden? What does this tell you about the Uppland “crater”? Note that the largest
confirmed crater in Sweden is the 55 km-diameter Siljan structure.
Using the same cratering flux, compute the maximum size of crater likely to be found
in the UK (now it is the student’s task to look up the area and mean geologic age of that
fair but damp archipelago). Note that a single, probably 4 km diameter, impact crater was
discovered in the North Sea by Stewart and Allen (2002) – the Silverpit crater!

6.4 Those blasted Martian rocks!

F. Hörz et al. (1999) argued that dish-shaped depressions on large rocks at the Pathfinder
landing site on Mars, along with split boulders, might be due to the impact of millimeter to
centimeter diameter meteorites at speeds of 300 m/s or more (corresponding to the muzzle
velocity of a high-speed rifle).
Evaluate the ability of the Martian atmosphere (surface pressure PS = 600 Pa) to stop
small meteoroids (average pre-atmospheric speed 7 km/s) by comparing the mass of the
projectile to the mass of atmospheric gases swept out of the cylindrical volume defined by
the meteoroid’s path (assume a straight trajectory for the purposes of this estimate).
The total mass of atmospheric gas above a unit area on the surface of a planet is PS /g,
where g is the acceleration of gravity at the surface. Perhaps surprisingly, you do not need
to know the density of the atmosphere as a function of height to do this problem (convince
yourselves of this!).
Using momentum conservation, estimate how much the meteoroid is slowed upon arriv-
ing at the surface as a function of its mass and density. Finally describe how reasonable the
Horz–Cintala proposal really is for both stony and iron meteoroids. Could this process be
effective on high mountaintops on Earth? Why or why not?

Extra credit
As the meteoroid traverses the atmosphere it encounters a ram pressure of ρav2 on its lead-
ing face ( ρa is the density of the atmosphere, 0.0104 kg/m3 at the surface of Mars), while its
rear is nearly a vacuum for supersonic flight. Estimate the maximum stresses thus acting to
crush the meteoroid and compare them to the typical strength of rock or iron for the mete-
oroids in the problem above. (NB, you do have to know the height distribution of density
for this problem – you may want to know that the scale height of the Martian atmosphere
is about 10 km.)

6.5 Titan gets its kicks!

One of the mysteries about Saturn’s large moon Titan is its (slightly) eccentric orbit,
e = 0.028. If Titan could dissipate tidal energy as Io or Europa does, its orbit would have
circularized long ago. It has been proposed that Titan’s Xanadu region is a gigantic impact

EBSCOhost - printed on 11/3/2023 9:56 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
Exercises 275

structure, 700 km in diameter (with an ejecta blanket bringing the diameter of the entire
feature to 1800 km). Using the impact crater scaling relation in Problem 6.1, estimate the
size of the object that could have made this crater, assuming a cometary impact velocity of
30 km/s (do not forget the difference between the transient and final crater!). If this object
struck the equator of Titan at the optimum place and angle to transform its linear momen-
tum to orbital angular momentum, could this impact have imparted enough momentum
to account for Titan’s present eccentricity? You may need to know that the orbital angular
momentum H of a planet of mass m circling a body of mass M with a semimajor axis of R
and eccentricity e is:

H = m GMR(1 − e2 ).

I will leave you with the task of looking up the values of the necessary parameters. Don’t
forget that you are looking for a change of orbit from circular (e = 0) to elliptical. Note that
this problem is akin to the oft-asked question of whether the K/Pg impact that killed the
dinosaurs could also have knocked the Earth “out of its orbit.” What do you think about
this possibility?

EBSCOhost - printed on 11/3/2023 9:56 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
7
Regoliths, weathering, and surface texture

It is generally accepted that the dynamic nature of the [lunar] regolith is

primarily the result of meteoritic bombardment. Although other contrib-

uting mechanisms have been suggested (e.g. electrostatic levitation …)
the formation of agglutinates, breccias, diaplectic glasses, and the pres-
ence of significant contamination of the regolith with meteoritic material
are collectively unambiguous evidence in support of the dominant role
meteoritic impact has played in regolith evolution.
Gault et al. (1974)

7.1 Lunar and asteroid regoliths: soil on airless bodies

Impact pioneer Gene Shoemaker and his colleagues introduced the modern concept of a
planetary regolith in 1967 (Shoemaker et al., 1967). Following the first soft landing on the
Moon’s surface by the Russian probe Luna 9 in 1966, Surveyors 1, 3, and 5 were the first
successful landers of the American Surveyor program. Detailed analysis of their images
by Shoemaker’s team was the first step toward clarifying the nature of an airless planetary
body’s surface. Geologist R. B. Merrill had coined the word “regolith” in 1897 to designate
the fragmental layer of rock debris that mantles the Earth’s surface. Compounded from the
Greek words for “blanket” (rhegos) and “stone” (lithos), the word had become obsolete by
1967, allowing Shoemaker to revive it in the context of the Moon’s surface.
Although the lunar surface layer is sometimes referred to as “lunar soil,” it is utterly dif-
ferent from the familiar agricultural soil of the Earth. Terrestrial soils typically contain large
amounts of organic carbon and weathered rock material that has been transported by wind
or water. Living organisms play a major role in creating Earth’s soil layers. In contrast, the
lunar surface is blanketed by a loose breccia composed of broken, angular rock fragments
that range in size from fine, submicron dust to meter-sized blocks. The lunar regolith has
never been disturbed by wind, water, or organic life. Instead, it is created and maintained by
the steady hail of meteoritic debris and radiation from open space (Figure 7.1).
Copyright 2011. Cambridge University Press.

The lunar regolith is still, by far, the best understood surface of an airless body thanks
to the many in situ investigations and sample returns by both manned and unmanned space
missions (Heiken et al., 1991). Although the ubiquitous regolith prevented the Apollo

276

EBSCO Publishing : eBook Academic Collection (EBSCOhost) - printed on 11/3/2023 9:57 AM via YDUQS
AN: 399390 ; H. Jay Melosh.; Planetary Surface Processes
Account: s4526441.main.estacio
7.1 Lunar and asteroid regoliths 277

Figure 7.1 Apollo 17 panorama showing details of the regolith at Station 1 on EVA 1. Note the
chaotic mixture of dust and rock fragments outcropping on the surface, and impact craters of all
sizes in the foreground. The hills in the background display the “elephant hide” texture typical of
lunar slopes. Wessex Cleft and the Sculptured Hills are visible in the background. Hasselblad photo
AS17–134-2048.

astronauts from directly sampling any bedrock exposures, it was quickly realized that 90%
of the rocks lying on and in the regolith are of local origin: The fact that the mare basalt/
highland contacts are still visible after 4 Gyr of meteoritic bombardment makes it clear that
lateral transport of regolith is very inefficient. Nevertheless, the remaining 10% of the rego-
lith is composed of material exotic to the collection site and represents debris thrown great
distances by large meteorite impacts. Up to about 2% is composed of meteoritic material
foreign to the Moon. A random regolith sample is, thus, much more informative about both
local and global geology than a comparable grab-bag sample of the Earth’s soil.
By terrestrial standards, lunar regolith is a most peculiar deposit. Its density is very low
at the surface, about 1400 kg/m3, but it rapidly compacts to about 2000 kg/m3 only a meter
below the surface (Figure 7.2). It is a gray, gritty, and extremely abrasive powder, full of poorly
sorted glass shards and angular rock fragments. Most of its particles range in size from around
40 μm up to a few hundred microns, with rare, larger rock clasts. Its electrical conductivity is
low and it easily acquires electric charges so that it sticks to everything: The astronauts found
that it coated any surface that it came in contact with and rapidly fouled the interior of their
spacecraft and the seals of their space suits. Even the loose surface layers have a small cohesive
strength, so that the astronaut’s bootprints retained vertical sides a few centimeters high. The
regolith has a large coefficient of internal friction, so the lunar surface has a considerable bear-
ing strength, allaying initial fears that landers would sink into a deep powdery surface layer.
On the Moon, the regolith ranges from 4 to 5 m deep on the mare to perhaps 20 m
deep in the highlands. Its depth is highly irregular even over short horizontal distances

EBSCOhost - printed on 11/3/2023 9:57 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
278 Regoliths, weathering, and surface texture

Depth, m

3
1000 1500 2000 2500
Bulk density, kg/m3

Figure 7.2 Calculated in situ bulk density of the lunar regolith as a function of depth below the
surface. The curve shown is a power-law fit, ρ = 1799 z0.056, where depth z is in meters. Simplified
after Figure 9.16 of Heiken et al. (1991).

Figure 7.3 Schematic cross section of the regolith (gray), indicating its irregular character. The
lunar regolith was created by repeated impacts of all sizes that generated a heterogeneous deposit
containing abundant impact melt (black) and broken fragments of bedrock (white).

(Figure 7.3). Drive tube cores reveal many layers whose thickness varies from about 1
to 10 cm. Spectrally, the lunar regolith is much darker than fresh rock of the same com-
position and reflects a larger fraction of long-wavelength (red) light. Spectral absorption
bands of minerals composing the bedrock are muted (Figure 7.4). Mineral fragments in the
regolith are riddled with tracks produced by energetic cosmic rays, and solar wind gases
are implanted below the surfaces of many grains. Although the Moon’s interior seems
to be nearly devoid of water, recent remote spectral observations have detected OH and
H2O molecules produced as solar wind protons bond chemically with oxygen in silicate
minerals on the surface (Sunshine et al., 2009). Neutrons liberated by primary cosmic rays

EBSCOhost - printed on 11/3/2023 9:57 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
7.1 Lunar and asteroid regoliths 279

0.6 fragmented breccia

0.5

0.4
Reflectance
0.3
mature regolith
0.2

0.1

0
0.5 1 1.5 2 2.5
Wavelength (µm)

Figure 7.4 Spectral reflectance of mature regolith and fragmented breccia. As regolith matures it
becomes darker (lower reflectance) and the prominent absorption bands at 0.9 and about 2 μm in the
fresh breccia disappear. Bidirectional reflectance spectra of noritic breccia 67455 showing pyroxene
absorption bands and mature regolith 68501, both collected by the Apollo 16 mission. After Figure 4
of Clark et al. (2002).

penetrate many meters into the surface, creating exotic isotopes whose presence records the
relentless irradiation and permits the total duration of exposure to be determined.
The regoliths of asteroids and other airless bodies are much less well understood than
that of the Moon. It was once believed that asteroids could not retain loose surface mater-
ial because meteorite impacts would eject it from their weak gravity fields. Surprisingly,
regoliths on the few asteroids for which thicknesses have been estimated are, if anything,
deeper than the lunar regolith. The regoliths of Ida and Eros are at least tens of meters thick,
while Phobos’ regolith may be as deep as 200 m. Although it is now clear that most aster-
oids are, in fact, mantled with loose surface material, asteroid regoliths seem to be much
less “mature” than that of the Moon, containing less glassy material. Nevertheless, asteroid
reflectance spectra clearly show the effect of “space weathering” that alters the way that they
reflect light and makes it difficult to connect astronomically observed asteroids with mete-
orites in the laboratory (Clark et al., 2002). The space environment also affects the spectra
of icy bodies in still unknown ways: Controversies presently rage over whether radiation
or hydrated salts control the spectra of icy satellites such as Europa. Organic “tars” on the
surfaces of comets and objects in the outer Solar System play similarly obscure roles.

7.1.1 Impact comminution and gardening

The major process creating and shaping the regolith is meteorite impact. When a fresh rock
is exposed on the surface of the Moon it is immediately pockmarked by small meteorites.
Indeed, the faceplates of the Apollo astronauts’ space suits recorded an extensive sample

EBSCOhost - printed on 11/3/2023 9:57 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
280 Regoliths, weathering, and surface texture

of micrometeorite impacts after only a few hours’ exposure to the lunar environment. As
time passes, more and more impacts accumulate on exposed rock surfaces. Surface rocks
are occasionally flipped over, so that small craters are found on their undersides. Small
impacts occur much more frequently than large ones (see the discussion of crater popula-
tions in Section 6.7) so that a surface is typically covered over with small impacts, before
it becomes more deeply excavated by large impacts. Boulders lying on the Moon’s surface
are eventually destroyed by an impact large enough to burst it into small fragments.
Crater populations are best described by the cumulative size-frequency distribution
Ncum(D), which is the number of craters per unit area of diameter equal to or greater than
D. Observation shows that this number density usually approximates a power-law (fractal)
relationship:
Ncum(D) = c D−b (7.1)
where c is equal to the number density of craters of diameter D = 1 km and the exponent b
indicates how sharply the density of craters falls as their diameter increases. If b = 2, craters
occupy the same fraction of the surface’s area, independent of their diameter (crater area is
proportional to D2, so when b = 2 the total area occupied by craters of diameter equal to or
larger than D is proportional to Ncum(D) D2, which is independent of D only when b = 2).
However, impact crater distributions at the sizes important for regolith formation typically
show b > 2: It ranges between 2.9 and 3.4 for lunar craters smaller than 4 km in diameter.
For distributions of this type, the surface is completely covered over by small craters before
it is covered with large craters. The case b < 2 does occur, but it is much more complicated
and the reader is referred to Section 6.8.2 for more information on this interesting situation.
Fortunately, this case seems important only for large impact craters.
To better understand regolith formation, pretend, for a moment, that all impact craters
are the same size. As these impact craters begin to accumulate on a freshly exposed surface,
their number grows steadily with time. Each new crater blasts out virgin rock and throws
it radially away from its rim, heaping it into an ejecta blanket outside the crater. As craters
become denser, their ejecta blankets begin to overlap and the ejecta from one crater partially
fills earlier craters. New craters then form in both virgin rock and in previously excavated
ejecta. As still more craters accumulate, less and less virgin rock remains and more craters
form in pre-existing ejecta. Eventually, the surface is entirely covered over with craters
and, on average, each new crater obliterates an older crater. When this occurs, the number
of observable craters ceases to grow and the population reaches “equilibrium.” New craters
continue to form, but the material they excavate is the ejecta from former craters. Because
craters of a given size fracture and excavate the rock to a depth of about ¼ of their diam-
eter, the bottom of this continually overturned layer is also about ¼ of the crater diameter
and its depth does not increase with time: The broken debris covering the surface is just
recycled as impact after impact accumulates. The overturn time of this layer is roughly the
time needed for the area of the accumulated craters to equal the total surface area.
Now recognize that craters of all sizes are actually forming. In a population with
b > 2, however, small craters cover the area faster than big craters, so the surface reaches

EBSCOhost - printed on 11/3/2023 9:57 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
7.1 Lunar and asteroid regoliths 281

obliterated

Log cumulative number per unit area

crater formation
Nprod = C D–b

crater equilibrium
Ncum Neq = Ceq D–2
observed
≥D

Deq
Log crater diameter, D

Figure 7.5 The population of impact craters observed on the lunar regolith is in equilibrium at small
sizes (below Deq) and reflects the production population at larger sizes for populations with slopes
b > 2. The heavy curve shows the observed population, while the extension of the production
population indicates the total number of craters that have struck the surface. The difference in these
two curves indicates how many craters have been obliterated.

equilibrium for small craters long before it equilibrates for big craters (Figure 7.5). While
the big craters are still blasting out virgin rock, small impacts simply recycle the shallower
surface layers (with an occasional addition of deeper-seated material from a larger impact).
The regolith layer now gradually deepens as it comes into equilibrium at progressively lar-
ger sizes, while the upper layers are recycled at rates that depend on their depth. This rapid
overturn at shallow depths and slow overturn for deeper material is known as “gardening,”
a term that aptly describes this process. It is analogous to a garden whose owner digs it up
once a year, hoes it weekly, and lightly rakes it every day.
Mathematical models of regolith growth were developed and successfully applied to the
Moon by D. Gault, E. M. Shoemaker, and their colleagues in the late 1960s (e.g. Gault,
1970). In these models the base of the regolith is the depth excavated by craters (of diam-
eter Deq), whose number is just sufficient to reach equilibrium. Of course the exact location
of each of these craters is determined by chance, so the base of the regolith is very irregu-
lar, with some areas cratered more than once by impacts that land on top of one another,
adjacent to areas that have, for the time being, entirely escaped cratering at this diameter.
One can show that the depth of such a regolith grows at the rate of (time)1/(b–2), assuming a
constant rate of impact (see Box 7.1). This rate is not linear unless b =3, even for a constant
impact rate. However, given that the observed values of b are close to 3, it is approximately
correct to say that the lunar regolith has accumulated at the rate of about 1.2 m/Gyr since
the era of heavy bombardment. This is a very slow rate and hardly suggests that lunar rego-
lith is a renewable resource for future lunar colonists.

EBSCOhost - printed on 11/3/2023 9:57 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
282 Regoliths, weathering, and surface texture

Box 7.1 Growth of the lunar regolith

The lunar regolith grows by the gradual accumulation of meteorite impacts. The earliest
theoretical studies of the growth of the regolith by E. M. Shoemaker et al. (1969) and all
subsequent studies are based on the evolution of crater populations with slopes b > 2. Although
b is generally less than 2 for large craters on the lunar mare, it typically ranges between 2.9
and 3.4 for craters smaller than 4 km in diameter, so the assumption that b > 2 is valid for the
post-mare lunar regolith and we shall consider only this case.
As shown in Figure 6.21, at any given time a b > 2 crater population is in equilibrium up
to some maximum crater diameter De(t), where De(t) increases as (time)1/(b–2) for a constant
rate of meteoritic bombardment. The crater population is in equilibrium with the production
rate at diameters D < De and at the smaller diameters many more craters have been obliterated
than presently can be counted on the surface. Shoemaker et al. (1969) realized that an initially
craterless surface subjected to this type of bombardment would rapidly develop a layer of
debris with a wide variation of depths. They supposed that the maximum depth likely to occur
is some fraction of the depth of the largest crater that has reached equilibrium. Although the
first crater of this diameter in a given region might leave an open cavity, the ejecta blankets of
subsequent craters would eventually fill it, leaving a layer of fragmental debris whose thickness
is locally equal to the crater depth. This maximum likely thickness heq is taken to be:

Deq
heq = . (B7.1.1)
4

The equilibrium crater diameter may be determined either directly, if it can be observed
as a kink in the crater population curve, or indirectly, by equating the cumulative number of
craters in the production population, Ncum = c D–b, to the number of craters in the equilibrium
population, Nceq = ceq D–2. In this case:
1
 c  b −2
Deq =   (B7.1.2)
 ceq 

where ceq is dimensionless and independent of diameter, although it may depend somewhat on
b. If it is supposed that equilibrium takes place at a crater density of about 4% of geometric
saturation, Equation (6.2.1), then ceq = 0.056. Conversely, the coefficient c describing the
production population may be expressed in terms of Deq and ceq:
b −2
c = ceq Deq . (B7.1.3)

I will use this representation of c in the rest of this section because it leads to more compact
formulas without fractional powers of units.
In addition to the maximum likely regolith thickness, Shoemaker and his colleagues
introduced the concept of a minimum regolith thickness. This is perhaps the least well-founded
concept of the model and does not agree well with the data. Unfortunately, it is also an integral
part of the model. The definition of the minimum thickness begins with a straightforward
integral: the fraction of the total area covered by craters with diameters between D and Deq,
defined as fc(D, Deq), is given by:

EBSCOhost - printed on 11/3/2023 9:57 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
7.1 Lunar and asteroid regoliths 283

Box 7.1 (cont.)

π bceq  Deq  
D b −2
π eq 2 dN com 
fc ( D, Deq ) = − ∫ D dD =   − 1 (B7.1.4)
4 D dD 4( b − 2 )   D  
 

where the minus sign in front of the integral is a consequence of the negative slope of the
cumulative number distribution: dN = –(dNcum/dD) dD.
When fc(D, Deq) exceeds some fixed fraction fmin at diameter Dmin the bottoms of all craters
with this diameter or less are presumed to be connected and the regolith is at least as thick
as this crater’s depth, hmin = Dmin/4. Shoemaker et al. argued that this occurs when the target
area is covered twice over by craters between Dmin and Deq, so that fmin = 2. Solving Equation
(B7.1.4) for fc(Dmin, Deq) = fmin = 2, they obtained:
1
−
 4( b − 2 ) f  b −2
hmin = heq  min
+ 1 . (B7.1.5)
 π bceq 

The probability P(h) of finding a patch of regolith of depth h, where h must lie between hmin
and heq, is given by the ratio of the fractional area covered by craters between diameters D = 4h
and Deq, to fmin:
fc (4h, Deq )
P (h ) = (B7.1.6)
fmin

After some algebraic manipulation P(h) can be written in the convenient form:

 (h / heq )b −2 − 1 
P (h ) =  b −2
 for hmin < h < heq . (B7.1.7)
 (hmin / heq ) − 1 

The median regolith depth <h > occurs when P( < h >) = 0.5. Solving the above equation, it
is easy to show that:
1
−
 h  b −2  b −2
< h > = 2heq  + 1
eq
. (B7.1.8)
 hmin  
 

Table B7.1.1 compares the predictions of this model to the observed regolith thickness
distributions for four areas on the moon. These thicknesses were obtained from the
morphology of small craters. This table shows that the model tends to overestimate the regolith
thickness by about 40%. Although some adjustments could be made to bring theory and
observation into better agreement (fmin = 3 would improve the fit greatly), such adjustments are
probably unjustified given the other assumptions of the theory. On this level, 40% agreement
is quite good. Although Equation (B7.1.7) is a good fit at the larger depths, it fits poorly at
shallow depths. This is mainly a consequence of the artificial choice of a minimum depth,
although it may also be due to neglect of the previously formed regolith in blanketing the
surface (Gault, 1970).

EBSCOhost - printed on 11/3/2023 9:57 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
284 Regoliths, weathering, and surface texture

Box 7.1 (cont.)

Table B7.1.1 Predicted vs. observed regolith thickness from Equation (B7.1.8)

Locationa Deqb (m) heq (m) hmin (m) <h > (m) Observedc < h > (m)

LO III P-11 80 20 2.3 4.5 3.3

LO III P-13b 120 29 3.4 6.6 4.6
LO II P-7b 180 45 5.3 10.0 7.5
LO V 24 410 100 12.0 23. 16.

a
Locations are designated by Lunar Orbiter mission and frame. For more detail see
Oberbeck and Quaide (1968).
b
Computed from Equation (B7.1.2) using ceq = 0.056. b = 3.4 for all sites and c = 25,
43, 79, and 250, respectively, for the four sites, where c is in units of m1.4.
c
Oberbeck and Quaide (1968)

Another important result of this model is that it predicts that the regolith thickness may
increase non-linearly with time even if the cratering rate is constant. Both heq and hmin are
related to Deq by constant factors for constant b, so the median depth <h> is also proportional
to Deq. Because Deq increases as (time)1/(b–2) for constant cratering rate, so do all these measures
of the regolith thickness. Only if b = 3 does the regolith accumulate at a constant rate. Data
on the crater population, regolith thickness, and ages of the underlying mare at the Apollo 11
landing site indicates b = 2.9, so that the lunar regolith there accumulated at a nearly constant
rate of about 1.2 /109 m/yr.
On the whole, the regolith evolution model of Shoemaker et al. (1969) gives an adequate,
first-order description of the growth of the regolith. It has the great advantage of being analytic,
so that the equations may be easily applied to many different situations, and it relies on the
observed crater population for its input parameters. Although it might be improved by a better
statistical treatment of the smaller regolith thicknesses, no one has yet published a revised
model.

These mathematical models predict regolith overturn times of about 105 yr for the top-
most centimeter, 3 × 106 yr for the upper 10 cm and 108 yr for the upper meter. These rates
are in good agreement with radiometric measurements of cosmic-ray ages of layers in
the Apollo and Luna core samples. This process predicts that many grains and rocks cur-
rently buried in the regolith once resided at the surface, a prediction amply verified by the
recovery of grains containing gases implanted from the solar wind. The lunar regolith has
served as a model for the formation of gas-rich meteorites, which are now recognized as
compacted regolith from their asteroidal parent bodies.
In addition to vertical mixing of the regolith, impacts also necessarily mix the regolith
laterally. The various lunar landers and Apollo astronauts noted gently sloping “fillets” of
regolith piled up around the base of larger rocks lying on the surface. Such accumulations
of debris are the result of horizontal motion of the regolith under the influence of small

EBSCOhost - printed on 11/3/2023 9:57 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
7.1 Lunar and asteroid regoliths 285

impacts. Most debris ejected by an impact crater falls within a distance of one or two crater
diameters from the rim, so lateral transport is not very efficient. Nevertheless, as shown
by the moon-spanning rays of fresh craters such as Tycho, a small amount of material is
transported very large distances. Indeed, the current ages of the large craters Tycho and
Copernicus are based on the premise that the Apollo 17, 15, and 12 astronauts collected
distal ejecta from these craters, whose rays cross the various landing sites.

7.1.2 Regolith maturity

The regolith is formed principally from the overlapping ejecta blankets of innumerable
impact craters, large and small. The upper layers are frequently overturned by small
impacts, whereas deeper layers are overturned less frequently by large impacts. This
mode of formation explains the frequent observation of layers of widely varying thickness
observed in the cores returned from the Moon. On the Moon most crater ejecta travels
only a few times the diameter of the crater that created it and it nearly all falls back to the
Moon somewhere. On smaller bodies such as asteroids, the lower escape velocity means
that more material may be ejected into space and so growth rates may be smaller for a
given impact flux.
Because the regolith is composed of multiply recycled crater ejecta, material in the
regolith is strongly affected by the cratering process. On the Moon the average impact
velocity is high, approximately 17 km/s, so that melting and vaporization are common.
The lunar regolith is filled with numerous fractured rock fragments, impact-generated
glass and vapor-deposited coatings. A common component is “agglutinate” consisting of
angular rock fragments welded together by irregular masses of glass. On average, aggluti-
nates account for about 30% of the lunar regolith. Agglutinates are typically smaller than
about 1 mm in size. Agglutinate glass itself is riddled with 1 to 10 nm blobs of metallic
iron. This “nanophase” iron is largely responsible for the optical darkening and redden-
ing characteristic of the lunar surface. Its origin has been much debated: Nanophase iron
was originally attributed to chemical reduction of ferrous iron Fe2+ in silicate minerals
by solar-wind-implanted hydrogen. More recently it appears that impacts (and impact-
analog laser-evaporation experiments) can reduce iron directly, so that no reducing agent
is necessary, although the recent observation of OH and H2O produced by the solar-wind
plasma argues that at least some chemical reduction must occur.
Mature lunar regolith is in equilibrium between impacts that simultaneously break down
bedrock and build up agglutinate from the broken fragments. The regolith’s grain size dis-
tribution is apparently in a stable balance between the processes that break the regolith into
smaller fragments and agglutinates that weld them together into larger particles. The aver-
age grain size of a mature lunar soil is about 60 μm. Less mature soils have a larger average
grain size and contain fewer agglutinate particles. They are spectrally bluer and mineral
absorption bands are more distinct.
Soil maturity is often gauged remotely by brightness and color: Dark, spectrally red
regolith is mature, whereas bright, blue surfaces are immature. Images of the lunar surface

EBSCOhost - printed on 11/3/2023 9:57 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
286 Regoliths, weathering, and surface texture

typically reveal dark, red intercrater areas interspersed with the bright, blue rays surround-
ing fresh craters that have penetrated the regolith to fresh bedrock. This same pattern is
revealed on asteroids, although with some puzzling differences: Images of the main-belt
asteroid 243 Ida by the Galileo probe revealed a color contrast between old surfaces and
fresh crater ejecta, but albedo contrasts are low. On the other hand, NEAR images of
the near-Earth asteroid 433 Eros show large albedo differences but low color contrasts.
Carbonaceous asteroid 253 Mathilde seems to be uniformly dark and does not exhibit
either albedo or color contrasts, perhaps due to its large carbon content. At the moment,
asteroid regoliths are not well understood. It is presumed that some combination of low
gravity and lower impact velocities in the asteroid belt (typically about 5 km/s) results in a
very different evolution of the regolith on these bodies, but a clear understanding may have
to await sample returns from asteroid surfaces.

7.1.3 Radiation effects on airless bodies

In addition to the impacts of solid bodies, the surfaces of airless bodies are exposed to ioniz-
ing radiation from several sources. The Sun emits X-rays and ultraviolet light that can ion-
ize atoms and induce chemical reactions among surface materials. Although this radiation
does not penetrate deeply, it can cause radiolysis of ices and carbonaceous compounds.
While such changes have been much studied in the context of planetary atmospheres, their
effects on surface chemistry are less well known, although the red color of Martian soil
has been attributed to oxidation by ultraviolet light that penetrates its thin atmosphere. The
positive indications of life by the Viking GCMS experiment are now attributed to peroxides
created on surface minerals by the intense ultraviolet radiation bathing Mars’ surface. The
present surface environment of Mars is extremely hostile to life because of this radiation
and the highly oxidized chemical species it generates.
In addition to X-rays and ultraviolet light, the surfaces of airless, or nearly airless, bod-
ies are affected by a variety of ionizing particle radiations. In order of increasing particle
energy, there is solar-wind plasma, solar-flare radiation, and galactic cosmic rays. Most of
this particle flux is composed of hydrogen nuclei (protons) and electrons, but an important
fraction is nuclei of heavier elements. Its depth of penetration depends critically on par-
ticle energy and varies from microns to meters. In addition to the primary radiation flux,
energetic particles also generate neutrons below the surface that can induce further nuclear
reactions.
The solar wind consists of ionized gas ejected from the Sun’s corona that impinges on
the surfaces of exposed bodies with energies between 0.3 and 3 keV/nucleon. It penetrates
only microns into the surfaces of exposed mineral grains, but can induce sputtering and
implant gases below their surfaces. Large fluxes of the solar wind may destroy the crystal-
line surfaces of minerals and create an amorphous rim. The solar wind is the main source
of the elements H, C, N, and the noble gases that are otherwise very rare in the lunar rego-
lith. Solar flares occur sporadically, but an individual event may emit large fluxes (up to 106
protons/cm2-s) of 1–100 MeV/nucleon nuclei from the Sun’s atmosphere. These particles

EBSCOhost - printed on 11/3/2023 9:57 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
7.1 Lunar and asteroid regoliths 287

105
0.5-10 MeV
10-100 MeV

Flux (particles/m2-s)
104
Thermal
neutrons
0.1-1 GeV

103

E>1 GeV/amu

102
0 1000 2000 3000
Mass depth (kg/m2)

Figure 7.6 Fluxes of galactic cosmic rays as a function of depth (here measured in terms of mass
above the given depth per unit area, kg/m2: 2 m of lunar regolith is about 3000 kg/m2). This plot
indicates both primary and secondary radiation. Thermal neutrons, in particular, are only created by
collisions of the primary radiation with nuclei in the regolith. After Fig. 3.20 of Herken et al. (1991).

penetrate about 1 cm into the lunar surface and produce visible tracks of displaced atoms
through mineral crystals, but do not create much secondary radiation.
Galactic cosmic rays are the most energetic and penetrating radiation that impinges on
the surface of an airless body. Energies range from about 0.1 GeV/nucleon to above 10
GeV/nucleon (rarely, much more than this) and penetration depths are measured in meters
for the lighter nuclei. Heavy nuclei in the cosmic-ray flux penetrate only centimeters, but
because they cause a great deal of ionization they severely damage mineral structures, not
to mention electronic components of spacecraft. These energetic particles produce abun-
dant secondary radiation when they strike nuclei below the surface, creating cascades of
neutrons and other exotic particles. A consequence of this cascade is that the total radiation
exposure at some depth below the surface is larger than that at the surface itself. The radi-
ation dose peaks at a depth of about 1000 kg/m2 and declines at greater depths (Figure 7.6).
Galactic cosmic rays produce visible tracks in minerals and their secondary neutrons induce
nuclear reactions that create a suite of exotic isotopic species. These isotopes are useful for
dating regolith exposure ages and overturn rates.
The direct impingement of ionizing radiation on airless bodies has proved very use-
ful for remote sensing of their surface composition. X-ray fluorescence has been used to
map elemental abundances on the lunar and asteroid surfaces. X-rays emitted from atoms
excited by particle radiation are similarly used as a kind of natural microprobe to determine
the identity and abundance of different elements. The neutrons and gamma rays emitted by
secondary nuclear reactions have revealed the presence and distribution of many elements

EBSCOhost - printed on 11/3/2023 9:57 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
288 Regoliths, weathering, and surface texture

in the regolith from orbit around the Moon, Mars, and the asteroid Eros. Slow neutrons
thermalized by hydrogen in the surface of Mars and the Moon permit mapping the dis-
tribution of water in the upper few decimeters beneath their surfaces. Similar instruments
aboard the Mercury MESSENGER mission will shortly be used to map elemental abun-
dances on that planet. This project was delayed during the initial flybys of Mercury due
to an unexpected lack of solar activity that reduced the Sun’s X-ray flux below the level
needed for clear detection by the fluorescence experiment.
The Galilean satellites of Jupiter are immersed in the most intense radiation environ-
ment in the Solar System, with surface radiation doses 100 to 1000 times more intense than
experienced by our Moon (Johnson et al., 2004). Plasma and energetic particles trapped in
Jupiter’s magnetosphere, along with solar ultraviolet radiation, strongly affect the surfaces
of these satellites down to depths of about a centimeter. Impact gardening then churns this
material into their meters-deep regoliths. Radiation damage of ice grains creates thin lay-
ers of amorphous ice that are detected spectrally, while the radiation itself breaks chemical
bonds producing exotic chemical species. Considerable research has gone into understand-
ing the effects of radiation on water ice. The predominant process is radiolysis of the water
molecules; ultimately producing oxidized species such as H2O2, HO2, and O2 after H is lost
from the surface by diffusion into space. Carbon, either implanted from the radiation itself
or created by decomposition of organics present in the surface, yields CO2, while sulfur is
oxidized to SO2. Intense radiation seems to brighten the surfaces of the icy satellites, per-
haps by surface condensation of water molecules sputtered from ice grains. An intriguing
suggestion is that Ganymede’s bright “polar cap” may be created by radiation funneled to
its surface by its internal magnetic field. This would explain both the coincidence of the
bright cap with the location of open magnetic field lines and the otherwise puzzling fact
that Callisto, which lacks an internal magnetic field, also lacks a bright polar cap.
Io’s volcanoes eject considerable amounts of sulfur, some of which is incorporated into
the plasma of Jupiter’s magnetosphere. This sulfur is then swept out to Europa and beyond,
where it preferentially impacts the trailing hemisphere of the outer Galilean satellites and is
implanted in their surfaces. The result is a clear “bull’s-eye” ultraviolet spectral anomaly on
Europa that is also observed on Ganymede and Callisto. In addition to the trailing hemisphere
sulfur implanted from the magnetosphere, there are other spectral indications of hydrated
sulfur compounds, especially on Europa. It is debated whether the origin of this sulfur sig-
nature is due to radiation-induced sulfuric acid or endogenic sulfate salts from Europa’s
interior ocean. At the moment the exogenic/endogenic origin debate is unresolved.

7.2 Temperatures beneath planetary surfaces

The surface temperature of a planet is one of the most important facts about its surface
environment. It is one of the first things that most people want to know about another
planet. Temperature sets the stage for which elements and chemical compounds are pre-
sent, in what form (solid, liquid, or gas), and the rates of chemical and physical changes
that may occur. Planetary habitability largely depends on the occurrence of surface tem-
peratures in the range where liquid water is possible.

EBSCOhost - printed on 11/3/2023 9:57 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
7.2 Temperatures beneath planetary surfaces 289

Table 7.1 Effective temperatures of the planets

Distance from Sun Effective temperature

Planet (AU) Albedo (K)

Mercury 0.39 0.058 442

Venus 0.72 0.71 244
Earth 1.00 0.33 253
Mars 1.52 0.17 216
Ceres 2.77 0.09 164
Jupiter 5.20 0.73 87
Saturn 9.54 0.76 63
Uranus 19.19 0.93 33
Neptune 30.06 0.84 32
Pluto 39.53 0.58 (average) 32
Eris 67.67 0.86 21

The approximate surface temperature of a planet is set by the balance between solar
radiation received and thermal radiation emitted. A standard calculation (Houghton, 2001)
leads to the “effective” temperature estimate:
1/ 4
 (1 − A) L 
Teff =  2 
(7.2)
 16 π σ R 
where L is the solar luminosity, the planetary albedo A measures the fraction of solar
radiation absorbed, R is the distance from the Sun and σ is the Stefan–Boltzmann constant.
This crude estimate assumes that the planet’s surface temperature is uniform and ignores
the effect of an atmosphere. Atmospheric modifications may be profound and are the sub-
ject of a huge literature of their own, but are outside the scope of this book. A good, con-
cise, introduction to this literature is Houghton (2001).
Planetary effective temperatures range from a high of 442 K on Mercury down to 32
K on Pluto (Table 7.1), falling roughly as the inverse square root of distance from the
Sun. Albedo plays a large role: Even though the effective temperature of Mars is a chilly
216 K, the nearly black surface of comet Tempel 1 registered a sweltering 300 K well out-
side the orbit of Mars when it was visited by the Deep Impact spacecraft in 2005. Earth
itself is pretty cold in terms of effective temperature: Its current habitable state is largely
due to atmospheric greenhouse warming.

7.2.1 Diurnal and seasonal temperature cycles

The effective temperature estimate is a global average that ignores diurnal, seasonal, and
equator-to-pole temperature differences. These differences can be profound: In spite of
Mercury’s very high effective temperature, polar temperatures are estimated to fall as low

EBSCOhost - printed on 11/3/2023 9:57 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
290 Regoliths, weathering, and surface texture

as 167 K, making it likely that ice is stable in high-latitude craters (Paige et al., 1992).
Mercury’s 59-day rotational period and the peculiar 2:3 resonance between its rotational
and orbital periods yield further extremes between different longitudes along its equator.
Daily maximum temperature estimates range from about 570 K at an equatorial “warm
pole” to 700 K at the “hot poles” located 90° in longitude from the warm poles (Morrison,
1970). Minimum temperatures drop to about 100 K at all longitudes. On the Moon, diurnal
(27-day) variations in temperature range from a high of 374 K down to 92 K at the Apollo
15 site (Heiken et al., 1991). Polar temperatures in shadowed lunar craters may fall as low
as 40 K.
Surface temperature cycles on planetary surfaces vary enormously in duration. On Earth
we are accustomed to the 24-hour day/night temperature fluctuation, but because of the
Earth’s obliquity we also experience large yearly seasonal cycles. Climatic oscillations
vary on timescales that range from tens of thousands of years (the waning of the last ice
age) to millions of years. Other planets experience a similar wide range of timescales: The
Martian day is very similar in length to the Earth’s and its seasons are approximately twice
as long, but its layered polar deposits suggest climatic cycles due to obliquity variations
with periods of millions of years or even longer when chaotic orbital variations are taken
into account (Carr, 2006). Seasonal cycles on Saturn’s large moon Titan last 29 yr, 165 yr
on Neptune’s moon Triton, and 249 yr on Pluto’s Charon. The high obliquities of Uranus
and Pluto suggest that seasonal temperature swings are larger than diurnal alternations.

7.2.2 Heat transfer in regoliths

Computation of subsurface temperatures from surface temperatures is a straightforward
process using Fourier’s heat conduction Equation (4.13). The major practical difficulty is
knowing the thermal conductivity of the surface layers, which can be very complex in a
realistic regolith, because it depends on composition, grain size, grain packing, the pres-
ence of atmospheric gases or liquids, and the temperature (Presley and Christensen, 1997).
In the idealized case of a harmonic surface temperature variation at the surface of a half-
space of uniform thermal conductivity, the temperature as a function of depth is given by
(Carslaw and Jaeger, 1959, p. 64ff.):
2π z
−   t z 
T ( z, t ) = Tm + A e λ
cos 2π  −   (7.3)
  P λ
where the thermal “skin depth” λ is given by:

λ = 4π κ P (7.4)

where κ is the thermal diffusivity. The surface temperature fluctuates about the mean tem-
perature Tm with amplitude A and period P. The origin of time t is chosen so that the surface
temperature is given by T(0) = Tm + A cos (2π t/P). This equation ignores the slow increase
of temperature with depth due to internal heat flow.

EBSCOhost - printed on 11/3/2023 9:57 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
7.2 Temperatures beneath planetary surfaces 291

Temperature, T/Tm
Tmin Tm Tmax Tmin Tm Tmax Tmin Tm Tmax
0.0
Depth, z/λ

0.5

1.0

(a) (b) (c)

Figure 7.7 Temperature in the lunar regolith as a function of time, computed from the linear diffusion
equation. The temperatures in the left panel are plotted at noon, those in the middle panel at dawn
or dusk, and those in the right panel are plotted at midnight. The dashed lines show the envelope of
the temperature extremes as a function of depth. Note that the mean temperature does not vary with
depth.

Equation (7.3) indicates that the amplitude of the temperature fluctuation decreases
exponentially with increasing depth below the surface (Figure 7.7). The surface tem-
perature variation not only decreases in amplitude going deeper, but the time at which
it reaches maximum is progressively delayed. The depth to which the thermal wave pen-
etrates depends strongly on the period P of the thermal fluctuations: Rapid temperature
variations remain in the shallow subsurface, while long-period variations penetrate deeply.
Thus, diurnal temperature variations on the Earth may penetrate only centimeters into the
soil, while seasonal variations are appreciable at depths of meters. It is interesting to note
that terrestrial heat-flow measurements in boreholes hundreds of meters deep must take the
cold temperatures of the last ice age into account, or errors will result: The 10 000-year
thermal wave is just now reaching this depth. Rapid erosion or sedimentation must likewise
be considered in many such measurements (an example is the USGS borehole in Cajon
Pass, CA, which initially gave erroneously high estimates of the heat flow because rapid
erosion at that location was neglected).
Handbook tabulations of thermal conductivity k = ρcPκ may suggest that it is approxi-
mately constant; however, this is far from the truth. Thermal conductivity depends upon the
mineral composition of the surface, porosity, grain size and sorting, presence of included
gases, and temperature (Presley and Christensen, 1997). Temperature dependence is espe-
cially evident in the lunar regolith where, at high temperatures, heat is transferred through
its pores mainly by radiation, not grain-to-grain conduction. The thermal conductivity of
the lunar regolith is usually fit by an equation of form (Langseth et al., 1973):
k(T) = kc + krT 3 (7.5)
where kc is (approximately) temperature-independent and kr measures radiative transfer
within the regolith. The strong temperature dependence of Equation (7.5) leads to a phe-
nomenon known as “thermal rectification.” When thermal radiation can be ignored, the

EBSCOhost - printed on 11/3/2023 9:57 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
292 Regoliths, weathering, and surface texture

Temperature, T/Tm
Tmin Tm Tmax
0

Depth, z/λ

Figure 7.8 Temperature in the lunar regolith for temperature-dependent thermal conductivity that is
dominated by radiation transfer. The heavy line indicates the temperature at noon. Note that the mean
temperature rises sharply with increasing depth, reflecting the process of “thermal rectification.” The
temperature extremes (dashed) are also strongly asymmetric, in contrast to Figure 7.7.

heat conduction Equation (4.13) is linear and the average temperature below the surface is
the same as the average temperature at the surface itself. However, the T3 term in Equation
(7.5) makes the conduction equation non-linear and the solution (7.3) no longer applies.
Thermal conductivity is high when the temperature is high, allowing heat during the day
to penetrate into the surface, but drops to low values at night, turning the regolith into an
insulator that retains much of the day’s heat (Figure 7.8). The net result is to raise the aver-
age temperature below the loose surface layer by as much as 40–45 K. This shift in average
temperature complicated measurement of the lunar heat flow from the shallow Apollo 15
and 17 heat flow probes. Geophysicists and any future lunar colonists must take this shift
in average temperature into account when computing the temperature below the surface of
the Moon. At temperatures near absolute zero, thermal conductivity also depends strongly
on temperature, so that care must be taken in computing subsurface temperatures of very
cold bodies far from the Sun.
A phenomenon closely related to thermal rectification is the “solid-state greenhouse.”
This concept was invented by Bob Brown and Dennis Matson (1987; Matson and Brown,
1989), who realized that on bright, high-albedo bodies like Europa, sunlight is not absorbed
right at the surface but penetrates some distance into it. The Sun’s heat is thus absorbed a
few centimeters below the surface, where it must reach the surface by conduction before
being radiated back into space. It is estimated that this effect may account for about a 10 K
rise in temperature over the case where sunlight is absorbed right at the surface. This effect
is important for any bright surface that does not readily absorb sunlight, so it may also
affect the near-surface temperatures of snowpacks or icy polar deposits.

EBSCOhost - printed on 11/3/2023 9:57 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
7.2 Temperatures beneath planetary surfaces 293

Heat and material may also be transported by advection in the regoliths of bodies with
substantial atmospheres. On the Earth, fresh snow is rapidly metamorphosed by the advec-
tion of water vapor within a snowpack. Water vapor evaporates from sharp-edged snow-
flakes whose small radii of curvature destabilize molecules on their surface. This water
recondenses into larger crystals within the mass of snow. It also migrates readily from
warmer to colder regions of the porous mass of snow, transferring heat and accumulating
into crystalline masses known as hoar frost. Known as “firnification,” this process is an
important step in the conversion of snow into ice in snowpacks and glaciers (Shumskii,
1964). On Mars, the seasonal cycling of atmospheric pressure drives atmospheric gases
into and out of the soil, strongly affecting the distribution of volatile species, such as water
vapor, in the near-surface regolith (Haberle et al., 2008). Such vapor-phase advection may
play a role in the distribution of ground ice on Mars, as well as having a major effect on
near-surface heat transport.

7.2.3 Thermal inertia

The response of different materials to temperature changes depends strongly upon their
thermal conductivity, density, and heat capacity. Thus, solid rocks lying on a planetary
surface are slow to warm up after sunrise, while loose, powdery soil heats up rapidly.
Conversely, after sunset rocks retain their heat for a long time compared to fine-grained
powders. This observation has led to the development of thermal emission spectroscopy,
which enables orbiting spacecraft (or Earthbound observers) to estimate grain sizes and
dust/rock ratios of planetary surfaces remotely, utilizing thermal infrared (IR) emissions
(Mellon et al., 2008).
The principal parameter that determines the rate at which a surface warms (or cools) is
its thermal inertia, I, which is defined as:

I= k ρ cP . (7.6)

This parameter controls how much heat energy is gained or lost from a surface during
changes of temperature. Its units are rather complex: J/m2-K-s1/2. Table 7.2 lists some rep-
resentative values of thermal inertia for different materials and surfaces.
Spacecraft observations of temperature changes in response to insolation variations are
most readily interpreted in terms of thermal inertia itself. Conversion of these observations
to composition, grain size, or surface density is highly model-dependent. Nevertheless,
such measurements have proved invaluable in sorting out the nature of different terranes on
Mars as well as other bodies, such as the Galilean satellites of Jupiter, where the thermal
response of their surfaces to eclipses has long been used to infer their nature.

7.3 Weathering: processes at the surface/atmosphere interface

Planetary surfaces are geological battlefields: They are the frontier between a planet’s
interior and its atmosphere or, in the case of airless bodies, open space. They are assaulted

EBSCOhost - printed on 11/3/2023 9:57 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
294 Regoliths, weathering, and surface texture

Table 7.2 Thermal inertia of selected materials

Density Heat capacity, cP Thermal conductivity, k Thermal inertia, I

Material (kg/m3) (J/kg-K) (W/m-K) (J/m2-K-s1/2)

Water ice (273 K) 915 1960 2.3–3.6 2030–2540

Water icea (173 K) 926 1389 3.48 2116
Dense snowb (273 K) 540 2100 0.46 722
New snow (273 K) 200 1962 0.078 174
CO2 ice (150 K) 1560 1083 0.6 1006
Basalt 2900 830–900 1.12–2.38 1642–2492
Tuff 1980 850 0.91–3.20 1237–2320
200 μm sand @ 6 mb 1650 850 0.044 248
10 μm sand @ 6 mb 1375 850 0.01 108
1 μm dust @ 6 mb 1100 850 0.004 61

Data based on Table 18.1 of Mellon et al. (2008).

a
Engineering Toolbox (http://www.engineeringtoolbox.com/).
b
Data from Clark (1966).

by meteor impacts, ionizing radiation, rapidly varying temperatures, and, especially for
the Earth, corrosive gases and reactive liquids (mainly oxygen and water). Rocks erupted
or exhumed from the deep interior come into contact with the atmosphere, entering a very
different chemical environment with which they are seldom in equilibrium.
The events that transpire in this environment are described by the term “weathering.”
Weathering transforms rocks from structures and compositions that are at home in the
planet’s interior to new forms that can survive for long periods of time in the surface envir-
onment (Figure 7.9). Depending upon one’s point of view, these events are either destruc-
tive or creative. The destructive aspect was most strongly emphasized early in the history of
geology (“The Earth in Decay” was a favorite theme of early geomorphologists), whereas
more modern authors look upon weathering as the source of our agricultural soil. The
lunar regolith is regarded as a potential resource and the Martian regolith may harbor life.
Indeed, the very origin of life probably required an active, kaleidoscopically changing
environment that only occurs on planetary surfaces.
Weathering processes are traditionally divided into “chemical” weathering, which
emphasizes chemical changes, and “physical” weathering that deals mainly with macro-
scopic cracks and changes in mechanical properties. Although this traditional division
is respected in this book, there are many processes, such as granular disintegration of
rocks caused by swelling of minerals undergoing chemical reactions, which cut across
the boundaries of this division. Most chemical reactions would not occur at geologically
significant rates if reactants could not penetrate rock masses through cracks created by
physical weathering. Classifications are useful, but should not obscure the fact that both
chemical and physical processes play essential roles in the overall group of processes that
constitute weathering.

EBSCOhost - printed on 11/3/2023 9:57 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
7.3 Weathering 295

}
} A
soil
} B

} C
decayed
bedrock

}
corestone

unweathered
R
bedrock

Figure 7.9 Schematic of a weathered soil profile showing the A, B, C, and R (unweathered bedrock)
horizons. Slightly weathered corestones are mixed with fine-grained, heavily weathered material in
the C horizon.

7.3.1 Chemical weathering

Chemical weathering focuses mainly on the chemical changes that occur in minerals in
the near-surface environment. The interiors of all of the terrestrial planets appear to be
chemically reduced compared to their atmospheres, so minerals from the interior typically
become oxidized in the surface environment. Although there is a long list of elements with
variable oxidation states that may be altered by surface exposure (C, P, S, V, Mn, Fe, Co,
Ni, and Cu, plus others), the most visible and distinctive changes occur for iron, which
may occur in the metallic Fe0 state, as ferrous Fe2+, or ferric Fe3+ ions. The ferrous/ferric
transformation is particularly significant, as minerals containing ferrous iron are typically
black or green to the human visual system and the ion is highly soluble in water, whereas
minerals incorporating the ferric ion are typically red or yellow (the Martian soil is a prime
example) and the ion is very insoluble.
We typically think of oxygen O2 as the eponymous oxidizer; however, H2O, CO2, and
SO2 are also significant oxidizers in different planetary environments (as are UV and ion-
izing radiation). Table 7.3 summarizes the atmospheric composition, surface pressure, and
temperature of the terrestrial planets, Titan, and Triton. Titan’s nitrogen/methane atmos-
phere is the only truly reducing environment.
Planetary atmospheres are not passive backgrounds to the chemical drama that unfolds on
their surfaces. The reactions that atmospheric gases undergo with surface rocks profoundly
affect the composition of the atmosphere itself, often determining the abundance of many
atmospheric species. Thus, the long-term abundance of CO2 in the Earth’s atmosphere may
be controlled by weathering reactions with rocks erupted from its interior. Earth’s oxygen
is almost entirely the result of photosynthesis by plants on its surface and many other minor
species are in equilibrium with the surface through planetary geochemical cycles that, on
Earth, are usually part of the biosphere. The abundance of CO in the atmospheres of Venus

EBSCOhost - printed on 11/3/2023 9:57 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
296 Regoliths, weathering, and surface texture

Table 7.3 Near-surface atmospheres in the Solar System

Temperature Pressure Major atmospheric Chemically active

Body (K) (MPa) constituents minor constituents

Venus 750 9.7 CO2 (96.4%) N2, SO2, H2O, CO

Earth 288 0.1 N2, O2 (78.1%, 21%) H2O, CO2
Mars 250 0.0008 CO2 (95.3%) N2, O2, H2O, CO
Titan 94 0.15 N2 (95%) CH4
Triton 38 1.4–1.9 × 10–6 N2 CH4 (trace)

and Mars may be determined by weathering reactions at the surface, and on Titan ephem-
eral lakes of liquid methane buffer the methane content of its atmosphere.
The detailed atmospheric composition of a planet is, thus, a reflection of weathering
reactions on its surface. The first indication that life is present on distant planets around
other stars may be spectroscopic detection of chemical species indicative of biological
activity. Closer to home, a recent report of methane detected over the northern hemisphere
of Mars has given rise to much excitement over its possible biological origin.
Expositions of chemical weathering unfortunately tend to devolve into rather dull lists
of chemical reactions that the student is expected to memorize by rote. The exposition
that, in my opinion, best avoids this pitfall is a rather obscure pamphlet published in 1957
(Keller, 1957). A more recent book (Bland and Rolls, 1998) does a good job of explaining
the background, but both of these treatments are entirely Earth-focused. There is not yet an
extensive description of chemical weathering applicable to the other planets, although indi-
vidual book chapters (Fegley et al., 1997; Gooding et al., 1992; Wood, 1997) have made a
start on this difficult subject (which, admittedly, is not yet constrained by much data). Your
author, along with everyone else, has not found a better way to present this topic, but in the
following section I attempt to put each set of reactions into its most appropriate planetary
context. The classes of reaction are discussed in rough order of their Gibbs energy change
and, thus, in some sense, of the ease with which they occur. This is not an exhaustive list of
weathering reactions, which would be very long and not very informative, but an exhibition
of the most characteristic ones.
The basic weathering reactions are solution, hydration, hydrolysis, and oxidation, along
with a few reactions that include carbonation and sulfonation, in rough order of their
Gibbs energy change. The values of the reaction Gibbs energy are not, however, listed here
because they depend on the pressure, temperature, and, most importantly, on the activities
(hence, concentrations) of gaseous or soluble species. These numbers are, thus, highly
sensitive to the particular environment in which the reaction occurs. The interested reader
is advised to consult standard tables of Gibbs energy (the NIST tables, Chase, 1998, for
example) to determine accurate values for reaction energies.
One of the lowest-energy reactions with water is simple solution. Ionic crystals, such
as halite, NaCl, dissolve readily in water because the polar water molecules neutralize the
ions’ electric charges.

EBSCOhost - printed on 11/3/2023 9:57 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
7.3 Weathering 297

NaCl  Na+ (aq) + Cl−(aq). (7.7)

This process allows water to dissolve a large number of other ionic salts, such as MgCl2
or MgSO4, although halite is one of the most soluble. Waters containing large amounts of
dissolved salts are known as brines and they play a major role in terrestrial geochemistry.
Brines may also be present beneath the surface of Mars and in Europa’s subsurface ocean.
A slightly more complex process that involves water and CO2 acting together accounts for
the dissolution of limestone:
CaCO3 + H 2O + CO2 → Ca 2+ + ( HCO3 )22.− (7.8)

Extensive karst terranes on Earth, where limestone rocks have been dissolved by
rainwater to form pits, hollows, and underground caves, are the result of this reaction.
This apparently simple reaction has many subtleties, depending on the concentration
and temperature of the water. A large body of literature exists on karst, which connects
the morphologic expression of these terranes to the chemistry of carbonate and water
(White, 1988).
Another low-energy reaction with water, not normally listed as a weathering reaction,
is clathrate formation. Non-polar molecules of CO2, CH4, NH3, and many others assem-
ble a cage of water molecules about themselves at low temperatures and form distinctive,
often stoichiometric, phases in ice. These cold mineral phases may play important roles in
the trapping and release of gases in icy bodies far from the Sun. Clathrate densities differ
from that of pure water ice and so may also play a role in the tectonics and volcanism on
icy satellites.
A slightly more energetic reaction with water results in a stoichiometric association
between a mineral and water molecules. This reaction, called hydration, is readily revers-
ible by heat at temperatures of 600–700 K. A common terrestrial reaction with anhydrite
(calcium sulfate) produces the evaporite mineral gypsum:
CaSO4 + 2H2O  CaSO4  2H2O. (7.9)
Hydrolysis, in which the water molecule dissociates and forms a new -OH bond with
the reacting species, is distinct from hydration, in which the water molecule remains intact.
Hydrolysis is illustrated here by the reaction of water with minerals common in igneous
rocks. We first consider a reaction that, until recently, was thought to be an example of
hydration. The iron mineral hematite reacts readily with water to form goethite:
Fe2O3 + H2O  2FeO(OH). (7.10)
A complex mixture of goethite, hematite, and variable adsorbed water forms limon-
ite, which imparts a yellow color to many types of sediment. Limonite is very stable in
the Earth’s surface environment and is considered the endpoint of weathering of iron-rich
rocks.
The second example of hydrolysis focuses on orthoclase, a feldspar common in granitic
rocks:
KAlSi3O8 + H2O → HAlSi3O8 + KOH(aq). (7.11)

EBSCOhost - printed on 11/3/2023 9:57 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
298 Regoliths, weathering, and surface texture

The hydrogenated silicate created by this reaction undergoes further reactions with water
to eventually form the clay mineral kaolinite Al2Si2O5(OH)4 while silica, SiO2, is released
into solution. Kaolinite is another of the end products of weathering in the terrestrial envir-
onment. The silica and potassium are carried off in water, later to precipitate or add to the
ionic content of briny waters.
A third, more complex, example of hydrolysis focuses on the mineral olivine. Olivine
is a major component of planetary mantles, chondritic meteorites, and an import-
ant constituent of many types of basalt. It typically occurs as a solid solution of the
iron and magnesium endmembers, Fe2SiO4 and Mg2SiO4. The complex suite of reac-
tions with water known as serpentinization begins with the hydrolysis of the iron-rich
endmember:
3Fe2SiO4 + 2H2O → 2Fe3O4 + 3SiO2 + 2H2. (7.12)
This reaction is coupled with a similar reaction of the magnesium-rich endmember that
absorbs the silica produced in the first reaction:
3Mg2SiO4 + SiO2 + 4H2O → 2Mg3Si2O5(OH)4. (7.13)
The final product, serpentinite, is closely related to kaolinite and is also a terminal weath-
ering product. Note that reaction (7.12) releases hydrogen gas. This hydrogen may react
further with carbon in the environment, reducing it to methane, CH4. This is an abiogenic
source of methane, one that has been suggested to account for the observation of methane
in the atmosphere of Mars without invoking biological activity.
Oxidation is the quintessential reaction on Earth. One-fifth of our atmosphere is cur-
rently oxygen gas, a biological by-product of photosynthesis. Oxygen is not only highly
reactive with rocks from Earth’s interior, it is also highly toxic to life itself. Early photosyn-
thetic organisms excreted it as a waste product. As these organisms became more abundant
and the concentration of oxygen rose in the early Earth’s atmosphere, complex biological
pathways developed to protect organisms from their own pollutants. Animals now require
oxygen as a high-energy chemical propellant, but the cells that use it must still handle it
carefully and keep it out of most of their biochemical machinery.
A typical reaction is the oxidation of wüstite, FeO, to hematite. Although wüstite itself is
a rare mineral, FeO is a common component of iron-containing igneous rocks and miner-
als, so that the reaction is representative:
4FeO + O2 → 2Fe2O3. (7.14)
Notice that the iron in this reaction changed from the +2 oxidation state in wüstite to the
+3 state in hematite.
Oxygen itself is not the only atmospheric gas capable of oxidizing iron. On Venus the
stable form of iron is believed to be the black mineral magnetite, Fe3O4, which contains a
mixture of iron in the +2 and +3 oxidation states. FeO in minerals erupted from the interior
of Venus is oxidized by reaction with CO2:
3FeO + CO2  Fe3O4 + CO. (7.15)

EBSCOhost - printed on 11/3/2023 9:57 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
7.3 Weathering 299

This reaction is postulated to contribute to the small abundance of CO observed in

Venus’ atmosphere.
Carbon dioxide can also react with other minerals without changing their oxidation
states. Such reactions are known as carbonation, for example the iron-rich olivine fayalite
reacts with carbon dioxide to produce siderite plus silica:
Fe2SiO4 + 2CO2 → 2FeCO3 + SiO2. (7.16)
Siderite is a dark, dense mineral (density 3800 kg/m3) that sometimes forms concretions
in sedimentary rocks.
Sulfur plays a large role in the surface weathering on Venus. Carbonates do not appear to
be stable on Venus, but react with SO2 in the atmosphere to form anhydrite:
SO2 + CaCO3  CaSO4 + CO. (7.17)
Pyrite, FeS2, from igneous rocks erupted on the surface of Venus, may play an important
role in its sulfur cycle as well, through reactions of type:

3FeS2 + 2CO + 4CO2  6COS + Fe3O4. (7.18)

The gas COS has been detected in the lower atmosphere at the level of about 28 ppm.
The overall importance of these reactions is currently contentious (Fegley et al., 1997;
Wood, 1997), although it seems likely that anhydrite and magnetite are two major weather-
ing products likely to be present on Venus’ surface.
A major mystery on Venus is the nature of the “snow” detected on the tops of Venusian
mountains. Radar images show that terrain above 4 km elevation is highly reflective, remin-
iscent of snow-capped mountains on Earth. Snow itself is not a possibility in view of Venus’
high surface temperatures (indeed, the best fit to the data was half-seriously identified as
the metal, Te). Various suggestions, such as pyrite deposited in the vapor phase by FeCl2 or
deposition of volatile metals, have been suggested, but no consensus has yet been reached
on this puzzle.
An important reaction that has not yet been mentioned is the so-called “Urey reaction.”
First proposed by Nobel prize-winning chemist Harold Urey (1893–1981) in his 1952 book
The Planets (Urey, 1952), this reaction continues to play a major role in discussions of
planetary habitability and the evolution of the Earth’s atmosphere and climate. Urey noted
that the abundance of carbon dioxide might be controlled by reaction with silicate rocks. He
illustrated his reaction with the calcium silicate mineral wollastonite, CaSiO3. Wollastonite
is not particularly common, but similar reactions apply to many common minerals and we
illustrate the reaction here with Urey’s own example:

CaSiO3 + CO2  CaCO3 + SiO2. (7.19)

This reaction occurs at appreciable speed only when liquid water is present, in which the
carbon dioxide forms bicarbonate ions and the acid water attacks the silicate.
The vast majority of Earth’s carbon dioxide is locked up in limestone, which, at present,
is mostly deposited by oceanic organisms. Earth and Venus have similar complements of

EBSCOhost - printed on 11/3/2023 9:57 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
300 Regoliths, weathering, and surface texture

carbon dioxide, but at Venus’ high surface temperatures, reaction (7.19) has proceeded far
to the left. At the lower temperatures prevailing on Earth this reaction proceeds to the right
and would have removed all of the CO2 from the Earth’s atmosphere but for plate tectonics,
which constantly releases the CO2 from subducted sediments and creates a stable balance
over a billion-year timescale (Kasting and Catling, 2003; Kasting and Toon, 1989).

7.3.2 Physical weathering

Physical weathering is mostly about cracks and joints. However, chemical considerations
are always present. Chemical reactions cannot occur if the reactants cannot meet, which
is why weathering reactions usually involve a liquid or gaseous phase. However, even
gases cannot react with a mineral they cannot reach. Cracks at both the macroscopic
and microscopic scales provide the means through which weathering agents reach their
targets.
Stress corrosion cracking. The growth of cracks themselves, however, often requires
chemical assistance. A small initial flaw or microcrack can grow only when the stress at its
tip exceeds the strength of the material. Regional stresses are concentrated and, thus, amp-
lified at crack tips, but intact rock or minerals are often too strong for the crack to propa-
gate. However, if chemically reactive fluids are present, they can move through the crack to
its tip. Reaching the already-stressed bonds of the host mineral, they preferentially attack
these bonds, weakening them and permitting the crack to propagate a little deeper into the
intact rock. This process, known as stress corrosion cracking (Atkinson, 1987), proceeds
slowly, but often leads to a great reduction in the effective strength of a material. Bridges
and buildings that, if kept dry and clean would never collapse, crumble under the onslaught
of rain and polluted air.
Rock fracture is a complex process that has its own extensive literature. Rock masses
near the surface of a planet are subjected to a variety of crack-producing stresses, of both
internal and regional (tectonic) origin. Volcanic rocks are erupted as hot liquids that solid-
ify and cool to ambient temperatures. After they solidify they continue to cool and con-
tract, developing large internal stresses that create cooling joints, like the columnar jointing
mentioned in Section 5.3.1. Deep-seated rocks exposed by erosion or tectonic faulting cool
as they near the surface, generating large stresses by thermal contraction. Another factor
in joint formation is uplift, in which deep-seated rocks are stretched as they move to larger
radii on a spherical planet (Suppe, 1985). It is an observed fact that every large mass of
rock on Earth is cut by cracks. High-resolution images of the surfaces of Mars and Venus
have shown that their rocks, too, are highly fractured, even where meteor strikes have not
shattered the rocks.
In addition to macroscopic cracks, all rocks contain microscopic cracks at the scale of
their individual mineral grains. The minerals composing igneous rocks expand and contract
by different amounts as the temperature changes. Most minerals also expand and contract
by different amounts in different directions with respect to their crystal lattices. As an igne-
ous rock cools, stresses develop between adjacent grains and generate microcracks along

EBSCOhost - printed on 11/3/2023 9:57 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
7.3 Weathering 301

(a) (b)

Figure 7.10 Sheeting joints form in response to horizontal compression, shown in the left side of
the figure. The right panel indicates that sheeting joints follow the profile of the land surface. Their
spacing is also typically closer nearer to the free surface due to unloading of the confining stress.

grain boundaries, creating both small amounts of porosity for invading fluids to occupy and
incipient flaws from which macroscopic cracks can grow.
Sheeting and exfoliation joints. Besides the ubiquitous, nearly vertical cracks (joints)
that cut most large masses of rock, horizontal sheeting joints develop near the free surface
of homogeneous rocks. Due to the relief of vertical stresses, the rock expands upwards
and outwards, separating into stacks of closely spaced slabs that tend to follow the con-
tours of the free surface (Figure 7.10). These cracks grow when the stresses parallel to the
surface are compressive, tending to buckle sheets of rock up and out toward the stress-free
surface. Spaced only centimeters apart right at the surface, the horizontal joints’ spacing
increases rapidly with depth, reaching a few meters 20 to 30 m below the surface where
vertical and horizontal stresses are more equal. Good examples of this kind of jointing are
displayed at Half Dome in Yosemite National Park and Stone Mountain in Georgia, both
in the USA.
Spheroidal weathering. A striking landscape may develop where weathering solutions
propagate readily along the cracks and disintegrate large jointed blocks from the outside in.
Called spheroidal weathering, this process creates landscapes that appear to be covered by
gigantic (meters to tens of meters) misshapen marbles. Extensive terrains covered by such
blocks occur in southern California’s Peninsular Ranges and many other places on Earth.
They are frequently set aside as tourist attractions, such as at Arizona’s Texas Canyon.
Below the surface the spherical “marbles” grade into inhomogeneously weathered rock,
where intact corestones are imbedded in a matrix of crumbling, disintegrated rock debris
(Figure 7.11). It is clear that the rock has been preferentially attacked along the joints, from
which a weathering front has moved gradually inward, rounding off the sharp corners and
leaving a near-spherical unweathered mass in the center. Recent images from the Spirit
Rover confirm that spheroidal weathering occurs on Mars as well as on Earth.
Frost shattering. One of the most widely cited types of physical weathering is frost shat-
tering by freezing water (McGreevy, 1981). A moderate number of people have actually
heard the sharp report of a rock or tree suddenly splitting during a cold snap; the physical

EBSCOhost - printed on 11/3/2023 9:57 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
302 Regoliths, weathering, and surface texture

Figure 7.11 Spheroidally weathered granite showing incipient corestones in a homogeneous mass.
The light bands are aplite dikes, fine-grained dikes that typically cut granite plutons during their final
stages of crystallization. These dikes are more resistant to weathering than the coarse-grained granite
itself. Outcrop on the north shore of Big Bear Lake, Southern California that was exposed by road
construction in the 1970s. Photo by the author. Corestones are about half a meter in diameter.

evidence of fractured rocks in cold climates is clear for anyone to see. This process is
widely attributed to the 9% volume expansion of water upon freezing. When freezing water
is completely sealed up in a rigid container it can develop truly enormous pressures – 207
MPa if cooled to –22°C – but the problem is, how can water be so completely confined
under natural circumstances? If volume expansion were the cause of frost shattering, it
would make it a rather special process, occurring only for liquid water, but not for most
other liquids, which contract upon freezing. Thus, methane on Titan would not be capable
of shattering solid ice “rocks,” presuming that it were able to freeze under ambient Titanian
conditions, because solid methane does not expand upon freezing. However, it may come
as a surprise to many readers of this book (who may even have read that frost shattering is
caused by the volume expansion of freezing water in otherwise reputable books on geo-
morphology) that frost shattering does not, in fact, have anything to do with the peculiar
behavior of water and that any freezing fluid can shatter a solid in which it crystallizes. The
process of frost shattering is closely connected to the similar processes of salt weathering.
The downside of our current understanding of frost shattering is that the explanation of
what really happens is longer and more complex than simple volume expansion.
Although a large mass of water freezes, by definition, at 0°C (at 1 bar pressure), small
volumes of water must be cooled to much lower temperatures before they freeze, a fact first
discovered by J. J. Thomson in 1888. This discrepancy is caused by the surface energy of
small masses of water (this is also true for any other substance). Similarly, a thin film of
water on rock possesses a surface contact energy that keeps it liquid well below 0°C. In fine-
grained silts, where water is both dispersed into tiny pores and gains considerable energy

EBSCOhost - printed on 11/3/2023 9:57 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
7.3 Weathering 303

T < 0° C

unfrozen growing ice

water films crystal

rock

Figure 7.12 Ice freezing in narrow cracks wedges rock apart by the force of crystallization, not by
its volume expansion. Thin films of water in immediate contact with the rock do not freeze until the
temperature falls well below the normal freezing point of water because of the chemical interaction
energy between water and silicate minerals. Hexagonal ice crystals typically grow fastest along their
c-axes, shown here as dashed lines, which are, thus, usually perpendicular to the plane of the crack.

from contact with the silicate minerals, some water may remain unfrozen even at −30°C. In
contrast, water in gravel fills large pores and is entirely frozen just a little below 0°C.
A thin film of water in a joint or crack in a rock on the surface may, thus, remain liquid
at temperatures well below those necessary to freeze a large mass of water. Now suppose
that, adjacent to this crack, there is an unusually large pore or widening of the crack. The
larger mass of water in this space freezes first, while the water film remains unfrozen
(Figure 7.12). Under these circumstances water flows from the film to the frozen mass,
which grows larger and begins to wedge open the space in which it is growing. This is
also the origin of the ice lenses frequently found in frozen soil, as was first convincingly
demonstrated by Taber (1929). Detailed analysis of the thermodynamics of this process
shows that the chemical potential of the ice mass is lower than that of the water film, driv-
ing this process forward (Dash et al., 2006). The greater the difference between the ambient
temperature and the freezing point, the greater is the chemical potential favoring crystal
formation. This chemical energy difference translates to a maximum difference in pressure
between the crystallizing solid Ps and the liquid, Pl, given by:

ρ s Lm
Ps − Pl = (Tm − T ) = C fh (Tm − T ). (7.20)
Tm

When the pressure difference between the solid and liquid is less than this maximum,
water continues to flow to the solid, which grows in volume. If the pressure difference is
greater, the water flow reverses and the solid melts. The “frost-heave coefficient” Cfh in
(7.20) is equal to the density of the solid ρs times the latent heat of melting Lm divided
by the melt temperature Tm. For water it equals 1.12 MPa/K, which is unusually large for
most substances (this coefficient is only 0.32 MPa/K for methane). Thus, ice lenses segre-
gating from a thin film that freezes at –10°C can exert pressures of about 10 MPa – large
enough to split most solid rocks. This analysis ignores crystallographic orientations of

EBSCOhost - printed on 11/3/2023 9:57 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
304 Regoliths, weathering, and surface texture

the ice crystal, which is not actually correct: Hexagonal ice crystals grow preferentially
along their c-axes and correspondingly exert their maximum pressure along this axis.
Nevertheless, the pressure given in (7.20) gives the correct order of magnitude for frost
heaving.
The process of frost shattering is most effective where the temperature makes frequent
excursions around the freezing point and ice masses alternately grow and shrink in the
interiors of wet rocks. On Earth, this occurs at middle and high latitudes in spring and fall.
Where frost shattering is effective the surface may be covered with broken shards of rock
that form extensive fields. Because it requires the presence of thin films of liquid water,
frost shattering is ineffective where the temperature remains well below freezing or liquid
water is absent. Mars, at present, may not be a candidate for this process because its atmos-
pheric pressure is below the triple point of water so that liquid water is unstable at its sur-
face, except for a small region at the bottom of the Hellas crater depression. Temperatures
on Titan are presently believed to always remain above the freezing point of methane,
so methane-shattering may not be effective there, although it appears that some process
on Titan shatters its ice “bedrock” and produces ice boulders that can be transported by
methane streams.
Salt weathering. Minerals crystallizing from solution exert rock-shattering pressure in
a similar manner. In this case it is not the temperature below the nominal freezing point
that counts, but the degree of supersaturation of the mineral (often a salt such as halite) in
solution. If the concentration of the crystallizing species at saturation is Cs, and the actual
concentration is C, then the supersaturation is defined as C/Cs. The mineral remains in
solution if C < Cs, but C must exceed Cs by some finite amount before precipitation can
start. Supersaturations seldom exceed about 10 when nuclei are present to begin the crys-
tallization process. Typically the liquid portion of a solution evaporates, gradually raising
the concentration of the mineral solute until it exceeds the saturation limit, at which point
crystallization begins. The maximum pressure that can be exerted by a growing crystal is
given by (Winkler and Singer, 1972):

R ρT C
Pxtal = ln (7.21)
m Cs

where R is the gas constant, ρ the density of the crystalline species, T the absolute tem-
perature and m is its molecular weight. Equation (7.21) ignores crystal anisotropy and so
can only be trusted for its order of magnitude. Table 7.4 lists the isotropic crystallization
pressures of a number of geologically common evaporite minerals at a supersaturation ratio
of 2. It is clear that many salts can exert very large pressures as they crystallize and “salt
weathering” is an important and effective process where evaporation occurs readily, such
as in the hot deserts of Earth. The Phoenix mission to Mars is believed to have detected
brines in the immediate subsurface, which suggests the possibility of salt weathering on
Mars as well. Hematite, listed in Table 7.4, has been found as concretionary “blueberries”
on Mars by the Opportunity rover. Concretions of this type can grow by displacing the
enclosing permeable sediment through crystallization pressure.

EBSCOhost - printed on 11/3/2023 9:57 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
7.3 Weathering 305

Table 7.4 Crystallization pressures of evaporite minerals

Crystallization pressure at C/CS = 2 and 273 K

Mineral Chemical formula (MPa)

Halite NaCl 55.4

Hematite Fe2O3 52.3
Anhydrite CaSO4 33.5
Gypsum CaSO4•2H2O 28.2
Kieserite MgSO4•H2O 27.2
Epsomite MgSO4•7H2O 10.5
Jarosite KFe3(SO4)2(OH)6 9.7
Natron Na2CO3•10H2O 7.8

Data and method for jarosite and hematite from Winkler and Singer (1972).

Frost shattering by freezing pore fluid and salt weathering by crystallizing salts are only
two special examples of a general phenomenon. Whenever a chemical alteration, driven
forward by a change in chemical potential Δμ, results in a volume change ΔV, the reaction
will go forward unless the work expended in accommodating the volume change, PΔV, is
greater than the gain in chemical potential. The maximum pressure the reaction can exert
is thus given by

∆µ
Pmax = . (7.22)
∆V
Equations (7.20) and (7.21) are merely special cases of (7.22), specialized to the case of
a liquid that freezes at a temperature below that of the solid and of a crystallizing solute.
Rock disintegration. Another common realization of this phenomenon is the disin-
tegration of weathered igneous rocks. Initially solid rocks, such as granite, contain a
variety of minerals with different susceptibilities to chemical weathering. When a par-
ticularly susceptible mineral (biotite or feldspar, for example) becomes hydrated and
swells in volume by only a few percent, its expansion generates large tensile stresses
that wedge the mineral grains apart and reduce the rock to a pile of coarse, sand-sized
debris. This process is so common on Earth that disintegrated granite has its own special
name: grus. The formation of grus is of especial importance because the quartz grains
thus released from granite are eventually freed of all other igneous minerals. Quartz is
uniquely stable in the Earth’s surface environment – it is possible to find quartz grains
that have been recycled through multiple generations of sandstone for a billion years.
This quartz forms the sand that is so abundant in our beaches, streams, seas, and sandy
deserts. But because silica-rich granite is an exclusive product of plate tectonics, quartz
sand may also be unique to Earth. The process that forms the sand-sized particles in the
extensive eolian dune fields on Venus, Mars, and Titan is still a mystery, but these dunes
are not made of quartz grains.

EBSCOhost - printed on 11/3/2023 9:57 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
306 Regoliths, weathering, and surface texture

Volume changes are caused by many other processes: The reversible swelling and
shrinkage of hydrating clays during wet/dry cycles is another such process, one that keeps
clay-rich soils slowly seething in perpetual movement. The consequences of this motion
will be explored in more detail later in this chapter and the next, but we end this section by
examining a much more controversial volume change: Thermal expansion.
Thermal stresses. Diurnal and seasonal temperature changes, as discussed above in
Section 7.2, are most pronounced at or near the surface. Rocks exposed on the surface regu-
larly warm up during the day and cool at night. Because the thermal skin depth may be only
a few centimeters, rocks larger than the skin depth experience strong gradients in tempera-
ture. But because of thermal expansion, different parts of the rock expand or contract by
different amounts and this generates internal stresses. It has long been understood that such
stresses might shatter the rock (Mabbutt, 1977). This process has therefore, been given the
name “insolation weathering.” There is little doubt that it actually occurs on Earth: The
orientations of cracks in split desert rocks bear clear relationships to the direction of max-
imum solar heating. The problem is, laboratory experiments have been unable to reproduce
this phenomenon and, worse yet, no sign of insolation weathering has been found among
lunar rocks, where temperature extremes are much larger than those on Earth.
Much of the doubt about the effectiveness of insolation weathering dates from a series of
experiments performed by geologist David Griggs (1936). Griggs cycled dry rocks through
an extreme temperature excursion of 110°C, 89 400 times, an equivalent of 244 yr of daily
cycles on the Earth. In spite of his very high rates of temperature change as well as extreme
temperature excursions, he found no sign of damage to the rock. On the other hand, when
small amounts of water were present the rock crumbled after only 1000 cycles. The impli-
cation is that temperature variations alone do not cause rock disintegration, but that water
is required, perhaps by a chemical attack through hydration or stress corrosion cracking
(Moores et al., 2008). “Insolation weathering” is, thus, not a purely mechanical process but
requires that diurnal temperature changes be accompanied by some other, probably chem-
ical, weathering process.
It is known that really extreme temperature changes, such as occur in brushfires on the
Earth, do cause flakes to spall off rock surfaces. However, even here, temperature changes
might not be acting alone, as quartz in silica-rich rocks undergoes a large volume change at
573°C in association with the alpha-beta solid-state phase transition.

7.3.3 Sublimation weathering

Sublimation is not a very important weathering process on Earth. Although it has been
often suggested in connection with the evolution of snow and ice surfaces, the latent heat of
sublimation of ice, 2.83 MJ/kg (at 0°C), is much higher than its latent heat of melting, 0.34
MJ/kg. Observational investigations of the relative role of melting and sublimation show
that melting almost always dominates over sublimation when it is possible.
However, elsewhere in the Solar System melting may not be possible, such as on Mars
where liquid water is not stable on the surface, or on Callisto where both the pressure and

EBSCOhost - printed on 11/3/2023 9:57 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
7.3 Weathering 307

Start deposition
sublimation

ice + rock mixture

Later
pure ice residual rock debris

Figure 7.13 Sublimation weathering on Callisto. Water vapor sublimates from exposed warm surfaces
and is deposited on cold surfaces, raising its albedo and promoting the separation of the surface into
patches of cold, bright ice and ice-free regions mantled by darker silicate dust that absorbs more
sunlight and becomes warmer than the bright ice. After Figure 17–24 of Moore et al. (2004).

temperature are too low for liquid water. On Mars, CO2 frosts are common and the triple
point of CO2 is even higher than that of water (its heat of sublimation is much lower than
water, 0.571 MJ/kg). Under these circumstances, solid volatiles pass directly into the vapor
phase, giving rise to a novel form of weathering.
By itself, sublimation promotes a net transport of material from warm, exposed, areas on
a surface to cooler, shady places. On a planetary scale, vapor is transported from equatorial
regions toward the poles, and bright polar caps may be the result. However, when a bright
volatile ice is mixed with darker, inert material, complex and characteristic landforms may
arise. A surface composed of a mixture of dark material such as silicate dust and ice is
inherently unstable. The dark material absorbs more sunlight and, thus, remains warmer
than the more reflective ice. A thin coating of dark material, thus, enhances the sublimation
of the underlying ice, although a sufficiently thick deposit of dark material suppresses
sublimation by shielding the underlying ice from the daily high extremes of temperature.
Warm, dark areas tend to lose mass and become darker, while cool, bright areas gain more
bright ice by vapor deposition (Figure 7.13). The surface becomes dominated by strong
positive feedback that produces a landscape of high, bright peaks separated by dark, low-
lying plains and hollows. Spires observed on the surface of comet Wild 2 by the Stardust
spacecraft and “hoodoos” on comet Hartley 2 observed by the EPOXI mission may have
originated by similar processes.
Earth does not have any large-scale analogs of this process, although “suncups” that
evolve into spire-like “penitents” on old snowfields provide small-scale examples, and
some features in the Antarctic dry valleys have been attributed to sublimation. Mars has
many probable examples of this kind of terrain in its deposits of polar volatiles, where

EBSCOhost - printed on 11/3/2023 9:57 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
308 Regoliths, weathering, and surface texture

irregular to circular pits are etched into ice-rich sediments on a scale of hundreds of meters
in the so-called “Swiss-cheese” terrain. Callisto provides the most extensive example of
such terrains, where dark, broad plains are dotted with steep-sided knobs and fretted scarps
(Moore et al., 2004). Pole-facing slopes are typically brighter than equator-facing slopes,
reflecting the tendency of volatiles to deposit in cooler environments.

7.3.4 Duricrusts and cavernous weathering

It is common to find that individual boulders and rock surfaces exposed in the Earth’s
deserts are “case-hardened” by a strong rind or crust a few centimeters thick. Beneath
the rind, the rock is often much softer than the intact rock. In some cases the interior has
entirely disappeared, leaving an empty shell (sometimes called “turtle” rocks, as they look
like a giant empty turtle shell). Such rinds, properly called “duricrusts,” are another result
of chemical weathering coupled with fluid transport.
When an infrequent rain wets a rock in an arid environment, the slightly acid rainwater
soaks into the rock only a small distance, typically a few centimeters. Once there, it initi-
ates hydration and hydrolysis reactions and is enriched in some of the soluble weathering
products. Later, after the rain stops and the Sun returns, the water evaporates from the rock
surface, drawing the now briny water back to the surface by capillary action. The brine
becomes concentrated near the surface and deposits its load of solutes as it evaporates com-
pletely. In some rocks, this process extracts salts that disintegrate the rock surface by salt
weathering. However, in rocks whose minerals are cemented together by CaCO3, SiO2, or
Fe2O3, the deposit may enhance the natural cement, hardening the outer layer of the rock
while softening the interior by removing its cement. The depth of penetration of the water
sets the scale for this process: Weathering rinds are thicker in permeable rocks such as
sandstones or tuffs, and thin or absent in very impermeable rocks.
Duricrusts may also form in soils subject to infrequent rainfall followed by evaporation.
The process is the same as in rock, with the exception that soluble salts are more likely
to harden the soil surface than to disintegrate it. Hard soil caps with names like gypcrete,
calcrete, silcrete, or ferricrete (for soil duricrusts cemented by gypsum, CaCO3, SiO2,
or Fe2O3, respectively) are common in many arid regions on Earth (Mabbutt, 1977). The
Viking 2 and Phoenix landers on Mars encountered hard layers in the Martian soil associ-
ated with enhanced abundances of SO4 and Cl that may be duricrusts, although in this case
they are more likely to be produced by evaporation of briny soil water rather than rainfall.
High thermal inertia measurements over broad regions of Mars suggest that some form of
duricrust may be globally distributed.
Duricrust formation is closely related to another striking weathering feature, cavernous
weathering. Many rock exposures in the semiarid American West are peppered with shallow
holes now termed “niches” or “tafoni” (Figure 7.14). Such holes astounded and perplexed
travelers from the humid Eastern states and led to many colorful explanations, ranging from
the miners’ claim that they were drilled by “rockpeckers,” an extinct giant woodpecker that
extracted burrowing mammals from solid rock, to an early geologist’s assertion that they

EBSCOhost - printed on 11/3/2023 9:57 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
7.3 Weathering 309

Figure 7.14 Cavernous weathering surface in sandstone. Image is about 1 m across, showing deep
pits and columns that have become detached from the mass of the rock behind them. This variety is
often called “honeycomb” weathering from its appearance. Canyonlands National Park, Utah, USA.
Photo courtesy of Ingrid Daubar-Spitale, 2010. See also color plates section.

are caused by wind (this hypothesis is still often encountered in textbooks). Although wind
may, indeed, play some role in their excavation, the correct explanation is linked to duri-
crust formation (Blackwelder, 1929; Mabbutt, 1977).
Case-hardening of rock surfaces comes at the price of a weakened layer a few centim-
eters beneath the surface. Thus, when a case-hardened surface is breached for any reason,
the underlying weakened rock easily disintegrates. The resulting debris either falls free of
the rock face or is removed by wind or rainwash. Once a hollow forms, water remains in its
shaded interior longer than on the adjacent rock face and promotes further chemical weath-
ering. Rock deeper in the cavity then disintegrates, creating flakes and small fragments that
fall to the floor and are also easily removed from the growing cavity. Where this process is
effective, broad vertical rock faces gradually become deeply pitted. As individual cavities
continue to grow they may meet behind the original rock face, creating long colonnades
sometimes known as “choir stall weathering,” some small examples of which can be seen
in Figure 7.14.

7.3.5 Desert varnish

Many rock surfaces in arid environments are covered with a thin amorphous coating of
black or reddish material known as desert varnish. Its thickness ranges from 0.03 to about
0.1 mm and it is composed mainly of clay and the hydrous oxides of iron and, especially,
manganese, which impart its red and black colors. It gradually thickens with time and often
possesses an internal stratigraphy. It can, thus, be used to establish relative dates of surfaces

EBSCOhost - printed on 11/3/2023 9:57 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
310 Regoliths, weathering, and surface texture

(the darker the surface, the older it must be), but its growth rate varies widely from one
location to another, so its thickness cannot be used for absolute dating. Occasional wetting
seems to be essential for the growth of rock varnish. Much of its material is derived from
wind-blown dust, not from the underlying rock, although it often forms on top of weath-
ered duricrust rinds. Black, desert-varnished rocks under the summer Sun can become too
hot to touch, so it was a surprise to discover that it has an important biological component
(Mabbutt, 1977). Lichens, algae, and soil microflora are active in chemically reducing the
oxides in desert varnish. Carbon compounds of biological origin are entombed in the coat-
ing. Because of the biological connections of desert varnish, it is considered a prime target
for searches for life on Mars (Perry and Kolb, 2003).

7.3.6 Terrestrial soils

Soil on the Earth has, quite naturally, received intense scientific study for as long as sci-
ence has existed. Human civilization depends crucially upon the quality of the agricultural
soil: Civilizations have risen and fallen in response to changes in the soil. It is impossible
to do justice here to the immense literature on the subject, but only to indicate some of the
important factors in soil formation and maintenance and the major differences between ter-
restrial soils and those of other planets.
Soils on Earth range in thickness from near zero to as much as 100 m, although a thick-
ness of only a few meters is typical. Rates of soil formation are widely variable: observed
rates vary from 3 cm/yr to 3 cm/5000 yr (Leopold et al., 1964). While the nature of the soil
depends to some extent on the bedrock from which it forms, climate (moisture and tem-
perature) appears to be the dominant determinant of soil type.
The most compelling aspect of terrestrial soil is its biological activity. Organisms from
all the kingdoms of life are at home in the soil, which is an immense reservoir of life ran-
ging from microbes to higher animals and plants. Charles Darwin was so impressed by
the importance of biological activity in the soil that he devoted his last book to the humble
earthworm and its role in the “economy of nature” (Darwin, 1896). The bulk of Earth’s
organic carbon resides in the soil; about 1500 Gton total (Amundson, 2001), and turnover
from one organism to another is rapid. Because of the presence of life, chemical activity is
intense and weathering is faster in the soil than almost anywhere else on the Earth. Indeed,
it is difficult to separate purely chemical weathering processes from biologically mediated
ones. Experiments on sterile weathering reactions have almost invariably shown that they
are slow compared to their naturally observed analogs, and the difference is usually due to
biological activity.
Another major factor in soil formation is rainfall. The landscapes of even the driest
portions of the Earth show traces of rainfall and no Earthly soil can escape at least an
occasional wetting from rain. Rain falling on the surface typically soaks in some distance
before it either joins the underground water table, wicks back to the surface or flows lat-
erally to a spring. The result is a two-level structure, where the upper portion is leached by
rainwater flowing to deeper levels, while the lower portion receives leachates from above.

EBSCOhost - printed on 11/3/2023 9:57 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
7.4 Surface textures 311

These two regimes, along with the organic-rich layer right at or just below the surface, are
reflected in the common division of the soil into layers or horizons denoted by letters. Thus,
surface organic material is denoted the O-horizon. The typically dark, organic-rich and
leached upper layer is termed the A-horizon, while the B-horizon receives solutes leached
from above. The C-horizon is composed of weathered parent material and the R-horizon
is unweathered material. Other horizon designations and subdivisions are in common use,
but for these the reader is referred to specialized texts (Birkland, 1974). Not all of the vari-
ous horizons may be present in a given soil. Soil naming and classification is a complex
affair and depends on composition, color, texture, and other factors that differentiate one
soil from another.

7.4 Surface textures

The immediate solid surface of a planet is what we can observe from space and is the
major target of any spacecraft imaging system. Although global maps are made of moun-
tain ranges, valleys, and impact craters, the fine-scale structure of the surface is too exten-
sive to receive such attention. Nevertheless, the detailed surface texture, at sizes ranging
from meters to microns, gives the surface its distinctive optical properties and has a major
effect on the ability of machines and, ultimately, humans to traverse the surface.

7.4.1 “Fairy castle” lunar surface structure

The brightness of a night with a full Moon can be astonishing. Careful observation of the
Moon over a lunar cycle shows that the full Moon is not just bright because the illuminated
area is large: Each portion of the moon is quite literally brighter during the full Moon than
at all other times. This phenomenon has become known as the “opposition surge” and it
tells us something about the arrangement of particles on the Moon’s surface.
Consider the path of the sunlight that we receive from the Moon’s surface. The light
starts out at the Sun, travels to a point on the Moon’s surface, is reflected (or absorbed
or scattered), and then travels in a straight line to our eyes (neglecting refraction in the
Earth’s atmosphere). The angle between the light ray incident on the Moon’s surface and
the reflected ray that we receive is called the phase angle. During a full Moon, when the
Moon is in opposition to the Sun, this phase angle is very small: When the Moon is half-
full this angle is close to 90° and it approaches 180° at new Moon. The total brightness
of the Moon as a function of this angle is plotted in Figure 7.15. It is clear that brightness
is neither constant nor a linear function of phase angle. In particular, it increases dramat-
ically (the “surge”) at angles less than about 5° (Buratti et al., 1996). It is interesting to
note that the full range of the opposition surge was not known before spacecraft obser-
vations: The moon as a whole cannot be observed from Earth at less than a few degrees
phase angle.
Mutual shadowing of surface irregularities causes the opposition surge. Consider an irregu-
lar, pitted surface. Looking from nearly any direction, one’s field of view includes both bright

EBSCOhost - printed on 11/3/2023 9:57 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
312 Regoliths, weathering, and surface texture

Normalized, disk-integrated brightness

1.0

opposition
surge

0.5

0.0
0 20 40 60 80 100 120 140 160 180
Solar phase angle

Figure 7.15 Opposition effect of the lunar surface. The Moon is much brighter when incident sunlight
is directly backscattered, resulting in a distinct brightening during the full Moon. The brightness
peaks at reflection angles less than 5° from exact opposition. Simplified after Figure 5 of Buratti
et al. (1996).

illuminated high points and shadows in low areas and hollows. However, in one direction,
that looking along the direction of illumination (that is, near zero phase angle), the shadows
disappear and all surfaces appear bright. One can see this phenomenon by looking at the
shadow of one’s head on an irregular surface, such as a grassy lawn, or more obviously by
looking out of an airplane window at the shadow of the airplane on the ground. The oppos-
ition surge is related to the phenomenon known as “heiligenschein,” which is the bright halo
that appears around the shadow of one’s head on dewy grass, although in that case the inten-
sity of the backscattered light is given an extra boost by reflection from dewdrops.
The range of angles over which opposition brightening occurs is related to the average
surface slope and the depth to which light penetrates into the surface. Strong opposition
effects are typically observed in fine (10 to 20 μm) powders. Detailed models of how
surfaces reflect light can be fit to the observed opposition brightening and yield informa-
tion about the particle size and distribution on the surface (Hapke, 1993). Application of
such models to the Moon long ago suggested to astronomers that the Moon’s surface is
extremely irregular on a fine scale, to which the name “fairy castle” structure was given.
Small particles may be sintered together to form tiny towers surrounded by minute deep
crevasses. The narrow angle and extent of the opposition surge suggests an extremely pit-
ted, fractal surface on a scale larger than the wavelength of light. This deduction was veri-
fied by in situ stereoscopic measurements of the undisturbed lunar surface by the Apollo
11, 12, and 14 astronauts (Gold, 1970).
Opposition surges are observed for asteroids, indicating a powdery regolith surface. This
effect makes optical detection of asteroids most easy at opposition, as their brightness
declines rapidly as they move only a few degrees away from this point. The opposition
surge is also apparent in Saturn’s rings, and is cited as one of the first direct demonstrations
that the rings are composed of a swarm of individual solid particles.

EBSCOhost - printed on 11/3/2023 9:57 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
7.4 Surface textures 313

(a) (b)

Figure 7.16 Boulder-strewn surfaces on Venus, Earth, Mars, and Titan. Panel (a) is from the Soviet
lander, Venera 13, image VG00261. Panel (b) near Yuma, AZ, looking north toward the Cargo
Muchacho Mountains from Indian Pass Road (photo courtesy of Mark A. Dimmitt). (c) is a panorama
from the Viking 2 lander. (d) The surface of Titan as viewed by the Huygens lander. The lander
evidently set down in a former riverbed, only the “rocks” in the foreground are water ice and the
liquid that transported them was liquid methane. The two “rocks” just below the middle of the image
are 15 cm and 4 cm in diameter, respectively, and lie about 85 cm from the Huygens probe. The dark
fine material on which the “rocks” lie is probably a mixture of water and hydrocarbon ice. Image
PIA07232. ESA/NASA/JPL/University of Arizona. See also color plate section.

7.4.2 Stone pavements: why the Brazil nuts are on top

Many desert surfaces on Earth, as well as the surfaces of Mars, Venus, and Titan (Figure 7.16),
are littered with stones and small boulders. In many places the stones covering these sur-
faces appear to be fitted together in an almost artificial-looking mosaic (Figure 7.17).
Digging into such surfaces, one is surprised to find that the stones form a single layer that
is confined to the surface itself: Underneath one finds mostly fine-grained material with
few, or no, large stones. Such surfaces are called “desert pavements” or “stone pavements.”
Their striking appearance has attracted many geologists’ attention and has generated a cur-
rently raging controversy about their origin.
For at least a century, nearly every writer describing desert pavements has attributed them
to deflation. They supposed that the original soil contained a mixture of stones and fine

EBSCOhost - printed on 11/3/2023 9:57 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
314 Regoliths, weathering, and surface texture

Figure 7.17 Desert pavement surface showing a surface layer of large and small stones apparently
“fitted” together into a layer only one stone thick. The stones are basalt fragments. From central
Iceland, photo by the author.

silt. When the wind blows over such a surface, the fine silt is exported from the region and
the stones remain behind as a lag deposit that eventually armors the surface against further
deflation. The presence of pavements was, thus, taken as evidence that the region is degrad-
ing, with preferential removal of the fine-grained material by wind or perhaps water.
Doubts about this explanation began to arise when pavements were found on surfaces that
are manifestly aggrading. A revisionist view that is now gaining wide acceptance attributes
most pavement formation to upward migration of stones relative to a fine-grained sub-
strate (Dohrenwend, 1987). The explanation of how this counter-intuitive process works
is closely related to what is called the “Brazil nut phenomenon” (Rosato et al., 1987). It is
easy to show that in a bowl of mixed nuts of different sizes, say, peanuts and Brazil nuts,
vigorously shaking the bowl brings the Brazil nuts to the top, where they remain through
further shaking (parents may have noted the same phenomenon in children’s toy chests:
Shaking always brings the larger toys to the top while the small ones disappear into the
depths). The tendency for large objects to rise is so strong that it can even invert the density
of the shaken objects: In a mixture of large steel ball bearings and small peanuts, shaking
still puts the much denser ball bearings on top. The reason for the preferential uplift of large
objects is simple: When a large object shifts upward, smaller objects readily fall into the
vacated space, whereas it is very unlikely that a number of small objects will shift so as to
leave a space large enough for the large object to fall into. As shaking continues, the large
object gradually ratchets up relative to the small ones until it emerges onto the surface.
If the Brazil nut phenomenon is the reason that stone pavements form, there must be some
process that agitates the large stones and permits them to jostle relative to the surrounding

EBSCOhost - printed on 11/3/2023 9:57 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
7.4 Surface textures 315

fine-grained sediment. Fortunately, such processes abound near the surface: Wetting and
drying of expansive clays, freezing that generates ice pillars beneath the stones followed
by thawing, and bioturbation (creatures from lizards to wombats have been invoked) are
all plausible and effective types of disturbance. Rapid movement is not necessary, only
small, repetitive shifts of the stones relative to the fines. Such agitation is possible only
close to the surface. Stones buried deeply enough lose the ability to shift and so become
unable to participate in the upward climb. However, the soil surface itself is an active
place and stones may continue to “float” on the surface for a long time: 770 000-year-old
Australasian tektites are sometimes found on late Pleistocene soils. Although this obser-
vation is sometimes used as an objection to the natural occurrence of these objects, this is
probably an example of the Brazil nut effect. Likewise, deep-sea manganese nodules with
ages of millions of years are found lying on top of contemporary mud surfaces. There is no
paradox here if one recognizes the disturbing role of marine bioturbation.
Stone pavements on desert surfaces can remain on the surface even as wind-borne sand
and dust accumulate. Fine-grained sediments fall between the stones and eventually dis-
appear beneath the surface as the stones slowly shift and offer them a larger space to fall
into. Blocks of basalt that once lay on top of their parent lava flows 12 Myr ago are now
separated from them by meters of wind-blown sediment in many places along Arizona’s
Mogollon Rim. Stone pavement formation is a universal process,which depends upon the
restless activity of a planetary surface,for creating organized patterns.

7.4.3 Mudcracks, desiccation features

If the Earth were barren of vegetation, mudcracks would be one of the most common fea-
tures on the land surface. As it is, they are ubiquitous on bare, exposed clay-rich sediments
and are often found as fossils in ancient rocks. Mudcracks appear as networks of open
cracks that surround polygonal, sometimes crudely hexagonal, plates. These inter-crack
plates range in size from a few centimeters up to hundreds of meters across. The ultimate
cause of this cracking is shrinkage upon drying. Clay minerals are especially susceptible to
volume changes upon wetting or drying: Water molecules are easily added or lost from the
filling of the silicate sandwiches that make up clay minerals.
One of the most surprising characteristics of many mudcrack-covered surfaces is their
crudely hexagonal pattern. It has been shown by many researchers that a triple intersection
of cracks meeting at mutual angles of 120° minimizes elastic strain energy. However, all
three cracks must initiate at the same time if this configuration is to arise at all: If a single
tension crack formed first, it would relieve the stress parallel to its length so that a second
crack growing to meet it would curve to intersect it at a 90° angle. The result would be a
pattern of rectangular plates, not hexagonal ones. One does sometimes see such a rectangu-
lar grid of plates, but triple intersections are more common.
The horizontal scale of mudcracks is a complicated function of the strength of the mud,
rate of desiccation, and depth of the cracks. Cracks cannot grow too closely to one another,

EBSCOhost - printed on 11/3/2023 9:57 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
316 Regoliths, weathering, and surface texture

as the formation of the first crack relieves the stress in its vicinity, out to a horizontal dis-
tance of several times its depth. Mudcracks are closely spaced where a thin layer dries
quickly, while giant mudcracks are occasionally observed in thick playa sediments that dry
over long periods of time.
Because the top of drying sheets of sediment typically contracts more than the deeper
portions, the plates between mudcracks sometimes break off the underlying sediments and
curl up into separate flakes. Such flakes are especially vulnerable to being picked up and
transported by the wind. Weakly cemented clumps of fine sediment can then be blown into
heaps known as “clay dunes.” This process could be the solution to the “kamikaze effect”
noted for grains transported by the winds on Mars, discussed in Box 9.1 of Chapter 9,
assuming that vapor-phase hydration and freeze-drying can play the same role as liquid
water in expanding clays.

Further reading
The best summary of lunar regolith properties is contained in the book by Heiken et al.
(1991), where the peculiar thermal properties of the lunar regolith are also discussed.
French (1977) offers a clear description of the regolith at a semi-popular level, written
just after the end of the Apollo missions. Regolith formation and gardening processes are
discussed extensively in Melosh (1989). The classic book on terrestrial weathering proc-
esses is Ollier (1975), while Bland and Rolls (1998) present a comprehensive overview of
weathering from a more fundamental point of view. Few of the other topics in this chapter
are treated at book length; however, I have added many references to individual papers in
the appropriate places in the body of the chapter.

Exercises

7.1 Gardening asteroids

Like the lunar regolith, asteroid surfaces are constantly bombarded by small meteorites that
create craters of all sizes. On average, the depth to which the surface is overturned once
in a given period of time is equal to ¼ of the diameter of craters that have, in some sense,
just “covered over” the surface. Suppose that “covered over” means that the total area of
craters in the diameter range Dc to 2Dc just equals the area of the surface itself. Using the
cumulative size-frequency distribution, Equation (7.1), show that the following equation
expresses this condition:

π 2 D 2 dN cum
c π bc  1  1
− ∫ D dD = 1− =1
4 Dc
dD 4 b − 2  2b −2  Dcb −2

so long as b > 2.

EBSCOhost - printed on 11/3/2023 9:57 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
Exercises 317

The cratering rate in the asteroid belt is not well known, so use the better-known lunar
cratering rate, Ncum(D) = 1.9 × 10–12 D–3.82 craters/km2 yr for craters < 2 km in diameter
(Hartmann, 2005), as an estimate for the cratering rate in the asteroid belt. Note that D is
in kilometers. Using your equation for Dc and constants from the lunar cratering rate, com-
pute the overturn time for asteroid regolith to depths of 0.01, 0.1, and 1 m. Watch out for
unit conversions!

7.2 Huddling close to the fire

Cool red-dwarf stars (M-type stars) vastly outnumber solar-type stars in our galactic neigh-
borhood. Typical M-class stars have a mass about ½ that of our Sun, a surface temperature
of about 3600 K and a luminosity of only 0.08 Le. The “habitable zone” is the range of
distances of a planet from its star throughout which liquid water can occur on its surface.
Using the concept of effective temperature, use Equation (7.2) to estimate the semimajor
axis of planets in the habitable zone about M-type stars. Assume a planetary albedo similar
to that of known planets, such as the Earth or Mars. What would happen to a planet’s sur-
face temperature if it became covered with highly reflective snow or ice?

7.3 Cometary heat transfer

During its close approach to Comet Tempel 1 on July 4, 2005, the infra-red spectrom-
eter aboard the Deep Impact spacecraft provided an upper limit to the thermal inertia I
of the comet’s surface, I < 50 J/K m2 s1/2 (Groussin et al., 2007). Although the comet was
just outside the orbit of Mars during the encounter (and also nearly at perihelion), the
maximum surface temperatures reached a toasty 340 K because of the comet’s excep-
tionally low albedo. Approximating the heat capacity of the comet’s surface material as
fresh snow, from Table 7.2, and using the subsequently determined mean density of 400
kg/m3(Richardson et al.,2007), estimate the thermal conductivity k of the comet’s surface
from Equation (7.6). Finally, use this thermal conductivity in Equation (7.4) to compute the
thermal skin depth λ during the comet’s 6-month-long perihelion passage. What does this
imply for the volatiles that drive cometary activity?

7.4 Harold Urey’s feedback on planetary habitation

Qualitatively discuss the implications of the Urey reaction, Equation (7.19), for planetary
temperatures in the light of the greenhouse effect. Assume that both silicates and carbon-
ates may be present on the surface of the planet. If water is required for this reaction to
proceed at an appreciable rate, what are the implications for planets that either possess or
lack oceans? What is the role that plate tectonic recycling of carbonates might play? See
Kasting and Catling (2003) for an extended discussion of this interesting problem.

EBSCOhost - printed on 11/3/2023 9:57 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
318 Regoliths, weathering, and surface texture

7.5 (Methane) frost shattering on Titan?

At present, temperatures on Titan are just a bit too high (by only a few degrees K) to permit
methane to freeze anywhere on its surface. Suppose, however, that at some time in the past
methane could experience freeze–thaw cycles on Titan’s surface. Use Equation (7.20) to
discuss the possibility of frost shattering on Titan’s surface. Note that the frost-heave coef-
ficient Cfh = 0.32 MPa/K for methane. Is it plausible that freezing methane could shatter
solid ice “rocks”? Could “methane heaves” form in Titan’s soil? What properties of the ice/
methane interface must be understood to make such arguments quantitative?

7.6 Heaving rocks

The New England region of the United States is mantled with a thin veneer of glacial
till deposited by the now-vanished ice sheets of the last continental glaciation. This till
is a mixture of fine sand, clay, and decimeter-sized (or more) boulders. Farmers in these
states note that every spring a new crop of boulders seems to surface in their fields (some
old-timers are convinced that rocks on their farms grow underground like potatoes). As an
alternative to the potato theory, consider that the thermal conductivity of solid rock is much
larger than that of soil. Coupled with the tendency of soil water to form ice lenses and pil-
lars at freezing fronts, explain how the motion of solid rocks upward with respect to loose
fine-grained soil might occur in the annual freeze–thaw layer below the surface. Might any
analogous process occur on the surface of Mars? Explain why, or why not, based on your
knowledge of Martian surface conditions compared to those in New England.

7.7 Sublime cometary surfaces

In the wake of the Deep Impact missions to comets Tempel 1 and Hartley 2, we know
that these comets are composed of fine powder consisting of an intimate mixture of water
ice (plus highly volatile CO2 ice below the surface), silicates, and organic tars. During a
comet’s perihelion passage, the sun’s heat raises low-albedo surface layers to temperatures
high enough to rapidly evaporate water ice at pressures well below water’s triple point, so
that liquid water never forms: Ice simply sublimates into vapor. One of the more bizarre
discoveries of the Deep Impact and Stardust missions (Stardust returned images of comet
Wild 2) was that of bright, nearly vertical pillars on the comets that range in height from
tens up to 100 m. Speculate on how such features might form, given the inherent instability
of a surface containing a bright (high-albedo) volatile and dark (low-albedo) non-volatile
material. You might also look up references on sun cups, penitentes, and ice pillars on ter-
restrial snowfields.

EBSCOhost - printed on 11/3/2023 9:57 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
8
Slopes and mass movement

Imagine a landmass upon which no rain falls, no rivers flow, no glaciers

form, no waves beat, no winds blow. Let chemical and mechanical disin-
tegration disrupt the rocks, and gravity exert its downward pull. On such
a landmass earth and rock will move ceaselessly from higher to lower
levels, slopes will soften, relief will fade. Given time enough, the whole
will be reduced to a featureless plane of disintegrated rock debris.
Douglas Johnson, foreword to the book Landslides and Related
Phenomena by C. F. S. Sharpe, 1937

In the summer of 1935 young C. F. S. Sharpe undertook an excellent adventure. Having

acquired a car, he drove 16 000 miles through 28 American states and 3 Canadian prov-
inces. His quest was unusual: He was out to demonstrate that mass movement of debris
over the Earth’s surface is an important geological process. After he returned he wrote up
his observations in a book (Sharpe, 1938) that forms the basis of our modern understanding
of gravity-driven mass motions for the evolution of the Earth’s landscape. The current era
of space exploration has greatly broadened the reach of the processes he described: Mass
motion is important on bodies ranging from tiny asteroids and comets only a few kilom-
eters in diameter up to the largest moons and planets. Its action has been observed on every
solid body in the Solar System.

8.1 Soil creep

The mantle of loose debris on slopes everywhere in the Solar System is slowly creeping
downhill. This insidious motion is usually too slow to appreciate on human timescales,
but the evidence is there for anyone who will look. On Earth, gravestones and old fence
posts gradually tilt downhill. Linear trails of fragments lead downslope from distinctive rock
outcrops and steep stream banks are gradually overridden by thick sheets of soil that often
include intact mats of vegetation and trees. Although slow, the sheer ubiquity of soil creep
makes it an important, if not dominant, player in the overall denudation of sloping hillsides.
Copyright 2011. Cambridge University Press.

Accurate measurement of creep is difficult, but surface velocities on Earth range from mil-
limeters to centimeters per year, declining gradually to zero, through soil depths ranging

319

Figure 8.1 Lunar Orbiter image of a steep lunar hillslope illustrating the irregular elephant hide
topography with a basal berm that is presumably composed of material that crept downslope and
accumulated at the slope base. Note the small number of craters on the slope compared to the adjacent
mare surface. Portion of the Flamsteed Ring near 2° 50’ S, 42° 40’ W. Framelet strip width 220 m.
LO III frame H199.

from tens of centimeters to many meters. Although creep is most frequently noted in damp,
clay-rich soils on Earth, slope streaks and filled craters on Deimos, the 6 × 8 km smaller
moon of Mars, indicate active creep within its regolith, despite its tiny gravity field (2.5 ×
10–4 g). Slow downslope movement of the regolith occurs everywhere on steep lunar slopes,
where it is marked by characteristic “elephant hide” slope textures, the absence of small
impact craters, and berms of accumulated debris at the base of steep hillsides (Figure 8.1).

8.1.1 Mechanism of soil creep

The mechanism of soil creep is complex: Many different processes lead to the same over-
all motion. Sharpe listed seven possible causes in his book and modern research has added
several more (Carson and Kirkby, 1972). Many, such as wedging by plant roots and dis-
turbance by burrowing animals (published descriptions range from earthworms to wom-
bats), are distinctly Earth-related. Others, however, are of more general application, such
as heating and cooling, wetting and drying of expansive clays, freezing and thawing. Still
others, such as meteorite impact, are distinctly extraterrestrial. What they all have in com-
mon is mechanical disturbance of the near-surface zone. Repeated disturbances that might
only lead to small periodic motions on a level surface are converted by gravity to slow
motion in one overall direction – downhill.
One of the earliest theories of downslope creep was put forward by Charles Davison,
Esq. (1888). He noted that, in 1855, Canon Moseley had shown that pieces of lead lying
loose on the wood roof of Bristol Cathedral moved downslope 18 inches in two years,
which Moseley attributed to the alternate expansions and contractions of the lead through
diurnal temperature cycles. Davison went further and performed experiments on a pair of
bricks and blocks of sandstone tilted at an angle of about 20°, the upper one of which he
observed creeping slowly downhill over a period of several months.

EBSCOhost - printed on 11/3/2023 9:57 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
8.1 Soil creep 321

ex
pa
α nd
ed
su
rfa
ce

∆z
∆x

or
su igin
rfa al
ce

Figure 8.2 The mechanism of creep on a sloping surface. The ground surface first expands
perpendicular to the slope and then later contracts along a more vertical direction. The result of many
cycles of such expansion and contraction is a net downslope displacement.

The basic mechanism is illustrated in Figure 8.2. Any process that causes the upper
soil layer to expand, whether by heating, wetting, or freezing, pushes the land surface up
slightly in the direction of least resistance – perpendicular to the sloping surface. Later,
when contraction returns the layer to its original thickness, gravity tends to make the soil
settle more vertically, which results in a small increment of downslope movement. Early
theories of soil creep supposed that the contraction phase was exactly vertical, predicting a
downslope movement Δx =Δz sin α for each cycle, where α is the angle of the slope from
horizontal and Δz is the vertical expansion of the ground surface. These theories predicted
creep rates that are much too large, so modern theories of this type assume that the settling
phase is not exactly vertical. Mathematically, this introduces a recovery factor r into the
equation, Δx =(1–r) Δz sin α, where if r = 1 the slope settles back into its original position
and if r = 0 the slope settles vertically. Naturally, r is an empirical factor that must be deter-
mined from observations.
This theory predicts that the creep rate is proportional to the sine of the slope angle,
which is borne out by much observational data, and to the number of expansion/contraction
cycles, which also seems to be roughly correct. When slopes approach the angle of repose
the sine dependence breaks down, and the creep rate becomes a non-linear function of the
slope (Roering et al., 1999).
Expansion and contraction is not the only process that causes downslope creep. As early
as 1909 G. K. Gilbert noted that the impact of raindrops on steep divides must round off
the slopes and give hilltops a convex upward profile (Gilbert, 1909). The extraterrestrial
analog of this process is meteorite impact (Figure 8.3). Although each impact initially
ejects material symmetrically around the crater, gravity imparts a small downslope incre-
ment to the ejecta in flight that results in a net downhill transport. Perhaps more important
is the associated effect of seismic shaking induced by each impact. Crater excavation is

EBSCOhost - printed on 11/3/2023 9:57 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
322 Slopes and mass movement

Figure 8.3 Rain splash or impact on a sloping surface ejects more material downslope than upslope,
on average. The net result of many such impacts is a net downslope motion of material over a depth
approximating the depth of the crater.

accomplished by a shock wave that severely shakes the immediately surrounding terrain.
This shaking may induce local movements that cause a net downslope movement of loose
surface material (Richardson et al., 2005). On small bodies, such as asteroids, this process
is especially effective because the seismic energy is unable to spread out over a large vol-
ume. It remains trapped in the asteroid, reverberating until internal friction finally degrades
it into heat. This process explains the apparent absence of small impact craters on small
asteroids such as 433 Eros (Richardson et al., 2004) or Mars’ moons Phobos and Deimos.
On steep lunar slopes, which are already vulnerable to downslope movement, seismic shak-
ing by impacts may similarly account for the low density of small craters.
The Moon’s surface would also seem to be liable to creep by thermal expansion and con-
traction. After all, this was the first creep process analyzed by Davison in 1888 and the tem-
perature excursions on the Moon’s surface are far larger than those on Earth. Surprisingly,
however, this expectation seems to be false. Experiments on strongly heating and cool-
ing basaltic powder at Caltech’s Jet Propulsion Lab failed to show any detectable creep
(J. Conel, 1988, personal communication). Perhaps the cohesion of particles in fine pow-
ders like the lunar regolith prevents this process from being effective on the Moon.
A more exotic mass transport process is electrostatic dust levitation (Lee, 1996). This
process is only effective on airless bodies exposed to both direct sunlight and the solar
wind. Dust levitation on the Moon was observed by a variety of lunar landers as a “horizon
glow” shortly after sunset and was the subject of the LEAM experiment deployed during
the Apollo 17 mission. It consists of a light-scattering haze of ~10 μm particles hovering
at altitudes up to a few tens of kilometers. This particle cloud is attributed to electro-
static lofting: The Moon’s surface acquires a positive charge due to ejection of photoelec-
trons by solar UV radiation. Dust particles on the surface acquire similar charges and are
repelled from the surface, rising until the solar-wind plasma screens the electric field at
a height comparable to the plasma Debye length. Although the total mass involved in the
dust cloud is very small (fluxes are estimated at about 3 × 10–11 kg/m2-s), it is possible that,
over geologic time, accumulations of dust deposited from this cloud might be significant

EBSCOhost - printed on 11/3/2023 9:57 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
8.1 Soil creep 323

Figure 8.4 Terracettes are likely the result of soil creep rather than trampling by cattle. These slopes
in Iceland are corrugated into small terracettes ranging from about 30 cm to almost 1 m in height and
several meters wide. Photo by Ellen Germann-Melosh.

on the surface. On asteroids, smaller levitated particles may be swept away by the solar
wind and lost, while larger (ca. 1–100 μm) particles may settle back onto the surface.
Smooth, level “ponds” on the surface of asteroid 433 Eros are attributed to deposition by
levitated dust.

8.1.2 Landforms of creeping terrain

Hillslopes dominated by creep are often distinctive. In profile they are typically convex
upward with smooth, rounded crests. The slopes themselves often display downhill undula-
tions of slope, forming alternate steep “risers” and less steep “treads” that tend to run par-
allel to contour lines (Figure 8.4). These features are known as “terracettes” or sometimes
“slope garlands” on unconsolidated material such as cinder cones. Streaks of distinctive
rocks run straight downhill on creeping slopes, and small landslides may scar the slopes.
Where other erosional processes do not remove material at the base of the slope, as on the
Moon (Figure 8.1), berms of slope-transported material accumulate.
Terracettes are one of the minor unsolved mysteries of geomorphology. Often attributed
to trampling by cattle or sheep, they are sometimes observed in circumstances that make
the cattle explanation very unlikely. C. F. S. Sharpe himself was rather skeptical of cattle
trampling, although other authors have found substantial support for it: There is no doubt
that when cattle or sheep are present they use terracettes and follow trails over them, but
it is less clear that animals create them. The observation of terracette-like, elephant-hide
features on lunar slopes (Figure 8.1) would seem to resolve the controversy in favor of
inanimate causes. Surprisingly little study has been devoted to surface textures on lunar
slopes, however, and there are currently no models that describe how this pattern forms.

EBSCOhost - printed on 11/3/2023 9:57 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
324 Slopes and mass movement

divide

∆z
lan slope =
ds ∆x
urf
ace
∆x
∆z
S
z(x,t) h

regolith
bedrock W

Figure 8.5 The fragmental layer of soil (or regolith) covering the land surface creeps slowly
downhill. Mass balance at any location on the hillslope must account for the influx of material from
farther uphill, the efflux of material downslope, and the conversion of bedrock into soil. Other mass
balance factors, such as mineral dissolution by groundwater, may also have to be taken into account
to produce a quantitative model of soil creep and erosion.

Hopefully, with the advent of high-resolution images of the lunar surface from the Lunar
Reconnaissance Orbiter mission, this situation will change.
The proportionality between soil creep rate and the sine of the slope angle suggests
an appealing analogy between landform degradation by soil creep and thermal diffusion.
Noting that the flux of regolith material is proportional to the gradient of the hillslope,
W. E. H. Culling (1960, 1963) proposed that soil creep obeys a two-dimensional ana-
log of Fourier’s heat equation. Extensions of this work by Carson and Kirkby (1972) and
many modern authors have incorporated the diffusion model into comprehensive models
of landform evolution (Pelletier, 2008). The basic idea (Figure 8.5) is to combine mass bal-
ance with the rate of material transport. In any small area on the slope, regolith is created
by weathering, transported into the area from upslope, or exported by flow downslope.
Interpolating this balance into an infinitesimal area, the mass balance equation becomes:

∂S ∂z
− ( µ − 1) W = − (8.1)
∂x ∂t
where S is the volume flux of regolith moving along the slope, W is the rate of conversion
of bedrock into regolith (weathering rate), μ is the ratio of the volume of regolith to the
volume of parent rock, z is the elevation of the ground surface and x is the distance downhill
from the uphill divide.
A useful deduction from this equation is the thickness of the soil mantling the surface,
h(x, t). Its time rate of change is the difference between the rate of surface elevation change
and the weathering rate of the bedrock:

∂h ∂z ∂S
=W + = µW − . (8.2)
∂t ∂t ∂x

EBSCOhost - printed on 11/3/2023 9:57 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
8.1 Soil creep 325

If the volume flux S is proportional to the sine of the slope angle, which for small angles
is approximately equal to the derivative of z with respect to x, then:

∂z
S = −K . (8.3)
∂x
Using this definition, Equation (8.1) becomes a diffusion equation:

∂2 z ∂z
K + ( µ − 1) W = (8.4)
∂x 2 ∂t
in which K is a generalized “topographic” diffusion coefficient with the usual units of m2/s.
The sign has changed because a negative slope implies a positive mass transport in the
coordinate system defined in Figure 8.5.
The advantage of a diffusion equation is that many solutions to equations of this type
already exist. Not only have entire books been written collecting such solutions (Carslaw
and Jaeger, 1959), but fast numerical algorithms have been devised to solve this equation
under any conceivable set of boundary conditions. This probably explains why the diffu-
sion model is so popular among geomorphologists. Nevertheless, reality intrudes when
slope angles approach the angle of repose and the equation acquires new, slope-dependent
terms that make it non-linear (Roering et al., 1999). One must also remember that many
creep processes depend on the number of freeze/thaw cycles or nearby impact events and
so are not direct functions of time, making (8.4) a useful first approximation that must be
applied only with due caution.
An interesting consequence of Equation (8.4) is that a hillslope that degrades without
change of form (that is, ∂z/ ∂t is negative and independent of x and for which the weather-
ing rate W does not depend strongly on distance from the divide) must be convex upward
(that is, the second derivative in (8.4) is negative). Hills in a landscape that is being eroded
by creep, thus, evolve to be convex upward. Another way of seeing this is to realize that
the farther one goes from a divide, the steeper the hill must become to transport the vol-
ume of material that has been eroded from all the bedrock between one’s position and the
divide. Creep-dominated landscapes look “melted,” as if a plate of ice cream balls had
been left out in the Sun for some time. The diffusive character of Equation (8.4) means
that sharp contours are rounded; steep scarps degrade to gentle slopes and deep incisions
are filled.
The landscape-softening character of diffusion is in sharp contrast to the effects of
fluvial erosion, as will be discussed in more detail in Chapter 10. Fluvial erosion has
the opposite effect: It creates incisions and sharpens soft contours. Because the ero-
sive agent in fluvial processes, runoff, increases in intensity as one moves away from
a divide, slopes dominated by rainwash are concave upward: They begin steeply near
the divide and become more shallow as one proceeds downslope. Landslide-dominated
slopes are intermediate in form: Because slope collapse depends on a fixed angle of
repose with respect to the horizontal, such slopes are straight. Figure 8.6 illustrates
these relationships.

EBSCOhost - printed on 11/3/2023 9:57 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
326 Slopes and mass movement

cliff face

co n
vex

Elevation, z
str ,c
aig re
ht, ep
lan in
ds g
nc lid

co
ave e
, flu
vial

base level

Distance, x

Figure 8.6 Slope profiles can be convex, straight, or concave-upward. Different processes lead to
different profiles: Creep usually produces a convex slope profile, landslides produce straight profiles,
and fluvial erosion produces a concave profile.

The concatenation of the opposite effects of creep and fluvial erosion leads to much of
the topographic variety we enjoy on the Earth’s surface. In contrast, the creep-dominated
lunar landscape is monotonous on a broad scale.

8.2 Landslides
Slow creep processes are not the only type of mass movement. Much more rapid move-
ments take place when the ability of a slope to resist the force of gravity is exceeded and
rock material accelerates downhill until it achieves a new balance with the forces tending
to level the landscape.
Some of the most important contributions to understanding landslides were made by
Charles-Augustin de Coulomb (1736–1806), who is otherwise famous for his contributions
to the electrostatic force law. Coulomb started his career as a military engineer and in 1773
published a memoir treating the conditions of stability of earth retaining walls. He sum-
marized his experimental and theoretical understanding of the resistance of earth materials
to collapse in 1776 with the equation (in modernized notation):
σ s = c + σ n tanφ (8.5)
where σs is the maximum sustainable shear stress, c an empirical constant called cohe-
sion, σn is the stress normal to the plane of shearing, and φ is the angle of internal friction.
Planetary materials can be classified according to the values of c and φ that describe them,
and any discussion of landslides is most logically organized around the various terms of
this equation.
In the years since Coulomb’s revolutionary formula, only one major addition has been
necessary: Austrian engineer Karl von Terzaghi (1883–1963) showed that when pore fluids,

EBSCOhost - printed on 11/3/2023 9:57 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
8.2 Landslides 327

Table 8.1 Angles of internal friction

Material Angle of internal friction

Basalt talus 45°

Granitic gneiss talus 31–36°
Alluvium 41–44°
Glacial till 37°
Shale grit 43°
Sand 33–43°
Silt 32–36°
Cold water ice (77–115 K)a 29°

Data from Carson and Kirkby (1972) unless otherwise noted.

a
Beeman et al. (1988)

such as water, gas, or oil, are present the pore pressure p must be subtracted from the nor-
mal stress (Terzaghi, 1943):
σs = c + (σn − p)tan φ (8.6)
The following sections analyze the effect of each of the terms in this equation.

8.2.1 Loose debris: cohesion c = 0

Loose rock debris has no intrinsic strength: Its cohesion c in Equation (8.6) is zero.
Nevertheless, such material is not “strengthless,” because of friction. So long as normal
stresses, confining pressure, or the weight of overburden act across potential slip planes,
the material resists deformation and a finite shear stress must be applied to make it slip.
Only when the internal friction φ is zero is the material truly without resistance to deform-
ing forces.
Measured values of internal friction are surprisingly uniform (Table 8.1): Even ice near
its melting point has an internal friction angle close to that of rock and this persists down
to very low temperatures. The internal friction of loose materials depends not only on the
mineralogical composition of the material, but also upon such factors as grain size, grain
shape, sorting, and degree of interlocking of the grains. Nevertheless, it is a good guess that
any geological material has an angle of internal friction between about 30° and 45°.
Angle of repose. The angle of internal friction is an intrinsic property of the given mater-
ial. For soil materials it is measured in a shear box in the laboratory. The angle of repose
of a rock slope is the result of the balance between the intrinsic forces of resistance and
the driving force of gravity. Imagine that we have arranged a rock slope so that its tilt can
be adjusted. Starting out with a mass of rock debris on a level surface, slowly increase its
tilt until rocks on the slope begin to slide. The critical angle at which instability sets in is
called the angle of repose, φr.

EBSCOhost - printed on 11/3/2023 9:57 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
328 Slopes and mass movement

mass
m

contact area
A

mg sinα

mg cosα mg α

Figure 8.7 A balance between driving forces and resisting forces determines the stability of a block
resting on a sloping surface. When the resistance is Coulomb friction, the weight of the block can be
decomposed into the force pushing it down the slope and the resisting force, which is proportional to
the force pushing it into the slope. The ratio of these two forces when sliding just begins is equal to
the coefficient of friction. Because this is a ratio, the acceleration of gravity (and mass of the block)
cancel out: The angle of repose is independent of the gravitational acceleration.

The angle of repose is closely related to the angle of internal friction. Referring to
Figure 8.7, suppose a block of rock is resting on a surface sloping at angle α from the hori-
zontal. The relevant portion of Coulomb’s strength Equation (8.5) is:
σs = σn tanφ. (8.7)
The shear stress acting on the base of the rock is equal to its weight mg, where m is its
mass, divided by its basal area A and multiplied by the sine of the slope angle: σ s = (mg/A)
sin α. The sine of the angle comes from resolving the weight of the rock into its downslope
component. The normal stress pressing the rock into the slope is the same expression, but
now multiplied by the cosine of the angle: σ n = (mg/A) cos α. Inserting these expressions
into (8.7), the common factor mg/A cancels out and we are left with the statement that when
instability occurs, α = φr, where:
φr = φ, (8.8)
that is, the angle of repose is equal to the angle of internal friction!
The fact that the acceleration of gravity cancels out of Equation (8.8) comes as a big sur-
prise to many people. This means that barely stable slopes on a low-gravity body such as the
Moon (or an even lower-gravity asteroid) stand at exactly the same angle as they do on Earth.
Space artist Chesley Bonestell painted widely publicized landscapes of the Moon for many
years, which portrayed lunar slopes much steeper than any slopes possible on Earth. A stick-
ler for accuracy in his paintings, he was reportedly horrified when, in his later years, he was
acquainted with the fact that the angle of repose is independent of gravitational acceleration!
The angle of repose need not be exactly equal to the angle of internal friction: Factors
such as interlocking of angular clasts and the difference between static friction and sliding

EBSCOhost - printed on 11/3/2023 9:57 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
8.2 Landslides 329

friction lead to differences of a few degrees. Nevertheless, the two are very close to one
another and laboratory measurements of friction are a good guide to the maximum slopes
observed in a planetary landscape. Now that we have good images and shape models of
irregular asteroids and comets it is possible to compute the local direction of gravitational
acceleration on their surfaces and compare this to the surface slope. In nearly all cases it is
found that the slopes stand at less than the angle of repose, indicating that asteroid regoliths
can be approximated as loose, cohesionless layers of rock debris.
Pore pressure. Pore-filling fluids in a loose mass of rock debris do not change the coeffi-
cient of internal friction, but they can change the angle of repose by supporting some of the
weight normal to potential sliding planes. Because fluids partially relieve the normal stress
but do not contribute to resisting shear stress, slopes that include fluids fail at shallower
angles than dry slopes. The relevant equation is:

σs = (σn − p) tan φ. (8.9)

The analysis of this equation for a fluid-saturated slope proceeds in the same manner as
for a dry slope, although more care is required in balancing all the forces. If the total dens-
ity of the saturated granular material is ρt and the density of the fluid is ρf, it can be shown
(e.g. Lambe and Whitman, 1979) that the angle of repose of the slope is:

 ρt − ρ f 
tan φr =   tan φ . (8.10)
 ρt

The term in brackets is always less than one, so that angle of repose of a fluid-saturated
slope is always less than that of a dry slope. For a water-saturated slope in typical terrestrial
soil, the angle of repose may be reduced by about a factor of 2: φ r ≈ φ /2.
The decrease of the angle of repose as a slope becomes saturated with water describes
the common observation that small landslides are common after heavy rainfalls. However,
the reason that slopes fail is not, as commonly stated, that the water “lubricates” the slope.
The coefficient of internal friction of wet rock debris is indistinguishable from that of dry
debris. The weight of the water also does not promote failure: it increases the shear resist-
ance σn by the same factor that it increases the shear stress σs. The entire effect of the water
is to increase the pore pressure, effectively “floating” the rock debris off its underlying
support and decreasing the shear resistance by decreasing the normal stress.
The importance of pore fluids is not confined to slopes on the Earth. Small dust ava-
lanches on Mars appear to have been triggered by airblasts caused by recent small impacts
on its surface (Burleigh et al., 2009). The probable cause of these avalanches is the rapid
excursions in atmospheric pressure over the sloping surface. When the external air pressure
drops, the Martian atmosphere trapped in the pore spaces of the dust exerts an uncompen-
sated pore pressure and partially lifts the weight of the dust, triggering slope failure. One
such event triggered nearly 100 000 small avalanches within a few kilometers of an impact
that produced a cluster of 20 m diameter craters.
Slope profiles. As shown in Figure 8.6, slope instability leads to straight slope profiles
that stand at the angle of repose. Straight slopes are common on many parts of the Earth.

EBSCOhost - printed on 11/3/2023 9:57 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
330 Slopes and mass movement

(a) (b)

Figure 8.8 Loose surface material drains into gaping fissures along slip surfaces whose slope is
determined by the coefficient of friction in the loose material. The result is subsidence pits whose
width and spacing are approximately equal to the thickness of the loose layer.

Screes and talus cones at the base of steep cliffs are typically straight, created as blocks
from the cliff face fall off and roll some distance downslope. Mountain slopes in seismic
areas of southern California, Taiwan, and Tibet, among others, exhibit nearly linear slopes
from ridge crest to valley bottom. Measurements after major earthquakes have shown that
earthquake-triggered landslides dominate slope degradation in these locations. The slopes
of constructional landforms, such as cinder cones and volcanoes, are often limited by slope
instability and display correspondingly linear slope profiles.
Subsidence pits and grooves on asteroids. Subsidence pits are observed on Earth when
the substrate underlying loose granular material collapses. This occurs naturally over col-
lapsed underground caves in karst terrain and has been observed in Iceland when fissures
open up underneath loose volcanic tephra. It has been noted more commonly in association
with man-made accidents such as the collapse of old mine excavations, water-main bursts,
and the collapse of cavities created by underground nuclear tests.
When loose material drains into a cavity below, it first forms a shallow pit, whose breadth
at the surface is comparable to the thickness of the mantle of loose material (Figure 8.8).
This is because the shear planes in a granular material form at an angle to the direction of
maximum compression that is equal to the angle of internal friction. In the case of draining
material the maximum compression axis is vertical, so the shear planes dip steeply, typic-
ally at angles near 60° from the horizontal. As drainage continues, the pit deepens while
widening slightly until the walls of the pit slope inward at the angle of repose and further
widening takes place as material slides down the steep walls of the pit.
Although the formation of drainage pits may seem rather esoteric, it is the favored
explanation of a very striking feature that appears to be a common characteristic of aster-
oid surfaces. Linear arrays of pits and troughs were first observed on Phobos, the larger
and inner moon of Mars. With widths of hundreds of meters and lengths stretching from
one extremity of Phobos to the other, these pitted grooves are one of Phobos’ major topo-
graphic features (Figure 8.9). Many theories were concocted to explain them: rolling or
bouncing boulders ejected by craters on Phobos (except that they are not all radial to cra-
ters and there are no boulders at the ends of the grooves), secondary craters from impacts
on the surface of Mars (but why are there no similar chains on the surface of Mars itself?),

EBSCOhost - printed on 11/3/2023 9:57 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
8.2 Landslides 331

Figure 8.9 Grooves on Phobos are likely to be the result of thick (100 to 200 m) regolith draining
into gaping fissures in the body of the moon. Grooves are up to 200 m wide. Mars Express image
20080723, ESA/DLR.

and a number of others. The most plausible and widely accepted interpretation is that they
are lines of drainage pits in regolith undermined by gaping fractures in the body of Phobos
(Thomas et al., 1978). Large parts of Phobos are currently in an extensional state of stress,
caused by tidal forces because it is inside Mars’ Roche limit. The width and spacing of the
grooves then indicates the depth of the regolith (Horstman and Melosh, 1989).
Although Phobos is in a rather special stress state, grooves have now been reported on
nearly every asteroid that has been imaged at high resolution, including 951 Gaspra, 243
Ida, 433 Eros, 2867 Šteins, and the 130 km diameter asteroid 21 Lutetia. None of these
is subject to tidal extension, but perhaps large impacts temporarily opened large fissures
into which regolith could drain. Once some regolith had fallen into the fissure, it would
have jammed the fissure open and more regolith could follow, creating a groove. It may
be significant that some of the largest pit chains are observed on Šteins, which was nearly
destroyed by a very large impact.

8.2.2 Cohesive materials c > 0

Without cohesion, vertical scarps cannot exist. Even the centimeter-high scarps at the sides
of the Apollo astronauts’ bootprints imply that the lunar regolith has some degree of cohe-
sive strength. The vertical and even overhanging walls of sand castles at the beach imply a
small degree of cohesive strength: Dry sand alone can, at best, stand at the angle of repose.
The secret of successful sand castles (known to every sand-castle builder) is that the sand
must be damp. The surface tension of the water films between sand grains bonds the grains
together and imparts a small cohesive strength to the mass that permits vertical walls. If

EBSCOhost - printed on 11/3/2023 9:57 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
332 Slopes and mass movement

density ρ
g
Hc

φ
45° + /2

Figure 8.10 Collapse of a vertical cliff in a homogeneous mass of material possessing both internal
friction (parameterized by the angle of internal friction φ) and cohesion. Failure of such a cliff takes
place preferentially along a plane dipping at an angle 45°+φ/2 with respect to the horizontal.

the sand becomes completely saturated with water the films disappear and the cohesive
strength (and sand-castle ability) disappears.
The most general kind of material strength includes both cohesion and internal friction.
There are materials that have cohesion but lack significant internal friction, but they are
relatively rare. The internal friction of clays can be neglected for rapid flows (soil engin-
eers call this the “undrained” condition) because clays are so impermeable that water can-
not be expelled from between the grains and the pore pressure then equals the overburden
pressure, so that the term (σn–p) in Equation (8.6) vanishes, canceling the dependence on
internal friction. The same clays, if deformed slowly (under “drained” conditions) show a
strong dependence on internal friction (Lambe and Whitman, 1979).
The failure of many metals is also nearly independent of pressure, hence, these also have
negligible internal friction. Although the naked cores of planetesimals could potentially
form iron–nickel metal planets, we have no examples yet of landscapes on such bodies.
Presumably they would resemble terrain on clay-rich soils, although on a much larger size
scale due to iron’s much larger strength. Astrophysicists might consider such landscapes on
the surfaces of white dwarfs or neutron stars.
Stability of vertical cliffs. Steep or vertical cliffs are the characteristic features of cohesive
materials. The most important concept is the relation between cliff height and cohesion.
Coulomb, in his 1773 memoir, was the first to analyze this problem in its full generality,
including internal friction (Gillmor, 1971). Coulomb, working as a military engineer at
that time, had been assigned to construct Fort Bourbon on Martinique and this problem had
immediate practical applications. The useful formula he derived is an example of science
taking advantage of military technology.
Coulomb’s analysis of the stability of a vertical cliff was a classic application of calculus
to a practical problem. His method is shown in Figure 8.10. Coulomb supposed that the
failure surface was most likely to be a plane extending from the foot of the cliff upward
to the surface at the top of the cliff (this supposition was motivated by his observation of
actual slope failures). He resolved the weight of the triangular prism above into shear and
normal components to the surface and compared these driving forces to the resistance
exerted by cohesion and friction on the plane, given by Equation (8.5). The ratio between
resistance and driving force is a function of the slope of the failure plane. Coulomb then

EBSCOhost - printed on 11/3/2023 9:57 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
8.2 Landslides 333

applied calculus to locate the minimum (that is, the plane on which failure is most likely),
which occurs at an angle of 45°+φ /2 from the horizontal. He used this angle to solve for the
height of a cliff at the limit of stability, Hc:

2c
Hc = tan ( 45° + φ / 2 ). (8.11)
ρg

As one might expect, increasing cohesion increases cliff height, as does increasing the
coefficient of friction. Higher density ρ and, especially, acceleration of gravity g decreases
the possible height of a cliff.
When formula (8.11) is compared to the heights of natural vertical cliff faces it is found
that it generally predicts cliffs that are too high. This is mainly because the cohesion of rock
in the laboratory is usually measured on unweathered, intact rock samples in an experiment
that lasts only seconds. In nature, rocks are filled with joints and other fractures, so their
bulk strength is less than that of a small laboratory specimen. Furthermore, the long-term
strength of rock is generally less than that measured on a short timescale because of slow
processes like stress corrosion cracking. A common procedure, then, is to turn Equation
(8.11) around and use the maximum heights of observed cliff faces to deduce the long-term
strength of jointed rock. Panama Canal engineer David Gaillard used this scheme in 1907
to deduce the safe angle for hillside cutbacks, a decision that has stood the test of time.
Equation (8.11) has already performed good service in the cause of planetary science.
Shortly after Voyager 1 discovered sulfur on Io, Clow and Carr (1980) used a generalized
version of this equation to show that the 2 km high scarps of an Ionian caldera were too tall
to be composed of pure sulfur and deduced that Io’s crust must be made of much stronger
material. On the other hand, many journalists were astonished in 2004 when the Stardust
spacecraft sent back images of ~100 m high vertical cliffs on comet Wild 2. After all, aren’t
comets supposed to be very weak? However, application of Equation (8.11) to Wild 2’s
gravity field indicates that its material need be no stronger than the weakest soufflé: A cliff
of the same material under Earth’s gravity would collapse if its height exceeded 3 mm!
Figure 8.11 illustrates the varieties of cliff collapse that have been noted on Earth, many
of which have now been observed at the base of steep Martian cliffs as well. Cliff collapse
varies from spalling of rock slabs that extend from toe-to-crest, to grain-by-grain disintegra-
tion. The style of degradation of any individual cliff depends upon its composition, weath-
ering characteristics, and the mechanical condition of the rock (or ice) composing it.
Rotational slumps. How is the stability of a scarp that is not vertical determined?
Figure 8.12 illustrates the general case of a scarp of height Hc but with a face sloping at
angle α. In this case, observation first indicated that the failure surface is approximately
a section of a cylinder, not a plane. The analysis of the stability of such a slope is much
more complex than for a vertical scarp and is accomplished by the “method of slices”
(Lambe and Whitman, 1979) or the still more sophisticated “slip line analysis” (Scott,
1963). The simplified results of Scott’s widely used analysis are shown in Figure 8.13.
The right-hand side of this plot, for slope angle 90° (a vertical cliff), agrees with Equation
(8.11), but for other slope angles there is no simple analytic formula. This plot makes it

EBSCOhost - printed on 11/3/2023 9:57 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
Slab failure Rock avalanche Rock fall Granular
disintegration

Figure 8.11 Collapse of actual cliffs varies from the ideal because of structures in the rock mass.
Slab failure occurs when vertical joints control the strength of the rock face. Rock avalanches are
influenced by shallow vertical joints. Rock falls occur where the rock face disintegrates into large
blocks and granular disintegration of rock faces occurs where weathering easily releases small rock
fragments or grains of sedimentary rocks.

Hc
x cg

Figure 8.12 Rotational slumps occur in material with cohesion but little internal friction, so that
sliding on the deep-seated failure surface is not inhibited by friction. The center of gravity of the
mass, the point marked cg, moves downward as the slide progresses.

100

50
ρgHc/c

φ = 25°
20°
15°
“Safety factor”

10 10°

φ = 0°
5
4

3
0 10 20 30 40 50 60 70 80 90
Slope angle α, degrees

Figure 8.13 Slope stability as a function of slope angle for straight escarpments of the type shown
in Figure 8.12 in a material with both internal friction and cohesion. The curves are labeled by the
internal friction angle of the material. Greatly simplified after Figure 9–20 of Scott (1963).

EBSCOhost - printed on 11/3/2023 9:57 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
8.2 Landslides 335

clear that rotational slumps are more likely in material with very low angles of internal
friction: The stability of a scarp increases rapidly as internal friction increases and slope
angle decreases. On Earth, this means that rotational slumps are mostly confined to soils
that contain large amounts of wet clay, in which the pore pressure cancels the contri-
bution of internal friction. One would, thus, not predict that rotational slumps would
be common on airless bodies such as the Moon, because no pore fluids are available. It
therefore comes as a surprise to find that rotational slumps are one of the most common
landforms on the Moon – but they are confined to the rims of complex impact craters.
Box 8.1 explores this anomaly in more detail.
The most characteristic feature of a rotational slump is its head scarp. As illustrated in
Figure 8.14, the steeply dipping failure surface is exposed at the uphill crest of the land-
slide. One also typically sees several small terraces downhill of the main detachment. The
original ground surface rotates backward on the tops of these terraces; it dips toward the
detachment and often creates small, closed depressions that can trap water or other liquid,
forming ponds. The small ponds of impact melt on top of slump terraces in Copernicus
crater on the Moon are good examples of this process.
As the slump rotates along the failure surface, its overall center of mass drops downward,
but at its toe the ground is usually upheaved. The material at the foot is pushed upward into
an irregular bulbous, hummocky mass that overrides the original surface.
Toreva block landslide. A distinctive variant of landslides in cohesive material is called
the Toreva block landslide, after a small town on the Hopi reservation in Arizona where
many such landslides occur. First described by Reiche (1937), this type of landslide devel-
ops where a strong layer overlies a weak one (Figure 8.15). As the weak layer is eroded it
undermines the strong unit above, which eventually collapses as a nearly intact block, often
rotating backward like the head of a rotational slump, and plowing up the weak material
below it. This kind of detachment may occur several times in succession, leading to stair-
step topography at the edges of mesas capped with resistant rock. The continued evolution
of escarpments depends crucially upon the activity of other processes to remove the debris
that has already collapsed off the cliffs: Otherwise, the mechanical support of the blocks
that have already slid off the face will continue to support the scarp.
Although Toreva block landslides are characteristic landforms of the Colorado Plateau in
the southwestern Unites States, they may occur wherever strong rock units overlie weaker
ones. In volcanic terrains it is not unusual to find strong lava flows interbedded with weak
tephra: Large blocks lying below the high scarp at the base of Olympus Mons on Mars may
have originated in this manner.
Avalanche chutes. Although not strictly landslides themselves, avalanche chutes and
slope flutes are characteristic landforms that develop on steep slopes dominated by rock-
falls. Repeated rockfalls eventually erode U-shaped troughs into the slope. Probably starting
as small indentations in the slope, these chutes channel further rockslides that then deepen
and lengthen the chutes by abrasion and plucking of blocks along their beds. As avalanche
chutes grow and encroach on one another, the head of a steep cliff becomes fluted with
near-vertical troughs whose spacing is often surprisingly regular. Such small-scale flutings,

EBSCOhost - printed on 11/3/2023 9:57 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
336 Slopes and mass movement

Box 8.1 Crater terraces as slump blocks

One of the most distinctive features of large, complex impact craters is the wreath of terraces
that step down from the rim into the interior of the crater. The terrace just below the crest of
the rim is usually the widest, and the scarp above it is the highest, of the series. This is the
only terrace visible in many craters because crater floor deposits bury structures closer to the
center. When they can be seen, the terraces become progressively narrower and the scarp offset
smaller as one proceeds from the rim toward the center of the crater.
Crater terraces show the backward-rotated block morphology typical of slump terraces in
landslides. Early Lunar Orbiter images of the Copernicus and Tycho craters on the moon reveal
level ponds of presumed impact melts trapped among its terraces, an observation that now
appears to be common in both lunar and Martian craters (Figure B8.1.1), similar to the water-
filled ponds that form at the head of large terrestrial rotational slump landslides.
Terraces in complex craters, thus, closely resemble the terraces that form at the head of
landslides in wet clay: That is, in materials which possess cohesion but negligible internal
friction. The story of how lunar rock, in the absence of either air or water, can behave like
water-saturated clay is related to the mobility of long-runout landslides and is described in
Sections 6.3.3 and 8.2.4. For the present, this box explores the implications of assuming that
the mechanical strength of a planetary surface shortly after crater excavation can be treated
as if the material possesses cohesion but no internal friction. This type of reasoning is called
a “phenomenological model”: We do not know why the strength behaves this way, but we can
show that, if we assume that it does have this behavior, the model’s predictions agree with the
observations.

Figure B8.1.1 Terraces beneath the rim of the Bürg crater on the Moon. Bürg is a complex crater
40 km in diameter with clearly defined wall terraces, a smooth floor, and central peak. The lack
of perfect symmetry probably reflects the pre-impact structure in the target. Portion of LROC
WAC monochrome context image M119666881ME, NASA/GSFC/Arizona State University.
Courtesy Mark Robinson.

EBSCOhost - printed on 11/3/2023 9:57 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
8.2 Landslides 337

Box 8.1 (cont.)

The events that immediately follow the impact of a meteoroid on a planetary surface are
described in Section 6.3. The result of crater excavation is a transient crater, a bowl-shaped
cavity with a depth/diameter ratio of about 1/2.7. This ratio implies an average internal slope
of about 20°. This is steep, but not steeper than the angle of repose of most rock materials. In
other words, such a crater would not normally be expected to undergo much collapse. For lunar
craters smaller than about 15 km in diameter, this expectation is borne out: The rim crumbles
and slides into the crater, forming a breccia lens on its floor and slightly enlarging its diameter.
However, for larger craters the rim slumps down, forming terraces and widening the crater,
while the floor of the crater rises, greatly decreasing its depth.
This collapse process can be analyzed by the method of slip line analysis, mentioned in
Section 8.2.2. Application of such an analysis to craters with a parabolic profile (Melosh,
1977) shows that parabolic craters are stable until the dimensionless combination ρgHt/c
exceeds about 5. This parameter is the same as a soil engineer’s “factor of safety,” in which
Htis the depth of the transient crater. Slope failure, in which a segment of the rim slides into the
crater, produces a single terrace for values of ρgHt/c between 5 and 10. When ρgHt/c exceeds
15, the floor beneath the center of the crater rises almost vertically upward as the rim slumps
downward (Figure B8.1.2). In summary:
0 ≤ Hf ≤ 5c/ρg stable (B8.1.a)

(a)

Slope
failure

(b)

Toe
failure

(c)

Floor
failure

Figure B8.1.2 Three degrees of collapse of crater-like depressions in cohesive material that
possesses little or no internal friction. (a) Simple slope or wall collapse in which a layer slides
into the crater depressions. This is typical of simple craters, in which the oversteepened rim
collapses into the bowl to form a breccia lens. (b) Toe collapse, in which the toes of the sliding
masses meet at the base of the crater and uplift a small plug. This is transitional to (c) floor uplift
in which the walls slump down and inward, forming terraces, while the floor rises as a semi-rigid
plug.

EBSCOhost - printed on 11/3/2023 9:57 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
338 Slopes and mass movement

Box 8.1 (cont.)

5c/ρg < Hf ≤ 15c/ρg slope failure (B8.1.1b)
Hf > 15c/ρg floor failure. (B8.1.1c)
Because the failure of purely cohesive material is independent of overburden pressure, the
collapse threshold is independent of the initial crater’s diameter – only its depth is significant.
As the crater collapses, material beneath it flows until the stress differences beneath all portions
of the crater floor drop below the cohesion, at which point collapse ceases. This model, thus,
predicts that the final depth Hf of all complex craters is a constant equal to the depth of a
transient crater at the onset of collapse – about 4.5 km on the Moon. The prediction that craters
in a cohesive material collapse back to a constant depth also accounts for the development of a
broad, flat floor in the interior of large, complex craters. Observationally, the depth of complex
craters does depend somewhat upon diameter, but it only varies from about 2.5 to 6 km over a
diameter range of 20 to 300 km. This variation could be explained by a very small coefficient of
internal friction or by a small dependence of the effective cohesion on crater diameter.
The pure cohesion collapse model also predicts the width of each individual terrace as it
forms. A simple analytic formula that gives a good approximation to the numerical slip line
results for the terrace width is:

c  1 + 16 λ 2 
w= . (B8.1.2)
ρ g  16 λ 2 

where λ is the depth/diameter ratio of the crater at the time when the terrace forms. The first
terrace to form, when the depth/diameter ratio equals that of the transient crater, is thus the
narrowest, whereas the last terrace to form, when the crater’s depth/diameter ratio is only
slightly larger than its final value, is the largest, in qualitative agreement with observation.
A quantitative test of this equation against the widths of the final terrace in a suite of lunar
and Mercurian craters (Leith and McKinnon, 1991; Pearce and Melosh, 1986) indicates an
effective cohesive strength of 2–3 MPa for the post-impact strength of both bodies and also
agrees with the effective cohesion at the onset of crater collapse.
The phenomenological description of the transient strength of terrestrial planetary surfaces
as purely cohesive, with negligible internal friction, thus, agrees well with observations.
Application of the same model to the icy satellites indicates that the model also works on
these bodies, but with an effective strength about 1/3 that of the rocky planets. This model
does not explain the formation of central peaks in complex craters, only the shallow floor. A
further mechanical property, viscosity, must be added to the model to explain these features.
The rheology needed to more fully model impact-crater collapse is thus more complex – it
approximates a Bingham fluid.

spaced tens to hundreds of meters apart, adorn the crests of many of the deep canyons on
Mars as well as the Earth (look ahead to Figure 8.17b for good examples).
Granular disintegration produces similar landforms on a much finer scale. The factors
that determine the spacing of avalanche chutes are presently unclear: A better understanding

EBSCOhost - printed on 11/3/2023 9:57 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
8.2 Landslides 339

slump terraces

bulged toe

rupture
surface

Figure 8.14 The ultimate result of a rotational landslide slump is a terraced head scarp, in which the
terraces often rotate the original surface backward into the slope (forming closed depressions), and a
bulging toe that pushes the original ground surface upward into an irregular mound.

strong

weak

Figure 8.15 Toreva block failure occurs where a mechanically strong layer overlies a weak layer.
When failure occurs, the weak layer is compressed and expelled from underneath the strong layer,
which subsides as a coherent block. Often multiple failures of this type produce a stair-step structure
on the hillside below a mesa capped by the stronger layer.

of this process might give some insight into the properties of the rock that forms the
cliff face.

8.2.3 Gravity currents

Once a slope collapses, the mobilized material may move long distances if it becomes
fluid. Dry landslide debris usually moves only a short distance before coming to a halt,
but when liquids or gases are involved an emulsion of solid and fluid may travel consid-
erable distances at high speed. The driving force behind such flows is gravity, so such
flows are known as gravity currents. There are many different types of gravity current,
ranging from flows in the atmosphere and the ocean as well as on land (Simpson, 1999).
Dry snow avalanches are one type of such flows, as are pyroclastic flows from explosive
volcanic eruptions or turbidity currents initiated by slope failures under the ocean. Lava
flows can be considered a very slow variety of gravity current. Inertial forces may or may
not be important in such flows, depending on their speed. When inertia dominates the flow,

EBSCOhost - printed on 11/3/2023 9:57 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
340 Slopes and mass movement

parabolic front

U
h

Figure 8.16 Currents of dense fluids flowing down a slope beneath a less dense fluid develop frontal
lobes with approximately parabolic profiles, ending in a sharp reversal where the Froude number
reaches a critical threshold near √2. Fluid behind this front flows in a dense sheet behind the front.
Currents of this type are seen both in air (dust storms called haboobs) and under water (turbidity
currents).

a characteristic bulge develops at the head of the flow (Figure 8.16). The head, of height
h1, corresponds to a hydraulic jump where the Froude number U / gh1 is approximately
equal to √2.
Debris flows. Debris flows are a variety of gravity current that has been analyzed in
detail by Arvid Johnson (Johnson, 1970). Debris flows are dense mixtures of mud, rocks,
and water that are produced by torrential rains on the Earth. After a heavy rain that saturates
and mobilizes the loose debris on a slope, the mixture may roar down a steep canyon with
the sound of a speeding freight train. When it emerges from the mouth of a canyon it con-
tinues downslope as a dense slug of debris, shedding material from its sides and leaving
two parallel ridges behind that are often so regular that they look artificial. It eventually
comes to a halt as the slope shallows, leaving a heap of coarse stones armoring its snout,
while a rush of muddy water may continue some distance downslope.
A dense mixture of mud, rocks, and water behaves as a Bingham fluid (Section 5.1.3)
that flows as a viscous liquid after its yield stress is exceeded. The velocity profiles in deb-
ris flows are, thus, very similar to those for lava flows, Figures 5.13 and 5.14, although the
Bingham yield stress and viscosity have different values (Rodine and Johnson, 1976).
Debris flows are not entirely confined to the Earth. Many features of the fluidized ejecta
blankets around Martian impact craters are suggestive of debris flows and may be due to
admixture of subsurface water with the ejected debris (Figure 6.16). The fact that such
craters are confined to low latitudes and must be more than a few kilometers in diameter
suggest that they have breached a subsurface water table, although this interpretation is not
universally accepted (Carr, 2006).

8.2.4 Long-runout landslides or sturzstroms

Mass movements that do not involve a fluid such as water or volcanic gases typically move
rather slowly. However, there is a class of landslide that that been observed to move at

EBSCOhost - printed on 11/3/2023 9:57 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
8.2 Landslides 341

speeds ranging from 50 to 100 m/s while dry. Although the deposits of such landslides have
been known for a long time, their remarkable mobility was not appreciated until a landslide
obliterated the Swiss village of Elm in 1881. The disaster was initiated by the villagers’
own mining operations, which undermined a steep slope overhanging the town. The occur-
rence of a landslide was, thus, not a surprise. However, after it began, the dry rock debris
traveled several kilometers across the valley floor and even climbed some distance up the
opposite slope, overrunning the town as it went. Eyewitnesses measured its velocity to be
approximately 45 m/s. The most surprising aspect of this landslide was that the mass of rock
debris did not so much slide as flowed “like a torrential flood,” as described by geologist
A. Heim (1882), who investigated the event shortly after its occurrence. Heim also coined
the German name sturzstrom (literally, “collapse river”) to emphasize the flow of the debris
stream. Despite the appearance of flow, geological study indicated that stratigraphic rela-
tions between rocks in the debris lobe were preserved throughout emplacement.
Occurrence and morphology. Heim’s general observations have been repeated many
times since for both historic (e.g. Frank, Alberta 1903) and prehistoric landslides. In the
meantime, examples of this sort of landslide have multiplied enormously and the sizes esti-
mated for these floods of rock debris have correspondingly increased (Collins and Melosh,
2003): From the 5 km long prehistoric Blackhawk landslide in Lucerne Valley, CA, to the
50 km Shasta terrain at the foot of Mount Shasta, CA, to gigantic submarine landslides 200
km in length off the coast of Hawaii. As geologists have explored further, the deposits of
long-runout landslides seem to be ever more prevalent.
All of these landslides appear to have started as ordinary, although large, rockfalls trig-
gered by a variety of causes: Earthquakes, volcanic eruptions, heavy rainfall, and human
activity, including underground nuclear explosions. The 2002 Denali earthquake in Alaska
unleashed some of the most recent (Figure 8.17a). All of these landslides are characterized
by velocities in the range of 50 to 100 m/s, lack of an obvious fluidizing agent, and preser-
vation of initial stratigraphy (implying laminar flow, in spite of their high velocity). Some
are characterized by longitudinal striations, but this is not universal: Transverse ridges
cross the surface of the Blackhawk slide.
Long-runout landslide deposits are also among the most ubiquitous features on the sur-
faces of solid planets. Wherever steep slopes occur, long-runout landslides are likely to be
found. They have been documented on Mars (Figure 8.17b), Venus, the Moon, Callisto, Io,
and even on Mars’ tiny moon Phobos (Collins and Melosh, 2003). The one other require-
ment is that they always involve large volumes of material: Whatever “magic” allows dry
rock debris to flow like a liquid, it does not work for volumes less than about 106 m3 (on
the Earth: the minimum is larger on Mars). This large-volume-only constraint makes it dif-
ficult to investigate these landslides in the lab, although the Soviet government of Russia
did create a few long-runout landslides during construction of debris dams on large rivers,
as well as at their Novaya Zemlya nuclear test site.
Low coefficient of friction. Another surprising feature of long-runout landslides is their
ability to travel over astonishingly low slopes. This accounts for their tremendous lengths
and implies that large volumes of fast-moving rock debris possess very low coefficients of

EBSCOhost - printed on 11/3/2023 9:57 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
342 Slopes and mass movement

(a)

(b)

Figure 8.17 Panel (a) is a landslide triggered by the 2002 Denali earthquake in Alaska, looking west
toward the divide of the Black Rapids and Susitna glaciers. Image courtesy Dennis Trabant and Rod
Marsh, USGS. See also color plate section. Panel (b) shows landslides on the south wall of Valles
Marineris of Mars. The image is 60 km across. Viking Orbiter image 14A30.

friction. Because the center of mass of these deposits before and after deposition cannot
often be determined, a common metric for describing them is an effective “coefficient of
friction” that is equal to the ratio between their height of fall H (measured from the head of
the scarp from which they fell to their extreme toe) to the length of runout L (also measured
from the head scarp to the toe). The arc-tangent of this ratio can be considered to be a kind
of friction angle. For typical small rockslides this ratio is close to 0.6 (implying an angle
between scarp and toe of about 30°, close to the normal angle of repose). However, as the
data in Figure 8.18 show for both Earth and Mars, this ratio declines as the volume of the
slide increases, obeying a crude power law.
H/L ∝ V−0.16. (8.12)
H/L drops to about 0.03 for the largest landslides (implying an effective friction angle
of only 1.7°). Note also that this ratio is about a factor of 3 smaller for Earth than it is for
Mars, suggestively similar to the ratio between their accelerations of gravity.
The extremely long runout for large landslides also means that they are extraordinarily
dangerous on an inhabited Earth. Current building codes do not take this long reach into
account, but it is sobering to note that the city of Osaka, Japan, is built upon the debris from
an ancient landslide of this type.

EBSCOhost - printed on 11/3/2023 9:57 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
8.2 Landslides 343

1
H/L = 0.6

Effective coefficient of friction, H/L

Earth Mars

(Fall height / Runout length)

0.1

V = 10 6 m 3

0.01
0.0001 0.001 0.01 0.1 1 10 100 1000 10000 100000
Volume, V (km3)

Figure 8.18 The effective coefficient of friction for large landslides depends on the landslide volume.
This coefficient of friction is defined as the ratio between the vertical drop H between the head of the
landslide and its toe, divided by the horizontal distance L between the head of the landslide and its
toe. The square points on this plot are terrestrial landslides and the triangles are Martian. Despite the
large scatter, this plot shows that landslides with a volume less than about 106 m3 are limited to H/L
= 0.6, but this decreases slowly for larger volumes. Large Martian landslides are less mobile than
terrestrial landslides by about a factor of 3. Figure from Collins and Melosh (2003).

Mechanism. The mechanism that permits long-runout landslides to attain such a high
fluidity has intrigued geologists and physicists since Heim first described them in 1882.
None of these flows shows evidence for excess water. Many writers have sought the action
of some fluid other than water, including air, steam generated during sliding, or carbon
dioxide evolved from calcined limestone blocks. Geologist Ron Shreve (Shreve, 1966a)
proposed the widely accepted idea that the Blackhawk and Sherman Glacier slides floated
out across low terminal slopes on a cushion of air trapped underneath the flowing rock deb-
ris, but the discovery of long-runout landslides on Mars, the Moon, and airless bodies such
as Callisto and Io tends to discount atmospheric gas as a lubricant. Physical mechanisms
such as basal melting during sliding are more universal, and glass is found mixed within
a few landslides, such as the Köfels slide in Austria. However, examinations of the bases
of many other long-runout landslides fail to find any evidence of heating, suggesting that
the glass may form during the stopping phase when normal values of rock friction reassert
themselves while the mass is still in motion.
I have proposed a process that I call “acoustic fluidization” to explain the mobility of
long-runout landslides (Melosh, 1979, 1983). The basic idea is that strong internal vibra-
tion in the moving debris builds up during the initial collapse phase to the point that it
effectively liquefies the rock debris. Vibration is widely used in industrial operations with
granular materials to enhance their flow, so this part of the process is not mysterious. As
long as the debris flows down a small slope it gains gravitational energy that can offset the

EBSCOhost - printed on 11/3/2023 9:57 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
344 Slopes and mass movement

dissipation of vibrational energy (which is lost mainly by radiation through the bottom of
the slide). The minimum volume limit arises because energy is lost mainly through the
surface of the slide: The moving mass must be big enough that the volume/area ratio is suf-
ficient to retain most of the energy released by sliding. Motion stops when the slide spreads
out and becomes too thin: On Earth, this happens when its thickness drops below 10 to 20
m. This mechanism does not require the presence of any fluid and should work well in a
vacuum. It leaves no melt or any direct evidence of the former presence of strong vibra-
tions in the rock debris, except perhaps for a peculiar fracture pattern that Shreve identified
and called “domino breccia.” It is, however, unfortunate that acoustic fluidization does
not leave more evidence, for it is difficult to demonstrate that it occurred in any landslide
deposit. No active landslide has yet contained the instrumentation necessary to demon-
strate the presence of acoustic fluidization, so debate continues about the mechanism that
imparts low coefficients of friction to long-runout landslides.
Implications. Long-runout landslides are an effective agent of landform degradation. It
appears that the so-called “sector collapse” of large volcanic cones is a major factor in the
ultimate leveling of volcanic constructs. These collapse events typically shed long-runout
landslides and transport volcanic debris long distances from their source. Landsat images
of the Llullaillaco and Socompa volcanoes near the border of Chile and Argentina show
broad sheets of landslide debris that stretch up to 100 km from their source volcanoes.
The Hawaiian volcanoes also appear to degrade by large catastrophic landslides that, in
this case, are under water and so escaped recognition for a long time. The steep scarp that
forms the north side of Molokai is one of its most impressive features. We now know that
this scarp is merely the head scar of an enormous landslide whose debris blankets the sea
floor for 200 km to the north (Moore et al., 1989).
Large long-runout landslide deposits are also seen at the foot of Venusian volcanoes.
However, the largest known debris landslides in the Solar System form the “aureole” at
the base of Olympus Mons on Mars (McGovern et al., 2004), Figure 8.19. These deposits
extend up to 750 km from the base of the volcano. They originated from the 10 km high
scarp at the base of Olympus Mons and, once they began, flowed horizontally down a slope
averaging only about 0.5°. McGovern et al. (2004) estimated the volume of a single lobe to
be about 8.8 × 104 km3: It was shed from the collapse of a block 10 km thick, 60 km wide,
and 150 km long at the northern base of the volcano.
Gravity-driven mass movement, which was an underrated process at the beginning of
the last century, is now regarded as a highly effective agent of landform degradation. It is
ubiquitous on sloping surfaces and ranges in speed from the imperceptibly slow creep of
a thin surficial layer to nearly sonic speeds in immense landslides that can transport debris
over significant fractions of a planet’s radius.

Further reading
The best quantitative discussion of mass wasting is the now classic book Carson and
Kirkby (1972). A more qualitative discussion is part of most books on geomorphology, but

EBSCOhost - printed on 11/3/2023 9:57 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
Exercises 345

Figure 8.19 Oblique view of Olympus Mons, shown topographically draped over a Viking image
mosaic. The topography shows the relationship between the volcano’s scarp and massive aureole
deposit that was produced by flank collapse. The base of Olympus Mons is 600 km in diameter and
the summit caldera is 24 km above the surrounding plains. The vertical exaggeration is 10:1. NASA/
MOLA Science team PIA02805.

a good, focused discussion is in Selby and Hodder (1993). The stability of hillslopes is the
traditional topic of rock mechanics (for strong rocks), of which an excellent introduction
is Jaeger et al. (2007), or soil mechanics (for less cohesive materials), for which I recom-
mend Lambe and Whitman (1979). Turbidity currents of all kinds are discussed in Simpson
(1999), although this book is very short on quantitative detail. Sturzstrom and rockfalls are
treated in Erismann and Abele (2001). For more information the reader is referred to the
references cited in Section 8.2.4.

Exercises

8.1 Creepy lunar regolith

The Apollo 17 heat-flow probe measured a “daily” (that is, 29.5 Earth day) temperature
variation of 300 K at the surface of the regolith, which fell to about 30 K at a depth of 10
cm (for the purposes of this exercise it is permissible to approximate the temperature vari-
ation as linear – for extra credit, how could you make this more realistic?). If the linear
coefficient of thermal expansion of the powdery regolith equals that of basalt (6 × 10–6 K-1),
estimate the rate of downhill creep expected for a 30° slope exposed to full sunlight on the
moon. Use the theory of creep described in Section 8.1.1 and assume a maximal creep rate
with recovery factor r = 0. Estimate the rate of slope movement in terms of mass per unit
width of the slope. Is this result reasonable in view of the slope in Figure 8.1?

8.2 Diffusing topography

Figure 8.1 shows a bench at the base of a steep-sided lunar ridge viewed at near-vertical
incidence. Use the Culling theory of landform degradation described in Section 8.1.2 to

EBSCOhost - printed on 11/3/2023 9:57 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
346 Slopes and mass movement

roughly estimate the topographic diffusion coefficient K for the Moon’s surface by assum-
ing that the scarp (part of the Flamsteed Ring) is about 4 × 109 yr old. Make the same type
of estimate for the bowl of Meteor Crater, Arizona, using the fact that the crater, which is
50 000 yr old, 1.2 km in diameter, and 180 m deep, has presently been filled to a depth of
30 m by post-impact sediments. Compare the topographic degradation rates of the Moon
and the Earth. Does this tell you why we see so few impact craters on the Earth, compared
with the Moon?

8.3 Splashback creep

Surface creep can be initiated by either rainsplash impact on the Earth or meteorite impacts
on the Moon. When either process excavates a crater of volume Vc, a similar volume of
ejecta is launched out of the crater and travels a distance of roughly one crater radius Rc
before splashing down onto the surface. The time it remains in flight between ejection and
splashdown is given roughly by the ballistic time of flight equation t f = 2 R / g for a par-
ticle to travel a distance of R in a gravity field g, assuming a launch elevation angle of 45°.
On a flat surface the center of mass of the ejecta lies at the center of the crater. However, on
a surface sloping at an angle θ, the center of mass of the ejecta moves downslope under the
acceleration of gravity while it is in flight. Derive an equation that expresses the distance
the center of mass of the ejecta moves downslope as a function of crater radius, gravity, and
the surface slope. Use this equation to derive the downslope volume flux of material (m3
per meter of slope contour distance) moved by a surface-covering barrage of craters whose
total volume just equals the volume of the slope’s surface layer down to a depth equal to ¼
of the crater diameter 2Rc. If the duration of this barrage is T, use Equation (8.3) to relate
the downslope volume flux to the topographic diffusion coefficient K. Finally, use the fact
that the Moon’s upper 1 cm of surface material is overturned by micrometeorite impacts
about once every 70 000 yr to estimate the numerical value of K in m2/s. At this topographic
diffusion rate, how long will a fresh impact crater 1 km in diameter and 200 m deep persist
on the surface? What does this tell you about the importance of impact-driven creep for the
Moon’s surface features?

8.4 Following a French cliff-hanger

Following Coulomb’s method of 1773, described in Section 8.2.2, derive Equation (8.11)
for the maximum height of a vertical cliff in a homogeneous mass of material that pos-
sesses both cohesion and internal friction. Apply this equation to compute the maximum
height of a vertical Ionian caldera scarp by using the estimate of Clow and Carr (1980) that
the cohesion of Io’s porous S-SO2 crust is about c = 0.3 MPa and its coefficient of friction
ff = 1.73.

EBSCOhost - printed on 11/3/2023 9:57 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
Exercises 347

8.5 Strong Martian lavas?

The enormous volcanic caldera at the summit of Olympus Mons on Mars is 80 km across
and is surrounded by near-vertical cliffs between 2.4 and 2.8 km high. Its floor stands 24
km above the surrounding plains. Use Equation (8.11) to estimate the minimum strength of
the lavas in its walls and compare this with the strength of terrestrial lava flows.

8.6 Groovey Lutetia

On 10 July 2010 the European Space Agency’s Rosetta spacecraft flew by asteroid 21
Lutetia. Lutetia is a rocky body that is irregular in shape, with dimensions about 132 by
101 by 76 km. Although Lutetia is a main-belt asteroid, its surface is creased by a series of
Phobos-like grooves. These grooves have a typically beaded appearance, similar to those
on Phobos, but with a width of approximately ½ km and stretching tens of kilometers in
length. From these approximate dimensions, estimate the thickness of Lutetia’s regolith.
Is this a surprise? Discuss what processes may have acted to create this loose blanket of
debris.

EBSCOhost - printed on 11/3/2023 9:57 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
9
Wind

Owing to the development of motor transport, it is possible to study in

the further interiors of the great deserts the free interplay of wind and
sand, uncomplicated by the effects of moisture, vegetation, or of fauna,
and to observe the results of that interplay extended over great periods
of time.
Here, instead of finding chaos and disorder, the observer never fails to
be amazed at a simplicity of form, an exactitude of repetition and a geo-
metric order unknown in nature on a scale larger than that of crystalline
structure.
R. A. Bagnold (1941)

Ralph Bagnold (1896–1990) founded our modern understanding of the interaction between
wind and sand and how that interaction produces dune-covered landscapes in the Earth’s
great deserts. He lived to see spacecraft images of the sand seas on Mars and contributed
to our understanding of how universally important wind-driven (eolian) processes are. He
would have been delighted to know about the extensive dune fields of tarry sand on Titan.
Bagnold was a professional soldier and the descendent of a long line of professional sol-
diers (Bagnold, 1990). After an engineering education at Cambridge, he was posted to Egypt
in 1926 and then to other locations in North Africa where he became fascinated by the land-
scape and decided to devote himself to the study of that region’s most abundant commod-
ity – sand. In addition to unprecedented trips deep into the deserts of Sudan and Libya, he
built a wind tunnel out of plywood at Imperial College, London, to further his understanding
of the interaction of wind and sand. His classic book was published in 1941.
One of his major insights about this process is best stated, once again, in his own
words,
After much desert travel, extending over many years, during which sandstorms of varying intensity
were frequently encountered, I became convinced that the movement of sand (as opposed to that of
dust) is a purely surface effect, taking place only within a metre of the ground.
Copyright 2011. Cambridge University Press.

(Bagnold, 1941)

Bagnold used this insight to justify his reliance on wind-tunnel observations. Much more
elaborate and expensive wind tunnels than Bagnold’s are now used to simulate sand

348

EBSCO Publishing : eBook Academic Collection (EBSCOhost) - printed on 11/3/2023 9:58 AM via YDUQS
AN: 399390 ; H. Jay Melosh.; Planetary Surface Processes
Account: s4526441.main.estacio
9.1 Sand vs. dust 349

transport under both Martian and Venusian conditions. Modern research has also moved
outdoors and focuses on the interaction of dunes with both local winds and the planetary
boundary layer, but many mysteries still remain and new insights are still needed to fully
understand the dynamics of how the wind interacts with granular materials on planetary
surfaces.
Following Bagnold’s lead, this chapter is more mathematical than any other in this book.
Bagnold quite appropriately put the word “physics” into the title of his book and insofar
as physics requires the language of mathematics, the study of eolian processes has, since
Bagnold, always relied heavily on that mode of expression. As we shall see, however, the
mathematics required does not go much beyond algebra: When really intricate analyses are
required, such as the structure of turbulent boundary layers, Bagnold himself resorted to
empirical models, and we shall do the same.
Eolian processes are concerned with a gas, the atmosphere, and granular solids. A discus-
sion of interactions between the atmosphere and a liquid is reserved for the next chapter.

9.1 Sand vs. dust

The mechanical distinction between sand and dust is simple: Dust is easily suspended in the
atmosphere while sand, under a moderately strong wind, hops along the surface. However,
quantifying this distinction for a wide variety of solid particles in different planetary atmos-
pheres requires some careful discussion, as does the meaning of a “moderately strong”
wind. The first step in this discussion is to understand how solid particles fall through air.

9.1.1 Terminal velocity

Students in freshman physics courses are encouraged to ignore atmospheric drag while
they learn Galileo’s formulas for falling bodies. However, every skydiver’s life depends
on the fact that these formulas do not accurately describe his or her descent toward the
surface of the Earth. A human body falling out of an airplane rapidly accelerates, initially
following Galileo’s rules, but within about 15 s achieves a constant velocity known as the
terminal velocity. This is about 200 km/hour for a human body in Earth’s atmosphere. If
this were the end of the story, our skydiver’s arrival at the Earth’s surface would still be very
uncomfortable and skydiving would not be a popular sport. However, the terminal velocity
is a balance between the weight of a falling body and the drag force exerted on it by the
passing air. This drag force can be greatly enhanced by increasing the area of the falling
body – which is what parachutes are intended to do.
Quantitative analysis of the terminal velocity requires some simplifications. In the best
tradition of physics, we now consider a spherical sand grain (which is not such a bad
approximation as either the traditional spherical cow or a spherical skydiver!). The force
accelerating the grain toward the Earth is its weight, w, equal to its mass m times the accel-
eration of gravity g. Expressing this in terms of the grain diameter d and density σ, the
driving force is:

EBSCOhost - printed on 11/3/2023 9:58 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
350 Wind

(a) (b)

ρ
ν

Figure 9.1 Terminal velocity force balance for (a) turbulent flow, where the weight of the grain is
balanced against the momentum change of air displaced by the motion of the grain and (b) laminar
flow, where the weight of the grain is balanced by viscous drag forces, here approximated as being
localized in a cylinder with a diameter three times larger than the grain diameter d. In this case the
velocity of the air equals the velocity of the grain at the grain surface but drops to zero (approximately)
at the surface of the cylinder.

π 3
w= d (σ − ρ ) g. (9.1)
6
In this equation ρ is the density of the air, so that the weight is really the immersed
weight, which takes account of the buoyancy of the fluid that surrounds the grain. This cor-
rection may seem to be negligible for quartz sand (density about 2650 kg/m3) in the Earth’s
atmosphere (density 1.2 kg/m3); however, we shall see that the equations that we derive for
sand and air apply almost without alteration for sand and water or any other liquid, where
the buoyancy correction may be substantial, so we will retain this distinction in the follow-
ing analysis.
The drag force is more complex and requires our first empirical injection. Simple con-
sideration of the momentum of the air deflected by the particle (Figure 9.1a) suggests that
it should depend on the projected area of the falling grain, πd2/4, the density of the air ρ
and the square of the relative velocity v between the grain and the air. However, the exact
drag force depends on the shape of the grain and its velocity in a more complicated way, so
this complexity is absorbed into a mostly empirical constant called the drag coefficient CD,
defined so that the drag force FD comes out as:
π ρ d 2 v2
FD = C D . (9.2)
4
Over a wide range of velocities, CD ≈ 0.4 for a sphere.
Equating the weight (9.1) and drag force (9.2) and solving for the velocity yields an
expression for the terminal velocity:

EBSCOhost - printed on 11/3/2023 9:58 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
9.1 Sand vs. dust 351

4 (σ − ρ ) d g
v= . (9.3)
3 CD ρ

As one might expect, objects fall faster if they are denser, bigger, or the acceleration of
gravity is higher. They fall more slowly if the air density or drag coefficient is higher.
Equation (9.3) does not hold for all grain sizes. In particular, it may give an extremely
poor estimate of the terminal velocity for small particles unless the drag coefficient is
changed rather substantially. This equation holds best in what is called the turbulent regime,
where drag forces are created by the deflection of the air stream. The vigilant reader may
also be surprised that there is no dependence on the viscosity of the air in this equation.
Viscosity is important only for very small or slow particles. The dividing line between tur-
bulent flow and the low-velocity laminar flow regime is determined by the Reynolds num-
ber Re, which is the ratio between inertial and viscous forces,
ρ vd (9.4)
Re = .
η
When the Reynolds number is low, inertial forces are small compared to viscous forces,
and viscosity, not the deflection of the air stream, determines the drag force. George G.
Stokes (1819–1903) first analyzed the full equations for viscous drag in 1851 and the ter-
minal velocity for a sphere is now called Stokes’ law. The full derivation is complex and
not very edifying, so I instead present an approximate derivation that captures the essence
of the equation.
Suppose that our small falling sphere is surrounded by a cylindrical can of diameter 3d
and height d (Figure 9.1b). We suppose that the velocity of the air is zero on the surface
of the can, but equals the velocity of the sphere at the sphere’s surface. This is not really
true: The air velocity falls off more gradually with distance away from the sphere, but most
of its decline is close to the sphere, so our rigid can is a good first approximation. The air
between the sphere and the can is, thus, sheared with a strain rate ε ̇ ≈ v/2d. Remembering
that the definition of viscosity, Equation (3.12), is σ s = 2ηε̇, where σ s is the shear stress, we
obtain the drag force by multiplying the shear stress times the surface area of the vertical
sides of the can, 3πd2. Equating this drag force to the weight of the grain, (9.1), and solving
for the velocity, we obtain the terminal velocity of a small particle for which Re << 1,

1 (σ − ρ )d 2 g
v= (9.5)
18 η

which happens to be exactly Stokes’ law, thanks to a clever choice of the dimensions of
our cylindrical can. Another way to achieve this result is to note that at low Reynolds num-
ber the drag coefficient is given by CD = 24/Re, which upon substitution into (9.3) yields
Stokes’ law.
The most notable features of Stokes’ law are its inverse dependence on gas viscosity and
its dependence on the square of the particle diameter. This means that the terminal velocity

EBSCOhost - printed on 11/3/2023 9:58 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
352 Wind

of very small particles is very low. Anyone who has wondered why the clouds, composed
of tiny water droplets with mean radii of about 10 μm, do not fall out of the sky can answer
this question for themselves by evaluating Equation (9.5) for a cloud droplet.
Stokes’ law is valid for Reynolds numbers less than about 10, while the turbulent drag
equation holds for Reynolds numbers between about 103 and 105. Between these regimes
empirical expressions for the drag coefficient must be used to compute the terminal
velocity.
Evaluation of the terminal velocities of small particles in the atmospheres of different
planets requires the viscosity of the gas. This can be derived from experiment or looked
up in a table. However, it is useful to note a few results on gas viscosity from the kinetic
theory of gases. Most importantly, the viscosity of a gas is nearly independent of pressure.
Thus, whether we are dealing with air at the Earth’s surface or air in the stratosphere, the
viscosity η is the same (at the same temperature). J. C. Maxwell first deduced this fact from
his kinetic theory of gases and, at first, he could not believe it: Checking this prediction
was the motivation for his 1867 experiments that also led him to invent the concept of a
Maxwell solid (Section 3.4.3). This surprising behavior comes about because the viscosity
of a gas is the consequence of the exchange of momentum between layers of gas moving
relative to one another. As pressure decreases there are fewer gas molecules to exchange
momentum, but their mean free path increases at the same time, so the exchange takes
place between layers of greater relative velocity. The two factors cancel one another and
the resulting viscosity is independent of pressure. Maxwell also predicted that the viscos-
ity of a gas depends on the square root of the temperature. This prediction is less accurate:
Measurements show that in most gases the viscosity depends more strongly upon tempera-
ture. Subsequent research connects the temperature dependence of viscosity to the forces
between molecules, so this dependence must usually be determined empirically.
Table 9.1 lists the terminal velocities of small spheres based on Stokes’ law for silicate
grains near the surface of Earth, Mars, and Venus, along with velocities for tarry organic
grains on Titan. It is clear that as particle size decreases the terminal velocity decreases
rapidly as well. For a given size particle the terminal velocities are surprisingly similar des-
pite the wide differences between the various bodies. The major determinant of terminal
velocity is grain size, not which planet it falls on. Note that for grains larger than 100 μm
Stokes’ law underestimates the terminal velocity and a more accurate expression for the
drag coefficient must be used.

9.1.2 Suspension of small particles

The terminal velocity alone is not enough to estimate whether a particle will be suspended
or sink to the surface. In a quiet atmosphere, or if the wind flow was purely laminar, parti-
cles of all sizes would eventually settle to the surface. However, the atmosphere of a planet
is almost never completely quiet or laminar. Winds are ultimately due to the spherical
shapes of planets and the fact that solar radiation is not uniformly distributed over their
surfaces. Because of inequalities of heating, currents arise in the atmosphere.

EBSCOhost - printed on 11/3/2023 9:58 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
9.2 Motion of sand-sized grains 353

Table 9.1 Terminal velocities of small particles

Particle Gas 100 μm grain 30 μm grain 10 μm grain

composition, viscosity diameter diameter diameter
Body density (kg/m3) (10–6 Pa-s) (m/s) (m/s) (m/s)

Venus Silicate, 2700 33.0 0.40 0.036 0.0040

Earth Silicate, 2700 17.1 0.88 0.079 0.0088
Mars Silicate, 2700 10.6 0.55 0.05 0.0055
Titan Organic tar, 1500 6.3 0.18 0.016 0.0018

The flow of all known planetary atmospheres is turbulent. The air does not travel
smoothly from one point to another. Instead, instabilities in the flow develop that lead to
wide fluctuations of the instantaneous velocity about the mean. Turbulent flow develops
when the Reynolds number, Equation (9.4), is large, and at the scale of planetary atmos-
pheres it is invariably very large. Turbulence is a broad and complex subject, which is
treated in many specialized books devoted to that topic (a good introduction is Tennekes
and Lumley, 1972). At the moment what it means to us is that the motion of the atmos-
phere can be divided into two components: an average wind speed that remains con-
stant for long periods of time, and a fluctuating component that varies widely over short
timescales.
The full story of how particles can be suspended in fluids is surprisingly complex, rely-
ing on the phenomenon of “bursting” in turbulent fluids to inject momentum from the sur-
face boundary layer into the body of the moving fluid. For the purposes of this book we will
bypass this difficult topic and apply Bagnold’s rule of thumb, which states that the average
velocity of turbulent eddies in a flow of air is equal to 1/5 of the mean velocity. Thus, for a
wind velocity of a few meters per second, the turbulence velocity is about 0.5 m/s, so that
the dividing line between sand and dust is about 100 μm on the terrestrial planets. It may
be somewhat larger on Titan.

9.2 Motion of sand-sized grains

Grains too large to be suspended in the atmosphere may, nevertheless, be quite mobile
on the surface under the influence of the wind. The hopping motion of sand grains along
a river bed was first described by G. K. Gilbert (1914), who observed the process while
studying the ability of water to transport river sediments. Gilbert called this motion “sal-
tation” following McGee (1908), who took the term from the Latin word saltus, “leap.”
Bagnold accepted this term and applied it to the much larger hops that sand grains executed
in his wind tunnel.
Modern studies of eolian transport distinguish four modes of wind-driven particle motion.
Suspended material is carried aloft by turbulent winds. Saltating grains hop from the sur-
face, travel for some distance (the “saltation length”) with the air stream, and then reimpact

EBSCOhost - printed on 11/3/2023 9:58 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
354 Wind

(a)

h
β

sand

(b)

Figure 9.2 (a) Trajectories of saltating sand grains start out steeply as the grain is ejected from the
surface. Near the apex of their hops, saltating grains attain a constant velocity and then follow a
sloping (at angle β) linear trajectory as they fall back to the surface at terminal velocity, where they
may initiate hops of other grains. The hop height is h and the length of each saltation hop is l. (b)
Impact creep occurs as objects too large to saltate are struck by saltating grains. The momentum
imparted by each impact pushes the larger grain downwind along the surface.

the surface (Figure 9.2a). Sand grains splashed out of the surface when saltating grains
touch down either initiate new hops of their own or slither downwind in a snakelike motion
called “reptation.” Larger sand grains and pebbles too large to hop may still move down-
wind under the impact of saltating grains, a motion called “impact creep” (Figure 9.2b)
that includes rolling along the surface.

9.2.1 Initiation of motion

All the modes of wind transport, except perhaps suspension, require sand grains to be in
saltating motion. The problem of how this motion starts from an initial state in which the
wind begins to blow over a motionless sand bed has turned out to be both complex and
revealing. Before we can begin a full discussion of this process, we need a better under-
standing of how the wind interacts with the surface.
Wind near the surface. The first real understanding of how a moving fluid interacts with
a surface grew out of the studies of German aerodynamicist Ludwig Prandtl (1875–1953).
Prandtl spent most of his career at Göttingen University. His major claim to fame is the con-
cept of the boundary layer, a zone of sharply increasing velocity at the interface between a
moving fluid and a solid surface. When Bagnold needed more information about how sand
grains could begin moving away from a sandy surface, he corresponded with Prandtl and
incorporated many of Prandtl’s ideas into his work.
The first important concept is the friction velocity v*. The actual velocity varies with
height above the surface in a complicated way that we will explore in a moment, but the

EBSCOhost - printed on 11/3/2023 9:58 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
9.2 Motion of sand-sized grains 355

friction velocity is meant to express the overall effect of the wind on the surface by a single
number that is independent of height. The friction velocity is defined in terms of the shear
stress τ that the wind exerts on the surface. You could imagine measuring this wind shear on
an ice-covered lake by cutting a small raft of ice free of the cover, then measuring the force
with which the raft is pushed downwind while the wind is blowing. The shear stress τ is
then the force divided by the area of the raft. The shear stress does not depend on the details
of the velocity distribution above the surface – it is a single overall measure of the surface
force of the wind. The friction velocity is then defined in terms of the shear stress as:

τ
v* ≡ . (9.6)
ρ

The friction velocity is a central concept in theories of wind transport because it is dir-
ectly related to the force the wind exerts on the surface (and vice versa). However, it is not
often practical to measure the shear stress directly. Fortunately, the friction velocity has a
simple, almost universal, relationship to the velocity above the surface. The following for-
mula is partly empirical and partly can be derived from Prandtl’s mixing length theory of
turbulence. It relates the mean wind velocity to height z through the friction velocity and a
factor known as “roughness,” z0 :

 z
v( z ) = 5.75 v* log   . (9.7)
 z0 

The roughness factor is somewhat empirical. It is proportional to the grain size in the sur-
face and approximately equal to 1/30 of the grain size when the grains are tightly packed.
For Earth it is typically about 0.2–0.3 mm and it is usually assumed to be about the same
on Mars. The roughness of many different surface types has been calculated by measuring
wind velocities at different heights above the ground. Given tables of roughness it is pos-
sible to convert a wind speed measurement at a single height into a prediction of its value
at any other height, as well as to obtain the friction velocity.
Equation (9.7) describes the velocity dependence some distance above the surface.
However, it cannot hold down to scales comparable to the roughness because the flow is
broken up by the irregularities of the surface. If the roughness is small, another distance
scale becomes important, one controlled by the molecular viscosity of the gas. A layer in
which turbulence is suppressed then develops next to the surface. Called the viscous sub-
layer, the thickness of this zone (also empirically determined) is:

η
δ ≈5 . (9.8)
ρ v*

The numerical factor is empirical, while the dimensional ratio is derived from turbulence
theory (Tennekes and Lumley, 1972). Within the viscous sublayer the velocity falls linearly
to zero at the surface. The overall dependence of wind velocity on height above the surface
is illustrated in Figure 9.3.

EBSCOhost - printed on 11/3/2023 9:58 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
356 Wind

ν(z)
z

Turbulent flow
Viscous
sublayer
δ
Roughness
z0

Wind shear τ = ρν*2

Figure 9.3 Mean velocity profile of wind over a sand surface. Turbulent eddies produce a logarithmic
profile of mean velocity above the surface. Very close to the surface, viscous forces dominate and
the fluid velocity falls linearly to zero through a viscous boundary layer of depth δ. The surface
roughness is characterized by the parameter z0, while the average wind drag on the surface is related
to the friction velocity v*.

The fluid threshold. When the wind blows over a sand surface where no grains are yet
in motion, it has to begin by plucking individual grains out of the surface before the wind
stream can accelerate them. Supposing that the grains protrude above the viscous sublayer,
each grain experiences a downwind force that is given by the shear stress τ times its pro-
jected surface area, πd2/4. This drag force is resisted by the weight of the grain (which
is decreased by the lift provided by the deflected wind stream), Figure 9.4 and Equation
(9.1). Whether the wind actually succeeds in plucking an individual grain out of the surface
depends on the geometry of its contacts and the way it deflects the wind, so an exact for-
mula for a real surface is not possible. However, the drag force must equal the weight times
some factor of order 1. Calling this factor A2 and replacing τ by its definition in terms of
the friction velocity (9.6), we solve the resulting equation for the threshold friction velocity
at which motion just begins:

σ − ρ
v*t = A  g d. (9.9)
 ρ 

The threshold velocity is proportional to the square root of the particle size, so this
expression predicts, not surprisingly, that larger grains are more difficult for the wind to
pick up than smaller ones.
Something new happens, however, when we consider very small grains. When the size
of the grains becomes smaller than the thickness of the viscous sublayer, Equation (9.8),
the wind drag is spread over a large number of particles: it is not localized on a single grain.
The grains, thus, become more difficult to pull out of the surface and the wind velocity

EBSCOhost - printed on 11/3/2023 9:58 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
9.2 Motion of sand-sized grains 357

ν∗

lift

wind
drag

≈d

weight

Figure 9.4 Forces acting on a grain resting on a surface of other grains of similar size. Each grain
is subject to a drag force from the surface wind, possible lift forces from the deflected wind, friction
between grains, and gravity holding it down onto the surface. Drag and lift forces must exceed friction
and gravity before a grain begins to move.

must be higher than Equation (9.9) predicts before motion can start. There is a similar extra
resistance if the grains stick together by cohesive forces, such as electrostatic or van der
Waals forces, which are important for very small grains.
A full analysis of the effects of sublayer resistance is not presently possible. Instead,
we must rely on empirical measurements of threshold velocities. Bagnold faced this prob-
lem and observed that, although the factor A in Equation (9.9) is roughly equal to 0.1 for
large sand grains, it increases rapidly with decreasing particle size below some threshold
(Figure 9.5). He found that A is a function of a dimensionless parameter that he called
the “friction Reynolds number” Re*. This Reynolds number is defined like the usual one,
Equation (9.4), except that the friction velocity replaces the fluid velocity. For a friction
Reynolds number less than about 3.5 the threshold velocity shoots up steeply (Figure 9.6)
and appears to depend on Re* to a large negative power. The appearance of the friction
Reynolds number makes a lot of sense in this context because comparison of the definition
of Re* with the thickness of the viscous sublayer indicates that d/δ = 5/Re*. All this limit
says is that the threshold velocity shoots up when the particle size is equal to about 0.7δ.
The approximate dependence of the threshold velocity on grain size can be derived from
the dependence of A on Re*, even though the power n is not well known. If A is a func-
tion of 1/Re*n, then it also depends on 1/v*tn dn (neglecting other terms in the relationship,
which are not important for this argument). Inserting this dependence into Equation (9.9)
and solving for v*t, we find that it must depend on grain size d to the power (1/2–n)/(n+1).
But for large n, this ratio approaches –1, so that we can say that for small grain sizes, v*t
depends approximately on 1/d. Clearly, as d decreases the threshold velocity rises. But for
Re*>> 3.5 we know that A = 0.1, and v*t is proportional to d . The threshold velocity,
thus, rises rapidly for both large and small particle sizes (Figure 9.6). So we must conclude

EBSCOhost - printed on 11/3/2023 9:58 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
358 Wind

1
A~
Ren*

0.1

1.0 10
Re*

Figure 9.5 Dependence of the friction coefficient A in Equation (9.9) on the friction Reynolds
number Re*. The coefficient is nearly constant with a value 0.1 until Re* falls below a critical limit,
below which it rises rapidly.

1
~d
Log (Friction velocity, v* )

~√d

impact
threshold
v*t

dt
Log (Grain diameter, d)

Figure 9.6 The threshold velocity for sand motion over an initially static sand bed has a pronounced
minimum as a function of grain size. Small grains are buried in the viscous boundary layer and so are
difficult to individually pluck out of the surface, while large grains are too heavy to move easily. This
results in a unique size that is most easily entrained by the wind, at a corresponding minimum wind
velocity. The threshold is much lower (the impact threshold) when sand grains are already in motion.
Inspired by Bagnold (1941, Figure 28).

that there is a minimum for some special grain size that lies between the large particle and
small particle limits.
The existence of a minimum in the threshold velocity curve has profound implications
for wind- (and water-) transported material on any planet. It means that there is a special
grain size, unique to that planet (and process, wind or water), which is most easily moved

EBSCOhost - printed on 11/3/2023 9:58 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
9.2 Motion of sand-sized grains 359

and which, therefore, characterizes all the deposits emplaced by that process. The diam-
eters of wind-transported sand grains on Earth nearly all lie within a narrow range, 0.1 to
0.3 mm (Ahlbrandt, 1979). This reflects the selection of the most easily moved grains by
winds whose strength may vary, but the size of the grains that move first and most often is
always close to the minimum of the curve. Water-deposited sands on beaches and in rivers
show a wider variety of sizes owing to their frequently complex histories. In principle, we
should be able to distinguish wind-blown sands from water-deposited sands because of
the difference in threshold diameter. In practice this is not usually possible for a variety of
reasons, and other criteria such as shape or surface texture are now used (Siever, 1988).
The actual value of the threshold diameter and velocity can be estimated from the infor-
mation already presented by setting A = 0.1 in Equation (9.9) and using Bagnold’s obser-
vation that Re* at the threshold equals 3.5 to determine v*t. The resulting equations can be
solved for the grain diameter at the threshold, with the result:
1/ 3
 η2 
dt = 10.7   . (9.10)
 ρ (σ − ρ ) g 

The corresponding threshold velocity is most easily derived from the Reynolds number
at the threshold and the above equation:

η
v*t = 3.5 . (9.11)
ρ dt

These equations only approximately determine the minimum, which is rather broad in
practice, so the precise values have to be taken with some skepticism, but using the same
formulas to compare the onset of wind transport on different bodies is revealing. Table 9.2
lists the threshold grain diameters and velocities for the wind surface conditions on those
Solar System bodies with substantial atmospheres. Where applicable, we also list the initi-
ation conditions for flow in liquids that may be present on the surfaces of these bodies.
In order of ease of transport by wind, we see that the progression is Venus, Titan, Earth,
and Mars. The threshold speeds are especially high on Mars, a fact that presents some
problems that will be discussed separately. Wind speeds exceeding the threshold have been
directly observed on Venus, Earth, and Titan. Dune fields have been observed on all three
bodies, consistent with the predictions. The fact that extensive dune fields are also observed
on Mars calls for special consideration.
The threshold grain size increases in the same progression, from very fine on Venus to
very coarse on Mars. The predicted size for Earth is in fair agreement with the sizes com-
monly observed for dune sands.
Liquid is much denser than gas and so it can move particles with greater ease than the
wind. The threshold equations also predict that it can move coarser particles (Mars, again,
is an exception).
Grains at the threshold size are not the only ones that can be moved. The minimum in
the curve is, observationally, rather broad (Greeley and Iversen, 1985) and winds higher
than the minimum often occur, so that a range of grain sizes is usually transported. Greeley

EBSCOhost - printed on 11/3/2023 9:58 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
360 Wind

Table 9.2 Threshold grain diameters and fluid velocities

Threshold Threshold Fluid

Viscosity diameter friction velocity velocity at
Body Medium (10–6 Pa-s) (μm) (m/s) 1 ma (m/s)

Venus Quartz in CO2 33.0 94 0.018 0.37

Earth Quartz in air 17.1 220 0.21 4.50
Earth Quartz in water 1540 560 0.01 0.21
Mars Quartz in CO2 10.6 1100 3.3 69.
Mars Quartz in water 1540 770 0.007 0.15
Titan Tar in N2 6.30 160 0.025 0.53
Titan Tar in liquid methane 184 410 0.004 0.080
Titan Ice in liquid methane 184 530 0.003 0.062

a
Assuming roughness z0 = 2 x 10–4 m.

and Iversen, in their book, tend to downplay the role of the viscous sublayer in raising
the fluid threshold for small particles and instead emphasize cohesive forces between the
grains. Electrostatic forces, however, tend to have the opposite effect and strong electric
fields in the saltating layer may actually lift small particles off the bed (Kok and Renno,
2009b). It is clear that, even after 70 yr of study, there is still more to be learned about the
fluid threshold.
The impact threshold. When sand grains are already in motion, they impact the surface at
the end of their saltation hops and often knock other grains into the air. In this case the wind
does not have to blow as fast as it does when the surface is quiescent. It is, thus, easier to
keep grains in motion, once they have begun moving, than it is to initiate their first motion.
This is an example of history-dependence of a process, or hysteresis. Of course, if the wind
speed drops too low all of the sand grains fall back onto the surface and the process stops,
but there is a range of wind speeds between the threshold speed and this stopping speed
where sand motion is possible. The minimum velocity to keep grains in motion is known as
the impact threshold and is indicated on Figure 9.6 by the line labeled “impact threshold.”
Bagnold’s experiments suggested to him that the impact threshold is given by Equation
(9.9), with A equal to 0.08, rather than 0.1 at the fluid threshold (Figure 9.6). This estimate
of Bagnold’s has been supported by subsequent research, which places A between 0.08 and
0.085 at the fluid threshold (Kok and Renno, 2009a). Naturally there is no upturn reflect-
ing the viscous sublayer, but the decrease in surface velocity due to the grains already
in motion may be significant. Saltating grains in water do not approach the surface with
as high a velocity as those in air and so impacts may not be a strong factor in this case.
However, other phenomena, such as turbulent bursting, may play a role in making the curve
different for sediment in motion versus that for clean fluid just initiating motion.
A recent and important contribution to this topic (Kok, 2010) suggests that the impact
threshold may be far more important for sand transport on Mars than it is on Earth. This

EBSCOhost - printed on 11/3/2023 9:58 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
9.2 Motion of sand-sized grains 361

is a consequence of the very high speeds needed to reach the fluid threshold in Mars’ thin
atmosphere (Table 9.2). Once in motion, grains driven by high winds eject an exceptionally
large number of grains from the surface. Taking account of the role of occasional gusts of
high speed in initiating grain motion, it seems that winds with speeds ten times slower than
the nominal fluid threshold prediction may, nevertheless, be effective in transporting sand
and explain the many observations of eolian features on Mars, despite the few instances of
winds high enough to reach the fluid threshold.

9.2.2 Transport by the wind

Once a few grains begin to move, the saltating grains knock others out of the surface, which
themselves knock still other grains free, until the entire sand surface becomes covered by
a low carpet of saltating sandgrains. The dimensions of this carpet and the amount of sand
that is in motion are determined by the speed of the wind blowing over the surface and the
properties of the atmosphere.
The wind over saltating sand. When Bagnold first tried to analyze this process, he found
himself in new territory: He corresponded with Prandtl, the foremost expert of his time, but
Prandtl’s theories apply to pure fluids and need modification before they can be applied to
a carpet of saltating sand. Bagnold, thus, developed a rough-and-ready series of estimates
that have mostly stood the test of time and are still used to estimate rates of sand transport,
with a few minor modifications. He supposed that the wind velocity over a saltating sand
carpet is given by:

 z 
v = 5.75 v′* log   + vt (9.12)
 z′ 0 

where v′* is the friction velocity when sand is in motion. It reflects the increased drag over
the flowing sand carpet and is, therefore, larger than the friction velocity when sand is not
in motion. The factor z′0 is a modified roughness, now known as the “aerodynamic rough-
ness,” whose nature puzzled Bagnold. Empirically it is about ten times larger than z0 and
Bagnold suggested that it might correspond to the height of ripples. Similarly, vt is a thresh-
old velocity at height z′0. These last two factors must be viewed as empirical fitting factors
of obscure physical interpretation. More modern estimates are equally empirical (Greeley
and Iversen, 1985). Most recently, however, substantial progress has been made in comput-
ing this wind profile using numerical computer codes (Kok and Renno, 2009a).
The flux of wind-driven sand. Bagnold’s analysis of the effect of the saltation carpet
on the wind near the ground begins with the hop of a single grain. Referring back to
Figure 9.2a, a saltating grain first leaps out of the surface and, as it reaches the crest of its
trajectory, it is accelerated by the wind to an average horizontal speed us. It returns to the
bed at its terminal velocity v, so that the angle β at which the saltating grain approaches the
bed is given by tan β = v/us. Each sand grain of mass m thus leaps out of the bed, acceler-
ates from near-zero velocity to final velocity us, and then re-impacts the bed a distance l

EBSCOhost - printed on 11/3/2023 9:58 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
362 Wind

downwind. This start-stop motion of each saltating grain removes momentum m us/l per
unit distance from the wind. If the net motion of the sand, comprising many sand grains,
is expressed as a mass flux, qs, per unit width perpendicular to the wind (kg/s-m), this
momentum loss per second per unit area equals the drag stress τ ′ required to keep the sand
moving,

qs us
τ′= ≡ ρ v ′*2 (9.13)
l
where v′* is the friction velocity in the presence of a carpet of saltating grains. In his
wind-tunnel measurements Bagnold found that us/l is approximately equal to the vertical
ejection velocity, w, divided by g, us/l ≈ w/g. He then supposed that w is proportional to
the friction velocity v′*. These suppositions lead to a useful scaling relation for the salta-
tion hop length l:

v ′*
l∝ . (9.14)
g

The hop height h is proportional to the length through the tangent of the descent angle β.
Inserting the relations in the previous paragraph into Equation (9.13), and solving for qs,
we obtain an expression for the mass flux of the sand in terms of the friction velocity over
a moving sand carpet:

 ρ v ′*3 
qs = C  (9.15)
 g 

where C is a “constant” that Bagnold found depends on the square root of the grain
diameter. There are many modern variants of Equation (9.15) for the sand flux: Greeley
and Iversen (1985) list 15 of them. An obvious improvement is to subtract a threshold
friction velocity from v′*, so that this equation does not predict non-zero fluxes for arbi-
trarily low velocities. Nevertheless, the important feature of this equation is its depend-
ence on the friction velocity cubed, and that is widely agreed to be at least approximately
correct.
The “constant” C in Equation (9.15) conceals factors only partially considered by
Bagnold. The rate of sand transport depends on the surface over which the sand saltates: It
is slower over sand surfaces, where the saltating grains lose most of their momentum at the
end of each hop, and much faster over stony surfaces, a factor that is important in the accu-
mulation of sand patches. The rate of transport also depends on the slope of the surface, a
factor that plays a role in modern theories of sand dune formation.
The major implication of Equation (9.15) is that the flux of sand is not a linear func-
tion of the wind speed, but depends on the speed raised to the third power. The most
important sand-driving winds are, thus, not the average winds, but the exceptional winds.
Meteorological plots of average winds may thus be very deceptive when the direction of
sand transport is of interest. Gentle winds may blow from one direction most of the year,

EBSCOhost - printed on 11/3/2023 9:58 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
9.2 Motion of sand-sized grains 363

but a month of strong winds from another direction may completely dominate the orien-
tation of a dune field. Modern analyses of the wind regime among sand dunes reflect this
non-linear dependence of sand transport on the cube of the velocity by plotting a quantity
known as the vector drift potential on maps of sand dunes (Fryberger, 1979).
Impact creep and reptation. Although most of the mass moved by the wind travels by sal-
tation, the impulse delivered by the impact of saltating grains contributes up to about 25% of
the total. As illustrated in Figure 9.2b, multiple impacts by saltating grains can propel peb-
bles that are otherwise too big to be moved by the wind. Such pebbles may be seen at the bed
of the saltating carpet irregularly jerking or rolling downwind. Recently, an additional type
of motion has been distinguished (Lancaster, 1995). Called reptation because the path of the
sand grains resembles the slithering of a snake, it describes the motion of coarse sand grains
on the bed that are splashed out at low velocity by the impacts of fast-saltating grains.

9.2.3 The entrainment of dust

One of the surprising aspects of wind interaction with the surface is the difficulty of entrain-
ing fine particles, dust. Once dust is suspended in the air, it may rise high into the atmos-
phere, travel over intercontinental distances (Pye, 1987), and may even play a major role in
the radiation balance of the atmosphere, as it does on Mars. However, as many observers
have noted, dust lying on the surface tends to stay on the surface until something, other
than the wind, disturbs it.
Anecdotes about the resistance of fine material to erosion by wind or water abound.
Mud mounds in Galveston’s harbor survived the catastrophic 1900 hurricane unchanged,
while seawalls built of meter-sized rocks were carried away. Gilded Age geologist Raphael
Pumpelly was astonished to observe roads in China’s loess region entrenched tens of meters
into the surface (Figure 9.7), gradually excavated as the dust stirred up by traffic was blown
away by the wind. Fine Martian dust that settled over the solar panels of the Spirit and
Opportunity rovers was expected to eventually terminate the mission, until dust devils or
major sandstorms blasted the dust away.
Bagnold attributed the immobility of dust to its small grain size compared to the thick-
ness of the viscous sublayer in turbulent flow. Greeley and Iversen invoke cohesive forces
between small grains. Whatever the cause, neither wind nor water can mobilize fine-grained
material without the assistance of some disturbing agency.
Where saltating sand advances over dusty surfaces, the impact of the saltating grains on
the surface mixes the dust with the air and thus mobilizes dust. Sand grains and saltating
mud chips may play this role in raising dust from terrestrial playas. Sandstorms would be
relatively innocuous if they were restricted to the thin carpet of saltating sand, but they are
frequently accompanied by thick clouds of dust that do much more damage to the lungs of
humans and animals.
On Mars, dust devils seem to play an important role in raising dust from the surface. This
has been attributed to the unusually high winds that develop near the core of the dust devil
and to the mobilization of low-density aggregates (Sullivan et al., 2008). However, another

EBSCOhost - printed on 11/3/2023 9:58 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
364 Wind

Figure 9.7 A road deeply

entrenched in the fine-
grained loess terrain in
China. This illustrates
the fact that fine-grained
sediments cannot be moved
by the wind until disturbed,
in this case by traffic. Once
entrained in the air, the
wind exports the sediment
and the road gradually
deepens, with the long-term
consequences shown here.
Reproduced from Pumpelly
(1918, p. 468).

process known as “backventing” or “reverse percolation” may play a role in both dust-devil
mobilization and in association with impact-crater blast waves. This process was first pro-
posed in connection with the “dark halos” that were noted in radar images of craters on Venus
(Ivanov et al., 1992). It has also been observed in the vicinity of many explosion experiments,
but most discussions of it occur in the “gray” literature (Rosenblatt et al., 1982).
Backventing usually accompanies the passage of a shock wave over a dusty surface. In
the shock, the pressure first rises above normal atmospheric pressure, then falls below it
during the “negative phase” before returning to normal (Glasstone and Dolan, 1977). As
the shock wave passes over a permeable surface, it first drives air into the ground, then,
during the negative phase, the air vents out of the surface, often carrying dust that has
been entrained by direct lofting in the vertical air stream. Optically dark halos up to 1 km
in diameter have been noted about the sites of many recent impacts on Mars (Malin et al.,
2006) and probably originate by this backventing process. It seems possible that the sud-
den drop in pressure in the core of a dust devil might also be responsible for initiating a
temporary flow of air out of the Martian soil that could entrain and lift dust off the surface,
mixing it into the atmosphere.

EBSCOhost - printed on 11/3/2023 9:58 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
9.3 Eolian landforms 365

9.2.4 Abrasion by moving sand

Pebbles and rocks in the path of saltating sand are frequently polished, faceted, and grooved,
sometimes leading to fantastic surface patterns. The name ventifact is given to a rock that
has been sculptured by the wind. The flux of moving sand is akin to industrial sandblasting
and, over time, considerable abrasion of obstacles may occur. R. P. Sharp (1911–2004),
in a series of well-known experiments (Sharp, 1964), found that a brick left in the path of
saltating sand was half eroded away after 6 yr of exposure. He also planted Lucite rods ver-
tically in the path of the sand and found that the abrasion varied considerably with height:
Whereas very little damage occurred above the normal level of the saltating sand, the great-
est losses of material were at about 20 cm above the ground, near the top of the saltation
layer, while less was removed close to the ground. Hard igneous rocks, on the other hand,
suffered little visible damage during the course of his 10.5-year study. Wind abrasion is
evidently slow for resistant rocks. Abrasion rates for different materials were quantified in
more detail by abrasion experiments conducted in a wind abrasion simulation apparatus by
Greeley and Iversen (1985). These experiments showed the widely variable susceptibility
of different rock types to wind abrasion at a given impact velocity.
Wind abrasion is a strong function of the impact velocity of individual sand grains, as
well as the total flux encountering an obstacle. Because of the lower threshold velocities
on Venus and Titan, wind abrasion may not be of great importance there. On Earth, while
wind-abraded pebbles adorn the pages of most geology textbooks, they are rare in the field.
Wind abrasion is not a widespread process on our planet except in few especially favored
locales. However, because of the much higher threshold velocities on Mars, wind abrasion
is expected to be more common and, in fact, it did not take long for the Mars Exploration
Rovers to encounter wind-fluted rocks (Greeley et al., 2006).

9.3 Eolian landforms

Landscapes dominated by the wind are rare on Earth: Where they do occur they are so strik-
ingly different from our normal experience that they have received a great deal of attention
from geomorphologists. While silicate sand or dust is the material most often moved by the
wind, blowing snow may also create similar landforms. Mars seems to be the planet most
favored by the wind, in spite of its thin atmosphere, although extensive dune fields have also
been imaged on Venus and Titan. In the following discussion we will use the term “sand” in
its eolian process sense: Sand is loose granular material that moves by saltation.

9.3.1 The instability of sandy surfaces

Much of the interest we find in eolian landscapes derives from their instability. Dune
shapes are constantly shifting and dunes are landforms in slow motion. The formation of
dunes themselves is due to the instability of surfaces on which sand and larger rocks are
mixed.

EBSCOhost - printed on 11/3/2023 9:58 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
366 Wind

The flux of saltating sand depends upon the surface across which it moves. Sand grains
saltating over sand on Earth leap only a few tens of centimeters off the surface. However, if
the sand moves over a stony surface (or a paved road), the grains rebound from the surface
with considerably more energy than if they had struck a loose, sandy surface. The saltation
height increases dramatically, the high-flying grains sense a faster wind velocity, and the
sand moves faster and farther than before. This rapid motion continues until the sand grains
reach another sandy patch, where they lose their energy and slow down, many of them
dropping out of the wind stream in the process.
A surface of uniformly mixed sand grains and stones thus quickly separates into sandy
patches and stony patches as a result of this feedback. This sorting of the surface into
sand and stones persists even as the sand accumulates into dune fields, and in spite of the
constant downwind migration of the dunes, as Sharp observed in the Algodones Dunes of
California (Sharp, 1979).
This tendency for instabilities to grow and create patterns from an initially uniform
landscape is presently recognized as a field of study in itself. “Self-organized” pattern for-
mation occurs when positive feedback regulates systems of this kind. Computer models
incorporating idealized versions of the laws of sand transport successfully reproduce many
features of the natural system, including sand dunes (Anderson, 1996; Werner, 1995).
Although such model building is not considered a part of traditional geomorphology, this
approach does expose the most important factors that create the observed forms. Pelletier
(2008) describes Werner’s model, its philosophy, and provides a computer code for imple-
menting it.

9.3.2 Ripples, ridges, and sand shadows

Ripples. The quintessential eolian feature is the sand ripple (Figure 9.8). Wind ripples
develop everywhere that loose sand is exposed, on sand dunes, beaches, and even in chil-
dren’s sand boxes. On Earth, they are typically a few centimeters high, spaced about 10
cm apart, and may extend many meters laterally. They are oriented perpendicular to the
predominant wind direction and move slowly downwind. Ripples form within minutes
in a strong wind. Nevertheless, even after decades of scientific study, their formation still
presents many puzzles, the most serious of which have been raised by the observations of
what appear to be wind ripples on Mars.
Bagnold noted a similarity between the saltation hop length and the spacing of ripples
and concluded that the regular spacing of ripples reflects the average saltation jump length.
He supposed that an initially flat sand surface became wavy under the bombardment of
the saltating particles, developing so that most saltation hops start and end on the upwind
(stoss) side of the ripple, driving intense upslope surface creep, while suppressing creep on
the downwind (lee) side
Although Bagnold’s identification of saltation length and ripple spacing has been widely
accepted, doubts have arisen. Sharp (1963) noted that during windstorms the ripple spa-
cing gradually increases with time, which seems inconsistent with control by the saltation

EBSCOhost - printed on 11/3/2023 9:58 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
9.3 Eolian landforms 367

Figure 9.8 Asymmetric wind ripples in sand. The pocketknife is ca. 10 cm long. Note poorly
developed subsidiary ripples at a steep angle to the main trend. Ripples are topped by somewhat
coarser sand grains than average. 2000 photo by Jason Barnes, from longitudinal sand dunes near
Tuba City, AZ.

length. He proposed instead that the ripple spacing is controlled by the size of the creeping
grains and described extensive observations that tend to support his view.
The origin of ripples has currently reached a crisis precipitated by the observations of
surface landers and rovers on Mars. By scaling the saltation length from Earth to Mars using
Equation (9.14), both the higher threshold friction velocity and lower gravity suggest that
a Martian saltation hop should be more than 100 times longer than on Earth, reaching tens
of meters or more. If Bagnold’s theory is correct, wind ripples on Mars should be spaced
tens of meters apart. However, images of the surface show what appear to be Earth-sized
“bedforms” everywhere (Mars scientists are reluctant to use the word “ripple” for these
features, even though they strongly resemble the terrestrial feature in size and spacing). It
seems that, despite the familiar appearance of Martian eolian features, nothing about them
fits at the moment. The grain sizes of the material on top of the ridges are too small (200 to
300 μm near the surface, down to 100 μm a few centimeters below the surface), the dust is
observed to move when the sand-sized particles do not, and the millimeter-sized particles
predicted by the threshold equation (see Table 9.2) are apparently absent (Sullivan et al.,
2008), although see Box 9.1 on this topic.
All of these Martian observations suggest that something is seriously amiss in our
understanding of what wind processes on Mars are doing. And if our theories are wrong
on Mars, are they wrong on Earth as well? New ideas and approaches are urgently
needed, such as the suggestion that a very low impact threshold for sand motion on Mars
might account for some of the observations (Kok, 2010), or that ripples are actually due
to aerodynamic interactions between the wind and roughness elements on the surface
(Pelletier, 2009).

EBSCOhost - printed on 11/3/2023 9:58 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
368 Wind

Box 9.1 Kamikaze grains on Mars

Wind transport of sand on Mars has long posed a major problem for scientific analysts.
Because of Mars’ thin atmosphere, the minimum wind velocity required to set particles in
motion at the fluid threshold, Equation (9.11) and Table 9.2, is enormous. Note that 69 m/s in
Table 9.2 is a wind speed of 250 km/hour one meter above the surface. Do such powerful gales
ever blow on Mars? Moreover, the particle size at the threshold of motion is also enormous, of
the order of 1 mm, about five times larger than on Earth.
The prospect of a Martian sandstorm is truly fearsome: Anything on the surface would be
bombarded by a fusillade of ball-bearing-sized grains flying along at nearly supersonic speeds.
Because of this daunting prospect, the cameras of the Viking landers peered out through
narrow slits that could be covered in the event of sandstorms. Fortunately, the Viking landers
completed their missions before being engulfed in any such destructive sandblasts. But Martian
sandstorms must also be highly abrasive to rocks lying on the surface. Indeed, it becomes
surprising that there are any intact rocks at all.
Not only would such sandstorms be destructive of hardware on the planet, it should also
be destructive of the flying grains themselves. Granting that saltation can get started on Mars,
it seems that no sand grain could survive even a single hop before being blasted into dust.
Significantly, it seemed, Viking soil-sieve measurements showed a deficit of grain sizes in
the millimeter range. This observation has, moreover, been confirmed by the Spirit rover’s
microscopic imaging system (Sullivan et al., 2008).
This problem became known as the “kamikaze effect”: Martian sand grains can hop once,
then they die. This begs the question of where sand grains come from in the first place. If this
is their fate on Mars, then there cannot be very many of them.
The difficulty of getting sand grains to either move or survive on Mars, coupled with the
undoubted observation of sand dunes, led to a variety of proposals for how Mars manages
to build sand dunes in spite of its thin atmosphere. So far, no definitive solution has been
found. The leading idea is that the sand grains that saltate are not solid particles at all, but
instead are aggregates of finer material somehow cemented into millimeter-sized pellets. The
lower density of such grains would make entrainment by the wind easier, although it would
exacerbate the survival problem. However, even if the grains are destroyed after one hop, their
fragments might be re-cemented at a later time to prepare them for another hop. Clay dunes
(lunettes) on Earth form in this way, and such dunes, while relatively rare on Earth, have
received a lot of attention from the Mars scientific community.
Another possibility is that the Martian sand dunes cannot, in fact, move under current
climatic conditions and they are all relicts from a time when Mars’ atmospheric pressure
was higher. As of this writing, the HiRISE imaging system has not observed any dunes to
move, at a resolution of about 30 cm. However, the Spirit rover reported that after strong
wind gusts, grains up to 300 μm in diameter appeared on its top deck. While these are smaller
than predicted at the fluid threshold, the fact that they were in motion at all is something of a
surprise. Because the megaripples investigated by Spirit at El Dorado crater are composed of
grains of this size, it seems that at least megaripple-forming materials are mobile during high
winds.

EBSCOhost - printed on 11/3/2023 9:58 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
9.3 Eolian landforms 369

Box 9.1 (cont.)

A very new idea, mentioned in Section 9.2.1, is that the fluid threshold might not be relevant
on Mars. The impact threshold may occur at a much lower velocity on Mars so that the 100 μm
to 300 μm particles that make up the megaripples could be the dominant grain size on Mars.
Once they are in motion, grains of this size require much lower velocity winds to sustain them
and, at this lower velocity, are not imperiled by impact destruction after one saltation hop.

Our understanding of the humble, ubiquitous ripple is certainly in need of improvement.

One of the main drivers in this urgent need is the crisis raised by observations made on
another planet than the Earth, highlighting the importance of comparative planetology even
for those interested only in the Earth.
Ridges. Bagnold also noted that many large ripples (he called them “giant sand ridges”)
had coarser grains at their crests and suggested that surface creep of the larger grains drove
them to the crest and then armored the crest sand against further deflation. Many subse-
quent observations of sandy ridges that are bigger than ripples but smaller than dunes have
agreed with his observation that their grain size distributions are typically bimodal, and
that the coarse grains collect at the crest of the ridge. These ridges are long-lasting and, like
ripples, form transverse to the prevailing wind.
Some of the puzzling Martian eolian ridges may be of similar origin: The Spirit rover has
confirmed the bimodal distribution of grain sizes in several ridges in Gusev crater. These
features should, thus, properly be considered ridges, not large ripples, in spite of the unfor-
tunate terminology: “Megaripples” lie everywhere on Mars. Megaripples form transverse
to the wind and range from tens of centimeters up to 3 m high and a few to tens of meters
apart (Figure 9.9). Many are light-colored, in contrast to the mainly dark dunes of basaltic
sand and may be composed of a different material, perhaps low-density dust aggregates.
Sand shadows. Actively moving sand requires wind to keep it in motion. When the wind
strength drops, sand accumulates. This simple fact accounts for the accumulations of sand
commonly found in the lee of large rocks, brush, and in hollows below the general level of
the surface. On Mars, this accounts for the wind-blown material in the lee of crater rims and
in their interiors. Small dune fields typically form in the interiors of Martian craters where
sand has accumulated and the wind speed is reduced.
Ripples often curve sharply when they cross sand shadows, their curvature indicating the
deflection of the surface wind by the obstacle: The wind direction is always perpendicular
to the trend of the ripples. One may, thus, note the divergence of the wind upwind of the
obstacle and its convergence behind. The antitheses of shadows, sand scours, often develop
upwind of obstacles where the wind speed accelerates.
A characteristic pattern of sand shadows and scours develops around small craters, illus-
trated in Figure 9.10. These organized zones of deposition and scouring are the result of
twin eddies shed downwind as the wind parts around the crater rim. Sand accumulations

EBSCOhost - printed on 11/3/2023 9:58 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
370 Wind

Figure 9.9 Megaripples on Mars. Ripples lie on the floor of a channel at 29.34° N and 299.83° W.
Image is 3 km wide. Upper portion of Mars Orbiter Camera (MOC) image 1200991.

wind
(a)

deposit

erosion

deposit
(b)

deposition

Figure 9.10 Sand shadows behind Martian crater forms. As the wind blows, vortices form and detach
from the upwind rim of the crater. Sand is deposited in less windy areas upwind of the crater, in its
interior, and immediately downwind of the rim, while the increased wind velocity in the vortices
leads to enhanced erosion along the sides and farther downwind of the crater. Interpretation inspired
by illustrations in Chapter 6 of Greeley and Iversen (1985).

EBSCOhost - printed on 11/3/2023 9:58 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
9.3 Eolian landforms 371

behind mountain ridges are called “falling dunes,” but frequently are merely immobile sand
that has collected in the lee of the ridge.

9.3.3 Dunes
Sand dunes are larger accumulations of sand that are too wide for saltating grains to hop
across. As dunes grow higher they create zones of quiet air or even reversed flow in their
lee and, thus, block the saltation flow across the surface. Such shadow zones lead to the
accumulation of more sand behind the crest, building the dune higher until the slope of the
lee face reaches the angle of repose. At this point a “slip face” develops where sand that has
arrived at the crest avalanches down the face.
Dunes are self-organized forms that grow spontaneously when conditions are right. They
may start simply as a patch of sand on a stony plain that tends to capture more sand, which
leads to the capture of still more sand in a strong positive feedback. Sand dunes would
grow to unlimited height if given an unlimited supply of sand, except that as they grow
they deflect the wind over their tops, increasing its velocity and eventually blowing sand
off their crests faster than it can continue to build upward.
Dune velocity. All the time that a sand dune is growing in volume it is also moving down-
wind. The speed of downwind motion is easy to calculate and the result is very instructive.
If qs is the mass rate of sand movement per unit distance perpendicular to the wind, the rate
at which it adds volume to the lee side of the sand dune (Figure 9.11) is qs/σs, where σs is
the density of loose sand in the dune. In each interval of time Δt the sand blown over the
brink fills a prism in the lee of the dune of height h, length Δx and volume h Δx. The vol-
ume of this prism is equal to the volume of the sand blown in, qs Δt /σs, so the dune creeps
downwind at a velocity:

∆x q
vD = = s . (9.16)
∆t σ s h

Naturally, the dune velocity increases as the rate of sand transport increases. The inter-
esting thing about this equation is that the velocity depends inversely on the dune height.
This does make a lot of sense: It takes more sand to build a taller dune one unit of distance
downwind than it does for a smaller dune. However, this means that the taller a dune grows,
the slower it moves.
In a dune field in which a number of both large and small dunes occur, the small dunes
race along while the big dunes lumber downwind. But when a small dune overtakes a big
one, it climbs up its upwind face and collapses down the slip face, feeding the big dune
while it is itself annihilated. The big dunes, thus, tend to grow at the expense of the small
ones, looming larger and slower until the accelerated winds at their tops blow sand off their
summits as fast as it accumulates and they reach a fixed height. Even this, however, does
not stop their inexorable movement – they continue to crawl downwind until some new fate
consumes their sand.

EBSCOhost - printed on 11/3/2023 9:58 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
372 Wind

wind

lux h
ndf
sa

∆x
ν
Figure 9.11 Downwind migration of sand dunes. The migration velocity of a sand dune is computed
by balancing the sand flux moving up the upwind face of the dune against the volume required to
build the dune one unit of length downwind. Sand is immobilized in the lee of the dune after it moves
over the crest, where it accumulates at the angle of repose.

An important concept for interpreting the formation of dunes is the timescale over which
an individual dune forms (Allen, 1974). In a regime of changing climate, large dunes may
reflect conditions from a different climatic era and their morphology may thus be out of
harmony with the regime prevailing at the time they are observed. A useful timescale is
derived by comparing the volume of a dune with its rate of growth or, equivalently, the
length of time required for the dune to move its own length,

λh σ
τD = = λ h2 s (9.17)
vD qs

where λ is the length/height ratio of the dune, typically equal to about 10. This “dune modi-
fication timescale” on Earth ranges from a few years for small dunes to 50 000 yr for large
star dunes.
Dunes and dune fields are classified in various ways, but a common division groups
them by their orientation with respect to the predominant wind. Dunes are, thus, considered
to belong among the transverse, longitudinal, or star classes, depending on whether they
are oriented perpendicular to the prevailing wind, parallel to it, or are heaped chaotically
with no particular direction evident. There are special classes of dune, such as parabolic
dunes, that are dependent upon vegetation to create their form. In this book such forms are
ignored, but they are very important in terrestrial geology, particularly along coasts where
vegetation is common. In addition to the “pure” forms described below, compounds of dif-
ferent types are almost universal.
Barchan dunes. The classic dune type is the barchan (Figure 9.12a). Shaped like a crois-
sant lying with its convex side upwind, barchan dunes form spontaneously on stony plains
where sand is in short supply. Barchans on both the Earth and Mars grow to an approxi-
mately constant size in a given area and maintain their characteristic shape as they migrate
downwind. On Earth, barchans range from about 3 to 10 m in height and extend about 30

EBSCOhost - printed on 11/3/2023 9:58 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
9.3 Eolian landforms 373

(a) (b)

(e) (f)

Figure 9.12 Different dune patterns are determined by both the wind regime and sand supply. (a)
Barchan dunes form where a unidirectional wind drives a limited supply of sand. (b) Barchanoid
ridges with wavy crests form transverse to the wind when more sand is available. (c) Full transverse
dunes develop where sand is abundant and the wind is unidirectional. (d) Linear or longitudinal dunes
form where the wind blows from more than one direction, but are generally parallel to the dominant
wind direction. (e) Star dunes form under conditions of multiple wind directions with none dominant.
(f) Reversing dunes are created by unidirectional winds that occasionally reverse themselves. All
images from McKee (1979, pp. 11, 13).

to 100 m downwind. Their upwind slopes are gentle, standing at angles of about 5° to 15°,
while the slip face stands at the angle of repose, near 32°. Surprisingly, barchan dunes on
Mars are about the same size, despite the difference in surface conditions. The upwind face
of barchan dunes is symmetrically arched in plan view, while an arcuate slip face develops
downwind and lateral horns grade into low mounds lacking slip faces. Because the slip
face traps migrating sand and thus prevents it from moving further downwind, barchans
shed sand only off their lateral horns. The barchan form seems remarkably stable and these
dunes are capable of migrating large distances downwind without change of form.

EBSCOhost - printed on 11/3/2023 9:58 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
374 Wind

As the sand supply increases, instead of increasing in size indefinitely, barchan dunes
shed sand off their horns, which may then organize itself into small new barchans, a repro-
ductive process that fascinated Bagnold, who described it in the introduction to his book as
“vaguely disturbing” in its “grotesque imitation of life.”
Barchan dunes are most symmetrical when the wind blows from a single direction. When
lateral winds occur, one horn grows larger and longer than the other, with the larger horn
developing on the upwind side with respect to the lateral wind direction. With persistent
lateral winds barchans may grade into linear dune chains.
Transverse dunes. With increasing sand supply and a constant wind direction, isolated
barchan dunes merge into long ridges oriented perpendicular to the wind. Such ridges
stretch many kilometers laterally and are repeated downwind by parallel ridges, creating
a landscape dominated by parallel ridges with undulating crests. When the crests of these
ridges are sinuous in plan they are often called “barchanoid” ridges (Figure 9.12b), but
are simply called “transverse dunes” when their crests are more linear (Figure 9.12c). On
Earth, there is a crude relation between the height and spacing of these dunes: Dunes a few
meters high are spaced about 100 m apart, while dunes 100 m tall are spaced about 1 km
apart (Lancaster, 1995).
The regular spacing of transverse dunes is likely due to aerodynamic flow patterns: The
presence of a ridge suppresses surface winds for some distance downwind. Such shadow
zones extend 12 to 15 times the height of the obstacle, in agreement with the observed spa-
cing. Reverse eddies may also develop between the ridges, and many observers have noted
weak winds blowing up the slip face of tall dunes while sand was blowing downwind over
the crest.
All of the dune fields imaged on Venus belong to the transverse type, as indicated by
their perpendicular orientation to wind streaks (Greeley et al., 1997). Magellan could only
resolve the largest dunes, which seem to be spaced about 0.5 km apart with ridges trending
5 to 10 km perpendicular to the wind direction. Magellan could not measure the heights of
the dunes.
The transverse dune type also dominates on Mars, although large areas are occupied by
sparsely scattered barchan dunes (Greeley et al., 1992). Martian transverse dune ridges
are typically spaced 300 to 800 m apart. Most of the large dune fields lie in the northern
circumpolar plains. Two varieties of transverse dune are observed in small clusters, mostly
lying within craters. The smaller variety is spaced 100–1200 m apart, while the larger var-
iety is spaced at 1600–4000 m. Martian dunes are typically dark and are believed to be
composed mostly of basaltic sand.
Linear dunes. Impressively long, linear dunes traverse large areas of the Earth and Titan
(Figure 9.13). On Earth, individual linear dunes range from about 20 km to 200 km long,
2–35 m high and are spaced about 200–450 m apart. Compound versions of these dunes
may be much larger: 50–170 m high and spaced 1600–2800 m apart. Dunes on Titan cover
a vast area of the satellite, occupying about 40% of the low-latitude half of Titan, a larger
fractional surface coverage than on any other body in our Solar System (Jaumann et al.,
2009). Titan’s dunes are estimated to be about 30–70 m high with an average spacing of 2

EBSCOhost - printed on 11/3/2023 9:58 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
9.3 Eolian landforms 375

Figure 9.13 Longitudinal dunes in the equatorial region of Titan, which likely consist of sand-sized
particles made of organic material. Notice that the dunes are deflected by the brighter (and presumably
higher) terrain, following the wind direction. The imaged area is 225 x 636 km from the T-25 pass of
the Cassini orbiter Synthetic Aperture Radar. North is to the right. PIA12037: NASA/JPL.

km. Chains of dunes can be traced hundreds of kilometers. The dominance of linear dunes
on Titan may be connected with its unique wind regime, in which the surface winds are
mainly driven by tides rather than thermal gradients.
Linear dunes, called seif dunes by Bagnold, form when the sand-moving winds blow
from more than one direction (Figure 9.12d). One direction must predominate, which
determines the trend of the dunes, but they require a crosswind from another direction for
at least part of the year. Because of this inconstant wind regime, the crests of linear dunes
are often irregular and small slip faces alternate from one side to another.
The impressively parallel trend and spacing of linear dunes may be maintained by the
formation of alternate helical eddies aligned with the prevailing wind. This idea was first
proposed by Bagnold and has been popular with many subsequent authors. However, direct
measurements of the sizes of helical eddies often do not agree with the observed dune spa-
cing. At the moment, there is no universally agreed process that controls the lateral spacing
of linear dunes.
Star dunes. Where no predominant wind direction exists, the direction of sand transport
shifts constantly and sand piles up into huge, complex heaps known as star dunes. Strictly
speaking, a star dune must have at least three arms containing slip faces (Figure 9.12e). The
tallest accumulations of sand on Earth are star dunes, with heights that reach up to 4 km,
but are more typically a few hundred meters high. Star dunes are spaced at distances equal
to about ten times their height, a common dimensional ratio among dunes.
Reversing dunes. In areas where the wind regularly reverses direction, slip face direc-
tions may alternate from one side of a dune to another. Strictly speaking, reversing features
occur at the crest of larger accumulations of sand: Because the reversal timescale is short,
the volume of the reversing portion of a dune cannot be large. Nevertheless, the crests
of reversing dunes are distinctively symmetrical, with triangular profiles (Figure 9.12f).
Barchan dunes will also tolerate an exact reversal of the wind direction without major
effects on their morphology, so long as one direction predominates.
Many other dune forms and features have been identified and named. Lunettes, for
example, develop around the downwind margins of playa lakes and are often composed

EBSCOhost - printed on 11/3/2023 9:58 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
376 Wind

(a) (b)

Figure 9.14 Yardangs on Mars. (a) Closeup view of yardangs showing the classic “inverted boat
hull” morphology, located near 1° N, 214.4° W. Scale bar at the lower left is 400 m. MOC image
PIA04677 NASA/JPL/Malin Space Science Systems. (b) Panoramic view of a soft rock unit dissected
into yardangs south of Olympus Mons. The three flat regions in the fore-, middle-, and background
measure about 17 x 9 km in this oblique view. Mars Express HRSC image, orbit 143, ESA/DLR/FU
Berlin. (G. Neukum). See also color plate section.

of fine-grained material, even clay, whose particles are normally too small to saltate as
individuals. However, when cemented by salts or after wetting, these fine particles become
weakly cemented into aggregates that can be moved short distances by the wind.
Climbing dunes, as opposed to falling dunes, are special forms that develop on steep
slopes upwind of resistant ridges over which the sand eventually passes, to accumulate into
falling dunes on the lee side. For more detail on the range of types recognized, the reader is
referred to the specialized literature, such as the book by Lancaster (1995).

9.3.4 Yardangs and deflation

Dunes are depositional landforms in which material accumulates. Yardangs are the char-
acteristic erosional form created by the wind (Blackwelder, 1934). On Earth, yardangs are
a minor geomorphic curiosity, seldom seen by anyone but travelers in arid regions. Where
they occur, they are usually small, only fractions of a meter to at most tens of meters
high and up to 20 m apart. They usually develop in weakly cohesive rocks such as clay,
silt, or weakly cemented sandstones. Yardangs appear as long, linear ridges with stream-
lined upwind edges whose shape many observers have compared to overturned boat hulls
(Figure 9.14a). Saltating sand erodes the troughs between yardangs only along their floors,
so the ridges are often undercut along their sides and upwind edges. The troughs are typic-
ally flat-floored or U-shaped.
Whereas yardangs are only minor features on Earth, they dominate some landscapes
Mars (Figure 9.14b). Weak, perhaps pyroclastic, rock layers near the Martian equator are
deeply entrenched by linear grooves aligned with the prevailing wind. The edges of mesas

EBSCOhost - printed on 11/3/2023 9:58 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
9.3 Eolian landforms 377

are often deeply etched by yardangs and crater ejecta deposits are dissected into linear
ridges. Yardangs on Mars are enormous by terrestrial standards, tens of kilometers long and
separated by troughs averaging 200 m wide.
Yardangs may also be extensive on Venus. Although difficult to identify unambiguously
from Magellan radar images, Greeley et al. (1997) described an extensive field of yardangs
about 500 km southeast of Mead crater. These yardangs, if that is what they are, average 25
km long by 0.5 km wide and are spaced from 0.5 to 2 km apart.
Dust-raising winds may also erode shallow, closed depressions, in strong contrast to
fluvial processes, which tend to fill in closed basins. Called deflation hollows or pans on
Earth, such depressions range from “buffalo wallows” only a few tens of meters across to
the Qattara Depression in Egypt, a shallow, crescent-shaped basin more than 100 km wide
and 200 km long that has been excavated 134 m below sea level. The Qattara Depression
may have formed as recently as the past 2 Myr, suggesting a very high rate of eolian ero-
sion, even by terrestrial standards. Deflation hollows have not yet been definitively identi-
fied on other planets, perhaps due to the difficulty of ruling out other processes that create
shallow depressions. Any persistent source of eolian dust must eventually evolve into a
shallow basin of this kind.

9.3.5 Wind streaks

Images of wind streaks returned by the Mariner 9 orbiter provided the first evidence for
wind action on Mars (aside from the global dust storms themselves). Seen as variable albedo
patterns in the lee of obstacles such as crater rims, Martian wind streaks puzzled early obser-
vers: some were bright and some were dark. Sometimes both bright and dark streaks formed
around the same feature (Figure 9.15). Martian wind streaks change with time: They are
especially likely to have changed after dust storms. We now believe that most Martian sand
materials are dark, of mainly basaltic composition, whereas the ubiquitous red or orange dust
is bright, a simple fact that resolves many puzzles involving albedo markings on Mars.
Wind streaks have no topographic expression. They result from the deposition or removal
of thin deposits of wind-blown material whose color or brightness contrasts with the under-
lying surface. They are common on Earth as well as Mars. They appear often in snow-
covered terrains, but more permanent streaks have formed behind obstacles such as cinder
cones. The well-studied wind streak downwind of Amboy Crater in southern California is
a good example that can be readily seen in Google Earth images. This particular streak is
more permanent and appears to result from differential trapping of light-colored sand on
dark desert pavements near the cinder cone.
Both bright and dark wind streaks were observed downwind of Venusian craters by
Magellan’s radar. In radar images, “bright” means “rough” (at the radar wavelength,
12.6 cm) and may, in this case, imply removal of fines by enhanced turbulence down-
wind of the crater. Dark streaks imply deposition of smooth material. Analysis of Magellan
images revealed almost 6000 wind streaks (Greeley et al., 1997) that were used to map
global atmospheric circulation patterns.

EBSCOhost - printed on 11/3/2023 9:58 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
378 Wind

Figure 9.15 Bright and dark wind streaks are simultaneously present around craters on Mars. These
presumably formed during different wind regimes and are the result of the differential vulnerability
of dark basaltic sand and bright dust to wind erosion or deposition. Image is located at 28° S, 245° W
in Hesperia Planum. Viking Orbiter frame 553A54. NASA/JPL.

9.3.6 Transient phenomena

Patterns resulting from eolian deposition can often be observed on the surfaces of planets.
Global dust storms on Mars redistribute dust on an almost annual basis, obscuring areas
previously cleared of dust by impact blast waves or dust devils. Dust-devil tracks a few tens
of meters wide and many kilometers long criss-cross dusty plains, appearing dark after the
removal of the bright dust.
Volcanic eruptions create plumes of airborne volcanic dust that settles out downwind
for distances that may reach thousands of kilometers on Earth. Close to the volcano where
deposits of volcanic tephra are thick, winds may heap the fine material into dunes.
The ejecta from impact craters also interact with the atmosphere to create deposits simi-
lar to those of volcanoes, although organized somewhat differently. Notable impact-related
features are the dark crater parabolas on Venus (Figure 9.16). Radar-dark, parabola-shaped
features surround about 60 fresh Venusian craters. With blunt ends facing almost due east,
parabolas stretch a few hundred to more than 2000 km from east to west and extend up to
1000 km from north to south at their widest extremities. These features are successfully
explained as the fine-grained ejecta deposits of impact craters (Schaller and Melosh, 1998;
Vervack and Melosh, 1992). Although the impact throws out ejecta basically symmetric-
ally, most of the distal ejecta fall back into the atmosphere after a short ballistic flight. After
re-entry it drifts with the wind as it settles toward the surface. The upper atmosphere of
Venus flows steadily from east to west at speeds up to 60 m/s at 50 km altitude. The ejecta
drifts downwind, traveling westward as it settles. Because the ejecta falling closer to the

EBSCOhost - printed on 11/3/2023 9:58 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
Further reading 379

Figure 9.16 Adivar Crater on Venus and its parabolic ejecta deposits. Adivar is 30 km in diameter,
located at 9° N, 76° E, and is surrounded by an inner radar-bright (rough) parabola, which is in turn
surrounded by a radar-dark (smooth) parabola. This image is 674 × 674 km, showing the enormous
extent of these deposits, which probably represent coarse ejecta close to the crater and finer material
that fell back into the atmosphere farther away and was blown westward by strong prevailing winds
in the upper atmosphere. Portion of Magellan Radar Image C2-MIDR.00N080;1. NASA/JPL.

crater is coarser than that falling farther away, the closest ejecta is deposited first, while the
more distal ejecta blows farther downwind, creating the parabolic shape. The details of the
parabolas’ shapes even permits the particle size to be inferred and yields a general relation
between ejecta particle size, ejection velocity, and crater size for large impact craters. This
information could probably not be gained in any other way, short of arranging for a series
of large impacts.

Further reading
The fundamentals of the transport of sand and dust by wind are well treated by, obvi-
ously, Bagnold (1941) as well as in a later USGS report (Bagnold, 1966). A more mod-
ern treatment from the planetary perspective can be found in Greeley and Iversen (1985).
Dust deposition and transport is the subject of Pye (1987), while the standard work (which
includes many spectacular images) is McKee (1979). A more modern treatment of dunes
is given by Lancaster (1995). Wind erosion and deposition by wind on the Earth is well
described in an older classic (Mabbutt, 1977). The “planetary connection” is made by a
series of long papers: For Venus, see Greeley et al. (1997); for Mars see Carr (2006). Our
knowledge of Titan is relatively recent, but a good summary of what we do know is in
Jaumann et al. (2009).

EBSCOhost - printed on 11/3/2023 9:58 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
380 Wind

Exercises

9.1 The looming clouds?

Clouds on Earth are composed of water droplets whose density, about 1000 kg/m3, is much
larger than the density of air (about 1.3 kg/m3 at the surface of the Earth). So why don’t the
clouds simply fall out of the sky? Discuss this problem in the light of Stokes’ law for the
rate of fall of small particles. You may find it useful to know that the average size of cloud
droplets is about 10 μm.

9.2 Blowing in the wind or falling like a brick?

Small meteorites (weighing a few milligrams) that enter the Earth’s atmosphere typically
evaporate at altitudes near 70 km as they flare up briefly as “shooting stars.” The refractory
silicates that compose most of these small meteorites then condense into “smoke” particles
with sizes that range from 1 to 3 nm. An estimated total of about 100 tons/day of such
material currently falls into the Earth’s atmosphere. Estimate how long particles of this size
might stay aloft in the atmosphere. Compare this to the time that a large, solid, meteorite
fragment, 20 cm in diameter, takes to fall to the surface.

9.3 Do the math!

Derive Equation (9.9) in the text for the threshold velocity of the wind to just move a sand
grain of diameter d lying on the surface.

9.4 Sand dunes on Triton?

Triton, Neptune’s largest moon, possesses a very thin atmosphere that is composed mainly
of N2 gas at a chilly 38 K. Nevertheless, geysers spout plumes 8 km high into the atmos-
phere. Suppose that loose “sand” grains of ice (perhaps from impact ejecta) lie on the sur-
face. How fast do the winds of Triton have to blow to just entrain such ice grains, and how
big are these grains? Compute both the minimum friction velocity needed to loft these
grains and the minimum wind speed 1 m above the surface. Compare this velocity to the
speed of sound in Triton’s atmosphere. What can you conclude about the probability of
finding “sand” dunes on Triton if it is visited by a spacecraft with an imaging system cap-
able of resolving such features? Facts that you may find useful: The viscosity of nitrogen
gas at 38 K is about 2.2 × 10–6 Pa-s and its density at Triton’s atmospheric pressure of 1.5
Pa is 1.3 × 10–4 kg/m3. The acceleration of gravity at the surface of Triton is 0.78 m/s2.

9.5 The marching dunes

The Algodones Dunes along the eastern margin of southern California’s Imperial Valley
exhibit slip faces up to 40 m high. Large complex transverse dunes alternate with interdune

EBSCOhost - printed on 11/3/2023 9:58 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
Exercises 381

flats that are typically about 500 m to 1 km wide. The dunes themselves are about the same
width as the interdune flats. R. P. Sharp (1979) measured the rate of the downwind migra-
tion of these dunes to be about 35 to 40 cm/yr. From this data estimate the average sand
flux along the dune chain and the lifetime of an individual dune in the field. The California
Department of Transportation is contemplating the construction of a new highway about
50 m downwind of a small outlier dune with a slip face 5 m high. Discuss the wisdom of
this plan.

EBSCOhost - printed on 11/3/2023 9:58 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
10
Water

All indurated rocks and most earths are bound together by a force of
All rights reserved. May not be reproduced in any form without permission from the publisher, except fair uses permitted under U.S. or applicable copyright law.

cohesion which must be overcome before they can be divided and

removed. The natural processes by which the division and removal are
accomplished make up erosion. They are called disintegration and trans-
portation. Transportation is chiefly performed by running water.
…A portion of the water of rains flows over the surface and is quickly
gathered into streams. A second portion is absorbed by the earth or rock
on which it falls, and after a slow underground circulation reissues in
springs. Both transport the products of weathering, the latter carrying
dissolved minerals and the former chiefly undissolved.
G. K. Gilbert, Geology of the Henry Mountains (1880)

The Earth’s surface is dominated by landforms that have been carved by running water.
Fluvial landforms are usually apparent in even the driest deserts. Running water is such
an effective agent of erosion because of its density: Almost 1000 times denser than air,
it exerts greater shear stress, buoys the weight of entrained particles, and is driven more
forcefully by gravity than an equivalent volume of air.
Where rainfall is possible, even small amounts of water trump any other agent of ero-
sion. Although rain is not possible on Mars under current conditions, its landscape plainly
bears the scars of rainfall in the distant past. Some things are different: Mars has seen enor-
mous floods that are comparable to the largest floods known on Earth, and groundwater
sapping plays (or played) a far larger role than it does on our planet. We still do not under-
stand how Mars’ floods originated or how such large volumes of water came to be suddenly
released. Nevertheless, once released, the water followed the same laws as water on Earth
and produced landforms for which terrestrial analogs exist.
Recent exploration of Titan reveals that active fluvial landscapes are not unique to Earth:
Although the fluid on Titan is liquid methane at 95 K and the rocks are hard-frozen water
ice, Titan’s landscapes of stream valleys, riverbeds, and enormous lakes are eerily familiar.
The materials are different but the processes are similar and can be analyzed by the same
Copyright 2011. Cambridge University Press.

methods as the Earth’s water-carved surface. In the following chapter, as in its title, the
word “water” is used to keep the discussion focused, but most of the concepts apply to any
fluid interacting with a solid substrate.

382

EBSCO Publishing : eBook Academic Collection (EBSCOhost) - printed on 11/3/2023 9:59 AM via YDUQS
AN: 399390 ; H. Jay Melosh.; Planetary Surface Processes
Account: s4526441.main.estacio
10.1 “Hydrologic” cycles 383

As we explore fluvial processes in this chapter we will emphasize general laws and
relationships that hold where any fluid interacts with a solid surface. Whether on Earth,
Mars, or Titan, similar physical laws lead to similar landforms. Process trumps contin-
gency here.

10.1 “Hydrologic” cycles

It is a familiar concept that rain falling on the land suffers a number of possible fates,
returning eventually to the sea or evaporating into the air to continue the cycle that made
rivers seem eternal to our ancestors. Rain that falls on the surface can infiltrate into the
ground, where it joins the volume of groundwater beneath the surface, evaporate imme-
diately back into the air, or run over the surface to collect into streams and possibly large
bodies of standing water. On Earth, transpiration by plants is also an important factor, but
plants, so far as we know, play no role on either Mars or Titan.
The relative quantities of these factors, along with the movement of water underground,
are the subject matter of the field of hydrology. We will be mainly interested in that part of
the cycle which results in runoff and can, thus, perform work on the landscape.

10.1.1 Time, flow, and chance

The climates of the Earth are usually characterized by their temperature and annual rainfall.
The annual rainfall statistic specifies how much water arrives as rain in an “average” year.
However, as everyone knows, the amount of rainfall in a given year can fluctuate greatly.
Moreover, the pattern in which the rainfall arrives is also important – clearly, 30 cm of rain
per year that arrives in several large, brief thundershowers is more effective in eroding the
landscape than 30 cm of rain arriving in gentle, continuous showers.
In general, most geologic work is done by large, relatively infrequent events. Most
mountain streams run clear over large boulders that they obviously cannot move. Such
boulders are transported only during extreme rains when the clear rivulet may become a
muddy, churning torrent for a few hours. The level of water in streams may reach as high
as their banks only once a year, but that annual flood determines the form of their channel.
Much less frequent floods may sculpt their higher floodplains.
It is important to realize the significance of fairly rare events in surface processes.
Consider the example shown in Figure 10.1. The upper panel depicts a typical relation-
ship between streamflow (which is proportional to the friction velocity) and the amount
of sediment moved (e.g. Equation (9.15) for wind-blown sand: Sediment in streams obeys
a similar equation). The middle panel is a probability distribution showing the likelihood
of a given streamflow. It peaks at the mean discharge. Discharges both lower and higher
than average are less likely. The curve becomes increasingly less certain as the discharge
increases: Very rare events that are, nevertheless, possible may never have been observed
within the time frame over which records are available. The product of the two curves is
shown in the lower panel. This yields the probability of a given sediment discharge.

EBSCOhost - printed on 11/3/2023 9:59 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
384 Water

Sediment discharge
3
qs ~ ν
*

stream velocity
Probability of

fat tail
thin tail

erosion
max
sediment discharge
Probability of

fat

thin

Stream velocity

Figure 10.1 The amount of geological work done by flowing water depends on the cube of the
velocity, as shown in the top panel. The probability of finding a given velocity peaks at some stream
velocity between zero and infinity, as shown in the middle panel, although the high-velocity behavior
is somewhat uncertain, depending on whether the probability falls exponentially (thin tail scenario)
or as a power-law function of the velocity (fat tail scenario). The lower panel shows that the product,
the probability of a given sediment discharge, peaks at a higher stream velocity than the maximum
velocity.

Because the amount of sediment moved by a given flood is a strongly increasing func-
tion of the streamflow, the peak of the lower curve is displaced away from the mean flow,
toward a higher than average streamflow. Thus, the event that is most likely to move a large
amount of sediment, and so, cause a large amount of geologic change, is not the annual
average but a rare event that corresponds to an unusual flood.
The extreme end of the probability curve is the subject of major debates at the moment.
Because it corresponds to very rare events, there is not a lot of data to tie down the precise
form of the right-hand end of the curve, and these uncertainties are greatly magnified by

EBSCOhost - printed on 11/3/2023 9:59 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
10.1 “Hydrologic” cycles 385

the large amount of sediment moved in extreme events. This debate is connected with the
general “fat tail” problem that haunts large, rare events. If one assumes that the shape of
the probability curve is Gaussian, the probability of events far from the mean drops expo-
nentially fast with distance from the mean (a “thin tail”). However, if the curve falls as a
power law (a “fat tail”), then the probability drops much more slowly. In an extreme case, if
the curve falls as a power law less steep than 1/v3, where v is the stream velocity, then there
is no maximum at all and erosion is dominated by the most rare, but largest events! Water
floods do not seem to work this way, although meteorite impact events do, because there is
no meaningful upper limit to the size of a potential impactor.
As an example, the mean annual discharge of the Columbia River is about 7800 m3/s.
It fluctuates during an “average” year from a low of about 2800 m3/s to a high of about
14 000 m3/s in the late spring. However, there are wide fluctuations about these means: The
lowest flow ever recorded is about 1000 m3/s, while the peak is about 34 000 m3/s. Nothing
in these records, however, prepares one for the fact that between 15 000 and 13 000 yr ago
the Columbia River drainage suffered a series of catastrophic floods that increased the dis-
charge to about 107 m3/s for a few days. These floods were associated with the collapse of
ice dams holding back the waters of Glacial Lake Missoula. The outpouring of water from
these floods scarred the surface of eastern Washington, digging huge channels, transporting
blocks of basalt tens of meters across, and leaving enormous deposits of gravel in a series
of catastrophic floods that did more geologic work in a few days than millions of years of
normal erosion could accomplish (Baker and Nummedal, 1978).
Geologist Gene Shoemaker reached a similar conclusion in 1968 after he repeated
Powell’s historic trip down the Colorado River through the Grand Canyon. Shoemaker re-
occupied 150 sites that Powell’s photographers had recorded in 1871–1872 and compared
the images of the canyon taken almost 100 yr apart. In looking at these images (Stephens
and Shoemaker, 1987), one is struck by the fact that, apart from predictable changes in
vegetation, very few differences are seen in most images. Individual rocks can be rec-
ognized in the same places as 97 yr previously. However, in a small number of compari-
sons, the scene has changed utterly: almost nothing has remained the same. These drastic
changes are due to unusual floods originating in side canyons of the river that cleared away
massive heaps of rock and deposited new piles of rock debris from further upstream.
The lesson Shoemaker learned about stream erosion is that it is episodic: little may
change for a long time until an unusual event comes along that makes sudden, large alter-
ations. The fluvial landscape does not change gradually, but evolves in a series of jumps
whose effects accumulate over time.
This kind of evolution is called catastrophic: The strict definition is that a catastrophe is
a large event that causes more change than all smaller events combined. Floods fulfill this
definition within limits: For floods, the curve in the bottom panel of Figure 10.1 eventually
turns over (although the Lake Missoula floods do cause some concern about the validity of
this claim). This is not the case for meteorite impacts. Similar probability curves have yet
to be established for many other processes.

EBSCOhost - printed on 11/3/2023 9:59 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
386 Water

rainfall

rain splash
(a)
runoff

Areal flow rate, m/s-m2 infiltration

rainfall
ti
infiltration
(b)

runoff = (rainfall - infiltration)

Time

Figure 10.2 When rain begins to fall on a porous surface it initially all soaks into the ground. Once
the capacity of the surface to absorb water is saturated, however, the water collects in a thin film on
the surface and runs off. Panel (a) shows this process schematically, as well as the disturbance of
the granular surface by the small impacts of raindrops. Panel (b) indicates the volumetric division of
rainfall into infiltrated water and runoff, which reaches a steady state after an infiltration relaxation
time ti.

10.1.2 Rainfall: infiltration and runoff

Rainfall is ubiquitous on the Earth wherever temperatures permit liquid water to exist,
although some regions may receive more, others less. Rain also seems to have fallen on
the ancient cratered highlands of Mars to create valley networks and, although it has not
been observed directly, must be occurring in near-current times on Titan (except that it is
methane drops that fall there, not water).
The first drops of water that fall on the ground surface are trapped in small surface
depressions and irregularities – no runoff is produced. However, as more rain falls, the cap-
acity for depression storage is usually exceeded after some period of time. This moment is
easily observed while watching rain fall on a bare soil surface, for as soon as a connected
sheet of water forms the entire surface seems to suddenly glisten and water begins to flow
down whatever slope is present. Some water soaks into the ground: The amount depends
upon the permeability of the soil, among other factors. The remainder begins to flow along
the surface. This runoff is responsible for most geologic work (Figure 10.2a).
Runoff is equal to the difference between the rate at which rainwater arrives on the sur-
face and the rate at which it infiltrates into the surface (Figure 10.2b). If the infiltration cap-
acity is high enough or the rain gentle enough, there may be no runoff at all. However, as

EBSCOhost - printed on 11/3/2023 9:59 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
10.1 “Hydrologic” cycles 387

the rainfall intensity increases, the amount of infiltration remains approximately constant
while the runoff necessarily increases.
The process of infiltration has received much study, and there are entire books devoted
to it (Smith et al., 2002). A great deal of this study has practical ends: It is often consid-
ered desirable to increase infiltration to both reduce erosion and increase groundwater sup-
plies. Because infiltration involves only the partly saturated flow of water into loose soil
that already contains air, the process is very complex and depends on many factors. Chief
among them is the intensity of the rainfall (volume of water per unit time per unit area)
and the duration of the rain, along with its integrated volume. Raindrops falling on a bare
surface usually beat down and compact the soil, reducing the infiltration capacity as time
goes on. Equilibrium between rainfall and infiltration is reached after a relaxation time ti,
after which the rate of infiltration becomes approximately constant, whatever the rainfall
intensity or duration might be.
One factor that most geologists take for granted is the ability of water to “wet” silicate
minerals. Capillary forces play an important role in infiltration and the surface contact
energy between water and minerals is crucial to the ability of water to flow into the pores
between mineral grains in the soil. But what about methane on Titan? Does liquid methane
flow into the pores between grains of cold ice? If methane beads up on the ice surface
like water on a well-waxed car, Titan might not have a subsurface hydrologic cycle at all.
Fortunately, recent experiments (Sotin et al., 2009) show that not only does liquid methane
wet ice, it soaks into the tiniest cracks and pores on contact. Quantitative data on surface
energy is still lacking, but the viscosity of liquid methane is only about 10% that of water
(Table 9.2), so that liquid methane may readily infiltrate into the Titanian surface.
On Earth, the infiltration capacity depends upon the condition of the soil surface.
Vegetation plays a big role here, as does the type of soil. Clays, for example, are very
impermeable and, thus, have a low infiltration capacity, which promotes runoff and hence
surface erosion. Gullying on clay-rich badlands is intense. On the other hand, coarse sands
and loose volcanic cinders on the sides of fresh cinder cones are highly permeable and
may entirely suppress runoff. A frequent observation is that fresh cinder cones stand for
long periods of time without any sign of gullies, in spite of being composed of loose,
often cohesionless, volcanic lapilli. However, with time, weathering eventually converts
the glassy lapilli to impermeable clay and wind-blown dust settles between the lapilli, fill-
ing the pores between them. When this has gone far enough, runoff finally begins and the
cone is removed in a geologic instant: In a field of cinder cones it is common to see ungul-
lied fresh cinder cones, but rare to see gullied cones. Instead, one finds lava flows whose
original vents lack cinder cones – they have been removed by fluvial erosion. Cinder cones
in volcanic fields on Venus are far more abundant than on Earth, presumably because of the
greater efficacy of fluvial erosion on our planet.
On Earth, other highly permeable deposits may display a strong resistance to erosion,
not because of intrinsic strength but because of their ability to soak up rainfall and prevent
runoff. Highly fractured lava plains often have very little runoff and may, thus, be very
long-lasting. Even gravel may be highly resistant to erosion for this reason (Rich, 1911).

EBSCOhost - printed on 11/3/2023 9:59 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
388 Water

Water that infiltrates into the soil may percolate down to join the water table, flow lat-
erally to seep out as springs or into streams (meanwhile initiating sapping erosion that may
undermine the slopes out of which it flows), or evaporate from the surface, depending upon
the climate, geologic structure beneath the surface, and rock permeability. The fate of rain-
fall is, thus, complex and depends upon many factors, so that simplified models of the type
presented here may not be realistic: Much current work in hydrology and geomorphology
is focused on overcoming the drawbacks of overly simplistic models.

10.2 Water below the surface

On the Earth, every vacant space in rocks underground is occupied by fluid: water, brine,
oil, natural gas, or, very close to the surface, air. Most of these fluids reside within the
upper few kilometers of the surface, but significant porosity and permeability may extend
to depths of tens of kilometers. When piezometric pressures differ in these fluids they tend
to flow from areas of high pressure to low pressure. Underground flows can be as import-
ant as flows above ground in moving dissolved substances, contaminants, and potential
resources. On Mars, where surface water is not stable under the present climate, a global
underground hydrologic cycle has been proposed to explain sapping and catastrophic out-
bursts of water on the surface. Seeping brines may account for some of the modern-day
gullies on crater walls.
An understanding of how underground water moves and interacts with the surface was
first achieved in 1856 by French engineer Henry Darcy (1803–1858), who investigated
the water supply of Dijon, France. Percolation of fluids through fractured rocks has been
vigorously studied ever since, driven largely by economic concerns with water supply,
oil exploration, and, at present, pollution remediation. In more modern times, M. King
Hubbert (1940) made vital contributions to this field.

10.2.1 The water table: the piezometric surface

When a well is drilled into the surface of the Earth, water is usually not immediately
encountered. Although some water is often found adhering to mineral grains, the zone in
which all the pores are filled with water is only reached at some depth. The top of this satu-
rated zone is known as the water table. It is defined by the level at which water stands in an
open well. Above this fully saturated zone is a thin, partially saturated layer (the “capillary
fringe”) in which capillary forces raise water a small distance against gravity.
If ground water is stagnant and suffers neither gain nor loss, the water table beneath the
surface is level, conforming to a surface of constant gravitational potential (Figure 10.3a).
However, when ground water is recharged by infiltration or drained by discharge, its upper
surface becomes complex, crudely reflecting the overlying topography in regions where the
permeability is uniform (Figure 10.3b).
To describe how water moves underground it is necessary to define the force that makes
it move. Simply citing a pressure is not enough, for the pressure varies with depth in a

EBSCOhost - printed on 11/3/2023 9:59 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
10.2 Water below the surface 389

face
su r (a)

capillary fringe

water table

rainfall (b)

well
infiltration
spring
river

groundwater flow

Figure 10.3 If there is no recharge of the groundwater, as in panel (a), the water table with its capillary
fringe relaxes to parallel an equipotential surface, whatever the topographic complexity above. When
rainfall recharges the groundwater, as in (b), the water table tends to mimic the topographic variations.
Springs and streams of water emerge from the ground where the water table intersects the surface.
Extracting water from a well, shown on the right of this figure, locally draws down the water table
and creates a “drawdown cone.”

stagnant fluid because of the weight of the fluid. The most meaningful measure of the
tendency of a fluid to move is its total energy per unit mass, which is the sum of its gravi-
tational potential energy, kinetic energy, and the work done by the pressure on the fluid
volume. Energy is defined only up to an arbitrary constant, Φ0. Assuming that we can treat
water as incompressible, the total energy of a unit mass of water underground is:

P v2
Φ − Φ 0 = gz + + (10.1)
ρ 2

where z is elevation above some datum (measured positive upwards), P is the pressure, ρ
the (constant) density, and v the fluid velocity. Because water underground usually moves
slowly, the last term can be dropped and the energy equation is written:

P
Φ − Φ 0 = gz + . (10.2)
ρ

In a stagnant fluid Φ is constant and the pressure P = P0 – ρgz, increasing as z becomes

more negative (deeper).
Hydrologists usually do not deal directly with fluid energy, but instead work with the
hydraulic head, h, which is the level above a datum to which a fluid would rise in a tube

EBSCOhost - printed on 11/3/2023 9:59 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
390 Water

inserted at the point of interest. By definition, h = (Φ − Φ0)/g. Variations in head express

variations in the potential energy capable of driving fluid flow. The head is constant if the
fluid is stagnant, but if the head varies laterally, the fluid tends to flow from regions with
higher head to lower head.
A piezometric surface is the hypothetical surface to which water would rise in a well
drilled into an aquifer. It coincides with the water table if the aquifer is unconfined, that is,
not closed in by impermeable layers. However, if impermeable layers do exist the piezo-
metric surface can be very different from the surface topography.
When two fluids are present in the same aquifer, for example, oil and water or fresh
water and brine, their potential energies are different even at the same elevation and pres-
sure because their densities in Equation (10.2) are different. Thus, oil will displace water
downward because it is less dense and brine will displace fresh water upward because it is
more dense. The density of a fluid may also vary for other reasons, such as the content of a
solute or temperature. Potentials thus exist for driving convective motions of the fluid due
to either compositional or thermal differences. There are numerous specialized treatments
for these cases and we do not consider them further in this book.

10.2.2 Percolation flow

It is important to distinguish the porosity of a rock from its permeability. Porosity meas-
ures the fraction of void space in a rock. It is usually designated by the symbol φ. Porosity
ranges from 0 (a perfectly dense rock) to 1 (a hypothetical “rock” that is all void). If the
density of a non-porous rock is ρ0, then the density of a rock with porosity φ is (1– φ) ρ0.
Permeability, on the other hand, measures how readily a fluid can move through a rock. A
rock must have some porosity to be permeable, but it is possible for a rock to be porous
without having any permeability at all – it depends on how well connected the pores are.
Henry Darcy first determined the relation that is now named after him. He measured
the rate at which water flows through a cylinder full of sand as a function of the difference
in head between the water at its entrance and exit and wrote an equation very similar to
Fourier’s law for heat conduction:

kρ kρg
Q == − ∇Φ = − ∇h = − K ∇h (10.3)
η η

where Q is the vector volume flux of water (which has the same units as velocity), η is
its viscosity, and k is the permeability, which has dimensions of (length)2. Permeability k
is defined by the rate at which a fluid (water) moves through a rock. The permeability is,
thus, a kind of fluid conductivity. K = kρg/η is known as the hydraulic conductivity. The
equations describing the percolation of a fluid underground can be put into the same form
as the heat conduction equation, which immediately makes a host of solutions to the heat
conduction equation applicable to the problem of fluid flow (Bear, 1988; Hubbert, 1940).
The magnitude Q of the volume flux of water in Equation (10.3) is not equal to the vel-
ocity of the water in the rock, despite having the same units. Sometimes called the “Darcy

EBSCOhost - printed on 11/3/2023 9:59 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
10.2 Water below the surface 391

velocity,” it is the average rate at which a volume of water flows through the rock. However,
because water fills only a fraction of the total rock’s volume, the water must move faster
than Q to deliver the observed flux. If all of the porosity φ contributed to the permeability,
the actual velocity of water in the rock would equal Q/φ. In practice, not all of the porosity
contributes to the permeability and the local velocity may be much higher. The local vel-
ocity may become sufficiently high in some circumstances that the flow becomes turbulent,
in which case the permeability depends on the flow rate in a complicated manner.
Equation (10.3) can be elaborated to describe the unsteady flow of fluids as well as co-
transport of heat and dissolved substances. Combined with other relations describing the
storage of fluid in the rock and the rate of chemical reactions, many geologic problems
can be related to the transport of water, or other fluids, through fractured rock (Lichtner et
al., 1996). Problems of this type are important for understanding the origin of ore bodies,
underground motion of contaminants, and the transformation of sediments into rock (dia-
genesis and metamorphism).
Computation of the permeability of a given rock is a very complex affair and many dif-
ferent equations for estimating the permeability of a rock exist. Indeed, much of the uncer-
tainty in the field of hydrology centers about permeability and its distribution. Permeability
depends strongly on the size and distribution of the connected passages through rock. A
simple model in which the rock is filled with a cubic array of tubes of diameter δ spaced at
distances b apart can be solved to give a permeability (Turcotte and Schubert, 2002):

π δ4
k= . (10.4)
128 b2
Note the strong dependence of permeability on the size of the narrowest passages through
the rock. This is a typical behavior: Permeability is a very strong function of the grain size
of the rock. Thus, it is large for coarse sands, but becomes very small for fine-grained silts
and clays. Table 10.1 lists some “typical” values for the permeability of various rock types
and indicates the enormous range of permeability found in nature. Note that permeability is
commonly listed in the more convenient units of “darcys”: 1 darcy = 9.8697 × 10–13 m2.
It is very important to note that, although most of the values for permeability in
Table 10.1 apply to “intact” rocks, the actual permeability of a rock mass may be very
different (higher) because of fractures or even macroscopic cavities. Permeability is usu-
ally measured for small specimens that are often selected for their integrity. A rock mass
in nature, however, is inevitably cut by large numbers of joints and faults that can have a
major effect on the permeability of the mass as a whole: open joints allow fluids to flow
along them easily, while faults may be lined with fine-grained gouges that inhibit fluid
motion. Extrapolations of permeability from a small sample to an entire rock mass may,
thus, give highly misleading results.
Where possible, permeability is measured directly in the field. The first such method
was developed by Charles Theis (1935) and involves measuring the level of water in obser-
vation wells adjacent to a test well from which water is actively pumped. Other meth-
ods involving suddenly displacing the water in a single well by a cylindrical “slug” and

EBSCOhost - printed on 11/3/2023 9:59 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
392 Water

Table 10.1 Permeability of rocks

Rock type Permeability k (m2)

Gravel 10–9–10–7
Loose sand 10 –11–10 –9
Permeable basalta 10 –13–10 –8
Fractured crystalline rocka 10 –14 –10 –11
Sandstone 10 –16 –10 –12
Limestone 10 –18–10 –16
Intact granite 10–20–10–18

After Turcotte and Schubert (2002), Table 9–1.

a
Lichtner et al. (1996)

measuring the recovery of the water level, or by injecting water between pressurized pack-
ers in a borehole.
The porosity and permeability of rocks generally decrease with increasing depth. Near
the surface, wide variations of permeability are common, which lends hydrology much of
its variety and complexity. However, with increasing depth, overburden pressure rises and
pores are crushed closed. Temperature also increases with depth and this promotes pore
closure by viscous creep, so that porosity and permeability in terrestrial rocks are essen-
tially absent at depths greater than a few kilometers in most places.
On the Moon and planets such as Mercury and Mars, large ancient impacts initially
fractured the surface rocks to great depth, but overburden pressure has closed most pores
at a depth that depends on the resistance of the rock to crushing. Seismic measurements
on the Moon suggest that most lunar porosity disappears at a depth of about 20 km, or at
a pressure of about 0.1 GPa. Under Martian surface gravity this implies that most porosity
is crushed out at a depth of about 6.5 km. The fractured Martian regolith is estimated to be
capable of holding the surface equivalent of between 0.5 and 1.5 km of water.

10.2.3 Springs and sapping

When water percolating through rock reaches the surface it leaves the fractures and pores
of the rock and flows out onto the surface, either joining a saturated body of water already
there (a streambed, lake, or the ocean), or flowing out onto the land surface as a spring.
Because of the low permeability typical of rocks, the flow of water underground is usually
slow and springs may continue to flow long after rains have ceased (see Box 10.1 to esti-
mate how long a streamflow may continue without recharge). Water underground consti-
tutes a reservoir that buffers changes in surface runoff. The total volume of groundwater
on the Earth is only about 1.7% of the total surface water (this includes the oceans), but
30% of the Earth’s fresh water resides underground. Other planets may have substantial
inventories of groundwater: The principal uncertainty in estimates of how much water is

EBSCOhost - printed on 11/3/2023 9:59 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
10.2 Water below the surface 393

Box 10.1 How long can streams flow after the rain stops?
The question of how streams can continue to flow even when they are not being fed directly
by rainfall is as old as the science of hydrology. Pierre Perrault (1608–1680) was the first to
show that the rain- and snowfall in the basin of the Seine River is about six times larger than
the river’s discharge and could, thus, more than account for the flow from all of the springs and
streams in the region (he neglected evaporation and transpiration by vegetation, which balances
the net water supply).
Some streams, particularly those in arid climates, do not flow unless they have been recently
filled by rain. The water in such streams rapidly infiltrates and percolates downward to a deep
water table. In more humid climates, however, streams may continue to flow for weeks or
months between rains. Such streams are fed by groundwater seeping into their beds. Because
groundwater moves slowly, it may take a considerable period of time for groundwater to move
from the area where it infiltrates to its emergence in the bed of a stream, during which time the
stream continues to flow.
Using Darcy’s law for a uniform permeability, we can estimate the relaxation time over which
an elevated water table continues to feed a stream or spring from which the water drains. Let h
be the height of the water table above its outflow point, located a horizontal distance L from the
groundwater divide (Figure B10.1.1). The total volume of water contained above the outflow
point is of order hLφ per unit length along the stream, where φ is the permeability of the aquifer.
If the width of the zone along which water seeps out is w, and the groundwater discharge is Q per
unit area, then water is discharged from the groundwater reservoir at a rate Qw per unit length
of the stream (perpendicular to the plane of Figure 10.B1.1). The rate at which the height of the
̇ = Qw, from conservation of volume. The discharge Q is, thus,
water table declines is thus hLφ
̇
related to the height of the water table (which is equal to the head in this case) by Q = h Lφ/w.
This expression can now be inserted into Darcy’s Equation (10.3) to give:

 φ
hL kρg h
Q= =− (B10.1.1)
w η L

L
ce
rfa
stream su water table
w
h

Figure B10.1.1 The capacity of subsurface water to support a surface streamflow depends on
the quantity of water available, measured by the height of the water table h above the elevation
of the stream and the horizontal extent of the water supply, L. This water flows to the stream,
discharging a volume flux Q over a stream of width w. The rate of flow is regulated by the
average permeability of the rock beneath the surface.

EBSCOhost - printed on 11/3/2023 9:59 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
394 Water

Box 10.1 (cont.)

where the gradient ∇h has been approximated by h/L. This equation can be rearranged into a
first-order differential equation for the height of the water table h:

 k ρ gw 
h = −   h. (B10.1.2)
 ηφ L2 

The solution to an equation of this type is well known:

h = h0e−t/τR (B10.1.3)
where the relaxation time, τR, is given by:

ηφ L2
τR = . (B10.1.4)
k ρ gw

Inserting some typical values, L = 1 km, w = 10 m, φ = 0.1, k = 10–10 m2 (sand), and the
density and viscosity of water at 20°C, we find a relaxation time of about 4 months. It,
thus, appears that even for relatively permeable aquifers like sand, the time for discharge
of groundwater may be a large fraction of a year and streams may continue to flow even
after months of drought. Conversely, recharge of a depleted aquifer may take a similarly
long period. The timescales for groundwater flow may be very long because of the small
permeability of many rocks that form aquifers.

present on Mars is its subsurface inventory of ice and liquid water. Ice appears to lie within
tens of centimeters of the surface at high latitudes, suggesting that much of Mars’ crustal
porosity may be saturated with water. Similarly, the volume of methane observed in Titan’s
lakes might be only a fraction of the total residing below the surface. The Huygens lander
detected methane that evaporated from the regolith beneath the warm lander, suggesting a
methane table just below the surface of the landing site.
Water flowing from springs may undermine the mechanical stability of the surface. Just
beneath the surface, the exit flow creates a pressure gradient that is connected to the fluid
discharge by Darcy’s law (10.3). This extra pressure adds to the pore pressure in the rock
and may greatly weaken it, for reasons described in Section 8.2. In addition, the flow of
water through the surface may carry fine-grained silt and clay out of the rock matrix, further
weakening it. This selective removal of fine-grained constituents or dissolution of grain-
binding minerals may disintegrate the rock or soil and excavate tunnels through which the
water flows out readily. Known as piping, such tunnels develop in regions of heavy rainfall
and the cavities it creates can undermine the soil surface (Douglas, 1977). Piping also plays
a major role in undermining dams through which water is seeping. Water seepage may, thus,
cause wholesale collapse of the rock face from which it flows. This erosion process is known
as sapping.
Sapping may produce large-scale landforms. Such landforms are rare on Earth, where
overland flow and stream transport usually cause erosion. However, in restricted locales

EBSCOhost - printed on 11/3/2023 9:59 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
10.3 Water on the surface 395

where a permeable layer is underlain by an impermeable zone, surface runoff is limited and
most of the outflow occurs at the interface between the permeable and impermeable layers.
This situation is common, for example, on the USA’s Colorado Plateau where permeable
eolian sandstones overlie clay-rich fluvial formations (Howard et al., 1988). Former inter-
dune areas in the sandstone units also create extensive impermeable layers. Springs form
everywhere at the contact between permeable and impermeable layers.
Long-continued spring flow at the base of cliffs undermines the overlying rock by dis-
solving the minerals that cement the sand grains into sandstone. The constant seepage at the
base of cliffs also encourages chemical weathering that further weakens the rock. Alcoves
form where the cliffs collapse and spring-fed streams carry the fallen debris away. As time
passes, the alcoves grow deeper because they serve as drains for the subsurface water flow
and, thus, intensify the flow at their heads as they lengthen. Given sufficient time, a can-
yon system develops that is characterized by stubby, sparsely branched tributaries with
steep amphitheater-like headwalls. Canyon de Chelly is a terrestrial example of this kind of
development. Many canyons on Mars also appear to have formed by groundwater sapping,
of which the Nirgal Vallis is a prime example.
Groundwater sapping is of particular interest for Mars because it does not require over-
land flow initiated by rainfall. However, excavation of a canyon does require some process
that removes collapsed material from the headwall. Streamflow cannot presently occur on
the surface of Mars and so even these valleys may be relicts from a time when Mars pos-
sessed a higher atmospheric pressure.

10.3 Water on the surface

The terrestrial hydrologic cycle is dominated by the flow of water over the Earth’s surface.
Water falls as rain or snow, of which a portion flows over the surface from high elevations
to low. In the process, liquid water entrains solid material, transports it, and eventually
deposits it at lower elevations. Because of the prevalence of this process, undrained depres-
sions on the Earth’s surface are rare and, where they occur, draw immediate attention as
indicators of an unusual geologic situation.
The processes and products of fluvial erosion, transport, and deposition are so familiar
to Earth-bound geologists that it comes as a shock to realize that no other body in our Solar
System, with the possible exception of Titan, is so completely sculptured by fluvial ero-
sion as the Earth. The “rock cycle” as envisioned by James Hutton is essentially a fluvial
cycle, in which rock debris is removed from the land by rainfall, washed to the oceans by
streams, and converted back into rock after deep burial by other sediments. No other planet
in the Solar System experiences a cycle of this type. Because of the importance of fluvial
processes to terrestrial geology, each part of this cycle has been examined in great detail by
many researchers. Most texts that deal with surface processes on the Earth devote one or
more chapters to each portion of the fluvial cycle. This book attempts to put surface proc-
esses into a broader Solar System context, so our coverage of fluvial processes is neces-
sarily more cursory, being compressed into a single chapter. References to more detailed
treatments are given at the end of this chapter.

EBSCOhost - printed on 11/3/2023 9:59 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
396 Water

rainfall

divide
r

q(x)

1
α ν z
x

Figure 10.4 Runoff from rainfall on a sloping surface accumulates as an increasing function of
distance x from the divide. As the volume q(x) of runoff increases the flow both moves faster and
becomes deeper. The flow velocity is regulated by an equation that incorporates both these factors.

10.3.1 Overland flow

When rain falls on a slope, that portion that does not infiltrate or evaporate flows off the
surface as runoff. An important characteristic of runoff is that it increases steadily with
distance downhill from the crest of the slope. The slope crest is an important location for
fluvial processes and in this context it is called a divide because runoff moves in different
directions on either side of it. If the rate at which runoff is generated per unit area (projected
onto a horizontal surface) is r, measured in units of m3/s per m2, or m/s, then the volume
flux of water flowing off the slope per meter of contour distance q (m2/s) is given by:
q(x) = r x = zV (10.5)
where x is distance from the divide measured along the slope gradient, z is the average
depth of the flow (measured perpendicular to the slope), and V is the mean velocity of the
flow (Figure 10.4). The runoff r is roughly equal to the rate of rainfall minus the infiltration
rate. The volume flux of water, or slope discharge, increases with distance from the divide
because the amount of water passing each contour of the slope includes all of the runoff
generated between that contour and the divide. Because of the increasing volume of water,
the potential for erosion also increases downslope. This is one of the distinguishing char-
acteristics of fluvial erosion: The farther one travels from the divide, the greater the ability
of water to erode. The consequences of this relationship for slope profiles will be explored
in more detail in Section 10.4, but it should already be clear that it leads to the concave-up
profile typical of landscapes dominated by fluvial erosion (see Figure 8.6).
Fluvial erosion. The process of sediment entrainment by overland flow was first ana-
lyzed by Robert Horton (1875–1945) in a landmark paper published just months before
his death and is now known as “Horton overland flow” (Horton, 1945). Horton assumed
that runoff begins as a thin sheet of nearly uniform thickness before concentrating into
rilles farther downslope. Although thin sheets of water may flow in this way, during heavy

EBSCOhost - printed on 11/3/2023 9:59 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
10.3 Water on the surface 397

rainfall on smooth slopes the flow organizes into waves, now known as roll waves, that
develop when the Froude number, v 2̅ /gd, exceeds approximately 4. Under the wave crests,
erosion proceeds faster than expected for a uniform flow. Horton described such waves, but
he neglected the process of rain splash, which can mobilize soil particles in the zone that
Horton considered to be erosion-free as well as generating turbulence in the flowing sheet
of water. Nevertheless, Horton’s model agrees qualitatively with observations and is still
widely used to explain the onset of water erosion.
Horton supposed that runoff close to a divide would form a sheet too thin to entrain
surface material and so a “belt of no erosion” would develop. Such belts are observed in
badlands that lack vegetation, although rain splash may actually entrain some material in
the flow as well as redistributing it through creep. Carson and Kirkby (1972) point out that
natural surfaces are seldom smooth enough to permit an actual sheet of uniform depth to
flow over the surface and that the flow very rapidly becomes concentrated into small rilles
and channels. When the flow is thin, however, rain splash and surface creep constantly
rearrange these channels so that persistent rilles do not form. Farther downslope, with
increasing discharge, rilles do form that evolve into drainage networks.
Nevertheless, as Horton described, there is a finite distance between a divide and the
location of the first permanent rille. This distance depends upon the infiltration capacity of
the surface: It is larger for materials with high infiltration capacity, such as sand and gravel,
and small for materials with low infiltration capacity, such as clay or silt.
The capacity of the flowing water to entrain surface material and erode the slope is pro-
portional to the shear stress exerted on the surface, as discussed in Section 9.2.1. The shear
stress τ exerted by a sheet of flowing water of depth z on a slope α is given by:
τ = ρgz sin α (10.6)
where ρ is the density of the fluid (water). Flowing water begins to erode its substrate
when the shear stress exceeds a threshold that depends on the grain size and cohesion of
the surface material as well as on the properties of the fluid. Modern research on this topic
has centered on the semi-empirical Shields criterion, originally proposed by A. Shields in
1936 (Burr et al., 2006). This criterion can be expressed in a non-dimensional form that is
applicable to any planet. The threshold shear stress is given in terms of a non-dimensional
parameter θt:
τt = (σ − ρ) gdθt (10.7)
where σ is the density of the entrained material and d is the particle diameter. Defining a
friction velocity v* in terms of the shear stress as in Equation (9.6), and a friction Reynolds
number Re* as in Section 9.2.1, the Shields threshold is given by (Paphitis, 2001):

θt =
0.188
1 + Re*
(
+ 0.0475 1 − 0.699e −0.015 Re . *
) (10.8)

This expression is good up to Re* ≈ 105. This curve lacks the steep upturn at small Re*
described by Bagnold (discussed in Section 9.2.1) and by Shields in his original work. This

EBSCOhost - printed on 11/3/2023 9:59 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
398 Water

(a)

(b)

Figure 10.5 Overland flow of a thin sheet of water over an erodible bed is not stable. Panel (a) shows
a uniform sheet flowing over the surface of the soil. Panel (b) indicates that any small, accidental,
increase of depth concentrates the flow, increasing the shear stress under the deeper flow (and at the
same time thinning the flow over the adjacent regions), which increases the erosion rate beneath the
deeper portion, producing a positive feedback that quickly concentrates the flow into channels.

is reportedly because this equation takes into account rare but intense turbulent fluctua-
tions in the boundary layer. It seems unclear at the moment whether this equation applies
to eolian transport as well as water transport: The data on which it is based becomes sparse
and somewhat contradictory below Re* less than about 0.1, although it is stated to be valid
down to Re* = 0.01. The experimental material may also have contained a mixture of grain
sizes. It seems that this equation should be applied with caution to very small grains until
this situation is clarified.
When the threshold stress is exceeded, water begins to entrain material and carry it away,
eroding the underlying slope. Erosion is enhanced by any factor that increases the shear
stress: Deeper flows and increased slopes both contribute to the erosion rate, in addition to
the erodibility of the material.
Rille networks. As runoff moves downslope it collects into channels, and the channels
merge into larger channels. This process has been observed both on natural slopes exposed
to gullying by vegetation removal and in experimental rainfall plots. Horton described
the evolution of rilles into a drainage network in detail. He began by asking the question
“Why, then, do rille channels develop?” He supposed that accidental variations in slope
topography first concentrate sheet flow into slightly deeper than average proto-channels.
Because these concentrations are deeper than average, they also exert greater than average
shear stress on their beds and, thus, grow still deeper. That is, a flowing sheet of water of

EBSCOhost - printed on 11/3/2023 9:59 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
10.3 Water on the surface 399

(a) (b)

Figure 10.6 Straight, parallel rilles flowing off of an initially uniform slope are themselves unstable.
Side branches develop that collect the water from adjacent rilles, deepening the main channel while
starving the adjacent rilles. This process, called “cross-grading,” operates by the principle that “the
rich get richer while the poor get poorer.” A dendritic stream network is the result of this kind of
natural selection, which can be seen as a variety of stream piracy. Redrawn after Carson and Kirkby
(1972, Figure 8.1).

uniform thickness on an erodible bed is unstable: Any slightly deeper pocket erodes faster
than the adjacent regions and becomes still deeper, soon concentrating all of the flow in one
channel (Figure 10.5). Horton observed, however, that the actual evolution of a system of
rilles is more complex, because deepening channels initiate lateral flow that diverts water
from adjacent rilles, “pirating” their headwaters in a process Horton termed “cross-grad-
ing.” This lateral evolution continues until a network of initially parallel rilles becomes a
branched network, illustrated in Figure 10.6.
The process of rille development and cross-grading proceeds at progressively larger
scales as the landscape evolves. Undrained hollows are filled in, steep scarps are eroded
down, and the stream network adjusts itself until, as described in 1802 by John Playfair
(1964),
Every river appears to consist of a main trunk, fed from a variety of branches, each running in a valley
proportioned to its size, and all of them together forming a system of [valleys], communicating with
one another, and having such a nice adjustment of their declivities that none of them join the principal
valley either on too high or too low a level.

EBSCOhost - printed on 11/3/2023 9:59 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
400 Water

Figure 10.7 Dendritic stream network on Mars. Warrego Valles at 19.1° S and 244.0° E. This is part
of the densest drainage network on Mars and strongly suggests that it was created by surface runoff
from precipitation. Image is 24.8 km wide. THEMIS image, PIA05662. NASA/JPL/ASU.

Because of the tendency of water to collect in channels, the landscape becomes dissected
into a fractal pattern of valleys and hillslopes on many different scales (Figure 10.7). In
strong contrast to soil creep, which tends to simplify contours and smooth slopes, fluvial
erosion sharpens irregularities.
Drainage basins. As stream networks develop, the landscape becomes divided into a
hierarchical set of units known as drainage basins. In a fluvial landscape, drainage basins
are the most natural geographic unit from the viewpoint of water supply or transport of
sediment and solutes. The drainage basin associated with some point on a stream or river is
that area of the land that contributes runoff to the stream. Divides – hillcrests down which
water flows in different directions – separate drainage basins. The size of a drainage basin
varies with the size of the stream: Small rilles may have drainage basins of only a few
hundred square meters, while major rivers have drainage basins that encompass most of a
continent.
One of the most regular quantitative relationships in fluvial geology relates drainage
basin area Ab to the length Lb of the basin. Over a range of 11 orders of magnitude, from
tiny rilles to continental drainages, the relation states (Montgomery and Dietrich, 1992):

Lb  3 Ab . (10.9)

Drainage networks are self-similar over this enormous range. That is, a map of a drain-
age network does not allow one to infer the actual scale of the image. However, this regular
relationship breaks down at the smallest scale where the “belt of no erosion” asserts itself

EBSCOhost - printed on 11/3/2023 9:59 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
10.3 Water on the surface 401

at a drainage length of a few tens of meters. Channels disappear below this scale. Another
way of looking at this is to note that there is a threshold for channelized stream erosion
that must be exceeded before network processes become important. Once this threshold is
crossed, however, the processes that shape drainage networks become self-similar and any
hint of a scale disappears from the system.

10.3.2 Streamflow
After his success in understanding eolian processes, Ralph Bagnold, our hero of Chapter 9,
spent the rest of his career working on sediment transport in streams, applying ideas of fun-
damental physics to streamflow. He wrote a series of classic papers that still form the basis
of our understanding of how fluids affect their beds (Bagnold, 1966). Before Bagnold, G. K.
Gilbert approached the problem of sediment transport in streams from an experimental per-
spective, having constructed an enormous flume on the campus of the University of California
at Berkeley to understand how streams transport sediment (Gilbert, 1914). Gilbert’s interest
had been piqued by the very practical problem of how to deal with the fluvial debris produced
by hydraulic gold mining in California’s Sierra Nevada. By 1905, that debris was advancing
down California’s rivers and had begun blocking the mouth of San Francisco Bay, creating a
classic confrontation between the interests of gold miners, farmers along the river margins,
and ocean shipping (Gilbert, 1917). Although Gilbert’s meticulous flume experiments pro-
duced data that are still cited today, he failed to come up with any simple laws describing the
interaction of sediment transport with the flowing water. His results were summarized by a
large number of empirical correlations that, to a large extent, still characterize this field.
Sediment transport. Once established, steams continue to erode their beds and transport
sediment delivered to them from upstream tributaries. The load of material transported by
a stream is divided into several components. The dissolved load is composed of chemically
dissolved species or colloids that are uniformly mixed with the water. The bedload consists
of coarse material that slides, rolls, or saltates along the bed. Finer sediment moves as a “sus-
pended load” that is concentrated near the bed but may be found higher in the water column,
while the “washload,” composed of still finer sediment, is uniformly mixed with the water. The
criterion that distinguishes these categories is based on the dimensionless ratio ζ between the
terminal velocity v of the sediment (Section 9.1.1) and the friction velocity of the flow, v*:

terminal velocity v
ζ= = . (10.10)
friction velocity v*

The transition between bedload and suspended load can be taken at ζ = 1.0 to 1.8, while
that between suspended load and washload occurs about ζ = 0.05 to 0.13 (Burr et al.,
2006). Figure 10.8 illustrates these transitions and the threshold of motion using the mod-
ern version of the Shield’s curve, Equation (10.8).
Stream velocity and discharge. A frequently asked question is, “What is the discharge (or
velocity) of a stream or river given its depth, width and slope?” This question has engaged

EBSCOhost - printed on 11/3/2023 9:59 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
402 Water

102 103
Quartz in H2O on Earth

101 102

Friction velocity, ν* (m/s)

Mean flow velocity (m/s)

d
oa
hl
100 ? lo a d 101

s
wa
d ed
? en
sp
? su loa
d
10–1
? bed 100
?
?
10–2 10–1
no motion
transport
10–3
10–6 10–5 10–4 10–3 10–2 10–1
Grain diameter d, m

Figure 10.8 The relation between friction velocity and grain diameter for quartz grains in water
on Earth. The heavy curve for grains moving along the bed has no minimum, although the queried
extension at small grain sizes indicates much uncertainty in this conclusion, because the data from
which the heavy curve was compiled may include sediment with a range of sizes and, thus, represents
the impact threshold. Other curves show the thresholds for suspending grains in the lower water
column or mixing it entirely through the water mass. The limits for either transport or no bed motion
at low velocities are also shown. Greatly simplified after Figure 3 in Burr et al. (2006).

hydraulic engineers for centuries and several widely used equations can be found in the
literature. French engineer Antoine de Chézy (1718–1798) gave the first useful answer to
this question in 1775. His formula, as written today, is:

V = C RS (10.11)

where V is the mean velocity of the flow (equal to the discharge Q divided by the cross-
sectional area A of the stream), S is the slope of the channel, equal to tan α, and R is the
hydraulic radius (equal to the area A divided by the perimeter of the wetted surface P). C
is a dimensional constant that Chézy determined by comparing the velocity of one stream
with that of another. Although this equation was adequate for Chézy’s canal-design efforts,
it contains many empirical constants and it is unclear how to scale this to a planet with a
different gravitational field and to fluids other than water.
The next major improvement in an equation for streamflow came from the Irish engineer
Robert Manning (1816–1897) in 1889. The “Manning” equation that we now write was
neither recommended nor even devised by Manning himself, who actually did include the
acceleration of gravity in his original formula. As usually written, the equation is:
kM
V= R 2 /3 S1 / 2 . (10.12)
n
Where kM is a dimensional factor equal to 1.49 ft1/2/sec or 1.0 m1/2/sec.

EBSCOhost - printed on 11/3/2023 9:59 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
10.3 Water on the surface 403

Table 10.2 Manning roughness for terrestrial rivers

Bed material Grain size (mm) Manning roughness n (m1/6)

Sand 0.2 0.012

Sand 0.4 0.020
Sand 0.6 0.023
Sand 0.8 0.025
Sand 1.0 0.026
Gravel 2–64 0.028–0.035
Cobbles 64–256 0.03–0.05
Boulders >256 0.04–0.07

Data from Arcement and Schneider (1989).

The factor n, called the “Manning roughness,” has dimensions of (length)1/6. This factor
is widely tabulated for different channel conditions (smooth concrete, gravel, rock, etc.;
usually in units of ft1/6). It, like the Chézy coefficient, does not indicate how to scale to other
fluids or planetary surface gravities. Table 10.2 gives some representative values of n for
terrestrial rivers in metric units.
The most fundamental approach to this problem makes use of the balance between driv-
ing forces and resisting forces through the Darcy–Weisbach coefficient f. Unfortunately, this
generality comes with the price of an equation that cannot be expressed as a simple analytic
formula. Julius Weisbach (1806–1871) was a German engineer who focused on fundamen-
tal equations in hydraulics. In 1845 he published his major work on fluid resistance.
Consider a straight section of a stream channel that we will, for the moment, take to be
rectangular in section with depth h and width w. The weight of the water per unit length of
channel is ρgwh. The component of the force acting downstream is (ρgwh) sin α, where
α is the channel slope. The shear stress on the bed and sides of the stream is just this force
divided by the area of the streambed and wetted banks, equal to the perimeter P = (w +2h).
The stress τb on the wetted bed of the stream is, thus:

ρ gwh sin α  wh   A
τb = = ρg  sin α = ρ g   sin α = ρ gR sin α (10.13)
w + 2h  w + 2h   P

where R is, again, the hydraulic radius. Note that for a stream much wider than its depth
the hydraulic radius is nearly equal to its depth. Weisbach related this shear stress, the driv-
ing force, to the flow resistance, which he expressed in terms of the mean velocity V of the
stream:

f ρV 2
τb = . (10.14)
4 2

EBSCOhost - printed on 11/3/2023 9:59 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
404 Water

Equating (10.13) and (10.14), then solving for V, we obtain the Darcy–Weisbach equa-
tion for the mean flow velocity:

8
V= g R sin α . (10.15)
f

If we ignore the difference between sin α and tan α for small angles, comparison of
(10.15) and (10.11) shows that the Chézy coefficient is:

8g
C= . (10.16)
f

This indicates how the average stream velocity depends on the acceleration of gravity,
but we still lack an expression for the Darcy–Weisbach friction coefficient f. Determination
of this coefficient requires the solution of a transcendental equation, for which the reader is
referred to a clear and detailed discussion in the book by Rouse (1978). The friction coeffi-
cient is a function of the Reynolds number of the streamflow, Re = ρVh/ η. For Re >> 1000
the usual practice is to relate f to the widely tabulated Manning roughness n and to use these
empirical tables to calculate f:

1 kM R1 / 6 (10.17)
= .
f 8g n

It may come as a surprise that, although this expression depends explicitly on the accel-
eration of gravity, it does not contain the viscosity of the liquid. The flow velocity does, in
fact, depend somewhat on the fluid viscosity, but only through the Reynolds number Re. At
very low Reynolds number this equation becomes equivalent to the expression for the vel-
ocity of a flowing viscous liquid, Equation (5.15), which depends on the inverse viscosity,
1/η, but at high Reynolds number the viscosity of the fluid does not matter much because
the resistance to fluid flow depends mainly upon the exchange of momentum by turbulence
in inertial flow.
Floods. The discharge of a stream or river varies enormously over a seasonal cycle
and from season to season. Furthermore, catastrophic floods have occasionally scoured
the surface of both Earth and Mars. Most sediment transport takes place during floods.
There is no simple rule that relates sediment concentration to discharge, but rivers do
typically carry more sediment in flood than during average flows. However, as illus-
trated in Figure 10.1, the peak sediment transport takes place during greater than aver-
age floods because the transport capacity increases non-linearly with increasing flow
velocity. The morphology of a river valley is, thus, controlled by large, rare events, a
fact that makes many fluvial features difficult to understand unless this is taken into
account.
The level of water carried in a river channel varies with the discharge. It is difficult
to estimate the depth of a natural flow of water from a given discharge without detailed

EBSCOhost - printed on 11/3/2023 9:59 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
10.3 Water on the surface 405

Figure 10.9 Ares Vallis is one of the large outflow channels on Mars. This image shows the transition
between the Iani Chaos region to the lower left and the plains of Xanthe Terra to the top (north). The
spurs between the individual channels have been shaped into crude streamlined forms by massive
floods of water. 10 km scale bar is at lower right. Mars Express images by ESA/DLR/FU Berlin. (G.
Neukum). See also color plate section.

knowledge of the topography. The Darcy–Weisbach equation, (10.15), can be written in

terms of the total discharge Q = VA and a geometric factor A R :

8
Q= A R g sin α . (10.18)
f

As the discharge increases in a flood, the product of channel area and the square root of
hydraulic radius increases, but to make definite statements about the depth of the flow we
have to know how the depth and width vary with each other. In a wide, shallow channel,
A R ≈ wh3 / 2, but unless the width is known we cannot solve for the depth as a function of
the discharge. In general, the width of the channel increases as its depth does, so all we can
say is that increased discharge leads to increased depth.
Catastrophic floods. It is somewhat easier to estimate the discharge in the aftermath of
a flood when we know both the depth achieved by the flow and its width. This method has
been applied to the outflow channels on Mars to estimate the discharge during the height
of the floods. Some of these channels are hundreds of kilometers wide (Figure 10.9) and, if
they were once completely filled, must have carried enormous volumes of water. Martian
channels present the problem that the depths of the flows are not known, but estimates
based on the elevation of water-modified surfaces adjacent to the channels suggest that
discharges ranged from 107 to 109 m3/s, compared with about 107 m3/s for the largest ter-
restrial floods (Carr, 1996). The total volume of water is estimated from the volume of sedi-
ment removed. Assuming a maximum sediment/water ratio (typically about 40%), one can

EBSCOhost - printed on 11/3/2023 9:59 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
406 Water

estimate the water volume. Combined with the peak discharge rate, this gives the duration
of the flood. For example, the Ares Vallis flood was estimated to have moved 2 × 105 km3
of material. If the flow was 100 m deep, then the flood must have lasted 50 days. If it was
200 m deep, then it lasted 9 days.
Floodplains. Apart from catastrophic floods of the type that created the Martian outflow
channels or the Channeled Scabland on Earth, much smaller floods occur regularly on ter-
restrial rivers and streams. During times of larger than average discharge the water spills
over the banks and floods the adjacent terrain, putting the excess water into temporary
storage. Unless steep canyon walls confine the channel, the water spreads out over a broad
area, where its velocity decreases. Sediment previously in suspension settles out into a thin
deposit of fine-grained material adjacent to the channel. The grain size of the deposited
sediment falls off rapidly with distance from the channel, grading laterally from coarse
near the former banks to fine silt farther away.
Repeated overbank flooding eventually builds up a gently sloping plain adjacent to the
channel, known as the floodplain. Gentle ridges and swales, traces of former channels, usu-
ally curve across floodplains as their parent rivers shift. “Oxbow” lakes (abandoned former
channels) and wetlands occur locally. The material that underlies floodplains is generically
referred to as “alluvium.” Alluvium is generally fine-grained material: gravel, sand, and
silt, that is only temporarily at rest. Deposited by floods, its fate is to eventually become
re-mobilized as the channel shifts and move on downstream, traveling inexorably toward
the sea by slow leaps and bounds.
Levees. After repeated cycles of floods, the relatively coarse deposits close to the channel
build up broad natural levees that stand above the level of the surrounding floodplain and
tend to confine the water to the main channel. In subsequent floods the water rises higher in
the channel and, when it eventually breaks through a levee, causes more violent floods. The
location where the water breaks through a levee is known as a crevasse and the fan-shaped
deposit of coarse material laid down after a breakthrough is known as a splay.
Floods are a normal part of the hydrologic cycle, although humans who have built struc-
tures on the floodplain often treat them as a major calamity. The floodplain is an active and
necessary part of a river system, but its operation is unfortunately sporadic, making it diffi-
cult for many people to appreciate its essential role in the river system.
Alluvial fans. When a sediment-laden stream debouches onto a land surface, as opposed
to a body of water, the flow spreads out, slows down, and the sediment is deposited in a
conical heap known as an alluvial fan. In plan view, the contours of alluvial fans form
circular arcs that center on the mouth of the stream. At any given time the stream flows
down a radial channel over the surface of the fan, but as sediment is deposited and the
channel builds up, the active stream eventually shifts to another direction, in time cover-
ing the entire conical pile. On alluvial fans in California’s Death Valley one can easily
recognize multiple deposits of different ages by the different degrees of desert varnish on
the fan debris. The slope of an alluvial fan is typically steep at its head and gentler with
increasing distance downslope, in the concave-upward pattern characteristic of fluvial
landforms.

EBSCOhost - printed on 11/3/2023 9:59 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
10.3 Water on the surface 407

The area of an alluvial fan Af is related to the area Ab of the drainage basin of the stream
that feeds it by a simple equation (Bull, 1968):

A f ≈ cAb0.9 (10.19)

where the dimensional coefficient c varies with the climate and geology of the source area.
This equation has been established both in the field and in small-scale laboratory experi-
ments. Alluvial fans are often fed by debris flows and, high up on the fan surface, one can
often recognize boulder-strewn debris flow levees. Lower down, the surface is covered
with finer silt where muddy splays of water separated from the boulders as the fan slope
decreased. When many alluvial fans form close together against a mountain front they may
merge into a sloping surface known by its Spanish name, a bajada.

10.3.3 Channels
Channels develop where the flow of water over the surface persists over long periods of
time. The morphology of the channel itself is often distinctive and, thus, indicates the
action of a fluid flowing over the surface and excavating the channel, although the nature
of the fluid itself is often unclear. When the Mariner 9 images of Mars first revealed giant
outflow channels in 1971, many planetary scientists did not believe that they could have
been cut by water, because Mars’ atmospheric pressure is too low to sustain liquid water
on the surface. Many different fluids were proposed, ranging from low-molecular-weight
hydrocarbons to ice or mixtures of ice and water. Although it is still not possible to entirely
rule out ice or brines, in the wake of the discovery of channel networks that seem to require
overland flow, most planetary scientists now accept the likelihood of liquid water on the
surface of Mars under climatic conditions that differ greatly from those prevailing today.
Channels on Titan (Figure 10.10) were probably cut by liquid methane and channels on
Venus were formed by lava flowing over its hot surface.
Channel features. Streamflow over a granular bed produces a variety of distinctive fea-
tures from the interaction of the fluid and the deformable bed. Streamflow in a fixed chan-
nel is difficult enough to analyze at high Reynolds number because of the complex nature
of fluid flow, but when coupled with the additional complexity of a deformable bed it
poses problems that have yet to be completely solved. Nevertheless, extensive experimen-
tal study by many researchers using flumes and field observations of streamflow and its
consequences has produced some understanding of how flow affects the bed of a stream
and what features are produced by the interaction of the streamflow and its bed.
The major factor in the formation of bed features such as ripples, dunes, and larger accu-
mulations of sediment is grain size. In addition, some measure of the velocity of the flow is
needed. Much research has shown that to understand channel features the best measure of
the flow is stream power (Allen, 1970), given by the shear stress on the bed τ* multiplied
by the mean velocity V, τ*V, which is measured in W/m2. Low-power streams do not trans-
port sediment at all, but as the stream power increases characteristic fluvial features such as

EBSCOhost - printed on 11/3/2023 9:59 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
408 Water

Figure 10.10 Mosaic of three images from the Huygens Descent Imager/Spectral Radiometer
showing a dendritic drainage system on the surface of Titan. Each panel of the mosaic is about 7.5
km wide. Image PIA07236 NASA/JPL/ESA/University of Arizona.

ripples and dunes develop as a function of both grain size and stream power (Figure 10.11).
A peculiar finding is that, at certain combinations of stream power and grain size, the bed is
flat, even though intense sediment transport may be occurring. There is a small such plane
bed field for intermediate stream power and large grains, and a much larger field at high
stream powers for all grain sizes. These are designated, respectively, the lower and upper
plane bed regimes.
Antidunes are unique to shallow, rapid flows. Unlike dunes, antidunes move upstream
as flow proceeds. Although the form itself moves against the flow, sediment continues to
move downstream: Only the wavelike shape of this feature moves counter to the current
direction. Antidunes form when the Froude number of the flow, V / gh , approaches 1.
Their wavelength is approximately 2πh, where h is the depth of the flow.
Ripples, dunes and antidunes can be distinguished in sedimentary deposits by means of
cross-bedding. The ability to “read the rocks” and infer the nature of a flow, whether fast or
slow, unidirectional or alternating, shallow or deep, is an important tool in the kit of a sedi-
mentary geologist (Allen, 1982; Leeder, 1999). Observations of apparent cross-bedding in
Martian sedimentary deposits by the Opportunity rover have led to the important inference
of the former presence of shallow lakes on the surface of Mars (Grotzinger et al., 2006b).
Bedforms develop even in streams that flow over solid rock. Channels often excavate
into the underlying bedrock by quarrying away small joint blocks through differential pres-
sures and cavitation behind obstacles. Gravel and cobbles carried along the bed may, over
time, gradually erode the bed by abrasion. The result is flutes and, where circular motion is
maintained over long periods of time, potholes. Potholes reach impressive depths of tens of
meters as they are ground into the bed by swirling cobbles and debris below the streambed.

EBSCOhost - printed on 11/3/2023 9:59 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
10.3 Water on the surface 409

plane bed
10

Stream power, W/m2

dunes

1.0

plane bed
ripples

0.1
no sediment movement

1 0.5 10
Grain diameter d, 10–4m

Figure 10.11 The type of bedform that develops beneath flowing water depends upon both stream
power and the grain size of the sediment. This figure applies to water flowing over quartz sand on
Earth, but it shows the variety of forms that can develop. No movement occurs below the lowest line.
Note that the bedform pattern is not unique: Plane beds occur for both high stream power (upper
regime flow) and low stream power over coarse sediments (lower regime flow). Simplified after Allen
(1970, Figure 2.6).

In addition to the abrasion of the bed, the material carried by a river or stream is itself
abraded as it travels downstream. Angular grains of sand become rounder, gravel becomes
finer and cobbles are smoothed, rounded and reduced in size as they move downstream.
Such downstream variations in the size of transported sediment make it difficult to associate
a particular grain size with fluvial processes, because grains that might begin at the threshold
of transportation move into suspension as they are broken and abraded while large, initially
immovable rocks are broken down and begin to slide and roll along the bed.
Streamlined forms. Large floods create characteristic bedforms. One that is considered
particularly diagnostic of floods are teardrop-shaped, streamlined islands that develop
behind obstacles that divert the flow. These features are bluntly rounded in the upstream
direction and taper to a point downstream. Once diverted by the obstacle, the flow closes
back in around it downstream, but because this takes some time, the obstacle shields a
tapered triangle from the flow, creating this shape. The faster the flow, the more gradual
is this closure and the longer the island becomes. Streamlined islands in the Channeled
Scablandsof Washington State are typically about three times longer than their maximum
width, whereas streamlined forms on Mars are a little longer, about four times their width
(Baker, 1982).
Hydraulic geometry. A widely used metric relates quantitative descriptors of the chan-
nel to its discharge. Discharge is chosen as the independent variable because it is believed
that, while a stream may adjust its width, depth, or velocity by moving sediment from one

EBSCOhost - printed on 11/3/2023 9:59 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
410 Water

Table 10.3 Hydraulic geometry of selected rivers

(Leopold et al., 1964, p. 244).

Location Type b f m

Midwest, USA Fixed station 0.26 0.40 0.34

Semiarid, USA Fixed station 0.29 0.36 0.34
Rhine River Fixed station 0.13 0.41 0.43
Midwest, USA Downstream 0.5 0.4 0.1
Semiarid, USA Downstream 0.5 0.3 0.2

place to another, the discharge is determined by the climate and area of the drainage basin
and so cannot be adjusted by the interaction between the stream and its bed. The method is
entirely empirical: Data are collected from many rivers and streams, plotted against log-log
axes, and lines are fit to the data that typically form fuzzy linear arrays. These fits must thus
be regarded as approximate, but they do indicate general trends. Because straight lines on
log-log plots indicate power laws, the relations for stream width w, depth h, and velocity
V are written:

w = aQ b 

h = cQ f  (10.20)
V = kQ m 


where a, c, k are dimensional fitting parameters and b, f, m are dimensionless exponents.

There are two constraints on these parameter sets because Q = whV: a c k = 1 and b+f+m = 1.
Because river discharge even at one location is not constant over time, there are two ways
in which these parameters are compiled: either at a fixed location as a function of time, or
at different locations downstream on the same river. The exponents, b, f, m are considered
the most significant parameters and they are tabulated in Table 10.3 for a small number of
river systems.
Ratios illustrate the utility of this parameterization. For example, the aspect ratio of a
river is the ratio between its depth and width. The aspect ratio h/w is proportional to Q f–b, so
at a fixed location the aspect ratio equals Q0.14 for rivers in both the Midwest and semiarid
USA. Thus, as discharge increases the relative depth increases slowly at a given station.
On the other hand, going downriver the aspect ratio decreases slowly with discharge as
Q–0.1 for Midwestern rivers: Near its mouth the Mississippi is much shallower for its width
than it is upstream. But, despite folklore, the Mississippi is not actually a “lazy” river – its
velocity continues to increase with discharge either downriver or as a function of time at
one location.
Meandering rivers. The tendency of rivers and streams to deviate from a straight course
has long fascinated observers. Rivers flowing without constraint over an erodible bed
quickly develop a meandering course. In an initially straight channel, the meanders begin

EBSCOhost - printed on 11/3/2023 9:59 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
10.3 Water on the surface 411

as gentle bends as the water swings from side to side. The meanders grow in amplitude
until they develop such extreme hairpin curves that they loop back on themselves until the
water finally cuts through the narrow neck, temporarily shortening the course of the river.
The abandoned meander loop forms an oxbow lake on the floodplain adjacent to the new
main channel. The wavelength of a meander is a function of the channel size. A careful
regression of meander wavelength λm and channel width w shows that they are nearly (but
perhaps not exactly) proportional to one another (Leopold et al., 1964):

λm = 10.8 w1.01 (10.21)

where the wavelength and width are both in meters. Although meander loops on a river
floodplain are continually growing and being cut off, there is correlation between meander
amplitude Am and width as well:
Am = 2.4 w1.1. (10.22)
In addition to meandering laterally, rivers also meander vertically: Rhythmic variations
of depth develop as deep pools alternate with shallow riffles with the same periodicity as
the lateral meanders. The pools develop on the outside of meander bends, while the riffles
form between them. These rhythmic depth variations develop even when the channel is
confined between rocky walls that suppress lateral meanders, such as in the Colorado River
confined within its canyon in Arizona.
One occasionally finds a river channel meandering through a valley that itself mean-
ders on a much larger scale. In such cases one can infer that the smaller stream (called
an “underfit” stream) carries a much smaller discharge than its former counterpart. This
relationship is often observed in channels that once drained the meltwater from retreating
Pleistocene ice sheets on Earth. A similar relationship, but in a sinuous lava channel, is
observed in Schröter’s Rille on the Moon.
The outside of meander bends is usually a steep bank that is often undercut and is obvi-
ously undergoing erosion. A gently sloping bar on which sand and fine gravel is deposited,
called a point bar, occupies the inside of the bend. As the channel shifts laterally, the flood-
plain is consumed at the outer part of the bend and rebuilt on the inner bend. Cross-sections
of the migrating channel show coarse-grained material (often gravel) at the former channel
floor, fining upward into sands where the point bar is deposited, then silts where former
point bars are buried by floodplain silts. Such fining-upward sequences, when they can be
recognized in ancient fluvial deposits, provide a direct indication of the depth of the former
river channel (Figure 10.12). From the depth, the correlations of hydraulic geometry yield
an estimate of the discharge of the ancient river that created the floodplain.
Meanders do not form simply because water flowing in the straight sections between
meanders impinges on the outside of the bend. Many authors, including James Thomson
(William Thomson’s brother) in 1876 and Joseph Boussinesq in 1883 independently dis-
covered the helical flow of water in meander bends. However, the most famous re-dis-
coverer of this effect was Albert Einstein, who perceived the effect while stirring a cup of
tea (Einstein, 1954). Originally publishing in 1926, Einstein noted, as many others have

EBSCOhost - printed on 11/3/2023 9:59 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
412 Water

Cut bank

}
Point bar (Erosion)
(Deposition) silt
fining-
sand upward
sequence
gravel
coarse bed load

Figure 10.12 Water on the outside of meander bends rises higher than the water on the inside of the
bend because of centripetal acceleration. This drives a secondary circulation that moves water from
the outside of the bend toward the inner bend to create an overall helical circulation as the water
moves downstream. Sediment is eroded from the outside of the bend and deposited on point bars
on the inside. At the same time, coarse material remains near the bed while sand is deposited higher
up on the point bar. Silt is deposited on top of the sand during overbank floods to produce a fining-
upward sequence of sediment sizes.

done, that the tealeaves in the bottom of a stirred cup of tea gather together at the center
of the cup. He inferred that the rotating liquid must create a helical flow in addition to its
rotation. This flow descends along the outside of the cup and ascends in its center, sweep-
ing the tealeaves into the center as it flows. Einstein realized that this flow could account
for river meanders and point bars by transporting sediment down along the outside of the
meander bends and depositing it on the inner point bar. The reality of such helical flows in
river bends has now been demonstrated many times and this spiral flow is an accepted part
of river hydraulics.
The helical flow is due to two factors. The first is centrifugal force. As the water flows
around the bend, it rises higher along the outside of the bend. By itself, this would not cause
a secondary flow if the water in the river were in rigid-body rotation. However, friction
reduces the water velocity along the contact between the water and the outside wall of the
bend and the excess pressure caused by the elevated water surface drives a flow downward
along the outside wall.
Meanders are not restricted to just rivers flowing over granular material. They are com-
monly observed in glacial meltwater streams flowing over solid ice and in lava channels
(they are called “sinuous rilles” in lava feeder channels on the Moon). In these cases one
must presume that the same helical flow enhances channel migration on the outside of
bends, perhaps by thermal erosion as the flow brings hotter material to bear against the
outside of bends, but the precise mechanism is currently unclear.
Braided rivers. Steeply sloping streams heavily loaded with coarse sediment do not
flow in well-defined channels. Instead, the flow divides into a complex network of shallow
short-lived channels that diverge and rejoin many times as the water and sediment move
downstream. Such channels are known as braided rivers.
Unlike rilles on a slope, the constantly shifting channels in the bed of a braided river
do not unite to create channels progressively more capable of carrying the available load
of sediment. Such streams are sometimes described as “overloaded,” in the sense that the

EBSCOhost - printed on 11/3/2023 9:59 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
10.3 Water on the surface 413

union of two sediment-laden channels is less capable of transporting the load than the indi-
vidual channels, so some of the load is deposited when channels join, creating a temporary
bar that eventually diverts the flow to a new location.
The factors that decide whether a stream channel is meandering or braided are still
poorly understood (Schumm, 1985). Several rivers have been observed to alternate in
style between meandering and braided or vice versa in historical time, a process termed
“river metamorphosis” by fluvial geologist Stanley Schumm. For example, the channel of
the South Platte River in Nebraska changed from braided in 1897 to meandering in 1959
in response to a large decrease in mean annual discharge due to irrigation projects that
extracted water from the river (Schumm, 1977, p. 161). Likewise, the lower reaches of
the Pleistocene Mississippi River were braided because of its greater slope down to the
lowered sea level during the ice ages, as well as to the greater discharge it carried as the
ice sheets melted. As the sea level rose and the ice sheets disappeared, its channel changed
from braided to meandering.
An often-cited criterion that divides meandering from braided rivers on Earth is expressed
in terms of channel slope S and bankfull discharge Qbf, where the discharge is in m3/s
(Schumm, 1985). The river is usually braided when the average channel slope exceeds:

S ≥ 0.0125 Q bf−0.44. (10.23)

Thus, any factor that increases either slope or discharge favors braiding over meander-
ing. The frequent observation that braided rivers typically carry coarse debris is impli-
cit in Equation (10.23) through the connection between average channel slope S and the
grain size of bedload: Rivers that carry coarse debris are steeper than those that carry fine
sediment.
The paleohydrologic hypothesis. Schumm also noted an apparent connection between
channel stability and the presence of vegetation. Vegetation growing on islands in shifting
channels tends to stabilize them. Root mats add cohesion to channel banks and hold fine
sediment in place until undercut by powerful currents. Both of these factors tend to favor
meandering channels rather than braided. Schumm noted the lack of evidence for meander-
ing river channels before the Late Paleozoic era when vegetation first covered Earth’s land
surface. He elaborated a “paleohydrologic hypothesis” that suggests that all river channels
were braided before the evolution of land plants. If this connection between vegetation and
channel form is correct, we should not expect to find meandering river channels on Titan
or Mars. On Titan, present resolution is too poor to be sure, but meandering channels do
appear to cross the surface of Xanadu. On the other hand, the boulder-strewn surface at the
Huygens landing site is consistent with a braided river channel (Figure 7.16d). On Mars,
there are now HiRISE images of indisputable meandering channels in Aeolis Planum (Burr
et al., 2009), Figure 10.13. It must, thus, be possible for meanders to develop in the absence
of vegetation (Howard, 2009), perhaps because of cohesion from clay or permafrost that
binds the sediment together. Channel meanders, while evidently not requiring the presence
of vegetation, may, nevertheless, indicate special mechanical conditions in the sediment
adjacent to the channel.

EBSCOhost - printed on 11/3/2023 9:59 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
414 Water

Figure 10.13 A meandering channel in Aeolis Planum, Mars, that belies the proposal that meanders
develop only when banks are stabilized by vegetation. These highly sinuous meanders actually stand
as ridges at the present time, an example of inverted topography. Gravel in the channel presumably
made it more resistant to erosion than fine-grained surrounding material. Portion of HiRISE image
PSP_006683_1740. Image is approximately 2.3 km wide. NASA/JPL/University of Arizona.

River terraces. Gently scalloped scarps are often found parallel to the active flood-
plains of large river systems. The downstream slope of the relatively flat surface behind
such scarps is similar to that of the active floodplain. These surfaces, which often look
like treads on a giant staircase stepping up away from the river, are known as river ter-
races. Because of their distinctive appearance and their importance for land use, ter-
races have received a great deal of attention in the terrestrial geologic literature (e.g.
Ritter, 1986).
River terraces are abandoned floodplains of a river system that has eroded deeper into
its valley. Geomorphologists distinguish paired terraces, which appear at the same eleva-
tion on opposite sides of the river, from unpaired terraces. Terraces are the result of the
lateral migration of the river channel back and forth across the floodplain as the channel
slowly erodes downward into the floor of the river valley. Terraces are important because
they indicate changing conditions, although they are not usually diagnostic of exactly what
conditions are changing. For example, the erosion they record could be due either to tec-
tonic uplift of the rock underneath the stream, or increased downcutting of the stream.
Downcutting can be due to increasing water supply, decreasing sediment load, or a lower
base level at which the river discharges.
Although it is sometimes stated that the existence of discrete terraces indicates that
downcutting must be episodic rather than steady, this is not necessarily true. Because a long
interval of time may separate the impingement of the main channel on one valley wall dur-
ing its slow lateral swings, the change in the level of the stream between two terrace-cut-
ting events reflects the accumulated erosion between cutting events. Discrete terraces form

EBSCOhost - printed on 11/3/2023 9:59 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
10.3 Water on the surface 415

even if the rate of downcutting is uniform because of this interval between terrace-cutting
events. Although the ages of river terraces on the Earth can now be determined through the
measurement of cosmogenic isotopes, it is still extremely difficult to discriminate episodic
versus steady downcutting from such data.
Tributary networks. The most familiar pattern of drainage networks is one in which
smaller channels join into larger ones that, in turn, join still larger ones, forming a network
formally called a tree or dendritic pattern (Box 10.2). This tributary pattern persists on the
Earth’s land surface over most of its area because of the increasing capacity of downstream
water to carry sediment. Rivers that branch downstream and then rejoin do occasionally
occur on Earth, but they are relatively rare. Such non-tree-like patterns are more common
in Martian channels, for reasons not currently understood.
The junction angle in tributary networks is such that the acute angle between links usu-
ally occurs upstream of the junction. This is presumably because the momentum of the
joining currents tends to carry both in the same direction – it is unusual for a tributary to
discharge its water upstream into the channel it joins.
Distributary networks. When a stream or river can no longer carry its sediment load, due
either to loss of water by infiltration into a substrate (as on an alluvial fan), or because of
a decreasing gradient (as when it encounters a lake or the ocean), its sediment is deposited
and a system of dividing channels develops. The branching pattern of such a distributary
network may resemble the tree-like form of a tributary network, but the slope in this case is
reversed: The largest channels are upslope of the smallest channels. The acute angle of the
junctions is downstream in this case, again tending to preserve the momentum of the div-
iding channels. Similar networks develop among the channels actively feeding lava flows
spreading over flat terrain.
Unusual networks develop where the flow direction alternates, such as in tidal marshes
where the surface is alternately flooded and drained. In this case the same channels serve
alternately as a distributary and a tributary network. In such networks the channels tend to
divide and rejoin frequently and junction angles are typically close to 90°, perhaps because
of the frequent collisions between incoming and outgoing streams of water.
Venusian channels. No one seriously expected to find fluvial channels on Venus. The
surface temperature is far too high to permit liquid water to flow over its surface. However,
images returned by the Magellan radar (Figure 10.14) reveal channels that are remark-
ably similar to those of terrestrial river systems. Meandering channels with natural levees,
streamlined islands within the channel, even crevasse splays and abandoned meanders can
all be identified in the images. On Venus, however, we must certainly be looking at chan-
nels that once carried lava, not water. Because of Venus’ high surface temperature and large
eruption volumes, lava cools relatively slowly compared to terrestrial lava flows; further-
more, the flows may have continued over such long periods of time that “fluvial” features
developed. Lava, like water, is capable of eroding its bed by plucking and of transporting
“sediment” in the form of more refractory minerals, so this may be a case in which simi-
larity of physical process promotes similarity of form, even though the materials involved
are very different.

EBSCOhost - printed on 11/3/2023 9:59 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
416 Water

Box 10.2 Analysis of stream networks

Rilles and gullies join to form larger streams, which join again to form still larger streams, and
so on up to large rivers. The result is a branching or dendritic (tree-like) network that extends
from the smallest rilles to the largest trunks: rivers like the Mississippi or the Amazon. Water
and sediment eroded from the land are flushed down these channels, eventually to be deposited
in the oceans.
Most sediment is fed into the network from overland flow at the level of rilles. Larger and
larger streams mainly serve to transport it. Most erosion, thus, occurs on the scale of small
drainage basins, grading into transportation at larger scales, although deeply incised rivers such
as the Colorado may be fed large amounts of material directly along the main channel by mass
movement.
Robert Horton (1945) brought order to stream network analysis by proposing a simple
numbering scheme, which has been slightly improved by other authors. Horton assigned the
smallest recognizable rilles to order number 1. When two order 1 channels join they become a
channel of order 2. The union of two order 2 channels is of order 3, and so on. When a channel
of lower order joins a channel of higher order, the order of the higher channel does not change.
Thus, when an order 3 channel is joined by an order 2 channel, it remains of order 3. See, for
example, the fourth-order network shown in Figure B10.2.1.
To appreciate the success of Horton’s idea of numbering from the smallest rilles to the
largest river, consider the opposite scheme in which the main trunk of a large river is assigned
order 1. Proceeding upstream to smaller and smaller tributaries, we would fined that the last
recognizable rilles in most basins had a different order, even though their function in the

1 1
1 1
1
2
1 2 1
1
1 1
1 2 1
3 1
2 2
1 2
3 1
2 1 1
1 3
4 1
1 2
1
1
1 1

Figure B10.2.1 A typical stream network, which illustrates the ordering of stream segments.
This network is of order 4. After Figure 10.1 of Morisawa (1968).

EBSCOhost - printed on 11/3/2023 9:59 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
10.3 Water on the surface 417

Box 10.2 (cont.)

hydrologic system is identical. Apparently similar rilles on opposite sides of a divide would, in
general, be assigned different orders.
Some North American examples of high-order river systems are the Mississippi of order
10, the Columbia of order 8, the Gila of order 8, and the Allegheny of order 7. Unfortunately,
tables of the orders of all the world’s rivers seem to be hard to find, although ordering is so
suitable for computer computation that the ArcGIS program includes a tool that assigns orders
to streams.
The major problem with this scheme is that the definition of the first-order rille is uncertain:
Depending on the map scale, this could be the smallest recognizable rille (as Horton
supposed), or the smallest perennial stream (which means its assignment depends on climate).
Inadequate resolution caused a problem with the initial ordering of Martian channel networks:
At the lower resolution of Mariner 9 and Viking images, Martian valley networks appeared
to have much lower drainage densities than terrestrial networks, for which a variety of causes
were cited (Carr, 1996). However, once higher-resolution Mars Orbiter Camera (MOC) images
became available, true first-order rilles could be recognized and it was realized that Martian
and terrestrial networks have similar densities (Carr, 2006). Given this history, no one has
seriously tried to assign orders to Titan networks (except at the Huygens landing site) because
of the low resolution at which they are currently seen.
Stream ordering would be an amusing but mechanical pastime, except that ordering clarifies
the statistical properties of networks for practical applications in flood wave prediction and
sediment yield estimates (as only two examples). There are useful quantitative relations
between landscape properties, such as average slope or stream discharge, as a function of
order. It was also originally hoped that the statistical properties of stream orders might be
diagnostic of network origin.
Some properties of a fourth-order drainage basin are listed in Table B10.2.1. Several
characteristics are clear upon inspection: The number of streams decreases sharply with
increasing order. In addition, the average channel slope decreases regularly with increasing
order, the average channel length and basin area increases with order, and the drainage density
is nearly independent of order.
Drainage density is defined as the sum of the lengths of all channels in a basin divided by
its area. Its inverse is approximately the distance between streams in the basin. For first-order
streams this is also the width of the belt of no erosion, as defined by Horton. The drainage density

Table B10.2.1 The Mill Creek, Ohio, drainage network (Morisawa, 1959, Table 12).

Stream Number of Average Average basin Average Stream density

order streams length (m) area (km2) channel slope (km/km2)

1 104 111 0.065 21.6° 3.39

2 22 303 0.313 7.01° 4.36
3 5 1046 1.505 2.23° 3.77
4 1 1915 6.941 0.57° 3.52

EBSCOhost - printed on 11/3/2023 9:59 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
418 Water

Box 10.2 (cont.)

is, thus, closely related to infiltration capacity: Large drainage densities imply a low infiltration
capacity, as often seen in badlands, whereas low densities imply that water sinks in readily.
Plots of the logarithms of different characteristics of drainage networks versus stream
order generally form straight lines. This suggests power-law relationships among the different
quantities. For example, the number of streams of a given order p, expressed as N(p), can be
written:
N(p) = rb(s−p) (B10.2.1)
where s is the order of the main stream in the network. The constant rb is known as a
bifurcation ratio. In the fourth-order network of Table B10.2.1, s = 4. There are five third-order
streams, and s – p = 4 – 3 = 1, so rb = 5. So far, this is nothing new. But now note that for p =
2 this formula predicts that there should be 52 or 25 second-order streams (there are really 22).
For p = 1 the formula predicts 53 or 125 first-order streams (there are really 104). The fit is not
perfect, but it is fairly close.
Similar power-law formulas can be constructed for other quantities, such as stream length
and slope. Over large, high-order drainage basins fits can be adjusted by least squares to obtain
best estimates for each of these quantities, which then describe the branching properties of a
drainage network.
This type of fitting was popular between 1945 and about 1970 when it was believed that
such fits and bifurcation ratios reveal important information about how a drainage network
develops and could, in some way (no one knew quite how) be related to the fluvial processes
that created the network. Unfortunately, geomorphologist Ron Shreve dashed most of
these hopes in 1966 when he showed that relations of this type develop in any dendritic
network, including networks generated by random walks in a computer (Shreve, 1966b).
Shreve’s arguments have been confirmed and extended by later work (e.g. Kirchner, 1993).
Nevertheless, many workers are convinced that stream networks statistics indicate something
about the organization of fluvial processes and attempts have been made to assign a kind of
entropy to stream networks and show that actual networks maximize that entropy (Rinaldo
et al., 1998) or expend the least work. Other, more complex, numbering schemes have been
developed that claim to have genetic significance, but their success is presently unclear.
Statistical descriptions of drainage networks do have practical value in estimating the
numbers and lengths of links at different levels without actually having to measure the entire
network, but it is not easy to relate the parameters to genetic processes. Networks developed by
sapping seem to be less branched and possess shorter links than those developed by overland
flow, but this type of distinction can probably be made without the aid of detailed statistical
analyses.

10.3.4 Standing water: oceans, lakes, playas

Running water tends inexorably downhill. When it reaches the lowest possible level it
accumulates into a body of standing water that may range in size from tiny temporary
ponds to global oceans. The most important geologic fact about running water is its abil-
ity to transport sediment from higher levels to lower. When it enters a large accumulation

EBSCOhost - printed on 11/3/2023 9:59 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
10.3 Water on the surface 419

Figure 10.14 Sinuous lava channel on the plains of Venus. The overall channel flows from Fortuna
Tessera in the north, south to Sedna Planitia. Channel is about 2 km wide and is interrupted by
streamlined islands. The channel pattern illustrates the formation of an alternative channel during
flow. Frame width is 50 km. Magellan F-MIDR 45N019;1, Framelet 18. NASA/JPL.

of water its velocity drops (although currents are never completely absent in any body
of water: Underwater gravity currents are discussed in Section 8.2.3) and its sediment is
dropped somewhere near the shore. In contrast to the land surface, bodies of standing water
are the locales of sediment accumulation rather than erosion.
Standing water, however, possesses its own distinctive ability to move material. This
ability depends on the action of waves, so that sediment transport occurs mainly at the level
of the water surface. Wave action produces distinctive landscape features that remain even
long after a body of water disappears.
Aside from the prevalence of coastlines on the Earth, ancient Mars may have possessed
extensive bodies of water whose former shorelines, if found, would demonstrate their pres-
ence and dimensions. Titan is now known to possess ephemeral lakes of liquid methane and
ethane, making beach and lake processes of prime interest to planetary geologists.
An appreciation of the landforms created by waves began with eighteenth-century
British geologists who initially attributed most former geologic activity to the waves that
they observed crashing around the edges of their sea-girdled isle. In the nineteenth century,
American geologist G. K. Gilbert took a large step forward with his study of the now nearly
extinct Lake Bonneville in Utah (Gilbert, 1890). The present Great Salt Lake in Utah is a
small remnant of a much greater lake that existed during the Pleistocene. When it drained
about 14 500 yr ago, it exposed the beaches, deltas, spits, and bars that formed during its
brief existence of about 17 000 yr.
Gilbert was impressed that most of these features are the consequence of waves breaking
against the shore. His research, as does much modern research, therefore focused on the
generation, propagation, and interaction of waves with the shore.

EBSCOhost - printed on 11/3/2023 9:59 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
420 Water

Waves on water. The generation of waves has received a great deal of attention for its
own sake and we can only touch on the basics in this book. The reader who wishes to go
further should consult the treatise by Kinsman (1965). Waves on the ocean, lakes, or even
ponds are created by wind blowing over the surface. William Thomson (who became Lord
Kelvin) and German physicist Hermann von Helmholtz were the first to understand how
wind can generate water waves by an aerodynamic instability, now called the Kelvin–
Helmholtz instability. The interface between two fluids, such as air and water, cannot
remain flat if the fluids move with different velocities. Waves develop on the interface,
beginning with small, short-wavelength waves for which the restoring force is surface ten-
sion, then growing into much larger waves that are dominated by the weight of the water,
called gravity waves. The overall wave-generation process transfers energy from the wind
to waves on the water surface.
Wind must blow over the surface of the sea for some time, and continue over some dis-
tance, before a fully developed set of waves develops. The size and wavelength of water
waves, thus, depend on the speed of the wind, its duration, and the distance, or fetch,
over which the wind acts. Higher wind speeds develop higher waves of longer wavelength
and period. These simple facts permitted Gilbert to understand why the beaches of Lake
Bonneville are best developed along the Wasatch Mountain front on the eastern side of
the former lake: The prevailing wind blows from west to east, to the extent that shoreline
features are hardly recognizable on Lake Bonneville’s western side where few waves ever
beat.
Although the shape of a water wave may move at high speed over the water surface,
anyone observing the motion of an object floating in the water knows that the water itself
moves very little. There are two velocities relevant to waves. They both depend on the
period T, wavelength L, and water depth H, in addition to the acceleration of gravity g for
a gravity-dominated wave. In general, wave speed c = L/T. The first important speed is the
phase speed, cp, which is the speed at which some part of the waveform, its crest or trough,
moves across the water. For waves of small amplitude (compared to their wavelength) the
general expression for this phase speed is:

gT 2π H
cp = tanh . (10.24)
2π L
In deep water, H >> L, this simplifies to cp = gT/2π. Similarly, in shallow water this
equation simplifies to c p = gH . Note that the speed of deep-water waves depends on
their period, so after traveling some distance long-period waves arrive earlier than short-
period waves. This explains why long, slow waves are the first to arrive at the shore after
a distant storm, followed by shorter, choppier waves. The dependence of wave speed on
period is called dispersion: Wave packets tend to disperse as they propagate, spreading out
and changing shape.
The other velocity associated with waves is called the group velocity cg. This velocity
determines how fast the energy associated with a packet of waves propagates. It can be
derived from the phase velocity by a simple derivative:

EBSCOhost - printed on 11/3/2023 9:59 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
10.3 Water on the surface 421

250

Asymptotic limit
200

Wave Velocity, m/s

Phase
Velocity
150

100 Group Velocity

0
0 20 40 60 80 100
Wavelength, km

Figure 10.15 Group velocity and phase velocity for water waves on the Earth in water 4 km deep,
equal to the average depth of the Earth’s oceans. Long-wavelength waves approach a “shallow-water
wave” limit when their length is much longer then the depth of the water. Short-wavelength wave
velocities are increasing functions of their length in the “deep-water wave” limit.

 4π H 
cp  L

cg = 1 + . (10.25)
2   4π H  
 sinh  L  
 

Figure 10.15 illustrates the relations between the phase velocity, group velocity, and
wavelength for waves in 4 km deep oceans on the Earth. The important feature to note is
that both wave velocities are highest in deep water, and that the group velocity is always
less than the phase velocity at a given wavelength. Thus, the energy from a disturbance on
the ocean propagates more slowly than the leading waves. Waves from deep water thus
slow down on approaching the shore. Because energy is conserved, energy piles up in shal-
low water. More energy means higher waves, so we can deduce that the wave height must
increase as the water shoals.
The energy E (per unit area of ocean surface) in a wave of amplitude A0 (one-half of the
vertical distance from crest to trough) is made up of equal contributions of the gravitational
potential energy and the kinetic energy of motion. Its magnitude is given by:

1
E= ρ gA 02 . (10.26)
2
Because energy propagates at speed cg, the energy flux P in a wave is given simply by
P = cgE.
The path of a particle of water as a wave passes by is approximately a circle of radius
A0 near the ocean surface. At greater depths below the ocean surface the amplitude of the

EBSCOhost - printed on 11/3/2023 9:59 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
422 Water

Figure 10.16 The orbits of particles of water in deep-water waves are nearly circular. The amplitude
of this motion decreases exponentially with greater depth. Although the wave appears to progress
from left to right, the water itself only moves in small circles, whose maximum diameter at the
surface is equal to the distance between the crest and trough of the wave.

circle decreases exponentially. If z is the depth below the surface, the wave amplitude A(z)
in deep water decreases as:
2π z
−
A( z ) = A0 e L
. (10.27)

Figure 10.16 illustrates the rapid decline in wave amplitude with increasing depth. In
water shallower than the wavelength, H < L, the velocity of the water on the bottom is
still appreciable. The orbits of the water particles in shallow water are ellipses, not circles,
which degenerate to a straight line parallel to the bottom on the seabed itself.
The rapid decrease of the amplitude of water oscillation with increasing depth below the
surface defines the geologic concept of “wave base.” It is well known that a submarine can
ride out even the most violent storm by submerging to a depth comparable to the wavelength
of the largest seas in the storm. In a similar manner, the seabed below some critical depth is
unaffected by waves that may break up rocky shorelines. The ability of waves to erode the
shoreline is, thus, limited in depth. The short-lived island of Surtsey off the southern coast
of Iceland provides a fine example of the limited power of the waves. Surtsey was built by a
series of volcanic eruptions in 1963. Well observed and widely reported in the news media,
Surtsey was immediately attacked by the waves and within a few years most of its original
area had disappeared below the waves. The eroded base of the island is still there, but it
was planed off by waves to a depth of about 30 m below the surface, a depth that repre-
sents the effective wave base at this location. In a similar manner, volcanic Graham Island
appeared in the Mediterranean in 1831. Its ownership was hotly disputed by Britain, Spain,
and Italy, but wave erosion cut it down below the sea surface by 1832. Normal waves can

EBSCOhost - printed on 11/3/2023 9:59 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
10.3 Water on the surface 423

move sand down to a depth of about 10 m, so the concept of wave base is somewhat fuzzy:
The exact limit of erosion depends on the frequency of storms, the wavelength (and, hence,
the exposure to wave-generating winds), and the wave amplitude. The important concept is
that waves act only close to the surface of a body of standing water.
As waves approach the shoreline the water shoals and, as mentioned before, the waves
increase in height, eventually breaking. Wave breaking is a complex phenomenon for which
many theories have been proposed. A good summary can be found in the book by Komar
(1997). A simplified way of looking at wave breaking is that it occurs when the velocity of
the water particles at the crest of the wave exceeds the phase velocity. When this happens,
the steepness of the wave front exceeds the vertical and the wave crest cascades over its
front, dissipating much of its energy as turbulence. There are several ways in which waves
break, each type distinguished by the steepness of the beach face. In order of increasing
beach slope, these styles are called spilling, plunging (the iconic breaking wave is a plun-
ging breaker), collapsing, surging, and, in the case of vertical seawalls, a reflected wave
that does not break at all.
The principal consequence of wave breaking is that the wave energy, ultimately originat-
ing from the wind, is focused on the beach, where it is dissipated in turbulence. Where the
waves impinge directly on rocky cliffs, the hydraulic pressures generated by the breaking
waves may drive air or water into joints, loosening joint blocks or abrading the rock by
dashing smaller boulders and sand against it. Wave action moves sand up and down the
beach face and alternately offshore into bars, then onshore onto the beach again. Beach
sand is suspended by each breaking wave and becomes vulnerable to transport by long-
shore or rip currents. Overall, wave energy makes the beach a highly dynamic environment
in which erosion, deposition, and sediment transport are all active processes.
Coastal processes have received a great deal of study and limited space prevents a
detailed treatment in this book: The interested reader is referred to a number of excellent
texts on this subject in the further reading section at the end of this chapter. For our brief
survey here the only other processes of major importance are wave refraction and long-
shore drift, as these are chiefly responsible for building beaches that might be seen from
orbiting spacecraft.
Wave refraction. Wave refraction refers to the bending of wave fronts in water of varying
depth. Once waves begin to “feel bottom” at a depth H equal to about L/2, the phase speed
is proportional to the square root of the depth, c p = gH . Thus, the shallower the water
becomes, the more slowly the wave fronts move. Consider a linear wave approaching a
uniformly sloping shoreline at an oblique angle (Figure 10.17a). Because the wave moves
more slowly in shallow water, the oblique wave gradually rotates to become more parallel
to the shoreline as it approaches. It cannot turn exactly parallel to the beach, but the angle it
makes with the beachfront is greatly decreased before it reaches the beach and breaks.
An oblique wave arrival means that the momentum transported in the wave is not com-
pletely cancelled when the wave breaks on the beach. A component of this momentum
remains and generates a current, the longshore drift, which moves sediment in the direction
of the acute angle between the wave front and the beach.

EBSCOhost - printed on 11/3/2023 9:59 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
424 Water

fast
2
deep
slow a)

3
shallow longshore
current

beach

b)
shallow shallow
deep

beach

deep deep
shallow c)

beach

Figure 10.17 Behavior of wave crests approaching a shoreline. Panel (a) illustrates an oblique
approach of the wave crests (wavy lines) to the shore. As the water becomes shallower, the wave
velocity decreases so that wave crests near the shore travel more slowly than those farther out in
deeper water. The net result is that the wave crests rotate and tend to become parallel to the beach
as they approach. The oblique convergence also transfers a component of momentum along the
beachfront to produce a longshore current. The dashed lines indicate the direction of energy flow
perpendicular to the wave crests. Panel (b) illustrates the refraction of wave energy away from an
offshore trough. The wave crest over the deep water in the trough moves ahead of the wave crests
over its shallow flanks, turning the wave crests away from the axis of the trough and directing energy
away from the trough and onto the adjacent portion of the beach. Panel (c) illustrates the opposite
effect, when the waves approach over a ridge perpendicular to the shoreline. In this case the waves
move more slowly over the shallow ridge and the wave energy is concentrated on the ridge crest. The
combination of the focusing actions shown in (b) and (c) tends to even out submarine irregularities
near the shore by wave erosion of ridges and filling of troughs.

If the bottom is not uniform, wave refraction acts to fill in hollows and erode protuber-
ances. Figure 10.17b shows how waves approaching a shore are refracted over a submar-
ine canyon running perpendicular to the beach. Because the canyon is deeper than the
surrounding seafloor, waves moving down its axis travel more rapidly to the shore. Waves

EBSCOhost - printed on 11/3/2023 9:59 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
10.3 Water on the surface 425

Figure 10.18 Liquid methane lakes near the North Pole of Titan imaged by the Cassini synthetic
aperture radar. Dark regions are smooth lake surfaces and brighter regions are the surface. Intermediate
brightness levels near the lake shores indicate some radar return from the lake bottoms. Image is
centered near 80° N and 35° W and the strip is 140 km wide. Smallest details are about 500 m across.
The radar strip was foreshortened to simulate an oblique view from the west. Image PIA09102.
NASA/JPL/USGS. See also color plate section.

to either side move more slowly and refract the wave fronts away from the canyon axis.
Because wave energy flows perpendicular to the wave crests, the wave energy is refracted
away from the canyon axis toward its edges. The margins of the canyon are, thus, more
heavily eroded by wave action than its axis and so sediment tends to accumulate in the can-
yon, evening out the bottom contours parallel to the shoreline.
On the other hand, a submarine ridge, which might reflect the presence of a headland on
shore, tends to focus wave energy over its crest, as shown in Figure 10.17c. The approach-
ing waves collapse onto the shore right over the ridge, leading to intense wave action and
erosion of the ridge, again evening out the bottom contours parallel to the shoreline.
Wave refraction, thus, tends to straighten out complex shorelines, focusing wave energy
on promontories and diverting it from inlets. A newly flooded landscape, such as might be
created by erecting a dam at the mouth of a river (Lake Powell on the Colorado River is a
good example), presents a fractal shoreline of great complexity with protruding spurs and
deep inlets everywhere. However, if the water level remains constant, in time these spurs
and inlets are battered back and filled up, leading to a much more even shoreline.
The highly convoluted, fractal shorelines of the methane lakes on Titan constitute a
puzzle (Figure 10.18). These shorelines show no sign of wave action; no bars, no spits, or
eroded headlands. On the other hand, an observation of sunlight reflected from the surface
of one lake appears to be perfectly mirror-like, showing no sign of the glitter typical of

EBSCOhost - printed on 11/3/2023 9:59 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
426 Water

reflections from wave-ruffled liquid surfaces. Do lakes on Titan lack waves? If so, why?
Titan’s equatorial region is notable for its broad expanses of sand dunes, so winds must
exist. Or are the levels of methane in the lakes constantly changing so that there is no time
for erosion to straighten out the shorelines? At the moment the answers to these questions
are unknown.
Longshore drift. Longshore drift is another powerful force that tends to straighten out
shorelines. Beachgoers often confound longshore drift with the along-beach motion of
sand particles in the back-and-forth swash of waves breaking on the beach face. This
motion does drive some sand in the general direction of the longshore current illustrated in
Figure 10.17a, but the current that flows parallel to the beach just offshore transports a far
larger flux of sand. This current is driven by the uncompensated component of the momen-
tum of the waves parallel to the beach. It is localized near the beach within the breaker
zone. Ocean bathers are often unaware of its existence until they suddenly notice that they
are far down the beach from where they thought they should be.
Sediment suspended by waves breaking in the surf zone is caught up in the longshore
drift and transported parallel to the beach. This sediment-laden current is a true “river of
sand” that moves large volumes of material along the shore. The direction of the longshore
drift varies with the shoreline topography and the local direction of approaching waves.
Coarse sediment deposited by rivers flowing into a lake or the sea is often caught up by the
longshore drift and moved “down drift” to nourish beaches and build bars across inlets or
spits out from headlands. G. K. Gilbert noted many gravely bars and spits created during
the high stands of Lake Bonneville. These bars and spits are now dry, level ridges standing
in the Utah desert to bear witness to the former existence of a large lake.
Many other currents and interactions occur close to the beachfront. Rip currents develop
outside of the surf zone to return water pushed up onto the beach by shoaling waves that,
unlike deepwater waves, transport water in addition to energy and momentum. Rip currents
are often spaced periodically along the beach, their spacing determined by the excitation
of edge waves, a variety of trapped wave that can exist only along the beach face. Beach
cusps are rhythmically spaced beach features whose origin is still debated, but appear to be
related to standing edge waves. All of these currents have complex interactions with tides
and the material that makes up the beach. For more information, however, the reader should
refer to the references at the end of the chapter.
Playas. Playas are shallow, ephemeral lakes that develop in regions dominated by inter-
ior drainage. Playas are flooded after rain falls in the drainage basins that discharge into
them. The flowing water carries fine silt and dissolved minerals into the lake basin and
then evaporates, leaving this non-volatile material behind. Playas are, thus, accumulation
points for evaporite minerals. These minerals often form concentric rings around the center
of the basin, ranging from the most soluble minerals in the center (usually halite and other
chlorides on Earth) to less soluble minerals at the edges (typically carbonates on the out-
side and sulfates in an intermediate position between carbonates and chlorides). The edges
of playas grade upslope into alluvial fans, which trap most coarse sediment moving from
adjacent mountain fronts.

EBSCOhost - printed on 11/3/2023 9:59 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
10.3 Water on the surface 427

topset
water level

foreset

bottomset

Figure 10.19 Formation of a delta near the mouth of a river or stream discharging into a larger body
of water. As the stream loses its momentum in the lake or a sea it deposits much of the sediment it
carries. This initially produces steep foreset beds of coarse material close to the shore and thin beds
of finer sediment farther away in deep water. As the delta continues to build outward, topset beds are
laid down on a shallow slope, while the foreset and bottomset beds build farther from the mouth of
the stream. After Figure 14 in Gilbert (1890) and based on Gilbert’s observation of deltas left behind
when Lake Bonneville drained.

Because the water that floods playa lakes is often only a few tens of centimeters deep, the
surfaces of playas deviate only very slightly from an equipotential surface – large playas
form some of the most level (but not flat!) surfaces on Earth. Playa surfaces are usually
devoid of rocks or coarse sediment, except in circumstances where other processes move
rocks across them. A famous example is Racetrack Playa in Death Valley, whose surface is
criss-crossed by the trails of boulders weighing many kilograms that somehow move across
the level playa. No one has yet seen these boulders in motion, but they do shift between
repeated surveys, perhaps during winter storms with high winds when the playa surface is
wet and slick.
Playas may also serve as sources of fine dust, as they do in the American southwest.
High winds drive sand grains and mud chips across their surfaces, raising dust that may be
exported in suspension from their immediate vicinity.
Deltas. Deltas form where a river transporting sediment enters a larger body of water,
decreases its velocity, and drops its sediment load near the shoreline. If this sediment is
not carried away immediately by longshore drift, it builds up into an accumulation called
a delta. The eponymous delta is that of the Nile River, which is triangular in plan like the
Greek letter Δ. Because of interaction with waves and currents, deltas can be of many dif-
ferent shapes and sizes ranging from a few meters across to hundreds of kilometers, but all
are sediment accumulations built out into a body of standing water.
The sedimentary layers that compose a delta are divided into three general types:
Bottomset beds that underlie the delta, foreset beds that compose most of its interior and
topset beds that cap it near water level (Figure 10.19). Bottomset beds, as their name
implies, are laid down at the foot of the delta. They are typically composed of fine-grained
sediment that formerly traveled in suspension and may be deposited by density currents
that carry their sediment load far out into the body of water. Bottomset beds are usually
thin and their sediments are graded from coarse at the bottom of each bed to fine near the
top. Foreset beds are laid down in more steeply dipping sets. They are composed of coarse

EBSCOhost - printed on 11/3/2023 9:59 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
428 Water

material originally carried as bedload that avalanches down the front of the delta. The dip
of foreset beds may approach the angle of repose in small lakes, but in the deltas of major
river systems they may dip as little as 1°. Topset beds are extensions of the floodplain. They
come to overlie the foreset beds as the delta advances into the lake or ocean. Topset beds
are composed of sand- and silt-sized material typical of the floodplain and typically show
cut-and-fill channel features.
Sediment deposition in Earth’s oceans is complicated by the mixing of fresh water from
rivers with salt water in the oceans. Fine-grained sediments such as clay particles floccu-
late upon mixing with salt water and settle out more rapidly then they would if they had
entered fresh water.
Turbidity currents. Upon arriving in a large body of quiet water, the suspended load from
rivers usually forms a mixture of water and sediment that is denser than the surrounding
water. If the time required for the sediment to settle out of the mixture is long compared to
the time for the mixture to flow down the face of the delta, it moves downslope as a density
current, called a turbidity current. Turbidity currents act somewhat as underwater rivers:
They gouge underwater channels that may possess levees and create distributary networks
on the lower parts of deltas or deep-sea fans.
Deep submarine canyons that head on the continental shelves off the mouths of major
rivers, such as the Hudson River of New York, were initially thought to require enormous
fluctuations in sea level when they were first discovered. Only after a great deal of research
was it realized that turbidity currents, not subaerial rivers, cut these canyons.
Turbidity currents usually flow episodically. After a period of accumulation near a sedi-
ment source, a threshold is passed and the sediment pile collapses, mixing sediment with
water and generating a muddy, underwater density current. Storms and earthquakes may
also trigger the release of large and powerful turbidity currents.
The deposits of turbidity currents are called turbidites. Turbidites are layered accumula-
tions of sediment that grade from coarse at the bottom of each layer to fine near the top. They
often show evidence of high-velocity deposition, such as incorporation of rip-up clasts and
upper-regime planar bedding. The thickness of individual turbidite beds is usually highly
variable, reflecting statistically random triggering processes. Although turbidites are usually
deposited in deep water, many of them incorporate shallow-water fossils and other debris
acquired in the near-surface source area, before being carried to much greater depths.

10.3.5 Fluvial landscapes

Long-continued fluvial transportation and erosion create distinctive landscapes. We have
discussed branching channel networks, but fluvial processes have additional characteris-
tics. Undrained depressions are rare on fluvial surfaces: Lakes and other depressions are
quickly filled by transported sediment. Terrestrial geomorphologists regard any undrained
depression as an anomaly needing explanation. Even our ocean basins are anomalous: If it
were not for plate tectonic recycling, all land surfaces would eventually be cut down below
the level of the sea (to wave base) and the ocean basins would be partially filled.

EBSCOhost - printed on 11/3/2023 9:59 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
10.3 Water on the surface 429

Timescales are important when considering fluvial erosion. Annual floods and large
multi-annual inundations adjust the forms of the channel and floodplain, but the landscape
as a whole responds on a much longer timescale. This timescale can be estimated by com-
paring the volume of material that can be eroded from a drainage basin to the rate at which
sediment is carried out of the basin. The sediment discharge has been measured for many
watersheds on Earth and the result is about 10 cm of land surface (averaged across the
entire basin) per thousand years for the pre-industrial Earth – present erosion rates are
much higher. These rates depend upon relief in the basin as well as climate, so this is a
very rough average. If we take the average elevation of the continents to be a few hundred
meters, the terrestrial erosion timescale is thus a few million years. This is roughly the time
required for fluvial erosion to strongly affect the Earth’s topography.
Base-level control. The base level for large fluvial systems on Earth is mean sea level.
This is the level to which long-term fluvial erosion tends to reduce the land, because ero-
sion below the level of the sea is very slow. Of course, base-level control on the Earth has
been complicated by hundreds of meters of sea-level change during the glacial cycles of
the past 3 Myr, a fact that must be taken into account when interpreting modern fluvial
landscapes. Local base levels may develop above long-lived lakes or other obstructions to
fluvial downcutting. The base level may occasionally change drastically in extraordinary
events, such as the nearly complete evaporation of the Mediterranean Sea about 6 Myr ago,
which caused rivers such as the Rhone, Po, and Nile to excavate kilometer-deep gorges that
are now buried by modern sediment.
The base level changes whenever a dam is built; we now have considerable experience
with the changes that such disturbances engender. Aside from artificial dams, landslides
and lava flows create natural dams that may block an existing drainage, creating a tempor-
ary lake and inducing changes in river flow that gradually work their way upstream.
Older discussions of fluvial erosion supposed that after some change takes place, condi-
tions remain constant for a long period and the landscape has time to adjust to the change. It
has become clear that the Earth’s landscape is too dynamic for such long-term equilibrium.
Changes usually occur on timescales that are short compared to the equilibration time, and
so the landscape is constantly adjusting to perturbations.
Graded rivers. The concept of the graded river was famously introduced by J. Hoover
Mackin (1948). Mackin proposed that, over the long term, the slope of a river is adjusted
to the volumes of both the water and the sediment it carries. The volume of water increases
downstream (for perennial rivers) as more and more tributaries feed their water into the
main trunk river. At the same time the size of the sediment carried usually decreases down-
stream, permitting more of the load to travel in suspension. Mackin’s idea was that the river
strives toward a balance between the load to be carried and the water that carries it, such
that the “long profile” of the river (its elevation as a function of distance from its mouth)
tends toward a final state that is steep near its sources and gentle near its mouth.
Mackin proposed that the long profile of a river is self-adjusting: If some reach of a river
should suffer a decreased slope for any reason, the sediment that was formerly in transit is
deposited, building up the bed of the river upstream while the water downstream, relieved

EBSCOhost - printed on 11/3/2023 9:59 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
430 Water

of its sediment load, erodes into its bed (this is currently happening around many artificial
dam sites). Both processes tend to increase the slope and oppose the original perturbation
of the profile. Similarly, if any reach becomes steeper for some reason, the capacity of the
river to erode its bed increases and the river cuts into its bed, forming a step in the long pro-
file that gradually propagates upstream until an equilibrium is again established.
A change in the nature of the sediment carried by a river has analogous effects. As
described above, hydraulic mining in California’s Sierra Nevada in the mid-1800s added
a large volume of coarse debris to the rivers flowing into the Pacific (Gilbert, 1917). The
rivers responded by steepening their gradients until the coarse gravel could move down-
stream. The extra load of gravel built up the riverbeds downstream, causing the rivers to
overflow their previous banks and deposit gravel on the adjacent farmland. If the injection
of coarse gravel into the headwaters had continued, the net result would have been a river
system with a much steeper gradient from the ocean to the mountains, although this would
have required burial of most of the interior valley of California – an outcome considered
highly undesirable by the residents of the Golden State, which is why hydraulic mining is
now strictly banned.
Landscape evolution. Theories of how the Earth’s landscape evolves under the influ-
ence of fluvial processes are as old as geology itself. James Hutton in 1795 attributed
river valleys to erosion by the streams that flow in them, an idea that did not sit well with
his contemporaries or even with his intellectual heir Charles Lyell nearly a century later.
Even after the fluvial origin of landscapes was accepted, ideas on how they evolved were
qualitative and made few testable predictions. American geographer William Morris Davis
(1850–1934) is widely remembered for his classification of landscapes as young, mature,
and old, based on a presumed rapidity of tectonic movements, which create initial land-
scapes that are later dissected by fluvial erosion during an era of tectonic quiescence. Davis
supposed that, following a long period of erosion, landscapes are reduced to a surface of
low relief near the base level that he christened a “peneplain.” Davis and German geolo-
gist Walther Penck (1888–1923) engaged in a bitter but somewhat fuzzy controversy over
whether landforms “wear down” (Davis), with slopes everywhere declining as time passes,
or “wear back” (Penck) who suggested that steep slopes retreat from their initial positions
while retaining their steepness.
Classical ideas on fluvial processes, to which the ubiquitous G. K. Gilbert made many
contributions, especially in his report on the geology of the Henry Mountains (Gilbert,
1880), divided the fluvial system into erosion, transportation, and deposition. Erosion takes
place mainly at the level of valley sides and first-order rilles, which yield sediment that
is carried through the stream system and is finally deposited in an alluvial fan or body of
standing water.
Research in the modern era has focused on quantitative descriptions of each of the parts
of this cycle. Until recently, most effort has gone into understanding individual processes,
such as Bagnold’s work on the physics of sediment entrainment and transportation, or have
focused on hillslope processes, of which the book by Carson and Kirkby (1972) is a fine
example. “Process geomorphology” has now become so large a subject, and the synthetic

EBSCOhost - printed on 11/3/2023 9:59 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
Further reading 431

landscape evolution models of Davis and Penck have such a reputation for imprecision,
that many recent textbooks avoid the topic of landscape evolution altogether.
Most recently, perhaps driven by the immense increase in computer power, quantitative
syntheses of fluvial processes into landscape models have become possible and are now
achieving impressively realistic results. These results are being subjected to quantitative
tests through our recent ability to date landscapes through cosmogenic isotope methods.
The first quantitative model of fluvial evolution of this kind was constructed by geophysi-
cist Clem Chase (1991), who built a model of landscape evolution on a two-dimensional
grid that evolved by simple rules suggested by fluvial processes. Simple as this pioneer
model was, it produced very realistic landscapes that evolved in ways similar to those
inferred for actual landscapes. Models of this kind are now reaching a high level of sophis-
tication: The time has clearly come for our detailed understanding of individual processes
to be synthesized into descriptions of how entire landscapes evolve. A recent review of pro-
gress in this area is by Willgoose (2005).
Although the evolution of fluvial landscapes has seemed a quintessentially terrestrial
process (Martian fluvial landscapes are clearly not highly evolved), Cassini images of inte-
grated drainage networks on Titan have created a new field for application of these models.
Many of the basic parameters that control fluvial processes on Titan are unknown: How
much precipitation falls and how it varies with time, what materials are being eroded, how
Titanian “bedrock” (very cold water ice) weathers to sand-sized particles, and many other
important facts are presently unknown. However, landscape evolution models themselves
may shed some light on these questions. For example, how much sediment must be moved
before an integrated drainage network forms? How much erosion does it take to turn a
densely cratered landscape into a fractal landscape of connected drainage basins?

Further reading
Hydrology is an enormous subject on its own. One of the founding papers that can still be
read with profit is Hubbert (1940). This paper takes a quantitative analytical approach that
is difficult to find even in the modern literature. A comprehensive look at the older literature
that still has much of value is Meinzer (1942). A full treatment of the percolation of fluids
through porous media is found in Bear (1988). Students looking for a quick but mathem-
atical overview of flow through porous media and its application to geodynamic problems
can do no better than to consult Turcotte and Schubert (2002).
Fluvial processes are also the subject of an enormous literature, although many of the
modern treatments focus more on societal problems such as pollution and water supply
than on fundamental science. Two of the classic, science-oriented treatments are Schumm
(1977) and Leopold et al. (1964). A thorough examination of the role of water in all surface
processes, not just rivers and streams, is by Douglas (1977). The fluid mechanics of flow in
open channels and an in-depth discussion of the various resistance formulas can be found
in Rouse (1978). The interaction between flowing water (and other fluids) and its channel is
covered by Allen (1970), while the best treatment of the physics of sediment entrainment is

EBSCOhost - printed on 11/3/2023 9:59 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
432 Water

still Bagnold (1966). The sedimentological aspects of transport by water are well reviewed
in the massive book by Leeder (1999). Coastal and wave processes in general are lucidly
discussed in Komar (1997), while the more geomorphological aspects of shorelines and
coasts are treated by Bird (2008). The modern synthetic approach to landscape evolution
is too new to have texts describing it: The interested reader is referred to the short review
of Willgoose (2005), but readers wishing insights into the history of landform analysis will
be delighted by Davis (1969), or for a shorter and more comprehensive introduction by
Kennedy (2006).

Exercises

10.1 Underground plumbing on Titan

During a particularly hot, dry spell on Titan, when temperatures rose to a balmy 97 K, over
a period of one Earth year about a meter of methane evaporated from a (hypothetical) 100
km wide lake. In spite of the methane loss, no change in the level of the lake was observed
(to a precision of ± 10 cm). If no surface methane flowed into the lake at this time, the
loss must have been compensated by subsurface flow. Estimate the minimum permeability
required in the subsurface to supply this loss. Comparing this to permeability of rocks on
Earth (Table 10.1), what can you infer about Titan’s subsurface?

10.2 Take Manning to new worlds

Show how the Manning equation (10.12) scales with the acceleration of gravity g by assum-
ing that the Darcy–Wiesbach coefficient f is independent of gravity. Using this equation,
compute the flow velocity of a flood on Mars that moved down a channel 30 km wide and
scoured hills up to 100 m high above the channel floor. MOLA measurements indicate that
the slope in this region was about 1 m/km. Assume that the material of the valley floor was
large cobbles more than 20 cm in diameter. Use the same method to calculate the velocity
of a methane stream on Titan that flowed in a channel 100 m wide and 3 m deep down
a slope of 0.3°. Estimate the grain size of the material using available data/reasonable
guesses.

10.3 Swing wide, lazy river

A meandering river on Earth flows through a channel 100 m wide and moves along at an
average velocity of 3 m/s. It swings through a meander bend with an inner radius of 0.5 km.
The centripetal acceleration in moving around the bend raises the level of the water on the
outside of the bend relative to that inside. What is this elevation difference and is this large
enough, in your opinion, to measure?

EBSCOhost - printed on 11/3/2023 9:59 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
Exercises 433

10.4 Entrenched meanders on Mars?

One of the major problems of analyzing ancient rivers is figuring out how much water
once flowed through them. Nirgal Vallis on Mars is a sinuous trough nearly 500 km long
and about 8 km wide in its lower reaches. In this area the sinuous undulations and alcoves
along the walls have a wavelength of about 15 km. Although Nirgal has been attributed to
sapping, suppose that its undulations were formed by a meandering river that gradually
eroded down into the surface. Use the information given here to estimate the original width
of its channel. Given this width, how might you estimate the discharge of the river that cut
this gorge?

10.5 Low-gravity surfing

Evidence is accumulating for the former presence of lakes and perhaps oceans on Mars.
Discuss, so far as you are able, how the acceleration of gravity affects waves on the surface
of standing bodies of water (considerations of wave velocity and energy transport are par-
ticularly relevant here). In particular, how do you think low gravity might affect shoreline
processes such as erosion and transport on Mars or Titan?

EBSCOhost - printed on 11/3/2023 9:59 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
11
Ice

My theory of glacier motion then is this – A glacier is an imperfect fluid,

or a viscous body, which is urged down slopes of a certain inclination by

the mutual pressure of its parts.
J. D. Forbes, Travels Through the Alps of Savoy (1846), p. 365

Scottish physicist J. D. Forbes (1809–1868) was an inveterate mountaineer, observer, and

scientific quarreler. His direct observations of glacier flow led him to describe glaciers as
highly viscous fluids, a conclusion that could not be reconciled with the mechanical under-
standing of his day. He spent many years ferociously defending his view against the oppos-
ing opinions of another mountaineering physicist, John Tyndall (1820–1893), an Irishman
who argued that glaciers deform by regelation, a process in which ice melts and refreezes
under pressure.
Both Forbes and Tyndall had become fascinated by the mechanics of glacial phenom-
ena at a time when Swiss paleontologist Louis Agassiz (1807–1873) was announcing his
deduction that Europe had once been smothered under immense glaciers during a former
age of ice. Although Forbes’ views are now accepted as being closer to the truth, a full
understanding of how crystalline ice can flow like a viscous fluid was not achieved until the
1950s. Tyndall’s regelation idea still finds a place in glacier mechanics, although it is now
relegated to sliding at the base of warm-based glaciers.
Ice on the surface is mirrored by ice in the ground. The surfaces of cold regions are often
underlain by frozen ground that, because of seasonal thermal cycling, is unusually active
and highly productive of distinctive organized patterns.
The slow flow of ice, whether frozen water or other substance, as an “imperfect fluid
or a viscous body” creates landforms that are different from those produced by wind or
water. Ice occurs either disseminated through the regolith, in small bodies properly called
glaciers, and in large ice sheets that can cover large areas.
Copyright 2011. Cambridge University Press.

11.1 Ice on planetary surfaces

Glaciers are the counterparts of rivers in a cold hydrologic cycle. Ice accumulates as snow
at high elevations, slowly flows down to lower elevations, and there melts, eroding and

434

transporting the adjacent rock as it moves. Glaciers on Earth currently underlie about 10%
of its surface and incorporate about 3% of its water, a number that is rapidly declining in
the modern anthropogenic world. During the Earth’s recent Ice Age glaciers and ice sheets
occupied 30% of the surface and locked up 8% of the water. Water-ice glaciers are appar-
ently active on Mars at the present time and, on Mars as on the Earth, seem to have been
much more extensive in the past, although the former Martian Ice Age occurred much
farther back in time than the Earth’s. Mars may also host glaciers composed of solid CO2,
which require temperatures lower than any achieved on Earth. So far, no glaciers have been
found elsewhere in the Solar System: Titan’s surface is just a bit too warm for methane
to freeze into ice. Ammonia glaciers are theoretically possible but no examples have yet
been discovered. Although some researchers have speculated about solid nitrogen glaciers
on Triton, no glacial features have yet been identified there. Earth, however, hosts a very
unusual type of “glacier” composed of salt (halite), NaCl, that flows by virtue of interaction
with small amounts of water.

11.1.1 Ice within the hydrologic cycle

The fluvial hydrologic cycle begins with water falling on the surface as rain, running down-
hill and picking up sediment, then flowing to an accumulation point where the sediment is
dumped into a delta or moved further by longshore and marine currents. In an analogous
manner, snow that falls on mountaintops or cold regions, where snowfall exceeds melting,
accumulates into a permanent snowfield. If this accumulation were to continue without
limit, the Earth’s water would soon end up frozen into huge ice mountains. However, snow
metamorphoses into ice, which can flow off the land surface as a thick, viscous liquid
and, thus, returns the water to lower elevations, where it melts back into water to close
the cycle.
Different portions of a glacier or ice sheet can, thus, be distinguished by their func-
tion in the overall mass balance as either accumulation areas or ablation areas, which
are connected by flowing ice (Figure 11.1). These regions are readily recognized in a
terrestrial glacier by color and texture (or thermal inertia) at the end of the summer
season: Accumulation areas are bright white, underlain by coarse snow (firn), which is
undergoing the transformation into ice, while ablation areas appear as dense blue ice that
is in the process of melting. Typical glacier flow velocities range from 0.1 to 2 m/day,
exceptionally up to 6 m/day. Large outlet glaciers from ice sheets may flow up to 30 m/
day. Some glaciers are observed to suddenly speed up to rates of 70 m/day in brief epi-
sodes known as surges. The mechanical basis of surges was once mysterious, but we now
know that surges are due to changes in the subglacial plumbing of warm-based glaciers,
discussed below.
The heads of glaciers and ice sheets are in areas where more snow accumulates each
year than melts, and they flow down to elevations where melting exceeds snowfall, often
to altitudes well below the snowline at which accumulation and melting are in balance. A
glacier is, thus, in a state of dynamic balance between accumulation and melting.

EBSCOhost - printed on 11/3/2023 9:59 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
436 Ice

accumulation
(gain) snowline

ablation
(loss)

melt

Figure 11.1 Schematic plan of a valley glacier, illustrating the accumulation zone at high altitude,
flow of glacier ice downslope, and the melting of the ice in the low-altitude ablation zone. Depending
on its temperature, the ice at the base of the glacier may or may not be able to slide over the bed.
Inspired by Figure 1 of Sharp (1960).

Accumulation may occur by means other than snowfall directly onto a glacier or ice
sheet. Avalanches from adjacent highlands or wind-blown snow (especially import-
ant on icecaps where precipitation is low) may deposit snow or even ice directly onto
the accumulation area. Valley glaciers, like rivers, may be fed by tributaries that them-
selves originate in merging ice streams, although the number of links in such networks
is usually small.
Once snow falls on a glacier, it undergoes a regular series of changes as it metamor-
phoses from new snow, to old snow, to firn (density about 550 kg/m3), then finally into gla-
cier ice (density 820 to 840 kg/m3). These processes, discussed in Section 7.2.2, involve the
interaction with seasonal liquid meltwater and vapor-phase transport within the snowpack.
Atmospheric gas bubbles, presently of great importance for measuring the composition of
Earth’s pre-industrial atmosphere, may be trapped in the process and preserved for many
thousands of years.
Ablation, or mass wasting, of the ice is usually by melting. Evaporation is generally
unimportant except for tropical glaciers and in the dry valleys of Antarctica. Tidewater
glaciers and continental ice sheets, however, may lose most of their mass by calving of
icebergs into the sea.

11.1.2 Glacier classification

A common classification of glaciers is based on their morphology. There are three gen-
eral types: Icecaps or ice sheets are continuous sheets of ice. Their flow is centripetal,
from a high-standing center toward their edges. Terrestrial examples are the Greenland and
Antarctic ice sheets or the former Pleistocene ice sheets. A former Martian ice sheet may
have covered large areas in its Southern Highlands. Valley glaciers are ice streams that
have heads in mountainous terrain. They are common in the Earth’s high mountains, such
as the Alps of Europe, Himalayas of Asia, or the northwest coast of North America. The
North American glaciers are noteworthy as most of them head at low elevations – often

EBSCOhost - printed on 11/3/2023 9:59 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
11.1 Ice on planetary surfaces 437

0°C 0°C

Air Temperature Air Temperature

Glacier Glacier
ice ice

Rock Rock

Cold glacier Warm glacier

(a) (b)

Figure 11.2 Thermal regimes of glacier ice. (a) Illustrates a cold-based (or polar) glacier, in which
the temperature is everywhere below the freezing point of water and geothermal heat is conducted
to the surface. (b) Illustrates a warm-based (or temperate) glacier in which the temperature is at
the pressure-controlled melting point of ice. Because of the inverse dependence of the ice’s melt
temperature on pressure, the temperature is slightly colder at the base than at the surface, which
implies that heat is conducted to the glacier bed. After Figure 5 of Sharp (1960).

only 2000 m – where they are nourished by the heavy winter snowfall from the adjacent
Pacific Ocean. Piedmont glaciers are ice sheets formed by the coalescence of several val-
ley glaciers on flat terrain at the base of mountains. An example is the Malaspina Glacier
in Alaska.
A second classification is based on temperature. Cold glaciers (alternatively called polar
glaciers) are easiest to understand. The temperature throughout the ice body of the gla-
cier is below the pressure melting point of ice (Figure 11.2a). The normal gradient of
the internal temperature of the Earth is continued through the ice, from warmer below to
colder near the surface, although the slope is different because the thermal conductivity of
ice is somewhat different from that of rock (Table 4.2). Cold glaciers are frozen to their
beds. Their motion occurs by internal deformation of the ice itself through solid-state creep
(Section 3.4.3). They are among the most slow-moving of terrestrial glaciers and they are
generally ineffective in eroding their beds. Often, where a polar glacier has melted away,
there is little evidence of its former existence.
Warm glaciers (alternatively called temperate glaciers) are at the pressure melting point
of ice throughout their mass. Because of the inverse slope of the melting curve of water ice,
dTm /dz = -0.65 K/km of ice, the temperature actually decreases with increasing depth in
the glacier (Figure 11.2b). This inverse gradient means that thermal conduction moves heat
from the surface downward, toward the glacier bed. At the same time, the normal geother-
mal gradient in the rock below the glacier moves heat upward toward the glacier bed with
a slope of about 30 K/km. This creates a thermal crisis for the glacier, which responds by

EBSCOhost - printed on 11/3/2023 9:59 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
438 Ice

(a) (b)

Figure 11.3 (a) A tongue-shaped flow on Mars located on the eastern wall of Hellas Planitia.
This flow is about 5 km long and 1 km wide. It is likely to be a Martian analogue of terrestrial
rock glaciers. Image PIA09594_fig 1, portion of HiRISE image PSP_002320_1415. NASA/JPL/
University of Arizona. (b) Jungtal rock glacier in the Swiss Alps (image courtesy of Dr. Jan-
Christophotto, 2011).

melting at its base, converting about 1 cm of ice into water each year. Warm-based glaciers
are saturated with liquid water, which is in equilibrium with ice throughout the body of the
glacier. Temperatures remain everywhere at the pressure melting point, but a great deal of
heat is, nevertheless, transferred in such glaciers by the latent heat from the conversion of
solid ice to water, which can flow readily from place to place transferring its latent heat as
it moves and then freezes.
Warm-based glaciers can slide over their beds and, with the aid of rocks and debris fro-
zen into the ice, are very effective at abrading and quarrying out the underlying bedrock.
These glaciers also deform internally: Typically about half their surface velocity is due to
internal deformation and half is due to basal sliding.
Temperature is not, however, always a good classification for the entire glacier, because
the thermal regime can change with position in the glacier. Thus, a glacier’s upper reaches
could be “cold” while its lower parts are “warm.” Moreover, parts of the Antarctic ice sheet
have meltwater near their beds, indicating a warm-based regime, while their surfaces are
cold, well below the pressure melting temperature.

11.1.3 Rock glaciers

Although sometimes not considered “proper” glaciers, rock glaciers are dense mixtures
of rock and ice that, despite being mostly composed of rock, nevertheless show clear evi-
dence of flow, albeit moving much more slowly than the mostly pure ice glaciers familiar
to glacial geologists. Rock glaciers creep along at rates of centimeters to meters per year,
but exhibit the lobate margins, drapery-like ridges, and lateral moraines typical of valley
glaciers. Their margins are typically steep, at or close to the angle of repose. They are
included here because recently discovered Martian glaciers may be rock glaciers, not solid
ice (Figure 11.3).

EBSCOhost - printed on 11/3/2023 9:59 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
11.2 Flow of glaciers 439

Mixtures of ice and rock in glaciers form a continuum, running from nearly pure ice,
through ice carrying small quantities of rock and debris, to rock glaciers, which are mainly
composed of rock debris. Some rock glaciers have ice-rich cores mantled with ice-free rock
debris, while in others the ice merely fills the interstices between boulders. Rock glaciers
have not received much study, partly because of their rarity and partly because of the dif-
ficultly of probing into their interiors: Unlike glaciers, one cannot simply melt boreholes
through them with electric heaters or hot steam.
The detailed mechanism by which rock glaciers deform internally is not well under-
stood, in spite of finite element modeling of their flow (see the review by Whalley and
Azizi, 2003). The slow creep of the rock/ice mixtures must be due to the included ice, but
models of how the heavy burdens of rock debris affect the flow rate and its dependence on
factors such as shear stress are not well developed.

11.2 Flow of glaciers

Many theories of how glaciers flow have been proposed since Agassiz brought the import-
ance of glaciers to the attention of geologists. These historical theories include crevasse
filling and refreezing, regelation within the mass of ice, and many others. The reason for
so many theories was the apparent paradox of a crystalline solid (you can easily see the
crystals in partially melted specimens of glacier ice) that, nevertheless, flows like a fluid.
We now understand that “solid” is a poor description of a crystalline material near its
melting point, because any material can flow, although the motion is perceptible only very
near the melting point. This was first demonstrated for ice by glaciologist J. W. Glen in
1955, although this sort of “creep” had been observed in metals and some rock materials
long before. Individual crystals flow by the generation and movement of a peculiar sort of
crystal defect known as a dislocation. Dislocations and their dynamics were some of the
most important discoveries of twentieth-century materials science.
Figure 11.4 shows how a dislocation (an “edge” dislocation in this figure) can move
across a crystal with minimal distortion of the crystal lattice and yet accommodate a net
shear displacement. This mode of deformation is common to all crystalline substances,
so all can deform plastically and, thus, creep. There is only a minimal threshold stress, so
no material has finite strength at high temperatures or over long time periods. Figure 11.4
shows how a dislocation can “glide” through a crystal lattice. In any real material many
dislocations are present at the same time and, by gliding across one another, they create
kinks in one another that effectively pin the dislocations at the crossing locations. When
this occurs, glide ceases after a few percent of strain. A new step is required to free the
dislocations from their pinning points. That step requires the bulk motion of atoms through
the lattice – diffusion. The process of dislocation glide coupled with diffusive untangling
of dislocations is known as dislocation climb. Because diffusion is a thermally activated
process, so is the rate of creep. The flow of crystalline solids is, thus, strongly temperature-
dependent, with an activation energy similar to that of the bulk diffusion of the solid. This
explains why creep is rapid only at high temperatures.

EBSCOhost - printed on 11/3/2023 9:59 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
440 Ice

defect

Figure 11.4 Motion of a dislocation through a crystal lattice leads to shear deformation of the crystal.
Starting at the left, a line defect is created when the atomic bond of one row of atoms in the upper half
of the crystal is shifted one lattice spacing to the right, creating a line of local disorganization known
as a dislocation. The atoms in the crystal shift partners as the dislocation moves to the right. After the
dislocation finally emerges from the crystal on the right, the upper half of the crystal is shifted one
lattice spacing to the right, accommodating an increment of shear strain. This is the mechanism by
which ice crystals deform in a creeping glacier.

11.2.1 Glen’s law

J. W. Glen (1955) was the first to make careful measurements of the relation between stress
and strain in polycrystalline ice and to apply it to the flow of glaciers. Unlike the flow of
viscous fluids, he found that the strain rate is not a linear function of stress, but depends
upon stress raised to a power n greater than 1. As described in Section 3.4.3, this is the
general behavior of creeping hot solids, materials whose flow is dominated by dislocation
motion. Glen expressed his rheological law in terms of the strain rate:
ε ̇ = Aσ n = B e−Q/RT σ n (11.1)
where Q is the creep activation energy, R is the gas constant, T the temperature in K and σ
is the applied shear stress. Glen found that the power n ranged between 2 and 4, with a pre-
ferred value of 3.2 (the modern value of n is 4). His estimate of the activation energy was
Q = 134 kJ/mol, less than the modern value of 181 kJ/mol listed in Table 3.3. His experi-
ments gave the constant B = 3.5 × 1020 MPa–3.2s–1. Nevertheless, Glen clarified the import-
ant differences between the flow of Newtonian viscous liquids and glacier ice (Forbes was
not completely right) and showed that very cold ice should flow less readily than ice near
its melting point. He also realized that the creep rate changes as the ice recrystallizes during
flow, a factor that is still not fully incorporated into modern creep laws.
The implication of a power-law dependence of strain rate on stress is that as stress
increases the strain rate increases faster than a direct proportion. Thus, doubling the stress
for n = 4 means that the strain rate increases by a factor of 16. Most of the strain, thus,
becomes concentrated in high-stress areas; at the bottom of an ice sheet on an inclined sur-
face, for example. Non-linear rheological laws such as Equation (11.1) generally resist easy
analytic solutions and require numerical methods to get quantitative results. Thus, the creep
law for an infinite sheet of power-law fluid flowing down a surface of constant slope can be
readily obtained, but it is quite remarkable that there is also an analytic solution for the much
less straightforward case of a power-law fluid flowing down a constant slope in a semicircu-
lar channel of constant width (Nye, 1952). We give this more interesting result here, leaving
the much simpler case of an infinite plane sheet as an exercise for the interested reader.

EBSCOhost - printed on 11/3/2023 9:59 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
11.2 Flow of glaciers 441

Velocity normalized for equal discharge

n=1

n=4
n = 10
1 n = 100

0
–1 –0.5 0 0.5 1
Distance across glacier, normalized

Figure 11.5 Flow profiles across different power-law fluids flowing down a trough-shaped channel
with a semicircular cross section. Newtonian fluids with n = 1 attain a parabolic profile, while fluids
with increasingly high n attain more plug-like profiles, with uniform velocity in the center and steep
gradients near their walls.

Let r be the radial distance from the centerline of a power-law fluid flowing through a
trough-like semicircular channel of radius R that slopes downhill at angle α. For this case,
the downstream velocity uz(r) is given by:

2A  R  n+1  r  n+1 
uz (r ) = ( ρ g sin α )n   −    . (11.2)
n +1  2   2 


For flow in a parabolic or elliptical channel the coefficient of the right-hand side changes
slightly, and the scales for vertical and horizontal velocity gradients are different, but the over-
all behavior is similar. Naturally, this equation also applies to Newtonian flow when n = 1.
Plots of this equation in Figure 11.5 show that as n increases the flow becomes more
plug-like. Comparison of this equation to the profiles of actual glacier velocity profiles in a
straight reach of Saskatchewan Glacier shows good agreement with n = 3 and are inconsist-
ent with a Newtonian, n = 1 curve (Meier, 1960). Deformation of vertical boreholes through
a glacier can also be compared against theoretical velocity profiles such as Equation (11.2).
Measurements of this kind show good agreement between the non-Newtonian flow theory
and observation (Paterson, 1999).
For more complicated channel geometries with varying cross sections, bends and
obstructions, numerical solutions to the flow equations must be constructed. A great deal
of progress has been made in the application of finite element methods to prediction of
glacier velocity patterns.
The actual rheology of ice is more complicated than Glen’s law alone would suggest.
Recent investigations indicate that a variety of mechanisms in addition to dislocation

EBSCOhost - printed on 11/3/2023 9:59 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
442 Ice

climb are important for the deformation of ice, especially at low stress. Such processes as
intra-grain diffusion, diffusion through grain boundaries, and grain boundary sliding all
contribute to the deformation of ice under various conditions of stress and temperature.
The interested reader will find an excellent summary of the complex rheology of ice with
planetary applications in the review by Durham and Stern (2001).

11.2.2 The plastic-flow approximation

Figure 11.5 shows that as the exponent n increases, the corresponding flows become more
plug-like. For very large n the result would approximate the flow of a perfectly plastic
material, one that remains rigid until a yield stress Y is exceeded, after which it flows to an
extent determined only by external constraints on the displacement. This observation led to
the idea that a power-law fluid with large n can be approximately represented as a perfectly
plastic material. Application of this idea to ice suggests adoption of an effective yield stress
of about 0.1 MPa for the “strength” of ice.
The plastic model of ice rheology gives a fairly good prediction for the height profile of
an ice sheet as well as the cross section of some glacier tongues, as discussed in Section
5.3.2 for lava flows and below for ice sheets. However, there is a serious internal problem
with this model and it should be used only with some caution and understanding of this
issue. This problem can be seen from dimensional considerations. The fundamental rheo-
logical law (11.1) for ice relates a strain rate to a stress, so the coefficient relating the two
has the dimensions of inverse stress (to the nth power) and inverse time. However, the
plastic law contains no dimension of time, only stress. These two relations, thus, do not
transition smoothly over into one another as n becomes large unless some quantity with
dimension of time is present. That is, one must quote a timescale τ in addition to a yield
stress for this relation to be meaningful. To see this explicitly, rewrite (11.1) as:
n
1 σ 
ε = . (11.3)
τ  Y 

This rewrite is always possible once Y is given. In that case τ is defined implicitly by
(11.1). In the limit n → ∞ this equation approaches the plastic yield condition:
ε → 0 σ < Y
ε → ∞ σ > Y. (11.4)

Equation (11.4) is commonly cited in connection with the plastic approximation (e.g.
Paterson, 1999), but without a timescale,τ, Equation (11.3) is not dimensionally correct.
The creep rate according to (11.1) is never really zero at low stress – it is only “almost as
good as zero” over some timescale that the user considers important.
The widespread neglect of this timescale in the glaciological literature is probably why
there is quite a bit of variation in estimates quoted for the yield stress, although it is usually
given as “about” 0.1 MPa in round numbers. One can make this approximation precise for
Glen’s flow law if τ = 1.4 yr is the relevant timescale over which deformation is considered

EBSCOhost - printed on 11/3/2023 9:59 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
11.2 Flow of glaciers 443

important. This is also the time required for a stress equal to the yield stress to produce a strain
of 100%.

11.2.3 Other ices, other rheologies

Deformation by dislocation climb and other mechanisms is not confined to water ice. Ices
of other substances, such as solid CO2, ammonia, and nitrogen, are all of interest for the
formation of glaciers on other Solar System bodies. Unfortunately, the rheology of such
ices is not as well explored as that of water ice. Table 3.5 lists recent data on the flow law
of solid CO2, but the flow rates of other ices can only be conjectured at the present time. It
is, thus, important that theoretical models of steady-state creep exist and, in many cases,
are successful in predicting the rheological behavior of complex substances. The detailed
examination of creep mechanisms is a large area of research that is beyond the scope of
this book. For more information, the reader is referred to the recent monograph by Karato
(2008). However, a few simple scaling arguments can be stated that allow crude estimates
of the relative creep rates of different substances.
Nearly all rheological creep laws depend upon thermally activated diffusion to permit
slow deformation of crystalline material. This explains the observed exponential tempera-
ture dependence of relations like Glen’s law (11.1). The actual temperature dependence is a
function of the binding energy of atoms in the material, the species diffusing, and whether
the diffusion occurs along grain boundaries or through the body of the crystal. The depend-
ence of strain rate on stress is a function of whether the deformation is by dislocation climb
or by pure diffusion. Dependence of the creep rate on grain size is also a function of the
mechanism: Dislocation climb rates do not depend on grain size, whereas diffusion creep
processes are functions of grain size because it determines the length of the diffusion path.
The individual equations for different creep mechanisms are, thus, functions of many
different variables, not all of which may be known for a new substance. Nevertheless, the
overwhelming influence of diffusion makes it possible to create order of magnitude esti-
mates of creep rates. We can crudely write the strain rate at a given stress as:

ε ̇ = (a bunch of complicated stuff) D0e−Q/RT σ n. (11.5)

The complicated term in the parenthesis is a function of the many variables we have
mentioned (and others that we have not). The second and third terms, combined, consti-
tute the temperature-dependent diffusion coefficient, while the last term is the shear stress,
raised to some exponent ranging between 1 (for pure diffusion creep) and about 5 (for gen-
eric dislocation climb).
We can now form the ratio between the creep rate of an unknown substance and that of
a known material (water ice, for example) and compare the two. It should be clear that the
relative creep rate depends mainly on the relative diffusion coefficients:

ε unknown D0unknown (Q −Q
= ( number of order 1) e ice unknown ) / RT. (11.6)
ε ice D0ice

EBSCOhost - printed on 11/3/2023 9:59 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
444 Ice

This allows us to evaluate the relative effectiveness of different substances as candi-

date glacier materials if we can make some guess about their diffusion coefficients rela-
tive to ice. We may, if we are desperate and in complete ignorance, take one more step
into the world of wild and wooly approximations (a world in which planetary scientists
must dwell all too frequently) and invoke Shewmon’s rule of thumb, which states that
the diffusion coefficient of any substance at its melting point is 10–12 m2/s (Shewmon,
1963). This leads one to expect that, at their melting points, all substances should flow
roughly like glacier ice. At lower temperatures, flow rates are slower and depend upon
the activation energy for whatever form of diffusion is most effective in permitting the
crystal to deform.
Interactions of ices with other substances, especially interstitial fluid, may greatly
enhance the rate of creep. Box 11.1 describes the strange case of salt glaciers on Earth,
in which the presence of small amounts of intergranular water greatly enhances the creep
rate and permits salt to form kilometer-long flows that look superficially like glaciers.
Similar enhancements might occur in water ice at low temperatures when small quantities
of ammonia are present, as occurs in the outer Solar System.

11.2.4 Basal sliding

The sliding of a glacier over its bed is a quintessentially water-ice process. Warm-based
glaciers with melt at their beds are a consequence of water’s nearly unique negative-slope
melting curve. The process of regelation that makes glacial erosion so effective in warm-
based glaciers also depends on this peculiarity of water’s melting curve.
When it can occur, basal sliding is an important contributor to the overall motion of
a glacier or ice sheet. Many estimates suggest that approximately half of the surface
velocity of a warm-based glacier is due to basal sliding. The warm-based Antarctic ice
streams similarly depend on basal slip for their high velocities. In these streams the basal
ice interacts with water-saturated deformable sediments, not rock. The mechanics of this
soft-sediment interaction is complex and not presently well understood. The mechanical
behavior of water-saturated sediments themselves is complicated and their confinement
beneath a moving ice sheet introduces complex feedbacks that are the subject of current
research.
Regelation. Basal sliding over a rigid bed is better understood. When moving ice at
the base of a glacier encounters an obstacle on the bed, such as a rock protuberance or
wedged boulder, the pressure on the upstream side of the obstacle increases, while that
on the downstream side decreases. Because of the negative slope of water ice’s melting
curve, this lowers the melting point on the upstream side. A small amount of ice melts and
the local temperature declines slightly to the pressure melting point. Upstream ice thus
melts, but not instantaneously: The rate of melting is regulated by the rate at which the
latent heat of melting, 334 kJ/kg, is supplied to the compressed ice by conduction from
the adjacent ice and rock. However, as heat is conducted to the upstream ice, the melt-
water flows around the obstacle and freezes behind it at a slightly higher than ambient

EBSCOhost - printed on 11/3/2023 9:59 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
11.2 Flow of glaciers 445

Box 11.1 Salt glaciers and solution creep

Salt on the surface of the Earth would seem to be one of the least likely materials to flow as
a glacier. Measurements of the creep of pure halite (NaCl) show that, although it does creep
more readily than most rocks, it still requires temperatures in the vicinity of 550 °C for it to
creep at a rate comparable to that of glacier ice.
Unlikely as it might seem, glaciers of salt several kilometers long were described from
the dry Zagros Mountains of Iran in 1929. The discoverers did not believe that salt at normal
surface temperatures could flow at rates comparable to glacier ice and supposed instead that
the salt had erupted hot, at temperatures near 300 °C, and that the glaciers are not moving at
present.
However, salt is highly soluble in water and a small amount of rain does fall in this region.
The theoretical possibility that small amounts of water could enhance the flow rate of salt
by the mechanism of pressure solution creep was investigated by Wenkert (1979). Pressure
solution creep occurs when a crystal subject to differential stress preferentially dissolves on
faces under compression and is deposited on faces under extensional stresses. The dissolved
crystal material diffuses much more readily through the solvent than through the body of the
crystal. The shear strain rate is given by:

V0C L DL f
ε = 21 σ (B11.1.1)
kTd 2

where V0 is the volume of the diffusing species, CL its molar solubility in the solvent, DL is its
diffusivity in the solvent, k is the Boltzmann constant, T the absolute temperature, d the grain
size in the solid, and f is the fraction of liquid wetting the solid.
Applied to salt glaciers, this equation predicts a strain rate 108 faster than that expected for
pure halite. This prediction was verified by both direct observations of the creep of salt glaciers
following rare rainfall events and by laboratory measurements of damp halite (Urai et al.,
1986).
Study of the creep of salt has attracted a large amount of attention because of its importance
for proposed nuclear waste storage in salt deposits. Aside from this practical application,
solution creep is expected to greatly enhance the flow of limestone in the Earth. It has also
been proposed as an agent in enhancing the creep of cold water ice in the outer Solar System
through the solution of ice in interstitial ammonia.

temperature, there releasing its latent heat, which is now available to be conducted to the
upstream face.
This process of melting under compression, followed by meltwater flow and freezing
in the adjacent low-pressure zone, all regulated by the conduction of heat from the freez-
ing water to the melting zone, is called regelation. Regelation is easily demonstrated in
the kitchen (or classroom) by hanging a wire loop weighted at both ends over an ice cube
supported at its ends like a beam. The wire very quickly slices through the ice cube and
emerges on the other side, leaving the ice cube apparently intact (actually, it is not quite

EBSCOhost - printed on 11/3/2023 9:59 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
446 Ice

intact – examination under polarized light shows that the ice along the path of the wire has
recrystallized).
Regelation at the base of a glacier is very efficient for small obstacles, through which
heat is rapidly conducted, but inefficient for large obstacles. On the other hand, the glacier
can easily deform around long-wavelength obstructions, but deformation is difficult for
small wavelengths, requiring high strain rates. There is, thus, some intermediate wave-
length that is maximally obstructive – the expectation is that this wavelength accounts for
most of the resistance to basal sliding. Estimates of the size of this most obstructive obs-
tacle indicate that it is about 10 cm.
Glacier Surges. Most glaciers move down their valleys at a sedate speed of a few
meters per day. However, a few glaciers are observed to suddenly accelerate to many tens
of meters per day in rapid advances known as glacier surges. A surging glacier rapidly
lengthens and thins, overrunning forests and roads in its path. Its surface breaks up into a
wilderness of crevasses separating large blocks that topple as the glacier moves, making it
nearly impossible to cross or even remain safely in one spot on the ice for more than short
periods of time. Because they are so difficult and dangerous to study, little was known
about the mechanics of surging glaciers until a heroic effort with massive helicopter sup-
port was mounted during the 1982–1983 surge of the Variegated Glacier in Alaska (Kamb
et al., 1985).
It was discovered that the immediate cause of the Variegated Glacier’s acceleration was
a large increase in the water pressure at its base, which occurred in conjunction with a
rearrangement of the system of subglacial cavities and tunnels that drain the glacier. This
increase of water pressure lifted much of the glacier’s weight off its bed, decreasing basal
friction and greatly enhancing the rate of basal sliding.
Surges are evidently restricted to warm-based glaciers and it may be that, given enough
time, all warm-based glaciers will exhibit surge activity. An interesting question is whether
the warm-based Antarctic ice sheet is also subject to surges and, if it is, what conditions
must be met to cause a surge.

11.3 Glacier morphology

Glaciers are tongue-like masses of ice that flow down valleys, whereas ice sheets are broad
plains of ice that spread centripetally from their high centers. Valley glaciers typically
carry masses of rock debris along their margins, material that has avalanched from the
valley sides onto their surfaces. Where ice streams meet, these lateral moraines merge into
long trains of debris within the body of the compound glacier and are known as medial
moraines.
The terminus of a glacier or ice sheet may remain at the same location for a long period
of time, but this does not mean that the ice is not moving. Instead, the ice is continually
pushing forward while it melts back at the same rate, making the terminus a dynamic loca-
tion that is constantly subject to small oscillations as the balance between flow and melting

EBSCOhost - printed on 11/3/2023 9:59 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
11.3 Glacier morphology 447

shifts slightly. Because new ice is constantly arriving at the terminus, debris frozen into the
ice melts out and gradually builds up into what may become a large heap – the terminal
moraine. Even when the ice front retreats because melting predominates over flow, the
internal movement of the ice is still downward: The flow velocity never reverses.

11.3.1 Flow velocities in glaciers and ice sheets

A widespread misconception supposes that the ice in a glacier is squeezed out by the
weight of the overlying ice, somewhat like toothpaste from a tube that has been acciden-
tally stepped upon. Called “extrusion flow,” this idea is imbedded in many older texts
on glacier flow. Unfortunately, it is not supported by observation: Intensive studies of
glacier deformation in boreholes, starting during the International Geophysical Year in
1957–1958, have uniformly shown that the maximum velocity in a glacier occurs at its
surface. Because of friction on the walls and bed of a glacier, the velocities near the con-
tact between ice and rock are lower than elsewhere. Velocity contours on a transverse
section across a glacier are concentric arcs around the maximum, which is on the surface.
In bends of the ice stream the position of the maximum shifts from the centerline toward
the outside of the bend.
The former beds of vanished glaciers sometimes slope uphill in what was obviously the
downglacier direction and observers wonder how a glacier could have been flowing uphill.
However, careful consideration of the equilibrium of a block of ice with surface slope
αs and basal slope αb show that the shear stress on the base of a glacier of thickness H is
given by ρgH sin αs. That is, the basal shear stress depends only on the surface slope of the
glacier ice and is independent of the basal slope – even of its sign. This concept is not as
paradoxical as it might seem: The beds of many rivers also slope uphill in reaches between
deep pools and shallow riffles, but as long as the surface slope is downhill, the water flows
in the expected direction. Thus, so long as the surface of a glacier slopes downhill, the
shear forces driving it along continue to urge it downhill, even though its bed might have
the opposite slope.
The vertical component of the ice velocity varies systematically along the course of a
glacier. In the accumulation area the vertical velocity is downward. A marker placed on the
surface of the ice is gradually buried by snow, sinking into the glacier as the snow meta-
morphoses into new ice and more snow accumulates on top of it. Once our marker (whether
it be a meteorite that has fallen onto the snow or the body of some unfortunate early moun-
taineer) moves into the ablation region, melting ice gives the vertical velocity an upward
component and markers once frozen into the ice emerge onto the surface.
The longitudinal velocity of an ice stream does not follow any simple rule, responding
instead to variations in the underlying topography and the local thickness of the ice. Where
topographic steps occur the ice may flow particularly fast and the surface becomes very
steep in reaches known as icefalls. Extensive crevasse systems as well as locally high vel-
ocities characterize icefalls.

EBSCOhost - printed on 11/3/2023 9:59 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
448 Ice

The downstream velocity of a wide ice sheet of thickness H and surface slope αs is given
by the expression (derived from Glen’s law):

2 A( ρ g sin α s )n
uz ( y) = ubs +  H n+1 − ( H − y)n+1  (11.7)
n +1  

where y is the height above the glacier bed, ρ the density of the ice and ubs is the velocity
of basal sliding.
As shown in Figure 11.5, Equation (11.7) predicts that as the creep power law n becomes
larger, the flow is more concentrated toward the base of the glacier where the shear stress is
higher. In the limit of very large n the flow approximates that of a perfectly plastic material
and all of the deformation occurs at the bed.

11.3.2 Longitudinal flow regime and crevasses

As the longitudinal velocity of an ice stream varies along the glacier due to variations in ice
thickness and bed slope, the overall strain rate at the surface of the glacier alternates from
compressional to extensional. These strain-rate changes are accompanied by correspond-
ing changes in the longitudinal stress. Stress variations might remain unknown to a visual
observer, except for the fact that extensional stresses open crevasses that are readily seen in
images. The crevasse pattern on the surface of a glacier thus contains clues about the stress
state and flow regime of the ice.
Crevasses are gaping open fissures that cut the brittle upper surface of an ice sheet. Their
depths are limited (unless they become filled with water) because increasing overburden
pressure eventually overcomes the tensile stress and squeezes the crack closed. A crude
means of estimating the maximum depth of crevasses derives from the plastic approxima-
tion to power-law flow. If the maximum stress in a glacier is limited by a plastic yield limit
of 0.1 MPa, then the maximum depth of a crack is reached when the plastic yield stress
Y equals the overburden pressure, divided by 2 (this factor of 2 comes from the relation
between shear stress and the unidirectional stress exerted by the overburden). Thus, the
maximum crevasse depth is roughly:

2Y
hcrevasse ≈ . (11.8)
ρg

Substituting numerical values for ice on Earth, this comes out to about 20 m. Measured
crevasse depths in glaciers seldom exceed 25 to 30 m, so this result is the right order
of magnitude. A more sophisticated approach incorporating the full rheologic equation is
given in Paterson (1999).
Glacier reaches where the flow rate accelerates are in extension, while those in which the
flow decelerates are under compression. Crevasses occur on the surface where the stress is
tensional, that is, where the flow is accelerating. They are rare where the flow is deceler-
ating. Near the snout of a glacier the flow is typically decelerating as the ice is thinned by

EBSCOhost - printed on 11/3/2023 9:59 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
11.3 Glacier morphology 449

=
(a)

constant flow

(b)

extending flow

(c)

compressing flow

Figure 11.6 The pattern of crevasses on the margin of a glacier indicates the state of longitudinal
strain. (a) A glacier flowing in a uniform channel with uniform longitudinal velocity. Friction
against the side walls creates shear stress that tends to open crevasses at 45° to the flow in the
direction of greatest extension (the inset shows how shear stress can be resolved into extensional
and compressional principal stresses). With time, these crevasses rotate down the glacier (dashed
lines). (b) Where the flow velocity increases downglacier (extending flow) the crevasses may extend
all the way across the glacier nearly perpendicular to its flow direction. (c) Where the longitudinal
velocity decreases downglacier (compressing flow) crevasses are suppressed in the center of the
glacier stream and curve upglacier. After Figure 9.8 of Paterson (1999).

ablation, the glacier surface is under compression and the enterprising mountaineer may
ascend the Gesundheitstrasse (German for “healthy route”) onto the glacier surface.
Diagonal crevasses often form along the margins of glaciers, where friction against the
wall creates shear stresses. Resolving the shear into its diagonal components of compres-
sion and extension, one expects the crevasses to form at a 45° angle to the wall of the
glacier, with the acute angle facing upstream. However, with time these 45°crevasses rotate
to become more transverse to the trend of the glacier because of the differential flow of the
ice. Figure 11.6 illustrates the typical crevasse patterns expected for glacier reaches where
the flow is (a) neither compressing nor extending, but wall friction is important, (b) extend-
ing, and (c) compressing. Moreover, if the glacier spreads out over a broad region, as in a
piedmont glacier, crevasses often form perpendicular to the direction of lateral expansion.

11.3.3 Ice-sheet elevation profile

The elevation profile of ice sheets is often well approximated by a parabola, or at least a
simple curve resembling a parabola. The parabolic form is a direct consequence of the

EBSCOhost - printed on 11/3/2023 9:59 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
450 Ice

(a)
net force

1 H
2
ρgH2 shear stress
τb = Y

P = ρgH L

(b)

3000
Elevation, m

2000

1000

0
0 250 500 750
Distance, km

Figure 11.7 The topographic profile of ice sheets can be approximately computed from a plastic-flow
model. (a) Illustrates the balance of forces on a mass of ice in which the pressure of the ice mass to
the left is balanced against the shear stress at its base, resulting in a parabolic relation between the
ice thickness H and the distance to the ice margin L. (b) Comparison between the plastic-flow model
(dashed line) and the profile of the East Antarctic Ice Sheet between Vostok and Mirny (circles). A
better fit, indicated by the solid line, incorporates a uniform accumulation of ice. After Figure 11.4
of Paterson (1999).

plastic approximation to power-law flow. This same model was used in Section 5.3.2 to
argue that the profile of a lava flow consisting of a Bingham fluid should be close to a par-
abola. As shown in Figure 11.7a, the argument proceeds by balancing the total force from
the base of the sheet, Y L, against the pressure driving it outwards, ½ ρgH2. The resulting
equation for the thickness of the sheet H as a function of distance from the edge L along a
line running radially outward from the center of the sheet is:

2Y L
H= . (11.9)
ρg

For example, for a point 750 km from the edge of the ice sheet and a yield stress of 0.1
MPa on the Earth, Equation (11.9) predicts an ice elevation of about 3900 m, not far from
the observed elevations of the East Antarctic Ice Sheet.
The plastic-flow model is only an approximation and better fits can be attained using
the full flow law, coupled with the recognition that temperatures in the upper part of the
Antarctic ice sheet are below 0°C and so the ice there is less fluid. Figure 11.7b shows the
elevations along a profile from Vostok to Mirny Stations along with a parabola (dashed)

EBSCOhost - printed on 11/3/2023 9:59 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
11.4 Glacial landforms 451

and a more exact model (solid line). The parabolic model gives a good first approximation,
but the more accurate treatment improves the fit considerably. The simple model also does
not take changes in mass balance into account, which is important for most ice sheets.
Because of concerns over the effects of global climate change, modeling of the Antarctic
and Greenland icecaps is reaching a high degree of sophistication and incorporates details
well beyond the scope of this book. The interested reader is referred to the recent mono-
graph by Greve and Blatter (2009).

11.4 Glacial landforms

Glaciated landscapes betray themselves to the knowledgeable viewer by a variety of char-
acteristic features. Glaciers and ice sheets are effective agents of both erosion and depos-
ition. The present landscapes in high-latitude regions of the Earth bear many scars of the
recent series of ice ages. Earth has experienced other episodes of glaciation even farther
back in time: during the Permian Era and in the Neoproterozoic. We are aware of these
ancient episodes by the changes they produced in rocks exposed at that time, although
these are certainly not as apparent as the changes dating from 12 000 yr ago.
Features that seem to indicate glacial erosion also occur on Mars. These appear to be
much more ancient than glacial features on the Earth, but thanks to the very slow rate of
surface modification on Mars they remain to betray their origin. Acceptance of widespread
ice sheets on Mars has been slow, but very recent discoveries of relict ice masses near the
equator of Mars point clearly to a former era of extensive ice. In addition, eskers and many
other landforms are consistent with a previous age of ice on Mars.

11.4.1 Glacial erosion

Valley glaciers modify the stream valleys they initially followed, grinding the original
V-shaped cross profiles into a U-shaped trough. Glacial erosion truncates spurs, creates
bowl-like cirques at the head of canyons, and leaves tributary valleys hanging high above
their normal level of junction. The longitudinal form of glacial valleys is converted into a
giant staircase of treads and risers. The treads frequently slope against the general trend of
the valley and, after the glaciers have melted away, trap small lakes called tarn lakes.
The process of glacial erosion proceeds largely by the removal of large blocks from
the bed of the glacier, a process called quarrying or plucking. As the glacier slides over
irregularities in its bed, it may move fast enough that the ice cannot close in behind the
obstacles, leaving open cavities in the lee that fill with meltwater. However, the pressure
in the meltwater is far below the overburden weight of the glacier, except perhaps during
glacier surges. The pressure gradient between the upstream side of the block on which the
ice is impinging and the downstream water pocket is often enough to slide the block out
of the bed and incorporate it into the ice. The blocks may be directly fractured from intact
bedrock by this pressure differential or may be pre-existing joint blocks. In either case,

EBSCOhost - printed on 11/3/2023 9:59 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
452 Ice

once the ice mobilizes these blocks they are incorporated into the basal ice and dragged
downglacier along the bed, contributing to further erosion by abrading the bedrock still in
place.
Abrasion occurs between rock debris already incorporated into the ice and the rock bed
of the glacier. Its importance can be judged from recently deglaciated surfaces, which are
typically striated, smoothed and, in places, even polished by the action of debris moving
along with the ice. Abrasion of solids is a reasonably well-understood process, at least in its
dependence upon the force and velocity of the grinding surfaces, so it was a surprise when
Geoffrey S. Boulton in 1974 discovered an unanticipated aspect of glacial abrasion that
goes far toward explaining some of the details of glacial bedrock erosion.
Boulton (1979) studied glacial erosion by inserting plates of rock directly on the beds
of several glaciers beneath which tunnels had been excavated to collect glacial meltwater
for water supply. He found, as one might expect, that the thicker the glacier, the harder any
rocks frozen into the glacier ice bore down on the bed and the more material was abraded
from the bed. However, when the overburden pressure of the ice exceeded about 2 MPa, the
increase in the normal pressure of rock on rock ceased, because the embedded rocks simply
punched back into the ice instead of transmitting more pressure. When the ice overburden
exceeded 3 MPa, the rocks in the glacier simply stalled against the bed and the ice flowed
around them, bringing erosion to a halt. At greater pressures the basal debris was deposited
beneath the glacier as till.
There is, thus, a limit to how much pressure ice-entrained debris can exert: Abrasion is
possible beneath ice about 300 m thick or less. This limit depends somewhat on glacier
speed, with the pressure at the peak of abrasion ranging from about 1 MPa at speeds of 5
m/yr up to about 3 MPa at speeds of 100 m/yr. Nevertheless, the qualitative limit to erosion
goes far toward explaining the U-shape of glacier valleys.
Consider a glacier initially flowing in a V-shaped fluvial valley. It is deepest at its center,
but if its overall thickness approaches the limit of abrasion, it is relatively ineffective at
eroding its deepest portion, while removing more material from the walls higher up. The
shape then gradually changes from a V to a U as the rate of erosion is equalized across the
valley and the glacier continues to grind deeper into the bedrock (Harbor et al., 1988).
The centers of continental ice sheets easily exceed Boulton’s abrasion limit, so that most
of their work in plucking and grinding their beds is done within a few hundred kilometers
of their margins, where the thickness of the ice is relatively low. This prediction accords
well with the observation that the continental ice sheets eroded most deeply near their
edges, for example excavating the Great Lakes and Finger Lakes beneath the edges of the
Laurentide Ice Sheet of North America. Streamlined, ice-shaped hills, such as roche mou-
tonnée are best developed near the former margins of the great ice sheets.

11.4.2 Glacial deposition

Moraines. Glaciers are unselective agents of transport. They can and do carry everything
from multi-meter blocks of rock to the finest silt. Valley glaciers transport any debris that

EBSCOhost - printed on 11/3/2023 9:59 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
11.4 Glacial landforms 453

happens to fall on their surfaces, an occurrence that is common because of rock avalanches
from their over-steepened walls. Moving ice of all types picks up material from its bed and
carries it along as it moves. When this material in transit reaches the terminus where melt-
ing exceeds the rate of ice motion, this debris is dumped in an unsorted heap. This material
is called glacial till and the landform it creates is called a moraine.
Besides containing a miscellaneous collection of boulders, pebbles, and silt, glacial till
is also compounded of rock flour, an unusual type of sediment unlike that produced by
other processes. Rock flour is finely pulverized but otherwise fresh bedrock. Chemical
weathering of rock flour is minimal because it is produced by grinding of rock upon rock
at low temperatures beneath a glacier. It is composed of grains mostly less than 100 μm in
diameter that are easily suspended in meltwater streams. The bluish, milky color of glacial
streams and lakes is due to heavy loads of this material. When deposited in front of a gla-
cier it is easily picked up by the wind and blown in dense clouds that make the terminus of
a glacier a dirty, gritty place to work. During the ice ages the entire atmosphere of the Earth
was laden with dust from rock flour. It was laid down in thick deposits known as loess in
extensive plains in China and the midwestern United States that are today valued for their
agricultural potential.
Moraines also contain large amounts of sand-sized material that may be mobilized by
the high winds that often accompany glacial climates. During the Earth’s recent ice age a
great sand sheet formed the Sand Hills of Nebraska, created by sand washed out of glacial
meltwater streams. The high winds, lack of vegetation, and abundant sand-sized sediment
in the Polar Regions led to the surprising development of dune fields in this environment.
Given the evidence for former ice sheets on Mars it seems possible that some of the sand-
sized material there has a glacial origin.
Glaciologists distinguish several types of moraine, depending on where they form.
Terminal moraines pile up at the ends of glaciers, becoming large during times when the
ice margin remains at a nearly fixed location. Lateral moraines form at the edge of ice
streams. Ground moraines form when the ice retreats rapidly, leaving a thin, loose deposit
on the surface. Lodgement moraines form beneath the glacier and their till is often strongly
compacted by the weight of the glacier.
While roche moutonnée are streamlined hills eroded from underlying bedrock that range
in size from kilometers to a few meters, drumlins are streamlined depositional forms molded
out of till, whose size range is similar. From morphology alone it is difficult to separate
these two features: Indeed, some roche moutonnée have downglacier tails of streamlined
till and so are hybrid forms. Flowing ice may also produce very elongated hills that grade
into fluted surfaces with alternating hills and troughs aligned in the direction of ice flow.
Most glacially deposited material is reworked to some extent by water, for melting and
runoff are ubiquitous in the vicinity of glacier ice. A large variety of names have been
applied to these deposits depending on the special circumstances of their formation: The
interested reader is referred to the more specialized discussions cited at the end of this
chapter. In this chapter we refer to only some of the most important features for planetary
observations.

EBSCOhost - printed on 11/3/2023 9:59 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
454 Ice

Kettles or kettle holes are small, sometimes circular, pits that form in the wake of retreat-
ing ice. They have occasionally been mistaken for impact craters, although they almost
always lack rims. They form around blocks of ice left stranded by the retreating ice front.
These ice blocks are then partially or completely buried by outwash. After the ice melts, a
pit remains.
Water at the base of a glacier or ice sheet flows as films along the rock–ice interface, fills
pockets and cavities bridged by the moving ice, and eventually collects into streams that
form a subglacial drainage network. Subglacial streams are far more dynamic than those
flowing over a landscape because the ice flows to fill cavities where the pressure is low. Any
tunnel drilled into the ice closes in rapidly until it meets resistance, so the water within a
glacier travels in tunnels that tailor themselves to fit the volume of the flow. Increased water
pressure opens the passage until it is in balance with the ice pressure, while a decreased
head causes the conduit to shrink, always keeping the water pressure in balance with the
pressure of the encasing ice.
Eskers. As water moves beneath the glacier it picks up silt and debris and carries it
along, depositing it when the current slackens. The deposits of such englacial or subglacial
streams are known as eskers. After the ice melts away, eskers stand as branching, sinuous
ridges on the land surface. The material that composes eskers is clearly water-laid, with
the graded bedding, bedforms, and sorting typical of fluvial deposits. The bedding planes
in eskers tend to be anticlinal in cross section, rather than horizontal, as a result of collapse
along their margins as the confining ice melted away. An apparently enigmatic feature of
esker deposits is that these river-like ridges can travel up and over hills, apparently paying
little attention to the slope of the land surface. This is partly because some eskers were
draped over the topography after the ice melted away, but a more important difference
between esker networks and those of open streams is that they flowed in pressurized con-
duits for which local slopes are less important than regional pressure (head) variations.
Possible eskers were first recognized on Viking images of Mars and with increasing
image resolution in subsequent missions the esker interpretation has become increasingly
secure (Figure 11.8). They have been found at both low and high latitudes in the south-
ern hemisphere, but those in the high-elevation Southern Uplands are the most prominent
(Banks et al., 2009).
Eskers lead out from underneath the ice sheet and sometimes can be seen to connect with
water-laid deposits that form an outwash plain in front of the ice margin. Outwash plains
are effectively low deltas or alluvial cones that represent the material deposited by streams
or rivers draining from the ice. They are complex deposits in their own right, with many
characteristics different from those typical of long-term fluvial deposition

11.5 Ice in the ground

Water in the ground must be frozen wherever the mean surface temperature is below freez-
ing. On the present Earth, about 35% of the land surface satisfies this condition. The mean
temperature over the entire surface of Mars is below the freezing point of water. If the

EBSCOhost - printed on 11/3/2023 9:59 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
11.5 Ice in the ground 455

Figure 11.8 Eskers on Mars. It is suggested that these 1 km wide sinuous ridges on the floor of the
Argyre Basin at 55.5° S and 40.2° W are eskers formed by subglacial streams that were deposited as
the ice sheet melted away. The degraded crater in the center is 7 km in diameter. Arrow points north.
Frame is 50 km wide. Viking Orbiter frame 567B33.

temperature remained permanently below freezing, water in the soil would simply stay
frozen and not much of interest would occur. Temperature fluctuations, particularly those
that cycle about the freezing point, are what produce most landforms and lend geomorphic
interest to frozen landscapes.

11.5.1 Permafrost
The strict, but rather pedantic, definition of permafrost is that permafrost is ground that
is below 0°C, whether or not water is present, independent of rock type. This is pedantic
because if no water is present then absolutely nothing of interest happens and there are no
special landforms to talk about. All of the following discussion focuses on what happens
when water is present and on the effects of freezing and thawing water in the soil. Most
of this discussion holds equally well for water and any other substance that undergoes a
liquid–solid transformation during temperature excursions that occur on a planetary sur-
face. Methane on Titan freezes at 91 K, just a few degrees below its average surface tem-
perature of 94 K. However, Titan’s massive atmosphere prevents temperature variations of
even a few degrees from this average, so, at the moment, we do not believe that periglacial
processes are relevant to Titan (unless some other common substance on its surface under-
goes freeze/thaw cycling).
The nature and behavior of permafrost were not well known in the United States until the
later years of World War II. At that time, a Russian-speaking geologist named Siemon W.
Muller was employed by the US Army to read and translate the extensive Russian literature

EBSCOhost - printed on 11/3/2023 9:59 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
456 Ice

0°C T 0°C T

min mean max min mean max

daily summer
frozen fluctuation thaw

frozen
seasonal
permafrost
fluctuation
“zero” “zero”
annual annual
amplitude amplitude

(a) (b)

Figure 11.9 Annual temperature variations below the ground surface. (a) Indicates a temperate
climate in which freezing takes place only in winter and mean temperatures are above freezing. (b)
Illustrates a cold climate in which the mean temperature is below freezing. Thawing takes place only
in summer. Ice is stable below the depth of the summer thaw, down to a maximum depth determined
by warming due to geothermal heat flow. Permafrost encompasses the entire range over which
temperatures remain below 0°C.

on the subject. After the end of the war he published a summary of his gleanings as a book
(Muller, 1947) that formed the basis for our modern understanding of permafrost. Although
very dated, this book can still be read with profit. Muller coined the word “permafrost”
during his research.
Thermal Regime. Figure 11.9a illustrates the subsurface temperature at a location where
seasonal cycles allow some freezing temperatures, but the mean temperature is above freez-
ing, and Figure 11.9b illustrates a location where the mean annual temperature is below
freezing and a permafrost zone is present. Although the top of the frozen zone is subject to
seasonal temperature variations, these become negligible below the level of “zero” annual
amplitude (temperature variations are never actually zero, but at this depth they are so small
that they can be neglected). Below this, the bottom of the permafrost zone is determined
by the planetary geothermal gradient. On Earth the geothermal gradient is about 30 K/km
and the thickness of the permafrost in Siberia ranges from 200 to 400 m at 70° N down
to a few tens of meters (where it is discontinuous) at 50° N. Unfrozen patches within the
permafrost are known as talik. Talik occurs for many reasons even deep within permafrost
zones; under deep lakes and rivers, for example.
Active Zone. The zone from the surface down to the level where the soil thaws annually is
known as the “active zone” for reasons that will shortly be apparent. Seasonal temperatures
still vary noticeably below this zone and, because not all the soil water is frozen at 0°C (or
even at −30°C; see Section 7.3.2), there is still some movement of liquid water even below
the permafrost table. The depth of freezing and amplitude of thermal fluctuations depend
sensitively upon the nature of the surface. Surface covers of insulating material such as
grass, peat, or snow have a large effect on the thermal regime of the ground underneath.

EBSCOhost - printed on 11/3/2023 9:59 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
11.5 Ice in the ground 457

Late summer Early winter

Frozen
Thawed Thawed,
trapped water

permafrost permafrost

Late winter

All
frozen

permafrost

Figure 11.10 Seasonal evolution of the active layer overlying a permafrost terrain. In late summer
this layer is completely thawed, although the permafrost below creates an impermeable layer that
prevents water from draining into the subsurface and so this layer is usually saturated with water. In
early winter the top of the active layer freezes, trapping water in the lower part of the active layer
between the frozen water above and the permafrost below, a circumstance that promotes many kinds
of instability as water pressures rise. By late winter the active layer is entirely frozen. When spring
arrives the layer thaws from the top down.

Because such covers depend upon the details of surface topography and exposure, large
lateral variations in thermal regime are common in permafrost areas.
Where summer temperatures rise above the freezing point of water, a seasonal cycle
of freezing and thawing develops that ranks permafrost terrains among the most unstable
on Earth. Permanently frozen ground is highly impermeable to liquid water: Any water
that reaches the permafrost table quickly freezes and seals any cracks through which it
may have originally entered. The soil overlying permafrost is, thus, commonly saturated
with water when temperatures are above freezing, leading to the concept of an “active
layer” (Figure 11.10). In late summer a warm thermal wave has propagated to its max-
imum annual depth. The soil overlying the permafrost table is thawed and often saturated
with water. It forms a sea of mud that ranges from tens of centimeters deep in the far north
to meters deep farther south. As winter arrives, the upper part of the active layer freezes
over, trapping water in the lower portion of the layer between the impermeable permafrost
table and the similarly impermeable frozen upper soil. On level ground the trapped water
is stable, unless the heavy tread of a caribou or human breaks through to the mud below.
However, on sloping ground the trapped water migrates laterally and high pressures can
build up only a few tens of centimeters below the surface. Should the surface layer rupture
for any reason, near-freezing water flows out in large volumes and quickly freezes in a low
mound or sheet on the surface. Such surface layers of ice are known as “icings.” Muller,

EBSCOhost - printed on 11/3/2023 9:59 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
458 Ice

in his book, delighted in showing pictures of cabins built directly on the ground in perma-
frost regions. Heating the cabin destroys the upper layer of ice, so in the early winter such
cabins filled suddenly with icy water that spilled out the windows before freezing solid.
Eventually, by late winter, all the water in the active layer freezes and this layer becomes
quiescent until melting begins again in the spring.
Solifluction. Soil creep is rapid in the active layer and freeze/thaw processes lower slope
angles quickly. High pore pressure in the active layer during early winter greatly enhances
the probability that thin landslides develop. Creep is caused by the alternate growth and
melting of ice crystals under the surface. Lenses of ice forming above the permafrost table
cause intense frost heaving. When large (tens of meters broad and meters high), these ice
lenses are called frost blisters: They may tilt overlying trees in the boreal forests and prod-
uce what the Russians fondly call “drunken forests.” All of these processes mobilize the
soil, which flows downhill in a process called solifluction or sometimes gelifluction. This
soil motion often organizes into lobe-shaped steps in the surface that range from tens of
centimeters to meters in height.
Pingos. Pingos are small, dome-like hills cored with ice. Internally they possess lenses
of more or less pure ice beneath a layer of soil. Their mechanics of formation resem-
bles that of igneous laccoliths, and they are sometimes called “hydrolaccoliths.” They may
reach heights of a few tens of meters (rarely a hundred meters) and diameters of nearly a
kilometer. They often exhibit gaping radial dilation cracks at their crests from the uplift
and stretching of the overlying sediment as they grew. Pingos are classified as either open-
system types, in which the growing ice lens is fed by water flowing from beneath the
permafrost layer, or closed-system types that develop where a former lake has frozen and
fed water into the near-surface ice lens. Pingos in which the ice lens has melted resem-
ble small volcanoes, with a central collapsed “caldera” surrounded by uplifted sediments.
Pingos have been reported on Mars, but it is extremely difficult to differentiate pingos from
small volcanic cones (“rootless cones” or hornitos) on morphologic characteristics alone
and so these identifications are presently somewhat dubious.
Permafrost on Mars. After many years of conjecture, the presence of permafrost (in
the extended sense of including frozen water) has been confirmed on Mars. Permafrost
should be stable down to depths of several kilometers in Mars’ polar regions, although
it is not expected to be stable over the long term near the equator. Mars Global Surveyor
studies of thermalized neutrons from primary cosmic rays striking the surface revealed
water (more strictly, hydrogen atoms, irrespective of chemical bonding) within a few
tens of centimeters of the surface in 2001. This near-surface ice extends poleward from
about 50° latitude. The thrusters of the Phoenix spacecraft, which landed at 68° N, dir-
ectly excavated an ice table about 5 cm below the surface. Fortuitously, five small clus-
ters of impacts imaged by the HiRISE camera aboard the Mars Reconnaissance Orbiter
also revealed rather pure ice close to the surface at five locations north of about 45°
N, which includes the Viking 2 landing site. Evidently, if the Viking lander had dug
just 10 cm deeper it might have uncovered water ice during its operational period from
1976–1979.

EBSCOhost - printed on 11/3/2023 9:59 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
11.5 Ice in the ground 459

11.5.2 Patterned ground

The repeated thermal cycling of the active layer in permafrost terrain affords surface fea-
tures many opportunities for self-organization. Freezing and thawing in the active layer
leads to a poorly understood kind of slow convective motion that sorts fine-grained silts
from rocks and organizes them into repeating patterns. Early explorers of permafrost ter-
rains on Earth were astonished at the regular patterns of polygonal troughs, sorted stone
circles, stripes, and other forms that develop with such regularity they often appear to be
artificial. The size scale of these features is of the order of a few times the depth of the
active layer, a few to perhaps ten meters. Larger-scale features are exceptional and require
special explanations.
Ice-wedge polygons. The best understood of these features are ice-wedge polygons,
thanks to the efforts of Arthur Lachenbruch, whose study of these ice wedges has become
a classic of geomechanics (Lachenbruch, 1962). The implications of his study extend far
beyond that of ice-wedge polygons themselves. His report should be read by anyone inter-
ested in the application of mechanical thinking to geology.
Ice-wedge polygons form networks that cover vast areas of permafrost terrain. The indi-
vidual polygons may have either high centers bordered by troughs, or low centers sur-
rounded by ridges that are split by troughs. In either case the distance across polygons
is typically a few to tens of meters. The intersections of the troughs may be either close
to 90°, in which cases the polygons are rectangular, or close to 120°, in which case they
approach a hexagonal shape, very much like mudcracks caused by desiccation. A wedge of
ice lies beneath the polygonal troughs in active terrain. These ice wedges may extend 30 m
below the surface and are a few tens of centimeters to a meter wide at their tops.
Lachenbruch’s explanation for how ice wedges form is illustrated in Figure 11.11. During
the coldest part of the winter the active layer is frozen. Cold snaps of many days’ duration
often occur, at which time the frozen ground contracts and strong tensional stresses develop
in a layer between the surface and the depth to which the thermal wave from the cold snap
extends. If the tensile stress becomes great enough a crack opens. This crack may propa-
gate several times deeper than the depth of the cooling from the cold snap itself, penetrating
into the permafrost below the depth where seasonal temperature variations are negligible.
The opening of a crack relieves tensional stresses in its vicinity out to a horizontal distance
comparable to its depth. A second crack is, thus, unlikely to form close to the first: Cracks
tend to be evenly spaced at distances comparable to their depths.
When the cold snap ends, the surface ice expands again, but the crack never quite closes:
Dirt and small stones prop it slightly open for the rest of the winter. Upon arrival of the
spring thaw, water from the active layer trickles down into the crack and freezes there,
forming a thin sheet of ice along the crack surface. During the next winter the cycle repeats,
but the crack is now a weak zone: Pure ice is weaker than frozen soil, so cracking during
subsequent cold snaps occurs preferentially along the first-formed crack.
The next summer a second layer of water flows into the crack and freezes. After hun-
dreds or thousands of seasonal cycles the thin crack grows into a massive wedge of ice. The

EBSCOhost - printed on 11/3/2023 9:59 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
460 Ice

(a) Winter cold snap (b) Late summer

tension frozen active layer “thawed”
permafrost permafrost
zero

(c) Winter again (d) Summer

Many years later

frozen thawed

permafrost permafrost

Figure 11.11 Formation of ice wedges, according to the theory of Art Lachenbruch (1962). (a)
During late winter a cold snap causes the ground to contract, creating enough tensional stress to
open a vertical crack that propagates some distance into the permafrost. The dashed “zero” line
indicates the depth below which annual temperature variations are negligible. (b) During the late
summer thaw, water percolates into the open crack and fills it, freezing at the subzero temperatures
in the permafrost. (c) The next winter another cold snap re-opens the same crack because water ice
is weaker than the surrounding permafrost. (d) After many such cycles of crack-opening and water-
filling a broad wedge of ice has grown in the original crack, slowly enlarging by forcing adjacent
sedimentary layers to deform as it grows.

soil stretches and thins over the opening wedge, which also presents a mechanical and ther-
mal contrast to the rest of the permafrost, being composed of nearly pure ice. Soil adjacent
to the growing wedge is slowly pushed aside by the wedge and heaped up into a flanking
ridge or thrust farther into the center of the polygon. Such soil deformation is frequently
noted in exposed sections of ice wedges.
If the climate changes and the permafrost warms and melts, melting the ice wedges with
it, relicts of the ice wedges still remain. As the ice wedges melt away, soil from the active
layer flows into the vacated wedge and may be recognized long after by the interruption
of the original stratigraphy of the permafrost, distorted layers, and textural differences in
the ice-wedge filling. Such fossil ice wedges are frequently discovered in former perma-
frost terrains and serve to indicate the extent of cold conditions during the Earth’s recent
ice age era.
The size scale of ice-wedge polygons reflects the depth of penetration of the thermal
wave from cold snaps (times a factor of a few to account for the deeper penetration

EBSCOhost - printed on 11/3/2023 9:59 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
11.5 Ice in the ground 461

(a) (b)

Figure 11.12 Ice-wedge polygons on Mars. (a) Patterned ground seen on the ejecta from Lyot Crater
at 54.6° N and 326.6° W. In this case the polygon margins are ridges on which lie large boulders.
This image is 3 km wide and is illuminated from the lower left. Image MOC2–564. NASA/JPL. (b)
Troughs are spaced 1.5 to 2.5 m apart near the Phoenix landing site at 68°N and 26° W. They are
believed to represent ice-wedge activity. On Earth, ice wedges may also be manifested by either
ridges or troughs. NASA/JPL/University of Arizona. See also color plate section.

of the cracks beyond the depth of actual tension). It is, thus, a combined function of
the duration and intensity of a cold snap and the thermal conductivity of the soil. One
might then wonder if more long-continued eras of cold create larger polygons. This is
precisely what many geologists thought when 5–10 km scale polygonal troughs were
discovered on Viking images of the northern plains of Mars. However, the theory of ice-
wedge formation, coupled with the rheology of ice (Glen’s law) show that this cannot
be the case.
Tensile stresses in ice due to contraction endure only so long as creep does not relax
them. Slow freezing that extends to great depths requires a long period of time, during
which the ice has time to flow under the applied stress and zero the stress. If the cold snap
is not quick, it cannot generate tensile stresses and deep cracks cannot develop. This argu-
ment was applied to the polygonal terrain on Mars to show conclusively that such large
polygons cannot have been created by thermal contraction (Pechmann, 1980). They may
instead be due to draping of compacting sediments over a pre-existing cratered terrain
(McGill, 1992). Only recently, with the advent of the very high-resolution imaging possible
with the HiRISE camera system and the Phoenix lander have true ice-wedge polygons been
observed on Mars (Figure 11.12). These polygons have the expected dimensions of 2–3 m
across and closely resemble terrestrial ice-wedge polygons.
The observation of apparently fresh ice-wedge polygons on Mars suggests that liquid
water on the surface is not, in fact, necessary for their formation. Perhaps the crack fillings
on Mars are dust that has drifted into gaping cracks, not frozen water, and we are really
looking at dust-wedge polygons in the Martian permafrost.

EBSCOhost - printed on 11/3/2023 9:59 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
462 Ice

Block fields. Block fields are enigmatic periglacial features that have been found in many
areas formerly occupied by permafrost. They appear as gently sloping, nearly planar sur-
faces up to many kilometers in extent, which are covered with boulders of roughly the same
size. Block fields lack any obvious matrix material, although when active they may have
contained interstitial ice. They may be related to frost heaving in some way as they show
little sign of lateral movement: The blocks in these fields seem to have formed in place
from jointed bedrock.

11.5.3 Thermokarst
Permafrost betrays its presence most clearly when it is about to disappear. The most dra-
matic landforms created by permafrost are formed as it is melting during a period of cli-
matic warming. Permafrost does not melt uniformly: Small variations in surface thermal
conductivity and exposure become amplified by positive feedback and are expressed as
topographic features. The most obvious result of melting permafrost is a volume change.
Thawed permafrost expels water and contracts, sagging downward into small ponds that
collect more water and enhance melting. Such thaw lakes are common, creating landscapes
packed with kilometer-diameter circular to elliptical ponds that are often aligned with the
prevailing wind. Such lakes constitute the infamous Carolina Bays, which impact-crater
enthusiasts persistently claim to be of impact origin in spite of the complete lack of evi-
dence for impacts. Extensive fields of active thaw lakes occur in lowland areas of northern
Alaska, Yukon Territory, and northern Russia. Depressions believed to be thaw lakes have
been recognized in the catastrophic outflow channels on Mars, suggesting at least one era
of warming on that currently chilly planet.
Once depressions are created by melting permafrost, the scarps that form at the interface
between the thawed and still-frozen ground are subject to rapid denudation that removes
the insulating surface layer and accelerates the disintegration of the permafrost. These
small scarps retreat rapidly, forming shallow cirques that cut into the frozen ground.
Asymmetric valleys are common in permafrost terrains and may be accentuated by its
decay. North–south asymmetry develops predominantly because of differences of exposure
to solar radiation. East–west asymmetry may also develop because of differing exposure
to the prevailing wind and the resulting differences in the depth of wind-drifted snow and
chilling by the wind.
The Southern Polar Cap of Mars is subject to another kind of weathering akin to thermo-
karst disintegration, but not involving liquid water. Thin layers (~10 m) of residual CO2
ice overlying water ice sublime away to create coalescing circular pits informally dubbed
“Swiss-cheese terrain.” These pits range from a few hundred meters up to a kilometer
across and are approximately 10 m deep.

Further reading
The classic American study of the geological effects of the Pleistocene continental ice
sheets is the fat book of Flint (1971). The basic physics of glacier flow, mass balance, and

EBSCOhost - printed on 11/3/2023 9:59 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
Exercises 463

ice sheet formation is Paterson (1999), a book that has gone through many editions (a fourth
has just appeared), but remains the most lucid of several such books. Readers seeking to
visually feast on glacial features and phenomena should peruse the picture book by Post
and LaChapelle (2000), which also contains much wisdom in addition to its magnificent
photographs. The ability of glaciers and ice sheets to create landscapes is treated in Sugden
and John (1976), a book that has unfortunately not been updated in recent years. The
mechanics of cold soil and its implications for landform evolution is the topic of Williams
and Smith (1991), while the more observational aspects of periglacial environments is well
covered by Washburn (1980). The geomorphology of both glaciated regions and periglacial
environments is the subject of a pair of books by Embelton and King (1975a, b).

Exercises

11.1 Rheology in space

The Maxwell relaxation time τM for ice at 273K (0°C) is 100 minutes. Use the approxi-
mately universal relation:
ε ̇ = A(σ )e−gTm/T
where g ~ 26 for non-metals. Compute τM for the following substances on the indicated
planetary body (suppose σ is the same as for ice at the τM given, and that all materials have
nearly the same shear modulus):
Methane, ammonia on Triton, surface temperature 45K
Ammonia, CO2 on Ganymede, surface temperature 145K
CO2, water on Mars polar caps, mean temperature ca. 170K
Salt (NaCl), olivine (forsterite) on Venus, surface temperature 750K.
Melting points are:
Methane Tm = 91K
Ammonia Tm = 196K
CO2 Tm = 216K
Salt (NaCl) Tm = 1100K
Olivine (Fo) Tm = 2200K.
Which of the two materials is more likely to show evidence of flow in surface deposits
(“glaciers”)? Compare your computed flow rates to those of a terrestrial glacier.

11.2 The inner heat

Compute the rate of basal melting of a warm-based glacier on the Earth and on Mars (use
some plausible means of estimating the heat flow of Mars, perhaps by assuming that the
rate of heat generation per unit volume on Mars is the same as on the Earth). The average
heat flow on the Earth is 80 mW/m2. Does this suggest a way of estimating the minimum

EBSCOhost - printed on 11/3/2023 9:59 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
464 Ice

rate of snow precipitation necessary to support warm-based glaciers? If so, what is this
minimum for Mars?

11.3 Infinite flowing ice

Derive the velocity profile for an infinitely wide sheet of power-law material (Glen’s law,
Equation (11.1)) of uniform thickness H creeping down a surface with a constant slope α.
That is, show how Equation (11.7) comes about in the case of a zero basal sliding velocity.
Using the data on Glen’s law given in Section 11.2.1, estimate the surface velocity of an
ice sheet 3 km thick with a surface slope of 0.01 (about half a degree) at a temperature of
0°C and at −40°C.

11.4 Only the surface matters!

Demonstrate the assertion in Section 11.3.1 that, in a straight reach of a glacier, the basal
shear stress depends only upon the surface slope αs, not the basal slope αb. Hint: Consider
the forces acting on the left and right sides of a block of downglacier length Δx, then get the
basal shear stress from the force per unit area (if you get really stuck on this, see Patterson,
1999, p. 241).

11.5 Permafrosty Mars

Estimate the thickness of the permafrost layer on Mars by computing the depth of the 273
K isotherm below the surface at the equator and at 50° N. Assume a global average heat
flow of 30 mW/m2 and a thermal conductivity of the Martian surface layers of about 1.5
W/m-K. The average surface temperature of Mars at the equator is 240 K, but falls to a
chilly 200 K at 50° N. Compute the depth to the bottom of the permafrost at these two
latitudes. How does this depth compare to the depth of the (Martian) seasonal temperature
fluctuation?

EBSCOhost - printed on 11/3/2023 9:59 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
References

Aharonson, O. and Schorghofer, N. (2006). Subsurface ice on Mars with rough topography.
All rights reserved. May not be reproduced in any form without permission from the publisher, except fair uses permitted under U.S. or applicable copyright law.

J. Geophys. Res., 111, 2005JE002636.

Ahlbrandt, T. S. (1979). Textural parameters of eolian deposits. In Desert Sand Seas, ed. E.
D. McKee. Washington, DC: US Geological Survey, pp. 21–51.
Allen, J. R. L. (1970). Physical Processes of Sedimentation. London: Allen & Unwin.
(1974). Reaction, relaxation and lag in natural sedimentary systems: General principles,
examples and lessons. Earth Science Reviews, 10, 263–342.
(1982). Sedimentary Structures: Their Character and Physical Basis. Amsterdam:
Elsevier.
Amante, C. and Eakins, B. W. (2009). ETOPO1 1 Arc-Minute Global Relief Model:
Procedures, Data Sources and Analysis. Report.
Amundson, R. (2001). The carbon budget in soils. Annu. Rev. Earth Planet. Sci., 29,
535–62.
Anderson, D. L. (1995). Lithosphere, asthenosphere, and perisphere. Rev. Geophys., 33,
125–49.
Anderson, E. M. (1951). The Dynamics of Faulting. Edinburgh: Oliver and Boyd.
Anderson, R. S. (1996). The attraction of sand dunes. Nature, 379, 24–5.
Araki, H., Tazawa, S., Noda, H., Ishihara, Y., Goossens, S., Sasaki, S., Kawano, N., Kamlya,
I., Otake, H., Oberst, J. and Shum, C. (2009). Lunar global shape and polar topog-
raphy derived from Kaguya-LALT laser altimetry. Science, 323, 897–900.
Arcement, G. J. and Schneider, V. R. (1989). Guide for selecting Manning’s roughness
coefficients for natural channels and flood plains (metric version). Report USGS Water
Supply Paper 2339.
Artyushkov, E. V. (1973). Stresses in the lithosphere caused by crustal thickness inhomo-
geneities. J. Geophys. Res., 78, 7675–708.
Ashby, M. F. and Sammis, C. G. (1990). The damage mechanics of brittle solids in com-
pression. Pure and Appl. Geophysics, 133, 489–521.
Atkinson, B. K. (1987). Fracture mechanics of rock. In Fracture Mechanics of Rock. San
Diego, CA: Academic, p. 534.
Bagenal, F., Dowling, T. E. and McKinnon, W. B. (2004). Jupiter. Cambridge, UK:
Cambridge University Press, p. 719.
Bagnold, R. A. (1941). The Physics of Blown Sand and Desert Dunes. London: Methuen
& Co.
Copyright 2011. Cambridge University Press.

(1966). An approach to the sediment transport problem from general physics. USGS
Prof. Pap., 422I, 1–37.
(1990). Sand, Wind and War. Tucson, AZ: University of Arizona Press.

465

EBSCO Publishing : eBook Academic Collection (EBSCOhost) - printed on 11/3/2023 10:00 AM via YDUQS
AN: 399390 ; H. Jay Melosh.; Planetary Surface Processes
Account: s4526441.main.estacio
466 References

Baker, V. R. (1982). The Channels of Mars. Austin, TX: Texas University Press.
Baker, V. R. and Nummedal, D. (1978). The Channeled Scabland. Washington, DC:
NASA.
Baldwin, R. B. (1963). The Measure of the Moon. Chicago, IL: University of Chicago
Press.
Banerdt, W. B., Golombek, M. P. and Tanaka, K. L. (1992). Stress and tectonics on Mars. In
Mars, ed. H. H. Kieffer, B. M. Jakowsky, C. W. Snyder and M. S. Matthews. Tucson,
AZ: University of Arizona Press, pp. 249–97.
Banks, M. E., Lang, N. P., Kargel, J. S., McEwen, A. S., Baker, V. R., Grant, J. A., Pelletier,
J. D. and Strom, R. G. (2009). An analysis of sinuous ridges in the southern Argyre
Planitia, Mars using HiRISE and CTX images and MOLA data. J. Geophys. Res., 114,
E09003, doi:10.1029/2008JE003244.
Bass, J. D. (1995). Elasticity of minerals, glasses, and melts. In Mineral Physics and
Crystallography: A Handbook of Physical Constants, ed. T. J. Ahrens. Washington,
DC: American Geophysical Union, pp. 45–63.
Bate, R. D., Mueller, D. D. and White, J. E. (1971). Fundamentals of Astrodynamics. New
York, NY: Dover.
Bear, J. (1988). Dynamics of Fluids in Porous Media. New York, NY: Dover.
Beatty, J. K., Petersen, C. C. and Chaikin, A. (1999). The New Solar System, 4th edn.
Cambridge, UK: Cambridge University Press.
Beeman, M., Durham, W. B. and Kirby, S. H. (1988). Friction of ice. J. Geophys. Res., 93,
7625–33.
Ben-Avraham, Z. and Nur, A. (1980). The elevation of volcanoes and their edifice heights
at subduction zones. J. Geophys. Res., 85, 4325–35.
Bergstralh, J. T., Milner, E. D. and Shapley, M. (1991). Neptune. Tucson, AZ: University
of Arizona Press, p. 1076.
Bingham, E. C. and Green, H. (1919). Paint, a plastic material and not a viscous liquid:
The measurement of its mobility and yield value. Am. Soc. Testing Materials Proc.
Pt. II, 641–75.
Bird, E. (2008). Coastal Geomorphology, 2nd edn. New York, NY: Wiley.
Birkland, P. W. (1974). Pedology, Weathering and Geomorphological Research. New York,
NY: Oxford.
Blackwelder, E. (1929). Cavernous rock surfaces of the desert. Am. J. Sci., 17, 393–9.
(1934). Yardangs. Bull. Geol. Soc. Amer., 45, 159–66.
Bland, M. T. and Showman, A. P. (2007). The formation of Ganymede’s grooved terrain:
Numerical modeling of extensional necking instabilities. Icarus, 189, 439–56.
Bland, W. and Rolls, D. (1998). Weathering. London: Arnold.
Bolt, B. A. (1970). Jottings from Australia: Australian seismographic stations, earthquake
engineering and the 1968 Meckering earthquake. Bull. Seis. Soc. Amer., 60, 675–83.
Bottke, W. F., Cellino, A., Paolicchi, P. and Binzel, R. P. (2002). Asteroids III. Tucson, AZ:
University of Arizona Press, p. 785.
Bougher, S. W., Hunten, D. M. and Phillips, C. B. (1997). Venus II. Tucson, AZ: University
of Arizona Press, p. 1362.
Boulton, G. S. (1979). Processes of glacier erosion on different substrata. J. Glaciology,
23, 15–38.
Brace, W. F. (1960). An extension of the Griffith theory of fracture to rocks. J. Geophys.
Res., 65, 3477–80.
Branney, M. J. and Kokelaar, P. (2002). Pyroclastic Density Currents and the Sedimentation
of Ignimbrites. London: Geol. Soc. London.

EBSCOhost - printed on 11/3/2023 10:00 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
References 467

Brown, C. D. and Grimm, R. E. (1997). Tessera deformation and the contemporaneous

thermal state of the plateau highlands, Venus. Earth Planet. Sci. Lett., 147, 1–10.
Brown, R. H. and Matson, D. L. (1987). Thermal effects of insolation propagation into the
regoliths of airless bodies. Icarus, 72, 84–94.
Brown, R. H., Lebreton, J.-P. and Waite, J. H. (2009). Titan from Cassini–Huygens.
Dordrecht: Springer, p. 535.
Bucher, W. H. (1933). Deformation of the Earth’s Crust. Princeton, NJ: Princeton University
Press.
Bull, W. B. (1968). Alluvial fans. J. Geol. Educ., 16, 101–6.
Buratti, B. J., Hillier, J. K. and Wang, M. (1996). The lunar opposition surge: Observations
by Clementine. Icarus, 124, 490–9.
Burchfield, J. D. (1990). Lord Kelvin and the Age of the Earth. Chicago, IL: University of
Chicago Press.
Burleigh, K. J., Melosh, H. J., Tornabene, L. L. and McEwen, A. S. (2009). Small impacts
trigger dust landslides on Mars. In 40th Lunar and Planetary Science Conference.
Abs. #1431. Houston, TX: Lunar and Planetary Institute.
Burr, D. M., Emery, J. P., Lorenz, R. D., Collins, G. C. and Carling, P. A. (2006). Sediment
transport by liquid surficial flow: Application to Titan. Icarus, 181, 235–42.
Burr, D. M., Enga, M.-T., Williams, R. M. E., Zimbelman, J. R., Howard, A. D. and
Brennand, T. A. (2009). Pervasive aqueous paleoflow features in the Aeolis/Zephyria
Plana region, Mars. Icarus, 200, 52–76.
Byerlee, J. (1978). Friction of rocks. Pageoph., 116, 615–26.
Cailleau, B., Walter, T. R., Janle, P. and Hauber, E. (2005). Unveiling the origin of radial
grabens on Alba Patera volcano by finite element modeling. Icarus, 176, 44–56.
Carey, S. W. (1953), The rheid concept in geotectonics. Aust. J. Earth Sci., 1, 67–117.
Caristan, Y. (1982). The transition from high-temperature creep to fracture in Maryland
diabase. J. Geophys. Res., 87, 6781–90.
Carr, M. H. (1996). Water on Mars. New York, NY: Oxford.
(2006). The Surface of Mars. Cambridge, UK: Cambridge University Press.
Carslaw, H. S. and Jaeger, J. C. (1959). Conduction of Heat in Solids, 2nd edn. Oxford:
Oxford.
Carson, M. A. and Kirkby, M. J. (1972). Hillslope Form and Process. Cambridge, UK:
Cambridge University Press.
Cathles, L. M. (1975). The Viscosity of the Earth’s Mantle. Princeton, NJ: Princeton
University Press.
Chandrasekhar, S. (1961). Hydrodynamic and Hydromagnetic Stability. New York, NY:
Oxford.
(1969). Ellipsoidal Figures of Equilibrium. New Haven, CT: Yale University Press.
Chapman, C. R. and McKinnon, W. B. (1986). Cratering of planetary satellites. In Satellites,
ed. J. A. Burns and M. S. Matthews. Tucson, AZ: University of Arizona Press.
Chase, C. (1991). Fluvial landsculpting and the fractal dimension of topography.
Geomorphology, 5, 39–57.
Chase, M. W. (1998). NIST-JANAF Thermochemical Tables, 4th edn. Woodbury, NY:
American Chemical Society.
Chopra, P. and Paterson, M. S. (1984). The role of water in the deformation of dunite.
J. Geophys. Res., 89, 7861–76.
Clark, B. E., Hapke, B., Pieters, C. and Britt, D. (2002). Asteroid space weathering and
regolith evolution. In Asteroids III, ed. W. F. Bottke, A. Cellino, P. Paolicchi and R. P.
Binzel. Tucson, AZ: University of Arizona Press, pp. 585–99.

EBSCOhost - printed on 11/3/2023 10:00 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
468 References

Clark, S. P. (1966). Thermal conductivity. In Handbook of Physical Constants, ed. S. P.

Clark. Denver, CO: Geological Society of America.
Clauser, C. and Huenges, E. (1995). Thermal conductivity of rocks and minerals. In Rock
Physics and Phase Relations, ed. T. J. Ahrens. Washington, DC: American Geophysical
Union, pp. 105–26.
Clow, G. D. and Carr, M. H. (1980). Stability of sulfur slopes on Io. Icarus, 44, 268–279.
Collins, G. S. and Melosh, H. J. (2003). Acoustic fluidization and the extraordinary mobil-
ity of sturzstroms. J. Geophy. Res., 108, doi: 10.1029/2003JB002465.
Collins, G. C., Head, J. W. and Pappalardo, R. T. (1998). The role of extensional instability
in creating Ganymede grooved terrain: Insights from Galileo high-resolution stereo
imaging. Geophys. Res. Lett., 25, 233–236.
Collins, G. S., Melosh, H. J. and Marcus, R. A. (2005). Earth Impact Effects Program:
A Web-based computer program for calculating the regional environmental conse-
quences of a meteoroid impact on Earth. Meteoritics and Planet. Sci., 40, 817–840.
Comer, R. P., Solomon, S. C. and Head, J. W. (1979). Elastic lithosphere thickness on the
Moon from mare tectonic features: A formal inversion. Proc. Lunar Planet. Sci. Conf.
10th, 2441–63.
Croft, S. K. (1980). Cratering flow fields: Implications for the excavation and tran-
sient expansion stages of crater formation. Proc. Lunar Planet. Sci. Conf. 11th,
2347–78.
Cruikshank, D. P. (1995). Neptune and Triton. Tucson, AZ: University of Arizona Press,
p. 1249.
Cruikshank, D. P. and Morbidelli, A. (2008). The Solar System Beyond Neptune. Tucson,
AZ: University of Arizona Press, p. 632.
Culling, W. E. H. (1960). Analytical theory of erosion. J. Geology, 68, 336–44.
(1963). Soil creep and the development of hillside slopes. J. Geology, 71, 127–61.
Currie, J. B., Patnode, H. W. and Trump, R. P. (1962). Development of folds in sedimentary
strata. Geol. Soc. Amer. Bull., 73, 655–74.
Cyr, K. E. and Melosh, H. J. (1993). Tectonic patterns and regional stresses near Venusian
coronae. Icarus, 102, 175–84.
Darwin, C. (1896). The Formation of the Vegetable Mould through the Action of Worms.
New York, NY: Appleton.
Dash, J. G., Rempel, A. W. and Wettlaufer, J. S. (2006). The physics of premelted ice and
its geophysical consequences. Rev. Mod. Phys., 78, 695–741.
Davies, A. G. (2007). Volcanism on Io: A Comparison with Earth. Cambridge, UK:
Cambridge University Press.
Davies, G. F. (1980). Thermal histories of convective earth models and constraints on
radiogenic heat production in the Earth. J. Geophys. Res., 85, 2517–30.
(1982). Ultimate strength of solids and formation of planetary cores. Geophys. Res.
Lett., 9, 1267–70.
(1999). Dynamic Earth: Plates, Plumes and Mantle Convection. Cambridge, UK:
Cambridge.
Davies, M. E., Colvin, T. R., Oberst, J., Zeitler, W., Schuster, P., Neukum, G., McEwen, A.
S., Phillips, C. B., Thomas, P., Veverka, J., Belton, M. J. S. and Schubert, G. (1998).
The control networks of the Galilean satellites and implications for global shape.
Icarus, 135, 372–6.
Davis, G. L. (1969). The Earth in Decay. New York, NY: American Elsevier.
Davison, C. (1888). Note on the movement of scree-material. Quater. J. Geol. Soc. London,
44, 232–8.

EBSCOhost - printed on 11/3/2023 10:00 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
References 469

Dodds, P. S. and Rothman, D. H. (2000). Scaling, universality and geomorphology. Ann.

Rev. Earth Planet. Sci., 28, 571–610.
Dohrenwend, J. C. (1987). Basin and range. In Geomorphic Systems of North America, ed.
W. L. Graf. Boulder, CO: Geological Society of America, pp. 303–42.
Dones, L., Chapman, C. R., McKinnon, W. B., Melosh, H. J., Kirchoff, M. R., Neukum,
G. and Zahnle, K. (2009). Icy satellites of Saturn: Impact cratering and age determin-
ation. In Saturn from Cassini-Huygens, ed. M. K. Dougherty, L. W. Espositto and S.
M. Krimigis. Berlin: Springer, pp. 613–35.
Douglas, I. (1977). Humid Landforms. Cambridge, MA: MIT Press.
Durham, W. B. and Stern, L. A. (2001). Rheological properties of water ice – Applications
to satellites of the outer planets. Annu. Rev. Earth Planet. Sci., 29, 295–330.
Durham, W. B., Kirby, S. H. and Stern, L. A. (1993). Flow of ices in the ammonia–water
system. J. Geophys. Res., 98, 17 667–82.
(1999). The rheology of solid carbon dioxide: New measurements. LPSC XXX, Abstract,
# 2017.
Ehlers, E. G. (1987). The Interpretation of Geological Phase Diagrams. New York, NY:
Dover.
Einstein, A. (1954). The cause of the formation of meanders in the courses of rivers and
of the so-called Baer’s law. In Ideas and Opinions, ed. A. Einstein. New York, NY:
Random House, pp. 249–53.
Eisenberg, D. and Kauzmann, W. (1969). The Structure and Properties of Water. Oxford:
Oxford University Press.
Embelton, C. and King, C. A. M. (1975a). Glacial Geomorphology, 2nd edn. New York,
NY: Wiley.
(1975b). Periglacial Geomorphology, 2nd edn. New York, NY: Wiley.
Engelder, T. (1993). Stress Regimes in the Lithosphere. Princeton University Press.
England, P. C., Molnar, P. and Richter, F. M. (2007). Kelvin, Perry and the age of the Earth.
Am. Scientist, 95, 342–9.
Erismann, T. H. and Abele, G. (2001). Dynamics of Rockslides and Rockfalls. Berlin:
Springer.
Ernst, R. E. and Buchan, K. L. (2003). Recognizing mantle plumes in the geological record.
Annu. Rev. Earth Planet. Sci., 31, 469–523.
Ernst, R. E., Grosfils, E. B. and Mége, D. (2001). Giant dike swarms: Earth, Venus, and
Mars. Annu. Rev. Earth Planet. Sci., 29, 489–534.
Evans, B. and Kohlstedt, D. L. (1995). Rheology of rocks. In Rock Physics and Phase
Relations: A Handbook of Physical Constants, ed. T. J. Ahrens. Washington, DC:
AGU, pp. 148–65.
Fegley, B., Klingelhöfer, G., Lodders, K. and Widemann, T. (1997). Geochemistry of sur-
face-atmosphere interactions on Venus. In Venus II, ed. S. W. Bougher, D. M. Hunten
and R. J. Phillips. Tucson, AZ: University of Arizona Press, pp. 591–636.
Ferrill, D. A., Wyrick, D. Y., Morris, A. P., Sims, D. W. and Franklin, N. M. (2004).
Dilational fault slip and pit chain formation on Mars. GSA Today, 14, 4–12.
Festou, M. C., Keller, H. U. and Weaver, H. A. (2004). Comets II. Tucson, AZ: University
of Arizona Press, p. 780.
Fleitout, L. and Froidevaux, C. (1982). Tectonics and topography for a lithosphere contain-
ing density heterogeneities. Tectonics, 1, 21–56.
Fletcher, R. C. and Hallet, B. (1983). Unstable extension of the lithosphere: A mechanical
model for basin-and-range structure. J. Geophys. Res., 88, 7 457–66.
Flint, R. F. (1971). Glacial and Quaternary Geology. New York, NY: Wiley.

EBSCOhost - printed on 11/3/2023 10:00 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
470 References

Fourier, J. (1955). The Analytical Theory of Heat. New York, NY: Dover.
Francis, P. and Oppenheimer, C. (2003). Volcanoes. New York, NY: Oxford.
Freed, A. M., Melosh, H. J. and Solomon, S. C. (2001). Tectonics of mascon loading:
Resolution of the strike-slip fault paradox. J. Geophys. Res., 106(20), 603–20.
French, B. M. (1977). The Moon Book. New York, NY: Penguin.
Frenkel, J. (1926). Zur Theorie der Elastizitätsgrenze und der Festigkeit kristallinischer
Körper. Z. Physik, 37, 572–609.
Fryberger, S. G. (1979). Dune forms and wind regime. In Desert Sand Seas, ed. E. D.
McKee. Washington, DC: US Geological Survey, pp. 137–69.
Garland, G. D. (1965). The Earth’s Shape and Gravity. Pergamon.
Gault, D. E. (1970). Saturation and equilibrium conditions for impact cratering on the lunar
surface: Criteria and implications. Radio Science, 5, 273–91.
Gault, D. E., Hörz, F., Brownlee, D. E. and Hartung, J. B. (1974). Mixing of the lunar
regolith. In Proceedings of the Fifth Lunar Science Conference. Houston, TX, pp.
2365–86.
Gilbert, G. K. (1880). Report on the Geology of the Henry Mountains. Washington, DC:
Government Printing Office.
(1890). Lake Bonneville. Washington, DC: US Government Printing Office.
(1909). The convexity of hilltops. J. of Geology, 17, 344–50.
(1914). The transportation of debris by running water. Prof. Paper 86, USGS.
(1917). Hydraulic-Mining débris in the Sierra Nevada. Prof. Paper 105, USGS.
Gillmor, C. S. (1971). Coulomb and the Evolution of Physics and Engineering in Eighteenth-
Century France. Princeton, NJ: Princeton University Press.
Gilman, J. J. (1969). Micromechanics of Flow in Solids. New York, NY: McGraw-Hill.
Glasstone, S. and Dolan, P. J. (1977). Effects of Nuclear Weapons. United States Departments
of Defense and Energy.
Glen, J. W. (1955). The creep of polycrystalline ice. Proc. Roy Soc. London A, 228,
519–738.
Goehring, L., Mahadevan, L. and Morris, S. W. (2009). Nonequilibrium scale selection
mechanism for columnar jointing. Proc. Nat. Acad. Sci., 106, 387–92.
Gold, T. (1970). Apollo 11 and 12 close-up photography. Icarus, 12, 360–75.
Goldreich, P. and Soter, S. (1966). Q in the Solar System. Icarus, 5, 375–89.
Golombek, M. P. (1979). Structural analysis of lunar grabens and the shallow crustal struc-
ture of the Moon. J. Geophys. Res., 84, 4657–66.
Gooding, J. L., Arvidson, R. E. and Zolotov, M. Y. (1992). Physical and chemical weather-
ing. In Mars, ed. H. H. Kieffer, B. M. Jakosky, C. W. Snyder and M. S. Matthews.
Tucson, AZ: University of Arizona Press, pp. 626–51.
Gordon, J. E. (2006). The New Science of Strong Materials, 2nd edn. Princeton, NJ:
Princeton University Press.
Greeley, R. and Batson, M. (2000). Planetary Mapping. Cambridge, UK: Cambridge
University Press.
Greeley, R. and Iversen, J. D. (1985). Wind as a Geological Process. Cambridge, UK:
Cambridge University Press.
Greeley, R., Lancaster, N., Lee, S. and Thomas, P. (1992). Martian aeolian processes, sedi-
ments and features. In Mars, ed. H. H. Kieffer, B. M. Jakowsky, C. W. Snyder and M.
S. Matthews. Tucson, AZ: University of Arizona Press, pp. 730–66.
Greeley, R., Bender, K. C., Saunders, R. S., Schubert, G. and Weitz, C. M. (1997). Aeolian
processes and features on Venus. In Venus II, ed. S. W. Bougher, D. M. Hunten and R.
J. Phillips. Tucson, AZ: University of Arizona Press, pp. 547–89.

EBSCOhost - printed on 11/3/2023 10:00 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
References 471

Greeley, R., Arvidson, R. E., Barlett, P. W., Blaney, D., Cabrol, N. A., Christensen, P. R.,
Fergason, R. L., Golombeck, M. P., Landis, G. A., Lemmon, M. T., McLennan, S. M.,
Maki, J. N., Michaels, T., Moersch, J. E., Neakrase, D. V., Rafkin, S. C. R., Richter,
L., Squyres, S. W., de Sousz, P. A., Sullivan, R. J., Thompson, S. D. and Whelley, P.
L. (2006). Gusev crater: Wind-related features and processes observed by the Mars
Exploration Rover Spirit. J. Geophys. Res., 111, E02S09.
Greenberg, R., Geissler, P., Hoppa, G., Tufts, B. R., Durda, D. D., Papplardo, R., Head, J.
W., Greeley, R., Sullivan, R. and Carr, M. H. (1998). Tectonic processes on Europa:
Tidal stresses, mechanical response, and visible features. Icarus, 135, 64–78.
Greene, M. T. (1982). Geology in the Nineteenth Century. Ithaca, NY: Cornell University
Press.
Gregg, T. K. P. and Fink, J. H. (1999). Quantification of extraterrestrial lava flow effusion
rates through laboratory simulations. J. Geophys. Res., 101, 16 891–900.
Greve, R. and Blatter, H. (2009). Dynamics of Ice Sheets and Glaciers. Berlin: Springer.
Grieve, R. and Therriault, A. (2000). Vredefort, Sudbury, Chicxulub: Three of a kind? Ann.
Rev. Earth Planet. Sci., 28, 305–38.
Griggs, D. T. (1935). The strain ellipsoid as a theory of rupture. Am. J. Science, 30, 121–37.
(1936). The factor of fatigue in rock exfoliation. J. Geology, 44, 783–96.
(1939). Creep of rocks. Jour. Geol., 47, 225–51.
Griggs, D. T. and Handin, J. (1960). Observations on fracture and a hypothesis of earth-
quakes. In Rock Deformation, Geol. Soc. Amer. Memoir 79, ed. D. T. Griggs and J.
Handin. Boulder, CO: Geological Society of America, pp. 347–64.
Griggs, R. F. (1922). The Valley of Ten Thousand Smokes. Washington, DC: National
Geographic Society.
Grotzinger, J., Jordan, T. H., Press, F. and Siever, R. (2006a). Understanding Earth, 5th
edn. San Francisco, CA: W. H. Freeman.
Grotzinger, J., Bell, J. F., Herkenhoff, K., Johnson, J., Knoll, A., McCartney, E., McLennan,
S. M., Metz, J., Moore, J. M., Squyres, S. W., Sullivan, R., Ahronson, O., Arvidson,
R. E., Joliff, B. L., Golombeck, M. P., Lewis, K., Parker, T. and Soderblom, L. A.
(2006b). Sedimentary textures formed by aqueous processes, Erebus crater, Meridiani
Planum, Mars. Geology, 34, 1085–8.
Groussin, O., A’Hearn, M. A., Li, J.-Y., Thomas, P. C., Sunshine, J., Lisse, C. M., Meech,
K. J., Farnham, T. L., Feaga, L. M. and Delamere, W. A. (2007). Surface temperature
of the nucleus of Comet 9P/Tempel 1. Icarus, 191, 63–72.
Haberle, R. M., Forget, F., Colaprete, A., Schaffer, J., Boynton, W. V., Kelly, N. J. and
Chamberlain, M. A. (2008). The effect of ground ice on the Martian seasonal CO2
cycle. Planetary & Space Sci., 56, 251–5.
Handin, J. (1966). Strength and ductility. In Handbook of Physical Constants, ed. S. P.
Clark, Jr. New York, NY: Geological Society of America, pp. 223–89.
Hapke, B. (1993). Theory of Reflectance and Emittance Spectroscopy. Cambridge, UK:
Cambridge University Press.
Harbor, J. M., Hallet, B. and Raymond, C. F. (1988). A numerical model of landform devel-
opment by glacial erosion. Nature, 333, 347–9.
Hartmann, W. K. (1972). Moons and Planets. Belmont, CA: Wadsworth Pub. Co.
Hartmann, W. K. (2005). Martian cratering 8: Isochron refinement and the chronology of
Mars. Icarus, 174, 294–320.
Hauber, E., Broz, P. and Jagert, F. (2010). Plains volcanism on Mars: Ages and rheology
of lavas. In 41st Lunar and Planetary Science Conference. Abstract #1298. Houston,
TX: Lunar and Planetary Institute.

EBSCOhost - printed on 11/3/2023 10:00 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
472 References

Haxby, W. F. and Turcotte, D. L. (1976). Stresses induced by the addition or removal of

overburden and associated thermal effects. Geology, 4, 181–4.
Heiken, G. H., Vaniman, D. T. and French, B. M. (1991). Lunar Sourcebook. Cambridge,
UK: Cambridge University Press.
Heim, A. (1882). Der Bergsturz von Elm. Zeit. der Deutschen Gesellschaft, 34, 74–115.
Heiskanen, W. A. and Vening Meinesz, F. A. (1958). The Earth and its Gravity Field. New
York, NY: McGraw-Hill.
Helfenstein, P. and Parmentier, E. M. (1983). Patterns of fracture and tidal stresses on
Europa. Icarus, 53, 415–30.
Hertz, H. (1884) Ueber das Gleichgewicht schwimmender elasticher platten.Wiedemann’s
Annalen der Physik und Chemie, 22, 449–55.
Hiesinger, H., Head, J. W. and Neukum, G. (2007). Young lava flows on the eastern flank of
Ascraeus Mons: Rheological properties derived from High Resolution Stereo Camera
(HRSC) images and Mars Orbiter Laser Altimeter (MOLA) data. J. Geophys. Res.,
112, E05011 10.1029/2006JE002717,2007.
Hill, R. (1950). The Mathematical Theory of Plasticity. Oxford.
Hobbs, B. E., Mühlhaus, H. B. and Ord, A. (1990). Instability, softening and localiza-
tion of deformation. In Deformation Mechanisms, Rheology and Tectonics, Special
Publication 54, ed. R. J. Knipe and E. H. Rutter. London: Geological Society, pp.
143–65.
Hon, K., Kauahikaua, J., Denlinger, R. and Mackay, K. (1994). Emplacement and infla-
tion of pahoehoe sheet flows: Observations and measurements of active lava flows on
Kilauea Volcano, Hawaii. Geol. Soc. Amer. Bull., 106, 351–70.
Hoppa, G., Tufts, B. R., Greenberg, R. and Geissler, P. (1999). Strike-slip faults on Europa:
Global shear patterns driven by tidal stress. Icarus, 141, 287–98.
Horstman, K. C. and Melosh, H. J. (1989). Drainage pits in cohesionless materials:
Implications for the surface of Phobos. J. Geophys. Res., 94, 12 433–41.
Horton, R. E. (1945). Erosional development of streams and their drainage basins. Geol.
Soc. Amer. Bull., 56, 275–370.
Hörz, F., Cintala, M. J., Rochelle, W. C. and Kirk, B. (1999). Collisionally processed rocks
on Mars. Science, 285, 2105–7.
Houghton, J. T. (2001). The Physics of Atmospheres, 3rd edn. Cambridge, UK: Cambridge.
Howard, A. D. (2009). How to make a meandering river. Proc. Nat. Acad. Sci., 106,
17245–6.
Howard, A. D., Kochel, R. C. and Holt, H. E. (1988). Sapping Features of the Colorado
Plateau. Washington, DC: NASA.
Howard, K. A. and Larsen, B. R. (1972). Lineaments that are artifacts of lighting. In Apollo
15 Preliminary Science Report: NASA SP-289, pp. 25–58 to 25–62.
Hubbert, M. K. (1940). The theory of ground water motion. J. Geology, 48, 785–944.
Hubbert, M. K. and Rubey, W. W. (1959). Role of fluid pressure in overthrust faulting I.
Mechanics of fluid-filled porous solids and its application to overthrust faulting. Geol.
Soc. Amer. Bull., 70, 115–66.
Hull, D. and Bacon, D. J. (2001). Introduction to Dislocations, 4th edn. Oxford, UK:
Butterworth-Heinemann.
Hurford, T. A., Beyer, R. A., Schmidt, B., Preblich, B., Sarid, A. R. and Greenbert, R.
(2005). Flexure of Europa’s lithosphere due to ridge-loading. Icarus, 177, 380–96.
Iess, L., Rappaport, N. J., Jacobson, R. A., Racioppa, P., Stevenson, D. J., Tortora, P.,
Armstrong, J. W. and Asmar, S. W. (2010). Gravity field, shape and moment of inertia
of Titan. Nature, 327, 1367–9.

EBSCOhost - printed on 11/3/2023 10:00 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
References 473

Ito, E. and Takahashi, E. (1987). Melting of peridotite at uppermost lower-mantle condi-

tions. Nature, 328, 514–17.
Ivanov, B. A. and Melosh, H. J. (2003). Impacts do not create volcanic eruptions. Geology,
31, 869–72.
Ivanov, B. A., Nemchinov, I. V., Svetsov, V. A., Provalov, A. A., Khazins, V. M. and Phillips,
R. J. (1992). Impact cratering on Venus: Physical and mechanical models. J. Geophys.
Res., 97, 16 167–81.
Jaeger, J. C., Cook, N. G. W. and Zimmerman, R. W. (2007). Fundamentals of Rock
Mechanics. London: Blackwell.
Janes, D. M. and Melosh, H. J. (1990). Tectonics of planetary loading: A general model and
results. J. Geophys. Res., 95, 21 345–55.
Jardetzky, W. S. (1958). Theories of Figures of Celestial Bodies. New York, NY:
Interscience.
Jaumann, R., Kirk, R. L., Lorenz, R. D., Lopes, R., Stofan, E. R., Turtle, E. P., Keller, H.
U., Wood, C. A., Sotin, C., Soderblom, L. A. and Tomasko, M. G. (2009). Geology
and surface processes on Titan. In Titan from Cassini-Huygens, ed. R. H. Brown, J.-P.
Lebreton and J. H. Waite. Dordrecht: Springer, pp. 75–140.
Jeffreys, H. (1950). Earthquakes and Mountains. London: Methuen.
(1952). The Earth, 3rd edn. Cambridge, UK: Cambridge University Press.
(1962). The Earth, 4th edn. Cambridge, UK: Cambridge University Press.
(1976). The Earth: Its Origin, History and Physical Constitution, 6th edn. Cambridge
University Press.
Johnson, A. M. (1970). Physical Processes in Geology. San Francisco, CA: Freeman,
Cooper.
Johnson, A. M. and Fletcher, R. C. (1994). Folding of Viscous Layers. New York, NY:
Columbia University Press.
Johnson, R. E., Carlson, R. W., Cooper, J. F., Paranicas, C., Moore, M. H. and Wong,
M. C. (2004). Radiation effects on the surfaces of the Galilean satellites. In Jupiter:
The Planet, Satellites and Magnetosphere, ed. F. Bagenal, T. E. Dowling and W. B.
McKinnon. Cambridge, UK: Cambridge University Press, pp. 485–512.
Kamb, B., Raymond, C. F., Harrison, W. D., Engelhardt, H., Echelmeyer, K. A., Humphrey,
N., Brugman, M. M. and Pfeffer, T. (1985). Glacier surge mechanism: 1982–1983
surge of Variegated Glacier, Alaska. Science, 227, 469–79.
Karato, S.-I. (1998). A dislocation model of seismic wave attenuation and micro-creep in
the Earth: Harold Jeffreys and the rheology of the solid Earth. Pure Appl. Geophys.,
153, 239–56.
(2008). Deformation of Earth Materials: An Introduction to the Rheology of Solid Earth.
Cambridge, UK: Cambridge University Press.
Kargon, R. and Achinstein, P. (1987). Kelvin’s Baltimore Lectures and Modern Theoretical
Physics. Cambridge, MA: MIT, p. 547.
Kasting, J. F. and Catling, D. (2003). Evolution of a habitable planet. Annu. Rev. Earth
Planet. Sci., 41, 429–63.
Kasting, J. F. and Toon, O. B. (1989). Climate evolution on the terrestrial planets. In Origin
and Evolution of Planetary and Satellite Atmospheres, ed. S. K. Atreya, J. B. Pollack
and M. W. Matthews. Tucson, AZ: University of Arizona Press, pp. 423–49.
Kaula, W. (1968). An Introduction to Planetary Physics. New York, NY: Wiley.
Keller, W. D. (1957). The Principles of Chemical Weathering. Columbia, MO: Lucas
Brothers.
Kennedy, B. A. (2006). Inventing the Earth. London: Blackwell.

EBSCOhost - printed on 11/3/2023 10:00 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
474 References

Kinsman, B. (1965). Wind Waves. New York, NY: Dover.

Kirby, S. H. and Kronenberg, A. K. (1987a). Rheology of the lithosphere: selected topics.
Rev. Geophys., 25, 1219–44.
(1987b). Correction to “Rheology of the lithosphere: selected topics.” Rev. Geophys.,
25, 1680–1.
Kirchner, J. W. (1993). Statistical inevitability of Horton’s laws and the apparent random-
ness of stream channel networks. Geology, 21, 591–4.
Knopoff, L. (1964). Q. Rev. Geophys., 2, 625–60.
Kok, J. F. (2010). Difference in the wind speeds required for initiation versus continuation
of sand transport on Mars: Implications for dunes and dust storms. Phys. Rev. Lett.,
104, 074502.
Kok, J. F. and Renno, N. O. (2009a). A comprehensive numerical model of steady state
saltation (COMSALT). J. Geophys. Res., 114, D17204.
(2009b). Electrification of wind-blown sand on Mars and its implications for atmos-
pheric chemistry. Geophys. Res. Lett., 36, L05202.
Komar, P. D. (1997). Beach Processes and Sedimentation, 2nd edn. Englewood Cliffs, NJ:
Prentice-Hall.
Kreslavsky, M. A. and Head, J. W. (2000). Kilometer-scale roughness of Mars. Results
from MOLA data analysis. J. Geophys. Res., 105, 26 695–711.
Lachenbruch, A. H. (1962). Mechanics of Thermal Contraction Cracks and Ice-Wedge
Polygons in Permafrost. New York, NY: Geological Society of America
Lambe, T. W. and Whitman, R. V. (1979). Soil Mechanics, SI Version. New York, NY: John
Wiley & Sons.
Lambeck, K. (1988). Geophysical Geodesy. Oxford: Oxford University Press.
Lancaster, N. (1995). Geomorphology of Desert Dunes. London: Routledge.
Landau, L. D. and Lifshitz, E. M. (1970). Theory of Elasticity, 2nd edn. Oxford: Pergamon.
Langseth, M. G., Keihm, S. J. and Chute, J. L. (1973). Heat flow experiment. In Apollo 17
Preliminary Science Report, ed. L. B. J. Space Center. Washington DC: NASA, pp.
9–1 to 9–24.
Lawn, B. R. and Wilshaw, T. R. (1975). Fracture of Brittle Solids, 2nd edn. New York, NY:
Cambridge University Press.
(1993). Fracture of Brittle Solids. Cambridge, UK: Cambridge University Press.
Lee, P. (1996). Dust levitation on asteroids. Icarus, 124, 181–94.
Leeder, M. (1999). Sedimentology and Sedimentary Basins. Oxford: Blackwell.
Leith, A. C. and McKinnon, W. B. (1991). Terrace width variations in complex Mercurian
craters, and the transient strength of the cratered Mercurian and lunar crust. J. Geophys.
Res., 96, 20923–31.
Leith, C. K. (1923). Structural Geology. New York, NY: Henry Holt & Co.
Leopold, L. B., Wolman, M. G. and Miller, J. P. (1964). Fluvial Processes in Geomorphology.
San Francisco, CA: Freeman.
Lewis, J. S. (1995). Physics and Chemistry of the Solar System, revised edn. San Diego,
CA: Academic Press.
Li, S.-I., Miller, N., Lin, D. N. C. and Fortney, J. J. (2010). WASP-12b as a prolate, inflated
and disrupting planet from tidal dissipation. Nature, 463, 1054–6.
Lichtner, P. C., Steefel, C. I. and Oelkers, E. H. (1996). Reactive transport in porous media.
In Reviews in Mineralogy, ed. P. H. Ribbe. Washington, DC: Mineralogical Society
of America.
Lilljequist, R. and Henkel, H. (1996). The Uppland structure: A suspected 300-km-diame-
ter impact crater in Sweden. Meteoritics and Planet. Sci., 31, A59.

EBSCOhost - printed on 11/3/2023 10:00 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
References 475

Lockner, D. A. (1995). Rock failure. In Rock Physics and Phase Relations, ed. T. J. Ahrens.
Washington, DC: Am. Geophys. Union, pp. 127–47.
Lowell, J. D. (1972). Spitsbergen tertiary orogenic belt and the Spitsbergen fracture zone.
Geol. Soc. Amer. Bull., 83, 3091–102.
Lundborg, N. (1968). Strength of rocks and rock-like materials. Int. J. Rock Mech. and
Mining Sci., 5, 427–54.
Mabbutt, J. A. (1977). Desert Landforms. Cambridge, MA: MIT Press.
Mackin, J. H. (1948). Concept of the graded river. Bull. Geol. Soc. Amer., 59, 463–512.
Mackwell, S. J., Zimmerman, M. E. and Kohlstedt, D. S. (1998). High-temperature
deformation of dry diabase with application to tectonics on Venus. J. Geophys. Res.,
103, 975–84.
Malin, M. C., Edgett, K. S., Posiolova, L. V., McColley, S. M. and Noe Dobrea, E. Z.
(2006). Present-day impact cratering rate and contemporary gully activity on Mars.
Science, 314, 1573–7.
Mandl, G. (1988). Mechanics of Tectonic Faulting. Amsterdam: Elsevier.
Mark, K. (1987). Meteorite Craters. Tucson, AZ: University of Arizona Press.
Matson, D. L. and Brown, R. H. (1989). Solid state greenhouses and their implications for
icy satellites. Icarus, 77, 67–81.
Matsuyama, I. and Nimmo, F. (2008). Tectonic patterns on reoriented and despun planetary
bodies. Icarus, 195, 459–73.
Maxwell, D. E. (1977). A simple model of cratering, ejection, and the overturned flap. In
Impact and Explosion Cratering, ed. D. J. Roddy, R. O. Pepin and R. B. Merrill. New
York, NY: Pergamon, pp. 1003–8.
Maxwell, J. C. (1867). On the dynamical theory of gases. Philosophical Transactions of the
Royal Society of London, 157, 49–88.
McBirney, A. R. and Murase, T. (1984). Rheological properties of magmas. Ann. Rev.
Earth Planet. Sci., 12, 337–57.
McGarr, A. and Gay, N. C. (1978). State of stress in the Earth’s crust. Ann. Rev. Earth
Planet. Sci., 6, 405–36.
McGee, W. J. (1908). Outlines of hydrology. Geol. Soc. Amer. Bull., 19, 193–200.
McGetchin, T. R. and Head, J. W. (1973). Lunar cinder cones. Science, 180, 68–71.
McGill, G. E. (1971). Attitude of fractures bounding straight and arcuate lunar rilles.
Icarus, 14, 53–8.
(1992). Origin of Martian polygons. J. Geophys. Res., 97, 2633–47.
McGovern, P. J., Smith, J. R., Morgan, J. K. and Bulmer, M. H. (2004). Olympus Mons aure-
ole deposits: New evidence for a flank failure origin. J. Geophys. Res., 109, E08008.
McGreevy, J. P. (1981). Some perspectives on frost shattering. Prog. Phys. Geog., 5, 56–75.
McKee, E. D. (1979). Desert Sand Seas. Washington, DC: Prof. Paper 1052, US Geological
Survey.
McKenzie, D. and Bickle, M. J. (1988). The volume and composition of melt generated by
extension of the lithosphere. J. Petrology, 29, 625–79.
McKinnon, W. B., Schenk, P. M. and Dombard, A. J. (2001). Chaos on Io: A model for for-
mation of mountain blocks by crustal heating, melting and tilting. Geology, 29, 103–6.
McPherson, G. J., Kita, N. T., Ushikubo, T., Bullock, E. S. and Davis, A. M. (2010). High-
precision 26Al/27Al isochron microchronology of the earliest Solar System. In 41st
LPSC, Abstract #2356, Houston, TX.
McSween, H. Y., Richardson, S. M. and Uhle, M. E. (2003). Geochemistry: Pathways and
Processes, 2nd edn. New York, NY: Columbia University Press.
Meier, M. F. (1960). Mode of flow of Saskatchewan Glacier. Alberta, Canada. Report.

EBSCOhost - printed on 11/3/2023 10:00 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
476 References

Meinzer, O. E. (1942). Hydrology. New York, NY: Dover.

Mellon, M. T., Fergason, R. L. and Putzig, N. E. (2008). The thermal inertia of the surface
of Mars. In The Martian Surface: Composition, Mineralogy and Physical Properties,
ed. J. F. Bell. Cambridge: Cambridge, UK, pp. 399–427.
Melosh, H. J. (1977). Crater modification by gravity: A mechanical analysis of slumping.
In Impact and Explosion Cratering, ed. D. J. Roddy, R. O. Pepin and R. B. Merrill.
New York, NY: Pergamon Press, pp. 1245–60.
(1977). Global tectonics of a despun planet. Icarus, 31, 221–43.
(1978). The tectonics of mascon loading. In Proc. 9th Lunar Science Conf., pp.
3513–25.
(1979). Acoustic fluidization: A new geologic process? J. Geophys. Res., 84, 7513–20.
(1980a). Tectonic patterns on a reoriented planet: Mars. Icarus, 44, 745–51.
(1980b). Tectonic patterns on a tidally distorted planet. Icarus, 43, 334–7.
(1983). Acoustic Fluidization. Am. Sci., 71, 158–65.
(1989). Impact Cratering: A Geologic Process. New York, NY: Oxford University Press.
(1990a). Giant impacts and the thermal state of the early Earth. In Origin of the Earth, ed.
J. H. Jones and H. E. Newsom. New York, NY: Oxford University Press, pp. 69–83.
(1990b). Mechanical basis for low angle normal faulting in the Basin and Range prov-
ince. Nature, 343, 331–5.
Melosh, H. J. and Dzurisin, D. (1978). Mercurian tectonics: A consequence of tidal despin-
ning? Icarus, 35, 227–36.
Melosh, H. J. and Ivanov, B. A. (1999). Impact crater collapse. Ann. Rev. Earth Planet.
Sci., 27, 385–415.
Melosh, H. J. and Janes, D. M. (1989). Ice volcanism on Ariel. Science, 245, 195–6.
Melosh, H. J. and Williams, C. A. (1989). The mechanics of graben formation in crustal
rocks: A finite element analysis. J. Geophys. Res., 94, 13 961–73.
Minton, D. A. (2008). The topographic limits of gravitationally bound, rotating sand piles.
Icarus, 195, 698–704.
Molnar, P. and Lyon-Caen, H. (1988). Some simple physical aspects of the support,
structure, and evolution of mountain belts. In Processes in Continental Lithosphere
Deformation, GSA Special Paper 218, ed. S. P. Clark, B. C. Burchfiel and J. Suppe.
Geological Society of America, pp. 179–207.
Montgomery, D. R. and Dietrich, W. E. (1992). Channel initiation and the problem of land-
scape scale. Science, 255, 826–30.
Moore, H. J., Arthur, D. W. G. and Schaber, G. G. (1978). Yield strengths of flows on the
Earth, Mars, and Moon. In Proc. Lunar Planet Sci. Conf. 9th. Houston, TX: Lunar and
Planetary Institute, pp. 3351–78.
Moore, J. G., Clague, D. A., Holcomb, R. T., Lipman, P. W., Normark, W. R. and Torresan,
M. E. (1989). Prodigious submarine landslides on the Hawaiian Ridge. J. Geophys.
Res., 94, 17 465–84.
Moore, J. M., Chapman, C. R., Bierhaus, E. B., Greeley, R., Chuang, F. C., Klemaszewski,
J., Clark, R. N., Dalton, J. B., Hibbitts, C. A., Schenk, P. M., Spencer, J. R. and
Wagner, R. (2004). Callisto. In Jupiter: The Planet, Satellites and Magnetosphere,
ed. F. Bagenal, T. E. Dowling and W. B. McKinnon. Cambridge, UK: Cambridge
University Press, pp. 397–426.
Moores, J. E., Pelletier, J. D. and Smith, P. H. (2008). Crack propagation by differential
insolation on desert surface clasts. Geomorphology, 102, 472–81.
Morisawa, M. E. (1959). Relation of the quantitative geomorphology to streamflow in rep-
resentative watersheds of the Appalachian Plateau province. Office of Naval Research,
Project NR 389–042, Tech. Rept. 20, p. 94.

EBSCOhost - printed on 11/3/2023 10:00 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
References 477

(1968). Streams. New York, NY: McGraw-Hill.

Morrison, D. (1970). Thermophysics of the planet Mercury. Space Science Reviews, 11,
271–307.
Mueller, K. and Golombek, M. (2004). Compressional structures on Mars. Annu. Rev.
Earth Planet. Sci., 32, 435–464.
Muller, S. W. (1947). Permafrost, or Permanently Frozen Ground and Related Engineering
Problems. Ann Arbor, MI: J. W. Edwards.
Murray, C. D. and Dermott, S. F. (1999). Solar System Dynamics. New York, NY:
Cambridge.
NASA (1978). Standard Techniques for Presentation and Analysis of Crater Size-Frequency
Data. Technical Memorandum 79730. Washington, DC: NASA.
Neumann, G. A., Zuber, M. T., Smith, D. E. and Lemonine, F. G. (1996). The lunar crust:
Global structure and signature of major basins. J. Geophys. Res., 101, 16 841–3.
Neumann, G. A., Zuber, M. T., Wieczorek, M. A., McGovern, P. J., Lemonine, F. G.
and Smith, D. E. (2004). Crustal structure of Mars from gravity and topography. J.
Geophys. Res., 109, E08002, doi:10.1029/2004JE002262.
Nevin, C. M. (1942). Principles of Structural Geology. New York, NY: J. Wiley & Sons.
Newsom, H. E. (1995). Composition of the Solar System, planets, meteorites and major
terrestrial reservoirs. In Global Earth Physics, ed. T. J. Ahrens. Washington, DC:
American Geophysical Union, pp. 159–89.
Newton, I. (1966). Principia (2 vols. Cajori version). Berkeley: University Calif. Press.
Nichols, B. L. (1939). Viscosity of lava. J. Geology, 47, 290–302.
Nimmo, F. and Pappalardo, R. T. (2006). Diapir-induced reorientation of Saturn’s moon
Enceladus. Nature, 441, 614–16.
Nye, J. F. (1952). The mechanics of glacier flow. J. Glaciology, 2, 82–93.
Oberbeck, V. R. and Quaide, W. L. (1968). Genetic implications of lunar regolith thickness
variations. Icarus, 9, 446–65.
Ollier, C. D. (1975). Weathering. London: Longman.
Paige, D. A., Wood, S. E. and Vasavada, A. R. (1992). The thermal stability of water ice at
the poles of Mercury. Science, 258, 643–6.
Paphitis, D. (2001). Sediment movement under unidirectional flows: An assessment of
empirical threshold curves. Coast. Eng., 43, 227–45.
Paterson, M. S. and Wong, T.-F. (2005). Experimental Rock Deformation – The Brittle
Field, 2nd edn. New York, NY: Springer.
Paterson, W. S. B. (1999). The Physics of Glaciers, 3rd edn. Oxford: Butterworth-
Heinemann.
Pauling, L. (1988). General Chemistry. New York, NY: Dover.
Pearce, S. J. and Melosh, H. J. (1986). Terrace width variations in complex lunar craters.
Geophys. Res. Lett., 13, 1419–22.
Pechmann, J. C. (1980). The origin of polygonal troughs on the Northern Plains of Mars.
Icarus, 42, 185–210.
Pelletier, J. D. (2008). Quantitative Modeling of Earth Surface Processes. Cambridge, UK:
Cambridge University Press.
(2009). Controls on the height and spacing of eolian ripples and transverse dunes: A
numerical modeling investigation. Geomorphology, 105, 322–33.
Perry, R. S. and Kolb, V. M. (2003). Biological and organic constituents of desert varnish:
Review and new hypotheses. In Instruments, Methods and Missions for Astrobiology VII,
ed. R. B. Hoover, R. R. Paepe and A. Y. Rozanov. Bellingham, WA: SPIE, pp. 202–17.
Petford, N. (1996). Dykes or diapirs? Trans. Roy. Soc. Edinburgh: Earth Sci., 87,
105–14.

EBSCOhost - printed on 11/3/2023 10:00 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
478 References

Phillips, R. J., Zuber, M. T., Solomon, S. C., Golombek, M. P., Jakowsky, B. M., Banerdt,
W. B., Smith, D. E., Williams, R. M. E., Hynek, B. M., Ahronson, O. and Hauck, S. A.
(2001). Ancient geodynamics and global-scale hydrology on Mars. Science, 291,
2587–91.
Pierazzo, E. and Melosh, H. J. (2000). Understanding oblique impacts from experiments,
observations and modeling. Annu. Rev. Earth Planet. Sci., 28, 141–67.
Pike, R. J. and Rozema, W. J. (1975). Spectral analysis of landforms. Annals Assoc. Am.
Geograph., 65, 499–516.
Playfair, J. (1964). Illustrations of the Huttonian Theory of the Earth. New York, NY:
Dover.
Poirier, J.-P. (1985). Creep of Crystals. Cambridge: Cambridge University Press.
(1991). Introduction to the Physics of the Earth’s Interior. Cambridge, UK: Cambridge
University Press.
Pollard, D. D. and Fletcher, R. C. (2005). Fundamentals of Structural Geology. Cambridge,
UK: Cambridge University Press.
Post, A. and LaChapelle, E. (2000). Glacier Ice. Seattle, WA: University of Washington
Press.
Presley, M. A. and Christensen, P. R. (1997). Thermal conductivity measurements of par-
ticulate materials 1. A review. J. Geophys. Res., 102, 6535–49.
Prockter, L. M. and Pappalardo, R. T. (2000). Folds on Europa: Implications for crustal
cycling and accommodation of extension. Science, 289, 941–3.
Project, B. V. S. (1981). Basaltic Volcanism on the Terrestrial Planets. New York, NY:
Pergamon.
Pumpelly, R. (1918). My Reminiscences. New York, NY: Henry Holt and Co.
Pye, K. (1987). Aeolian Dust and Dust Deposits. London: Academic Press.
Quitté, G., Latkoczy, C., Halliday, A. N., Schönbächler, M. and Günther, D. (2005). Iron-60
in the Eucrite parent body and the initial 60Fe/56Fe of the Solar System. LPSC XXXVI,
Abstract 1827.
Ramberg, H. (1967). Gravity, Deformation and the Earth’s Crust. London: Academic
Press.
Ramsay, J. G. (1967). Folding and Fracturing of Rocks. New York, NY: McGraw-Hill.
Ranalli, G. (1995). Rheology of the Earth, 2nd edn. London: Chapman & Hall.
Rayleigh, Lord (1916). On convection currents in a horizontal layer of fluid, when the
higher temperature is on the under side. Philosophical Magazine, 32, 529–46.
Reiche, P. (1937). The Toreva block, a distinctive landslide type. J. Geology, 45, 538–48.
Rice, J. R. (1976). The localization of plastic deformation. In 14th International Conference
of Theoretical and Applied Mechanics, pp. 207–20.
Rich, J. L. (1911). Gravel as a resistant rock. J. Geol., 19, 492–506.
Richardson, J. E., Melosh, H. J. and Greenberg, R. (2004). Impact-induced seismic activity
on asteroid 433 Eros: A surface modification process. Science, 306, 1526–9.
Richardson, J. E., Melosh, H. J., Greenberg, R. and O’Brien, D. P. (2005). The global
effects of impact-induced seismic activity on fractured asteroid surface morphology.
Icarus, 179, 325–49.
Richardson, J. E., Melosh, H. J., Lisse, C. M. and Carcich, B. (2007). A ballistics analysis
of the Deep Impact ejecta plume: Determining comet Tempel 1’s gravity, mass and
density. Icarus, 190, 357–90.
Rinaldo, A., Rodriguez-Iturbe, I. and Rigon, R. (1998). Channel networks. Ann. Rev. Earth
Planet. Sci., 26, 289–327.
Ritter, D. F. (1986). Process Geomorphology. Dubuque, IA: Wm. C. Brown.

EBSCOhost - printed on 11/3/2023 10:00 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
References 479

Rodine, J. D. and Johnson, A. M. (1976). The ability of debris, heavily freighted with
coarse clastic materials, to flow on gentle slopes. Sedimentology, 23, 213–34.
Roering, J. J., Kirchner, J. W. and Dietrich, W. E. (1999). Evidence for non-linear, diffusive
sediment transport on hillslopes and implications for landscape morphology. Water
Resources Res., 35, 853–70.
Rosato, A., Strandburg, K. J., Prinz, F. and Swendsen, R. H. (1987). Why the Brazil nuts
are on top: Size segregation of particulate matter by shaking. Phys. Rev. Lett., 58,
1038–40.
Rosenblatt, M., Hassig, P. J. and Orphal, D. L. (1982). Soil–air interactions during airblast-
induced ground motions. Report. DNA–TR–81–103.
Rouse, H. (1978). Elementary Mechanics of Fluids. New York, NY: Dover.
Rubin, A. M. (1993). Dikes vs. diapirs in viscoelastic rock. Earth Planet. Sci. Lett., 119,
641–59.
(1995). Propagation of magma-filled cracks. Annu. Rev. Earth Planet. Sci., 23,
287–336.
Ryan, M. P. (1987). Neutral Buoyancy and the mechanical evolution of magmatic sys-
tems. In Magmatic Processes: Physicochemical Principles, ed. B. O. Mysen. The
Geochemical Society, pp. 259–87.
Sagan, C. (1970). Planetary Exploration. Eugene, OR: Oregon State System of Higher
Education.
Sanchez-Levega, A. (2010). An Introduction to Planetary Atmospheres. Francis and Taylor.
Schaller, C. J. and Melosh, H. J. (1998). Venusian ejecta parabolas: Comparing theory with
observation. Icarus, 131, 123–37.
Schenk, P. M. and Bulmer, M. H. (1998). Origin of mountains on Io by thrust faulting and
large-scale mass movements. Science, 279, 1514–17.
Schenk, P. M., Wilson, R. R. and Davies, A. G. (2004). Shield volcano topography and the
rheology of lava flows on Io. Icarus, 169, 98–110.
Schmincke, H.-U. (2003). Volcanism. Berlin: Springer.
Scholz, C. H. (1990). The Mechanics of Earthquakes and Faulting. Cambridge, UK:
Cambridge University Press
Schramm, D. N., Tera, F. and Wasserburg, G. J. (1970). The isotopic abundance of 26Mg
and limits on 26Al in the early Solar System. Earth Planet. Sci. Lett., 10, 44–59.
Schubert, G. (1975). Subsolidus convection in the mantles of the terrestrial planets. Annu.
Rev. Earth Planet. Sci., 7, 289–342.
Schubert, G., Turcotte, D. L. and Olson, P. (2001). Mantle Convection in the Earth and
Planets. Cambridge University Press.
Schulte, P., Alegret, L., Arenillas, I., Arz, J. A., Barton, P. J., Bown, P. R., Bralower, T. J.,
Christeson, G. L., Claeys, P., Cockell, C. S., Collins, G. S., Deutsch, A., Goldin, T.
J., Goto, K., Grajales-Nishimura, J. M., Grieve, R. A. F., Gulick, S. P. S., Johnson,
K. R., Kiessling, W., Koeberl, C., Kring, D. A., MacLeod, K. G., Matsui, T., Melosh,
J., Montanari, A., Morgan, J. V., Neal, C. R., Nichols, D. J., Norris, R. D., Pierazzo,
E., Ravizza, G., Rebolledo-Vieyra, M., Reimold, W. U., Robin, E., Salge, T., Speijer,
R. P., Sweet, A. R., Urrutia-Fucugauchi, J., Vajda, V., Whalen, M. T. and Willumsen,
P. S. (2010). The Chicxulub asteroid impact and mass extinction at the Cretaceous-
Paleogene Boundary. Science, 327, 1214–18 10.1126/science.1177265.
Schumm, S. A. (1977). The Fluvial System. New York, NY: Wiley.
(1985). Patterns of alluvial rivers. Ann. Rev. Earth Planet. Sci., 13, 5–27.
Sclater, J. G., Jaupart, C. and Galson, D. (1980). The heat flow through oceanic and contin-
ental crust and the heat loss of the earth. Rev. Geophys., 18, 269–312.

EBSCOhost - printed on 11/3/2023 10:00 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
480 References

Scott, R. F. (1963). Principles of Soil Mechanics. Reading, MA: Addison-Wesley.

(1967). Viscous flow of craters. Icarus, 7, 139–148.
Segatz, M., Spohn, T., Ross, M. N. and Schubert, G. (1988). Tidal dissipation, surface heat
flow, and figure of viscoelastic models of Io. Icarus, 75, 187–206.
Seidelmann, P. K., Abalakin, V. K., Bursa, M., Davies, M. E., DeBergh, C., Lieske, J. H.,
Oberst, J., Simon, J. L., Standish, E. M., Stooke, P. and Thomas, P. (2002). Report of
the IAU/IAG working group on cartographic coordinates and rotational elements of
the planets and satellites: 2000. Celest. Mech. Dyn. Astron., 82, 83–110.
Selby, M. J. and Hodder, A. P. W. (1993). Hillslope Materials and Processes, 2nd edn. New
York, NY: Oxford University Press.
Self, S., Keszthelyi, L. P. and Thordarson, T. (1998). The importance of pahoehoe. Annu.
Rev. Earth Planet. Sci., 26, 81–110.
Sharp, R. P. (1960). Glaciers. Eugene, OR: Oregon State System of Higher Education.
(1963). Wind ripples. J. Geology, 71, 617–36.
(1964). Wind-driven sand in the Coachella Valley, California. Geol. Soc. Amer. Bull., 75,
785–804.
(1979). Interdune flats of the Algodones chain, Imperial Valley, California. Geol. Soc.
Amer. Bull., 90, 908–16.
Sharpe, C. F. S. (1938). Landslides and Related Phenomena. Paterson, NJ: Pageant
Books.
Shelton, J. S. (1966). Geology Illustrated. San Francisco, CA: Freeman.
Shewmon, P. G. (1963). Diffusion in Solids. New York, NY: McGraw-Hill.
Shoemaker, E. M. (1962). Interpretation of Lunar craters. In Physics and Astronomy of the
Moon, ed. Z. Kopal. New York, NY: Academic Press, pp. 283–259.
Shoemaker, E. M., Hackman, R. J. and Eggleton, R. E. (1963). Interplanetary correlation
of geologic time. Adv. in Astronaut. Sci., 8, 70–89.
Shoemaker, E. M., Batson, R. M., Hold, H. E., Morris, E. C., Rennilson, J. J. and Whitaker,
E. A. (1967). Surveyor V: Television pictures. Science, 158, 642–52.
Shoemaker, E. M., Batson, M., Holt, H. E., Morris, E. C., Rennilson, J. J. and Whitaker, E.
A. (1969). Observations of the lunar regolith and the Earth from the television camera
on Surveyor 7. J. Geophys. Res., 74, 6081–119.
Shreve, R. L. (1966a). Sherman landslide, Alaska. Science, 154, 1639–43.
(1966b). Statistical law of stream numbers. J. Geology, 74, 17–37.
Shumskii, P. A. (1964). Principles of Structural Glaciology. New York, NY: Dover.
Siever, R. (1988). Sand. New York, NY: Freeman.
Sigurdsson, H. (1999). Melting the Earth. Oxford: Oxford.
Simpson, J. E. (1999). Gravity Currents: In the Environment and the Laboratory, 2nd edn.
Cambridge, UK: Cambridge University Press.
Smith, R. E., Smetten, K. R. J., Broadbridge, P. and Woolhiser, D. A. (2002). Infiltration
Theory for Hydrologic Applications. Washington, DC: American Geophysical
Union.
Smrekar, S. E., Kiefer, W. S. and Stofan, E. R. (1997). Large volcanic rises on Venus. In
Venus II, ed. S. W. Bougher, D. M. Hunten and R. J. Phillips. Tucson, AZ: University
of Arizona Press, pp. 845–78.
Soderblom, L. A. (1970). A model for small-impact erosion applied to the lunar surface. J.
Geophy. Res., 75, 2655–61.
Solomon, S. C. and Head, J. W. (1982). Mechanisms for lithospheric heat transport
on Venus: Implications for tectonic style and volcanism. J. Geophys. Res., 87,
9236–46.

EBSCOhost - printed on 11/3/2023 10:00 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
References 481

(1984). Venus banded terrain: Tectonic models for band formation and their relationship
to lithospheric thermal structure. J. Geophys. Res., 89, 6885–97.
Sonder, L. J. and Jones, C. H. (1999). Western United States extension: How the West was
widened. Ann. Rev. Earth Planet. Sci., 27, 417–62.
Sonnett, C. P., Kvale, E. P., Zakharian, A., Chan, M. A. and Demko, T. M. (1996). Late
Proterozoic and Paleozoic tides, retreat of the Moon, and rotation of the Earth. Science,
273, 100–4.
Sotin, C., Mielke, R., Choukroun, M., Neish, C., Barmatz, M., Castillo, J., Lunine, J. I.
and Mitchell, K. (2009). Ice–hydrocarbon interactions under Titan-like conditions:
Implications for the carbon cycle on Titan. LPSC Conference, 40, Abs. # 2088.
Squyres, S. W., Janes, D. M., Baer, G., Binshadler, D. L., Schubert, G., Sharpton, V. L. and
Stofan, E. R. (1992). The morphology and evolution of coronae on Venus. J. Geophys.
Res., 97, 13 611–34.
Stacey, F. D. (1992). Physics of the Earth, 3rd edn. New York, NY: John Wiley & Sons.
Stephens, H. G. and Shoemaker, E. M. (1987). In the Footsteps of John Wesley Powell.
Boulder, CO: Johnson Books.
Stern, S. A. and Tholen, D. J. (1997). Pluto and Charon. Tucson, AZ: University of Arizona
Press, p. 728.
Stesky, R. M., Brace, W. F., Riley, D. K. and Bobin, P. Y. (1974). Friction in faulted rock at
high temperature and pressure. Tectonophysics, 23, 177–203.
Stevenson, D. J. (1981). Models of the Earth’s core. Science, 214, 611–19.
Stewart, S. A. and Allen, P. J. (2002). A 20-km-diameter multi-ringed impact structure in
the North Sea. Nature, 418, 520–52.
Strom, R. G. (1964). Analysis of lunar lineaments I: Tectonic maps of the Moon. Comm.
Lunar Planet. Lab, 2, 205–16.
(1984). Mercury. In The Geology of the Terrestrial Planets, ed. M. H. Carr. Washington,
DC: US Govt. Printing Office, pp. 13–55.
Strom, R. G. and Sprague, A. L. (2003). Exploring Mercury: The Iron Planet. Dordrecht:
Springer.
Strom, R. G., Malhotra, R., Ito, T., Yoshida, F. and Kring, D. A. (2005). The origin of
planetary impactors in the inner Solar System. Science, 309, 1847–50.
Sugden, D. E. and John, R. S. (1976). Glaciers and Landscape. New York, NY: Wiley.
Sullivan, R., Arvidson, R. E., Bell, J. F., Gellert, R., Golombeck, M. P., Greeley, R.,
Herkenhoff, K., Johnson, J., Thompson, S., Whelley, P. and Wray, J. (2008). Wind-
driven particle mobility on Mars: Insights from Mars Exploration Rover observations
at “El Dorado” and surroundings at Gusev Craters. J. Geophys. Res., 113, E06So7.
Sunshine, J. M., Farnham, T. L., Feaga, L., Groussin, O., Merlin, F., Milliken, R. E. and
A’Hearn, M. A. (2009). Temporal and spatial variability of lunar hydration as observed
by the Deep Impact spacecraft. Science, 265, 565–8.
Suppe, J. (1983). Geometry and kinematics of fault-bend folding. Am. J. Science, 283,
684–721.
(1985). Principles of Structural Geology. Englewood Cliffs, NJ: Prentice-Hall, Inc.
Taber, S. (1929). Frost heaving. J. Geology, 37, 428–61.
Tennekes, H. and Lumley, J. L. (1972). A First Course in Turbulence. Cambridge, MA:
MIT Press.
Terzaghi, K. (1943). Theoretical Soil Mechanics. New York, NY: John Wiley & Sons.
Theis, C. V. (1935). The relation between the lowering of the piezometric surface and
the rate and duration of discharge of a well using ground water storage. Trans. Am.
Geophys. Union, 16, 519–24.

EBSCOhost - printed on 11/3/2023 10:00 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
482 References

Thomas, P. (1988). Radii, shapes and topography of the satellites of Uranus from limb
coordinates. Icarus, 73, 427–41.
Thomas, P., Veverka, J. and Duxbury, T. (1978). Origin of the grooves on Phobos. Nature,
273, 282–4.
Thomas, P., Burns, J. A., Rossier, L., Simonelli, D., Veverka, J., Chapman, C. R., Klaasen,
K. P., V., J. T. and Belton, M. J. S. (1998). The small inner satellites of Jupiter. Icarus,
135, 360–71.
Thomas, P. C. (1989). The shapes of small satellites. Icarus, 77, 248–74.
Thomas, P. C., Burns, J. A., Helfenstein, P., Squyres, S. W., Veverka, J., Porco, C., Turtle,
E. P., McEwen, A. S., Denk, T., Giese, B., Roatsch, T., Johnson, T. V. and Jacobson,
R. A. (2007). Shapes of the Saturnian icy satellites and their significance. Icarus, 190,
573–84.
Tilling and Dvorak, J. (1993). Anatomy of a basaltic volcano. Nature, 363, 125–33.
Tjia, H. D. (1970). Lunar wrinkle ridges indicative of strike-slip faulting. Geol. Soc. Amer.
Bull., 81, 3095–100.
Todhunter, I. (1962). A History of the Mathematical Theories of Attraction and the Figure
of the Earth. New York, NY: Dover.
Tozer, D. C. (1965). Heat transfer and convection currents. Philos. Trans. R. Soc. London,
Ser. A, 258, 252–71.
(1970). Factors determining the temperature evolution of thermally convecting earth
models. Phys. Earth Planet. Interiors, 2, 393–8.
Turcotte, D. L. (1979). Flexure. Adv. in Geophys., 21, 51–86.
Turcotte, D. L. and Schubert, G. (2002). Geodynamics: Applications of Continuum Physics
to Geological Problems, 2nd edn. Cambridge, UK: Cambridge University Press.
Turcotte, D. L., Willemann, R. J., Haxby, W. F. and Norberry, J. (1981). Role of membrane
stresses in the support of planetary topography. J. Geophys. Res., 86, 3951–9.
Turtle, E. P. and Melosh, H. J. (1997). Stress and flexural modeling of the Martian litho-
spheric response to Alba Patera. Icarus, 126, 197–211.
Urai, J. L., Spiers, C. J., Zwart, H. J. and Lister, G. S. (1986). Weakening of rock salt by
water during long-term creep. Nature, 324, 554–7.
Urey, H. C. (1952). The Planets: Their Origin and Development. New Haven, CT: Yale
University Press.
USGS (2003). Color-Coded Contour Map of Mars, M 25M RKT.
Van Schums, W. R. (1995). Natural radioactivity of the crust and mantle. In Global
Earth Physics, ed. T. J. Ahrens. Washington, DC: American Geophysical Union, pp.
283–91.
Vening-Meinesz, F. A. (1947). Shear patterns of the Earth’s crust. Trans. Am. Geophys.
Union, 28, 1–61.
Vervack, R. and Melosh, H. J. (1992). Wind interaction with falling ejecta: Origin of the
parabolic features on Venus. Geophys. Res. Lett., 19, 525–8.
Wallace, M. H. and Melosh, H. J. (1994). Buckling of a pervasively faulted lithosphere.
PAGOPH, 142, 239–61.
Washburn, A. L. (1980). Geocryology: A Study of Periglacial Processes and Environments.
New York, NY: Wiley.
Watson, E. B. (1982). Melt infiltration and magma evolution. Geology, 10, 236–40.
Watters, T. R. and Schultz, R. A. (2010). Planetary Tectonics. Cambridge, UK: Cambridge
University Press.
Watts, A. B. (2001). Isostasy and Flexure of the Lithosphere. Cambridge University
Press.

EBSCOhost - printed on 11/3/2023 10:00 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
References 483

Watts, A. B., Cochran, J. R. and Selzer, G. (1975). Gravity anomalies and flexure of the
lithosphere: A three-dimensional study of the Great Meteor Seamount, NE Atlantic. J.
Geophys. Res., 80, 1391–8.
Weast, R. C. (1972). Chemical Rubber Handbook. CRC Press.
Weertman, J. (1971). Theory of water-filled crevasses in glaciers applied to vertical magma
transport beneath oceanic ridges. J. Geophys. Res., 76, 1171–83.
(1979). Height of mountains on Venus and the creep properties of rock. Phys. Earth
Planet. Int., 19, 197–207.
Weertman, J. and Weertman, J. R. (1992). Elementary Dislocation Theory. New York, NY:
Oxford University Press.
Weissel, J. K., Anderson, R. N. and Geller, C. A. (1980). Deformation of the Indo-Australian
plate. Nature, 287, 284–91.
Wenkert, D. D. (1979). The flow of salt glaciers. Geophys. Res. Lett., 6, 523–6.
Werner, B. T. (1995). Eolian dunes: Computer simulations and attractor interpretation.
Geology, 23, 1107–10.
Whalley, W. B. and Azizi, F. (2003). Rock glaciers and protalus landforms: Analogous forms
and ice sources on Earth and Mars. J. Geophys. Res., 108, 10.1029/2002JE001864.
White, W. B. (1988). Geomorphology and Hydrology of Karst Terrains. New York, NY:
Oxford.
Wieczorek, M. A. (2007). Gravity and topography of the terrestrial planets. In Treatise on
Geophysics, ed. G. Schubert and T. Spohn. Amsterdam: Elsevier, pp. 165–206.
Wilhelms, D. E. (1987). The Geologic History of the Moon. Report 1348.
(1989). To a Rocky Moon. Tucson, AZ: University of Arizona Press.
Willgoose, G. (2005). Mathematical modeling of whole landscape evolution. Ann. Rev.
Earth Planet. Sci., 33, 443–59.
Williams, H. and McBirney, A. R. (1979). Volcanology. San Francisco, CA: Freeman.
Williams, P. J. and Smith, M. W. (1991). The Frozen Earth. Cambridge, UK: Cambridge
University Press.
Willis, B. (1893). The mechanics of Appalachian structure. In Thirteenth Annual Report
of the United States Geologic Survey, 1891–1892, ed. J. W. Powell. Washington, DC:
Govt. Printing Office, pp. 211–82.
Winkler, E. M. and Singer, P. C. (1972). Crystallization pressure of salts in stone and con-
crete. Geol. Soc. Amer. Bull., 83, 3509–14.
Wise, D. U., Golombek, M. P. and McGill, G. E. (1979). Tharsis province of Mars: Geologic
sequence, geometry and a deformation mechanism. Icarus, 38, 456–72.
Wolfe, E. W. and Bailey, N. G. (1972). Lineaments of the Apennine Front – Apollo 15 land-
ing site. In Proceedings of the Third Lunar Science Conference. Houston, TX: MIT
Press, pp. 15–25.
Wood, J. A. (1997). Rock weathering on the surface of Venus. In Venus II, ed. S. W.
Bougher, D. M. Hunten and R. J. Phillips. Tucson, AZ: University of Arizona Press,
pp. 637–64.
Woodward, N. B., Boyer, S. E. and Suppe, J. (1989). Balanced Geological Cross-Sections.
Washington, DC: American Geophysical Union.
Woronow, A. (1977). Crater saturation and equilibrium: A Monte-Carlo simulation. J.
Geophys. Res., 82, 2447–56.
Wyllie, P. J. (1971). The Dynamic Earth. New York, NY: Wiley.
Yoder, C. F. (1995). Astrometric and geodetic properties of Earth and the Solar System.
In Global Earth Physics, ed. T. J. Ahrens. Washington, DC: American Geophysical
Union, pp. 1–31.

EBSCOhost - printed on 11/3/2023 10:00 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
484 References

Zahnle, K., Schenk, P. M., Sobieszczyk, S., Dones, L. and Levison, H. F. (2001).
Differential cratering of synchronously rotating satellites by ecliptic comets. Icarus,
153, 111–29.
Zoback, M. D., Zoback, M. L., Mount, V. S., Suppe, J., Eaton, J. P., Healy, J. H.,
Oppenheimer, D., Reasenberg, P., Jones, L., Raleigh, C. B., Wong, I. G., Scotti, O.
and Wentworth, W. (1987). New evidence for the state of stress on the San Andreas
fault system. Science, 238, 1105–11.

EBSCOhost - printed on 11/3/2023 10:00 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
Index

aa, 208–09 Apollo 17, 152, 153, 277, 322

accretion of planets, 268 heat flow, 292

acoustic fluidization, 343 Appalachian Valley and Ridge, 129
adiabat, 180, 181 Arden corona, 139
adiabatic gradient, 123, 173–74 Ares Vallis, 405
Adivar crater, 379 Ares Vallis flood, 406
Aeolis Planum, 413, 414 Argyre Basin, 455
Agassiz, Louis, 434, 439 Aristarchus crater, 216
age, solar system, 3 Aristarchus plateau, 195, 206, 216
ages of planetary surfaces, see impact craters: Artemis Chasma, 16
populations: dating Artemis Corona, 93
agglutinate, 285 Ascraeus Mons, 207
Ahmad Baba crater, 225 asteroid, 4
Airy isostasy, 87, 88 atmospheres
Alba Patera, 157, 159 retention, 6
albedo, 289
Aleutian volcanic chain, 189, 195 Bagnold, R., 348, 374, 397, 401, 430
Algodones dunes, 366 Baltis Vallis, Venus, 217
alkali basalt, 184 Barringer, D. M., 223
Allegheny River, 417 basalt, 179, 196
Alpha Regio, Venus, 216 Basin and Range province, 136
Alphonsus crater, 155 batch melting, 187
Alpine Fault, 149 Beta Regio, 91
Alps, 436 Big Bear Lake, CA, 302
Amazon River, 416 Bingham, E. C., 64, 185
Amboy crater, CA, 377 biotite, 305
ammonia clathrate, 179 Blackhawk landslide, 341
Amontons, Guillaume, 68 blueberries on Mars, 304
amorphous ice, 288 Bonestell, Chesley, 328
Anderson, Don, 94 Borealis plains, 17
Anderson, E. M., 145 Boulton, G. S., 452
Andes mountains, 189, 195 Boussinesq, J., 411
andesite, 182, 184, 196, 197 Brazil nut effect, 314
anhydrite, 297, 299 Brent Crater, 241
Anorthositic Gabbro, 10 Bridgeman, Percy, 70
Antarctic ice sheet, 436, 438 Bristol Cathedral, 320
Antarctic ice streams, 444 Brown, R., 292
aplite dikes, 302 Bürg crater, 336
Copyright 2011. Cambridge University Press.

Apollo 11, 284

Apollo 15, 143, 144, 216, 217 Cajon Pass, CA, 291
heat flow, 292 Callisto, 4, 18, 20
Apollo 16, 98, 279 polar cap, lack of, 288

485

EBSCO Publishing : eBook Academic Collection (EBSCOhost) - printed on 11/3/2023 10:00 AM via YDUQS
AN: 399390 ; H. Jay Melosh.; Planetary Surface Processes
Account: s4526441.main.estacio
486 Index

Callisto (cont.) Darcy, unit, 391

sublimation weathering, 307, 308 Darwin, Charles, 310
Valhalla basin, 20 dating planetary surfaces, see impact craters:
Caloris basin, 15, 195, 227, 236 populations: dating
Canyon de Chelly, 395 Davis, William Morris, 430
Canyonlands National Monument, UT, 309 Davison, Charles, 320, 322
Cargo Muchacho mountains, 313 Death Valley, CA, 427
Carolina Bays, 462 Debye length, 322
Cassini, Jacques, 27 Deccan Plateau, 194, 195
catastrophism, 268–69 Denali earthquake, 341, 342
Central Atlantic Magmatic Province, 194 Devil’s Postpile, 210
Ceraunius Fossae, 150 diamonds, 234
Ceres, 4 diapirs, 188
Channeled Scablands, 385, 406, 409 Diemos, 320
Chapman, C. R., 266 dikes, 188–92
Chase, Clem, 431 critical length, 192
chemical potential, 177 pressure vs. depth, 191
Chézy, A., 402 surface grabens, 151
Chicxulub crater, 271 velocity of fluid, 192
Chugach Mountains, 438 Discovery Scarp, 149
Clapeyron, É, 171 dislocations, 439
Clausius–Clapeyron equation, 174 drag coefficient, 350
cliff Dunes, 371–76
maximum height, 333 Barchan, 372–74
climate Barchanoid ridges, 373
carbon dioxide control, see weathering: chemical: clay, 316, 368
carbon dioxide climbing, 376
coesite, 234 falling, 376
Colorado Plateau, 335, 395 linear, 373, 374–75
Colorado River, 385, 411, 416 longitudinal, 372, 375
Columbia Plateau, 194 lunettes, 375
Columbia River, 385, 417 migration, 372
Columbia River Basalt Province, 214 modification timescale, 372
columnar joints, 141 parabolic, 372
comet, 5 reversing, 373, 375
composition seif dunes, 375
eutectic, 179 slip face, 371
icy bodies, 176 star, 372, 373, 375
rocky planets, 175–76 transverse, 372, 373, 374
solar system, 175 velocity, 372
Conel, J., 322 height dependence, 371
constitutive relations, 56
Copernicus crater, 14, 226, 285, 335, 336 Earth, 3, 22–24
Coronae, 195 atmosphere, 23
Coulomb, Charles, 68, 326, 332 carbon dioxide, 299
creep, of soil, see soil:creep center of figure offset, 35
Cretaceous era, 271 conductive temperature gradient, 174
crevasse, 188, 447, 448–49 continents, 23
crust convection velocity, 124
thickness, from gravity, 98 crater clusters, 254
cryovolcanism, 169, 182–83, 218 crust, 22, 39
Culling, W. E. H., 324 flattening, 27
geoid anomalies, 99
da Vinci, Leonardo, 68 glaciers, 435
Darcy equation, 187 hypsometric curve, 39, 42
Darcy law, 393 Ice Ages, 435, 451, 460
Darcy, Henry, 388, 390 karst, 297

EBSCOhost - printed on 11/3/2023 10:00 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
Index 487

large igneous provinces, 217 fluid threshold, 358, 360

linear dunes, 374 impact threshold, 358, 360–61
moment of inertia, 29 impact threshold on Mars, 360
oblateness, 27 minimum threshold velocity, 358, 359
oceanic lithosphere, 106 threshold diameter, 359
oceanic plate subsidence, 120 threshold speeds on planets, 359
oxygen, 295, 298 Van der Waals forces, 357
plate tectonics, see plate tectonics megaripples, 368, 369, 370
primordial heat, 127 Reynolds number, 351
rock cycle, 23 ridges, 369
rotational pole, 43 ripples, 367, 369
salt glaciers, 435, 445 roughness, 355
shield volcano, 206 sand
stone pavement, 313 definition, 365
temperatures sand grain
ice age effects, 291 weight, 349
thermal boundary layer, 124 sand shadows, 370, 371
tidal dissipation, 114 sand surface stability, 366
topographic power spectrum, 46 sand vs. dust, 349
East African Rift, 151 Stokes’ law, 351–52
East Antarctic ice sheet, 450 transient phenomena, 378–79
eccentricity, orbital, 2 transport, 348, 361–63
Egypt, 348 exceptional winds, 362
Einstein, Albert, 411 impact creep, 354, 363
Einstein–Roscoe formula, 185 kamikaze effect, 316, 368–69
ejecta blanket, 244 laminar regime, 350
El Dorado crater, 368 reptation, 354, 363
Elastic constants saltation, 353–54
bulk modulus, 56 hop length, 354, 362
Poisson’s ratio, 56 sand carpet, 361
shear modulus, 56 sand flux, 363
Young’s modulus, 56 suspension, 352–53
elevation terminal velocity, 350, 352
definition, 36 turbulent regime, 350, 351
geodetic reference, 36 Bagnold’s rule of thumb, 353
spectral power, 45 wind velocity dependence, 362
Elm, Switzerland, 341 ventifact, 365
Elsinore corona, 139 wind streaks, 377, 378
Enceladus wind structure, 356
geysers, 205 laminar sublayer, 356
reorientation tectonics, 155 near surface, 354–55
enthalpy, specific, 201 over saltation carpet, 361
entropy, 173, 177 velocity vs. height, 355
eolian processes, 349 viscous sublayer, 355, 356
abrasion by wind, 365 yardangs, 376–77
aerodynamic roughness, 361 equilibrium surface, see equipotential surface
backventing, 363–64 equipotential surface, 29
deflation, 377 Eris, 3
drag coefficient, 350 Eros, 286
dunes, see dunes crater chains, 331
dust devils, 363 ponds, 323
friction Reynolds number, 357, 358 escape velocity, see velocity: escape
friction velocity, 355 Eucrite meteorites, 180
hysteresis, 360 Euler, Leonhard, 130
initiation of motion, 354–61 Europa, 18, 19
dust entrainment, 363–64 chaos, 19
electrostatic forces, 360 ice shell, 19

EBSCOhost - printed on 11/3/2023 10:00 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
488 Index

Europa (cont.) drainage basins, 400–01

lithospheric flexure, 97 Earth-centrism, 395
ocean, 19 efficacity, 382
ridges, 19 erosion, 396–98
solid–state greenhouse, 292 extreme events, 383–85
spectrum, 279 floodplains, 406
strike-slip faults, 149 fining-upward sequence, 411, 412
sulfur implantation, 288 floods, 404–05
tidal distortion, 33 catastrophic, 404–06
evaporite minerals, 426 discharge, 405
exosphere, 6 graded rivers, 430
Horton overland flow, 396
fayalite, 177, 299 hydraulic geometry, 409–10
feldspar, 179, 305 infiltration, 386–88
fillets, 284 capacity, 387, 418
Finger Lakes, 452 levees, 406
firn, 435, 436 long profile of rivers, 429
firnification, 293 longshore drift, 423–26
Flamsteed Ring, 320 methane wetting capacity, 387
flattening, see shape, planetary overland flow, 396–401
flexural parameter, 92, 93, 105 belt of no erosion, 397, 400, 417
flexural rigidity, 94, 132 roll waves, 397
flexure, 91–93 piracy, 399
Europa lithosphere, 97 playas, 426–27
flexural parameter, 95 rilles, 400
fourth-order equation, 94 rip currents, 426
maximum stress, 93 runoff, 386, 396
neutral sheet, 95 cinder cones, 387
plate deflection by load, 95, 96 permeable surfaces, 387
profile, 96 sediment transport, 401
fluvial processes initiation, 402
abrasion of bedload, 409 sheet flow, 396
alluvial fan, 406–07 instability, 398
alluvium, 406 stream velocity, 402–04
base level, 429 Chézy coefficient, 404
beach cusps, 426 Chézy equation, 402
bedforms, 409 Darcy–Wiesbach equation, 404
antidunes, 408 Manning equation, 402
cross-bedding, 408 Manning roughness, 403
potholes, 408 streamflow, 407
stream power, 407 streamlined forms, 405, 409, 419
channelization, 398, 399 subsurface flow, 393, see subsurface water
channels, 407–15 turbidity currents, see turbidity currents
braided, 412–13 wave action, 419, 423
distributary, 415 wave base, 422–23
drainage density, 417 wave swash, 426
meanders, 410–12, 414 waves, see waves
Einstein tea-leaf effect, 412 Forbes, J. D., 434
helical flow, 412 forsterite, 177
network analysis, 416–18 Fort Bourbon, Martinique, 332
paleohydrologic hypothesis, 413 Fortuna Tessera, 419
river terraces, 414–15 Fourier, Joseph, 116, 290
tributary, 415 Frank, Alberta, 341
underfit, 411 Frenkel, Yakov, 65
Venusian, 415 friction velocity, 355
cross grading, 399 frost heaving, 304
deltas, 427–28 frost-heave coefficient, 303

EBSCOhost - printed on 11/3/2023 10:00 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
Index 489

Froude number, 340, 397, 408 glaciers

ablation zone, 435, 436, 447
gabbro, 182, 194 accumulation zone, 435, 436, 447
Gaillard, David, 333 basal melting, 437, 438
Galilean satellites, 18, 20, 288 classification, 436–38
Callisto, see Callisto cold-based, 437
Europa, see Europa crevasses, see crevasse
Ganymede, see Ganymede flow velocities, 435
Io, see Io heat transfer inside, 438
Galileo Galilei, 1, 222, 349 hydrologic cycle, 434, 435–36
Galveston harbor, 363 North American, 436
Ganymede, 4, 18, 19 Piedmont, 437
grooved terrain, 20, 136, 137 rock, see rock glaciers
multiring basins, 227 salt glaciers, 445
palimpsest crater, 20 surges, 435, 446
polar caps, 20, 288 terminus, 446, 453
garnet, 179 thermal regime, 437
gas giant, see Jovian planets warm-based, 434, 437–38
Gaspara Glen, J. W., 439, 440
crater chains, 331 goethite, 297
Gault, D. E., 223, 263, 264, 276, 281 grabens, see tectonics: faults: grabens
geoid, 29, 37 Graham Island, 422
geoid to topography ratio, 99 Grand Canyon, 385
geologic processes, 7, 8 granite, 182, 196
Gesundheitstrasse, 449 Gratteri crater, 250
Giant’s Causeway, 210 Grätz number, 213
Gibbs, J. W., 177 gravity anomalies, 97
Gibbs’ Free Energy, 177 Bouguer, 98
Gifford, A. C., 230 free-air, 97
Gila River, 417 geoid, 99
Gilbert, G. K., 49, 193, 194, 321, 353, 382, 419, 426, stress exerted by, 98
430 Great Lakes, 452
Glacial landforms, 451–54 Great Salt Lake, 419
abrasion erosion, 452 Greenland ice sheet, 436
cirques, 451 Griffith, A. A., 142
drumlins, 453 Griggs, David, 70, 143, 149, 306
eskers, 454 Grimaldi, 11
kettle holes, 454 grooves on asteroids, see mass movement: subsidence
moraines, 452–53 pits
plucking erosion, 451–52 grus, 305
Roche moutonnée, 452, 453 Gusev crater, 369
rock flour, 453 gypsum, 297
U-shaped valleys, 451, 452
glacier flow, 439–46 habitability, planetary, 288
basal sliding, 446 Hadley Rille, 217
regelation, see regelation Half-Dome, Yosemite, 301
crevasse pattern, 449 halite, 296
dislocation creep, 440 harmonic analysis, 44–46
extrusion flow, 447 harmonic tremor, 197
Glen’s law, 442, 443, 448 Hartley 2, 307
non-ice solids, 443–44 Hawaii, 195
nonlinear flow laws, 441 submarine landslides, 341
plastic approximation, 443 Hawkins Salt Dome, 156
semicircular channel, 441 Heart Mountain Fault, 154
uphill motion, 447 heat capacity, 173
velocity, 448 heat generation, 116
yield stress, 443 accretion, 113–14

EBSCOhost - printed on 11/3/2023 10:00 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
490 Index

heat generation (cont.) hydraulic mining, 401, 430

effect of differentiation, 118 hydrology, 383–88
gravitational energy, 113 hypsometric curves, 38–41
inverted accretion temperatures, 114
Q of Kelvin–Voight solid, 115 Iani Chaos, 405
Q of Maxwell solid, 115 ice lenses, 303, 458
radioactive elements, 118 ice sheet topography, 449–51
radiogenic, 127 ice wedge polygons, 459–61
tidal dissipation, 114–16, 128 Iceland, 314, 323, 330, 422
heat transfer icy satellites, 5
advection, 116 Ida, 286
conduction, 116, 171 crater chains, 331
cooling of oceanic plates, 120 ignimbrite, 203
conduction time constant, 119 IGY (International Geophysical Year), 447
convection, 112, 116, 171 Imbrium basin, 11, 233
adiabat, 123 impact basins, see impact craters: multiring
adiabatic temperature gradient, 123 basins
boundary layer, 123, 124, 171 impact craters
convection cell dimensions, 122 angle, impact, 228
convective heat flux, 124 atmospheric dispersion, 254
convective velocity, 124 atmospheric entry, 254–55
critical Rayleigh number, 122 atmospheric erosion, 255
internal heating, 122 breccia, 237
Nusselt number, 122, 124–25 breccia lens, 224, 239
parameterized convection, 125 catastrophism, 268, 269
plume, 137, 180 central pits, 229
plume velocity, 125 circularity, 223
Rayleigh number, 123 complex, 224–26, 239, 240, 336–38
stress, 125 rim diameter, 226
viscosity regulation, 126 structural uplift, 225
planetary, 172 concentric craters, 228
ternary diagram, 171, 172 cratering rate
volcanic, 170 Earth, 269
heiligenschein, 312 Mars and Venus, 270
Heim, A., 341, 343 Moon, 269
Hellas basin, 17, 224 satellites of Jupiter and Saturn, 270
Hellas Planitia, 438 cryptovolcanic, 223
Helmholtz, Hermann, 420 degraded, 229
hematite, 297, 298, 304 discovery, 222
hemispheric asymmetry, 35 on Mars, 223
Henry Mountains, UT, 430 downhill creep, 321
Herschel crater, 84, 86 ejecta, 224, 246
Hertz, Heinrich, 91, 94 Archean deposits on Earth, 270
Hesperia Planum, 378 asymmetric, 228
Hevelius Formation, 11 ballistic sedimentation, 247
Himalaya, 436 block size, 246
hoar frost, 293 butterfly-wing pattern, 228, 251
hoobabs, 340 curtain, 247
Hooke, Robert, 55 depth of excavation, 238
Hooke’s law, 56 fluidized, 228, 248, 249, 340
Horton, Robert, 396, 398, 416 provenance, 238
Hugoniot, P. H., 230 secondary craters, 250
Humboldt crater thickness, 245
floor uplift, 155 Venus parabolas, 378, 379
Humorum basin, 98 extinction of life, 271
Hutton, James, 23, 395, 430 herringbone pattern, 226, 250
Huygens lander, 313, 408 landscapes, 255–62

EBSCOhost - printed on 11/3/2023 10:00 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
Index 491

microcraters, 224 jetting, 233

multiring basins, 11, 12, 20, 225, 226–27, 239, modification, 238–44
244 Bingham fluid model, 239
mantle uplift, 244 ring tectonics, 241
ring tectonic theory, 245 planar impact approximation, 231–33
oblique, 251 pressure decay, 233
peak ring, 226, 240, 241 residual velocity, 236
populations, 256–61 seismic shaking, 235–36
cumulative distribution, 257–60, 259, 280 shock metamorphism, 234
dating, 262–66 shatter cones, 234
equilibrium, 261, 263, 264, 280, 281, 282 shock velocity–particle velocity relation, 230
evolution, 261–62, 263, 265 spallation, 235
geometric saturation, 259, 261, 262, 265 vapor plume, 237
incremental distribution, 257 Z-model, 242–43
landscape, 255 inclination, orbital, 3
Late Heavy Bombardment, see Late Heavy intercrater plains, 14
Bombardment Inverness corona, 139
leading/trailing asymmetry, 266–67 Io, 18, 19, 170, 171
Moon post-mare, 259 caldera scarps, 333
production distribution, 260–61, 263, 265, 281 crustal resurfacing, 128
R-plot, 260 lava properties, 214
size-frequency distribution, 258, 268 mountains, 19, 159
time history, 269 resurfacing, 218
rampart craters, 248 sinking lithosphere, 160
scaling, 251–54 sulfur, 19
depth/diameter ratio, 252, 253 sulfur emission, 288
diameter, 252 thrust faults, 159
energy scaling, 252 tidal distortion, 33
melt volume, 253 volcanic plume, 203
secondary craters, 246, 250–51 volcanoes, 19
seismic shaking, 322 iridium, 271
simple, 224, 225, 239, 241 isostasy, see topography: support: isostatic
simple–complex transition, 224 Airy, 89
size-frequency distribution, see impact craters: Pratt, 89
populations
slumping, 337 Jeffreys theorem, 58–60, 62
terraces, 335, 336–38 Jeffreys, Sir Harold, 29, 31, 32, 47, 49, 50, 58, 60, 62,
width, 338 79, 101, 128, 129, 213
transient crater, 224, 237 Jovian planets, 5
volcanism, 270–71
impact mechanics kaolinite, 298
atmospheric erosion, 237 Kaula, William, 34
ballistic sedimentation, 248 Kelvin, Lord, 79, 80, 128, 420, see Thomson,
collapse timescale, 239 William
contact and compression, 232, 233 Kelvin–Helmholtz instability, 420
crater growth rate, 237–38 Kepler, Johannes, 2
detached shock, 236 Kepler’s laws, 2
ejecta Kilauea caldera, 193
curtain, 237 kimberlite pipe, see volcano: maar
equation of state, 230 Köfels landslide, 343
excavation, 232, 233–38
duration, 233 laccolith, 194, 207
Hugoniot equations, 230 Lachenbruch, Arthur, 459, 460
impact-explosion analogy, 230 Lake Bonneville, 419, 420, 426
impedence matching, see Impact mechanics: planar Lake Missoula, 385
impact approximation Lakshmi Plateau, 16
isobaric core, 233 Lampson, C. W., 251

EBSCOhost - printed on 11/3/2023 10:00 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
492 Index

landscape evolution, 428–31 temperature, 106

erosion rates, 429 thickness, 106, 108
slope profiles on Moon, 157
concave upward, 325 thickness, from admittance, 99
convex upward, 325 Venus, 108
straight, 325, 330 Llullaillaco volcano, 344
landslides, see Mass movement:landslides loess, 453
Late Heavy Bombardment, 10, 269–70 longitude, 41
latitude, 41 IAU convention, 43
IAU convention, 43 prime meridian, 43
Laurentide Ice Sheet, 452 long-runout landslide, see mass movement:
lava, 208 landslides: sturzstrom
Bingham yield stress, 211, 212 Lowell, Percival, 143
block, 209 lunar grid, 13, see tectonics: faults: planetary grids
channels on Venus, 415, 419 Lutecia
columnar joints, 209–10, 300 crater chains, 331
effusion rate, 213–14 Lyell, Charles, 430
flow, 208–18 Lyot Crater, 461
edge shape, 210, 211
length, 213 MacDougal crater, 204
levees, 212 Mackin, J. Hoover, 429
mechanics, 210–14 Magellan, 374, 377, 415
thickness, 210 Magma
hornitos, 209 ascent, 187–94
inflation, 209, 214 eruption volume, 193
levee, 212 expansion velocity, 199
pahoehoe, 209 intrusion vs. extrusion, 194
pancake domes, 215, 216 neutral buoyancy, 193, 207
pillow, 194, 208 neutral buoyancy level, 194
plateau, 217–18 percolation flow, 187–88
pumice, 198 percolation time, 187
rheology, 214 physical properties, 183–86
rootless cones, 209 pressure, 191
rubbery surface, 208 quantized volume, 191–92
silica content, 214 role of contact angle, 187
sinuous rilles, 215–17 segregation, 187
thermal erosion, 217, 412 solidification, 197
tubes, 209 solubility of water, 196
tumuli, 209 standpipe ascent model, 188, 189–90
vesicular, 198, 199 subduction zone, 195
viscosity, 213 vapor exsolution, 198
leading/trailing asymmetry, see impact craters: viscosity, 183, 184
populations: leading/trailing asymmetry role of crystals, 185
Leith, C. K., 149 role of water, 183
Libya, 348 water solubility, 196
limestone, 297 magnetite, 298, 299
limonite, 297 Malaspina glacier, 437
lineament, 143, 144 Manicouagan crater, 267
illumination effects, 143 Manning, R., 402
Linné crater, 225, 228 mare basalts, 10
lithosphere, 80, 91, 105 mare ridges, see tectonics: faults: mare ridges
defined by Maxwell time, 92 Mare Serenitatis, 153
definition, 100, 106 maria, lunar, 10
Earth, oceanic, 106 asymmetry, 11
planetary curvature, 158 Mariner 9, 189, 377, 407, 417
shell approximation, 110 Mars, 3, 16–18, 382
stress guide, 105 albedo markings, 377

EBSCOhost - printed on 11/3/2023 10:00 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
Index 493

atmosphere, 17 mechanism, 321

Barchan dunes, 373 debris flow, 340
blueberries, 304 rheology, 340
Borealis Plains, 17 gravity currents, 339–40
catastrophic floods, 405 hydraulic jump, 340
cinder cones, 205 landslides, 326–44
climate, 18 cohesionless material, 327–31
crater clusters, 254 cohesive material, 339
crustal dichotomy, 39 Coulomb failure law, 326
dark halos, 364 maximum cliff height, 333
duricrusts, 308 pore pressure, 327, 329
dust devils, 363, 378 rotational slumps, 335
eskers, 455 slope stability, 334
flattening, 29 sturzstrom, 340–44
geoid, 99 air lubrication, 343
glaciation, 451 domino breccia, 344
glaciers, 435, 438 friction coefficient, 343
global dust storms, 378 low friction, 341–42, 344
grabens, 150 Olympus Mons, 345
hemispheric asymmetry, 35 speed, 341
hypsometric curve, 39, 42 volume, 341
ice wedge polygons, 461 Toreva block, 335, 339
kamikaze effect, 368–69 vertical cliffs, 332–33, 334
mare ridge, 135, 152 rotational slump, 334, 339
megaripples, 369, 370 scree, 330
methane, abiogenic, 298 seismic, 321, 330
obliquity variations, 290 subsidence pits, 330–31
outflow channels, 405, 407 talus cone, 330
permafrost, 458 terracettes, 323
pit chains, 151 turbidity currents, see turbidity currents
polygonal troughs, 461 Masursky, Hal, 189
sapping, 395 Mathilde, 286
shield volcano, 206 Matson, D., 292
spheroidal weathering, 301 Maxwell Montes, 16, 91
stone pavement, 313 Maxwell relation, 173
streamlined channel forms, 409 Maxwell solid, 352
Swiss-cheese terrain, 308, 462 Maxwell time, see rheology
topographic power spectrum, 46 Maxwell, D, 242
topographic reference system, 38 Maxwell, J. C., 76, 352
transverse dunes, 374 Mead crater, 16, 377
underground water, 388 mean radius, 27
vapor-phase advection, 293 Meckering, Australia, 152
water, 18 melting
wind streaks, 370, 377, 378 complex, 179
yardangs, 376 decompression, 174–75
Mars Global Surveyor, 458 eutectic, 178, 179
Mars Orbiter Camera, 417 flux, 174, 180–82, 207
Mars Reconnaissance Orbiter, 458 ice, 176
mascon, lunar, 11, 98 liquidus, 177, 180, 182
role of curvature, 110 peridotite, 181
tectonics, 157, 158 planetary, 171
maskelynite, 234 pressure-release, 174, 180
mass movement rocks, 177
angle of repose, 327–29 role of carbon dioxide, 182
avalanche chutes, 339 role of water, 182, 184
creep, 319–26, 458 solid solution, 178
diffusion model, 324–25 solid solutions, 177–79

EBSCOhost - printed on 11/3/2023 10:00 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
494 Index

melting (cont.) daily, 290

solidus, 177, 180, 182 polar, 290
temperature, 171, 180 tidal evolution, 32
depression by water, 180 topographic power spectrum, 46
ice, 175 topographic reference system, 38
Mercury, 3, 14–15, 104 Moseley, Canon, 320
despinning tectonics, 154 Mount Hadley, 143, 144
flattening, 30 Mt. Fuji, Japan, 194
hot pole, 15, 290 Mt. St. Helens, 197, 204
hypsometric curve, 39, 42 Muller, S. W., 455
ice at poles, 290 multiring basins, see impact craters: multiring
intercrater plains, 218 basins
lobate scarps, 104, 146, 149
magnetic field, 15 nanophase iron, 285
planetary grid, see tectonics: faults: planetary Neptune, 3
grid Neptunian Planets, 5
rotation, 15, 290 Nevin, C. M., 149
temperature, 289 Newporte crater, 267
warm pole, 15 Newton, Isaac, 2, 25
Merrill, R. B., 276 Newton’s gravitational constant, 2, 27
Meteor Crater, AZ, 223, 224, 233, 248 Nile River delta, 427
meteoroid flux, see impact craters: populations: Nippur Sulcus, 137
production distribution Nirgal Valles, 395
meteors, see impact craters: atmospheric North Africa, 348
entry North American glaciers, 436
methane, 298 Novaya Zemlya, 341
Midwest USA rivers, 410 Nusselt number, see heat transfer: convection
Mimas, 84
Miranda, 139 obliquity, 3
coronae, 138 oceanic lithosphere, 160
Mirny station, 450 Odysseus crater, 84, 86
Mississippi River, 37, 410, 413, 416, 417 olivine, 176, 177, 179, 180, 298
Mogollon rim, AZ, 315 Olympus Mons, 17, 189, 205, 335, 376
Mohr, Otto, 144 aureole, 344, 345
molecular velocity, see velocity: molecular Oort cloud, 5
Molokai, 344 Opportunity rover, 304, 363, 408
moment of inertia, 29, 35 opposition effect, 312
principal, 35 opposition surge, 311
Mono Craters, 215 orbits, planetary, 3
Moon, 3, 4, 10–14 orders of relief, 25, 45
center of figure offset, 34 Orientale basin, 11, 225, 226, 239, 244
cinder cones, 205 orthoclase, 297
hemispheric asymmetry, 35 Ovda Regio, 129
horizon glow, 322 oxbow lake, 406, 411
hypsometric curve, 39, 42
impact velocity, average, 285 Pallas, 5
lineaments, 155 Panama Canal, 333
mare basins, 217 parachutes, 349
opposition effect, 311 pattern formation, 366
orbital evolution, 114 patterned ground, 459–62
origin by giant impact, 269 Penck, Walther, 430
regolith, see regolith, lunar penitents, 307
rille percolation flow, 187–88
arcuate, 150 peridotite, 180, 181
sinuous, 216, 412 period, orbital, 2
straight, 150 permafrost, 454–62
temperature active zone, 457, 458

EBSCOhost - printed on 11/3/2023 10:00 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
Index 495

block fields, 462 Q, quality factor, 114

definition, 455 Kelvin–Voight solid, 115
ice–wedge polygons, 459–61 Maxwell solid, 115
icings, 457 Qattara Depression, 377
Mars, 458
north–south asymmetry, 462 racetrack playa, 427
patterned ground, 459–62 rainsplash, 397
pingos, 458 Ramberg, Hans, 136
solifluction, 458 Rayleigh distillation, 187
talik, 456 Red Wing Creek crater, 267
thermal regime, 456 regelation, 434, 439, 444–46
thermokarst, 462 regolith
permeability, 187, 388, 390, 391 bearing strength, 277
depth dependence, 392 biological importance, 294
fractures, 391 definition, 276
Perrault, Pierre, 393 density, 277
Phobos, 341 depth, 283
crater chains, 151, 330, 331 growth rate, 281
orbital decay, 114 formation, 280–84
Phoenix lander, 304, 458, 461 gardening, 285
phreatic explosion, see volcano: maar heat transfer, 293
Pincacate Mountains, Sonora, 204 lateral mixing, 277, 284–85
pingo, 458 lunar, 13, 276–88, 294
planetesimal, 267 agglutinate, 285
plastic material, 64 density, 278
plate tectonics, 22, 107, 160–61, depth, 277
195 gardening, 284
oceanic plates, 120 grain size, average, 285
outer rise, 95 growth rate, 281
strain rates, 128 meteoritic contamination, 277
subduction zones, 90, 160 nanophase iron, 285
playas, 426–27 radiation, 287
Playfair, John, 399 thermal conductivity, 291
plumes, 195 maturity, 279, 286
Pluto, 3 neutrons, thermal, 288
plutonic rocks, 207 particle size, 277
porosity, 388, 390 radiation effects, 286–88
depth dependence, 392 stone pavements, 313, 314, 315
Mars, 392 deflation, 313
Moon, 392 upward migration, 314
Powell, J. W., 385 surface texture, 311–16
Prandtl, L., 354 elephant hide, 277, 320, 323
Pratt isostasy, 87, 88 fairy-castle structure, 311–12
pressure, 54 mudcracks, 315–16
lithostatic, 59 opposition effect, 312
principal stresses, 55 temperatures
Prinz, 216 depth dependence, 290, 291
Prometheus Patera, Io, 206 skin depth, 290
Prometheus Plume, Io, 199 solid-state greenhouse, 292
Ptolemaeus crater, 155 thermal inertia, 293
pumice, 199 thermal rectification, 291,
Pumpelly, Raphael, 363, 364 292
pyrite, 299 vapor-phase advection, 293
pyroclastic flow, 199, 203 reorientation of satellites by impacts,
pyrolite, 179 267
pyroxene, 176, 179 Reynolds number, 351, 404
pyroxene absorption bands, 279 rheidity, see rheology

EBSCOhost - printed on 11/3/2023 10:00 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
496 Index

rheology, 56, 64 semimajor axis, 2

Bingham, 185–86, 239, 338, 340 semiminor axis, 2
Bingham fluid, 450 serpentinite, 298
Bingham stress–strain rate, 186 shape, planetary
Bingham yield stress, 186, 210 center of figure offsets, 35
creep, 74–80, 439 flattening, 27, 28
dislocation climb, 439 Kaula scaling law, 34
dislocations, 439, 440 oblateness, 27
Griggs experiment, 75 spherical, 26
ice, see glacier flow tidal deformation, 30
non-ice solids, 444 triaxial, 30, 31
time dependence, 75 Sharp, R. P., 365, 366, 381
Kelvin experiment, 80 Sharpe, C. F. S., 319, 320, 323
Kelvin–Voight model, 80 Shasta terrain, 341
lava, 186 Shewmon’s Rule of Thumb, 444
lithosphere, see lithosphere Shields criterion, 397
Maxwell time, 78, 127 shock metamorphism, see Impact mechanics: shock
planetary structure, 105–07 metamorphism
pseudoviscous flow, 74 Shoemaker, E. M., 222, 248, 262, 276, 281, 282, 385
rheidity, 78 Shreve, R., 343, 418
solution creep, 445 Siberian flows, 195
stress–strain relation, 57 siderite, 299
viscoelasticity, 76–80 Sierra Nevada, 401, 430
viscous, see viscosity Sif Mons, 194
rhyolite, 184, 197 silica, 298
Ries Crater, 223, 229, 241 sill, 194
Ringwood, A. E., 179 sinuous rilles, 215–17
rock cycle, 395 slopes
rock glaciers, 438–39 angle of repose, see mass movement: angle of
rotation, principal axis, 36 repose
excited states, 36 angle on the Moon, 328
Explorer I, 36 definition, 40
Mars, 36 Hurst exponent, 40
Vesta, 36 median differential slope, 41
roughness, see topographic roughness profiles, see landscape evolution: slope profiles
Roza flow, 214 vertical cliffs, 332–33
rubble piles, 67 SNC meteorites, 235
snow metamorphism, 436
Sagan, Carl, 1, 24 Socompa volcano, 344
San Andreas Fault, 149, 154 soil
San Francisco Bay, 401 creep, 326
sand castles, 331 electrostatic levitation, 322
Sand Hills, NE, 453 mechanics, 320–23
sand transport, see eolian processes terracettes, 323–24
sapping, 394–95 thermal creep, 322
Saskatchewan glacier, 441 lunar, see regolith, lunar
Saturn, 3 Mars
schizosphere, 141 oxidation, 286
Schröter, J. H., 223 peroxides, 286
Schröter’s Rule, 245 terrestrial, 276, 311
Schröters Rille, 411 formation rate, 310
Schumm, S., 413 horizons, 295, 311
Sculpured Hills, 277 organic carbon, 310
Sedna Planitia, 419 thickness, 310
Seine River, 393 solar wind, 286
Selu Corona, 138 solid-state greenhouse, 292
semiarid USA rivers, 410 South Platte River, 413

EBSCOhost - printed on 11/3/2023 10:00 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
Index 497

South Pole Aitken basin, 11 in a stack of thin layers, 59

space weathering, 279 in an elastic layer, 59
spectral power, 45 limit theorems, 61
spherical harmonic functions, 44 location of maximum, 60
spheroidal weathering, see weathering: physical: second stress invariant, 55
spheroidal shear from flowing water, 397
spider structure, 234 tectonic, see tectonics: stress sources
Spirit rover, 363, 368, 369 Strindberg crater, 225
stagnation pressure, 254 sublimation, see weathering: sublimation
steins latent heat
crater chains, 331 carbon dioxide, 307
stishovite, 234 ice, 306
Stokes, G. G., 351 subsurface water, 388–95
Stokes’ law, 351–52 aquifer, 390
Stone Mountain, GA, 301 capillary fringe, 388
strain, 52–53 Darcy law, 390
stream Darcy velocity, 391
networks, 416 depletion of aquifers, 393–94
strength drawdown cone, 389
brace construction, 70 Earth abundance, 392
brittle–ductile transition, 73 energy per unit mass, 389
broken rock, 67 hydraulic conductivity, 390
Byerlee’s law, 72 hydraulic head, 389
cohesion, 73, 331 percolation flow, 390–92
Coulomb, 326 piezometric surface, 388–90
effect of pore pressure, 73, 327, 329 sapping, 394
Frenkel limit, 66 springs, 389, 395
frictional, 67–69, 328 water table, 388–90
ideal plasticity, 70 Sudan, 348
impact crater cohesion, 338 Sudbury crater, 267
linear law, 71 Sunbury, PA, 129
long-term, 333 suncups, 307
Lundborg, 70, 71 Surtsey island, 422
of rocks, 74 Surveyor Program, 276
planetary profiles, 81 Sykes crater, 204
pressure dependence, 74
role of defects, 66 tectonics
size dependence, 142 boudinage, 135–36
strength profile, 80–82 Ganymede, 136
tensile, 73 brittle–ductile transition, 140
ultimate, 64, 65, 66 buckling, 133
stress, 53–55 elastic, 130–32
beneath loads, 61 embedded layer, 131
buckling, 129, 131 Europa, 133
buoyancy, 111 Indian plate, 133
Castigliano’s Principle, 60 multilayer, 132
convection, 112, 125 viscous, 132
deviatoric, 55 convection, 112
flexural, 93 definition, 104
from gravity anomalies, 98 diapirs, 136–38
from phase changes, 111 domes, 156
from surface loads, 110 ductile flow, 140, 141
from volume change, 111 faults, 145
Griffith crack, 142 Anderson theory, 143–49
horizontal, at failure, 148 brittle–ductile transition, 141
in a despun planet, 109 conjugate planes, 145
in a planetary interior, 63 definition, 143

EBSCOhost - printed on 11/3/2023 10:00 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
498 Index

tectonics (cont.) terrae, lunar, 10

detachment, 134, 152–53 terrestrial planets, 5
dip, 147–48 Terzaghi, K., 326
flower structure, 153 Tethys, 84
graben dip, 151 Texas Canyon, AZ, 301
graben refraction, 157–58, 159 Tharsis Rise, 17, 29, 105, 156, 195, 217
grabens, 152, 227 tectonics, 157
high fluid pressure, 153 volcano heights, 189
mare ridges, 135, 152, 153 Theophilus crater, 225
strike-slip component, 152 thermal conduction, see heat transfer: conduction
mascon, lunar, 158 thermal conductivity, 106, 117, 290, 291
Mohr circles, 144 lunar regolith, 291
normal, 146, 147 thermal convection, see heat transfer:convection
planetary grid systems, 155 thermal diffusivity, 117, 214, 290
Riedel shears, 152 thermal expansion coefficient, 111, 173
slickenside, 145 thermal inertia, 293
strain ellipsoid theory, 149 thermal rectification, 291
strength paradox, 153–54 thermokarst, 462
strike-slip, 149 tholeiitic basalt, 184
strike-slip fault paradox, 157 Thomson, J. J., 302, 411
thrust, 146–48 Thomson, William, 79, 420
thrust fault, 149 tidal deformation, 31
transpression, 152 Moon, 32
flexural domes, 155–57 tidal dissipation, 30, 114–16, 128
folding, 128–35 Earth rotation, 31
fault bend, 134 Q, 114
fractures, 139 tidal flexing, 19
Griffith cracks, 142–43 Timocharis crater, 244, 246
Io’s sinking lithosphere, 160 Titan, 4, 20–22, 348, 382, 408, 431
joints, 139, 141–43, 300 atmosphere, 20
sheeting, 301 channels, 407
lineaments, 143, 155 frost-shattering, 304
localization, 139–40 glaciers, lack of, 435
mascons, 157 linear dunes, 374
necking, 135 longitudinal dunes, 375
neutral sheet, 133 methane, 21, 455
planetary expansion or contraction, 111 methane lakes, 425
plate tectonics, 160–61 seasons, 290
plates, 107–08 stone pavement, 313
rates, 127–28 subsurface methane, 394
role of planetary curvature, 110 surface temperature, 21
schizosphere, 141 Xanadu, 21
shrinking Earth, 128 Titus–Bode law, 3
stick-slip, 140 topographic diffusion coeficient, 325
strain-softening, 140 topographic roughness, 40–41
stress sources, 112 Kaula’s second law, 46
despun planet, 109 topography
reoriented planet, 109 collapse without strength, 51
tidal deformation, 109 dynamic, 90–91
tensile cracks, 142 friction limit, 68
tensile fractures, 142 isostatic compensation, 89
Venus, 108 maximum in solar system, 69
tektites, 255 strength limit, 63
Tempel 1, 289 support, 58, 100
temperature flexure, 91–93
effective, 289 isostatic, 87–90
terracettes, 323–24 viscous relaxation, 82–87

EBSCOhost - printed on 11/3/2023 10:00 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
Index 499

of craters, 84, 85 pyrite, 299

on Venus, 86 sulfur, 299
Trans-Neptunian Objects, 3 wind streaks, 377
Triton, 4 yardangs, 377
geysers, 205 vesicles, 199
seasons, 290 Vesta, 5, 169, 180
Tsilokovsky crater, 255 crust formation, 188
tuff magma percolation time, 188
airfall, 202, 203 Viking, 286, 313, 368, 417, 458
welded, 202, 203 viscosity, 56, 57–58, 351
tuff ring, see volcano: maar basaltic melts, 183
Tunguska explosion, 254 Bingham, 186
turbidity currents, 339, 340, 428 dependence on crystal content, 185
Tycho crater, 14, 226, 285, 336 derived from topographic relaxation, 83
Tyndall, John, 434 effective, 75
gas, 352
Ubehebe crater, 204 pressure independence, 352
undrained depressions, 395, 428 granitic melts, 183
Uranus, 3 Jeffreys equation, 213
Urey, Harold, 299 magma, 183
Utopia basin, 17 Newtonian, 57
of paint, 185
Valhalla basin, 20, 227, 245 of the Earth, 83, 84
Valles Marineris, 17, 342 of the Moon, 84
Varenius, 169 self-regulation, 125–26
Variegated Glacier, 446 silicate melts, 184
velocity volcanic ash, 198, 199, 206, 378
escape, 6 volcanic bombs, 199
molecular, 6 velocity, 201
volcanic ejecta, 200–02 volcanic eruptions
Venera 13, 313 deep-sea, 197
Vening-Meinesz, Felix, 108, 152 explosive, 195–204
Venus, 3, 15–16, 170 fire-fountain, 197
“snow” on mountaintops, 299 gas-rich, 196
carbon dioxide, 299 Impact-induced, see impact craters: volcanism
channels, 407, 415, 419 mechanics, 194–204
cinder cones, 205 quiescent, 197
coronas, 138, 157 Venusian, 197
arachnid, 157 volume rate of, 170
nova, 157 volcano
crater clusters, 254 accordant summits, 189, 190
dark halos, 364 caldera, 204, 207
fold belt, 129 cinder cone, 205
geoid anomalies, 99 lunar, 205
hypsometric curve, 39, 42 composite, 206–07
isostatic compensation, 99 ejection height, ballistic, 199
pancake domes, 215, 216 ejection height, buoyant, 199
parabolas, 378 ejection speed, 200–02
shield volcano, 205, 206 eruption mechanics, 200, 203
splotches, 255 explosive eruption, 198
stone pavement, 313 fire-fountain, 205
topographic power spectrum, 46 Hawaiian, 193
topographic reference system, 38 laccolith, see laccolith
topography, 16 maar, 204–05
transverse dunes, 374 magma chamber, 193, 194, 205,
weathering 207
oxidation of iron, 298 plume, 203

EBSCOhost - printed on 11/3/2023 10:00 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
500 Index

volcano (cont.) spheroidal, 302

buoyant, 203 sulfur, 299
pyroclastic flow, 199, 339 Urey reaction, 299
sector collapse, 344 desert varnish, 309–10
shield, 205–06 duricrust, 308–09
spatter rampart, 206 physical, 300–06
volume chemical potential, 305
expansion coefficient, 120 cracks, 300
expansion of water, 302 crystallization pressure, 303, 304
sphere, 27 frost shattering, 303, 304
triaxial ellipsoid, 27 grus, 305
Vostok station, 450 insolation weathering, 306
Vredefort crater, 267 pressure of reaction, 305
rock disintegration, 306
Warrego Vallis, 400 salt weathering, 302, 305
Wasatch Mountains, 420 sheeting joints, 301
water spheroidal, 301
expansion on freezing, 302 stress corrosion cracking, 301
waves, 420–26 supersaturation, 304
amplitude, 422 thermal stress, 306
breaking, 423 radiation, 287, 288
energy, 421 space weathering, 279, 285–87
energy flux, 421 sublimation, 307, 308
generation, 420 Weertman, Johannes, 188
gravity, 420 Wessex Cliff, 277
group velocity, 421 Westerly granite, 140
particle orbits, 422 Wiesbach, J., 403
phase velocity, 420, 421 Wild 2, 307, 333
refraction, 423–26 Willis, Bailey, 128, 129
weathering, 293–311 wind tunnel, 348
cavernous, 308–09 wollastonite, 299
honeycomb, 309 wrinkle ridges, see tectonics: faults: mare ridges
tafoni, 308 wüstite, 298
chemical, 295–300
carbon dioxide, 299 Xanadu, 21, 413
carbonation, 299 Xanthe Terra, 405
clathrates, 297
hydration, 297 Yakima Fold Belt, 152
hydrolysis, 298 Yellowstone, 195
iron oxidation, 295 Yukon Territory, 462
oxidation, 295, 299 Yuma, AZ, 313
reactions, 296 Yuty crater, 249
serpentinization, 298
solution, 297 Zagros Mountains, 445

EBSCOhost - printed on 11/3/2023 10:00 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
All rights reserved. May not be reproduced in any form without permission from the publisher, except fair uses permitted under U.S. or applicable copyright law.

Figure 1.4 The topography of the Moon referenced to a sphere with a radius of 1737.4 km. Data
were obtained from the Lunar Orbiter Laser Altimeter (LOLA) that was flown on the mission Lunar
Reconnaissance Orbiter (LRO). The color-coded topography is displayed in two Lambert equal-area
images projected on the near and farside hemispheres. Courtesy Mark Wieczorek, 7 August 2010.

volcanoes. Olympus Mons rises to the upper left. The gigantic trough of Valles Marineris extends to
the right center. The northern lowlands and the Borealis plains dominate the upper half of the right
hemisphere. The deep circular basin to the lower left is Hellas and the smaller basin near the center
is Utopia. NASA/JPL image PIA02820.

HU
60°
?

M
?

BO
?

LD
50°
50°

TI
AN
? ?

UM
?
40°
40°
?

?
?
30°

?
? 30°

?
? IMBRIUM

?
20° Apollo 20°
SERENITATIS
15 + Ma
?

?
P1 P2 P3 + Apollo 17
CRISIUM

?
10°
? 10°

?
In ?
Tr
? Luna
?

0° ? 20 0°
?
+Apollo 14

ii
?

Smyth
Grimaldi Fe

?
FB ? ?
+

?
?
? Apollo
NECTARIS
10° 16 10°

?
BK
ORIE

Nu WA
HUMORUM
NTAL

20° ? 20°

?
?
E

?
30°
30°

MV
Sc
40° hill 40°
ME er- ale
ND Zu
str

cch
Au

EL ius
-R 0° 10° 0° 10° 20° 30°4
?

YD PH °40° 30° 2 0°5 50°

Bailly60° 50 0°60
?

BE °70
°

RG °
0
°7

South 60°
80

Pole- Aitken
80

60°
°

70° 70°
NEARSIDE 80° AG 80°
SOUTH POLE
WEST EAST
LONGITUDE LONGITUDE

Figure 1.3 The major multiring basins of the Moon and the extent of their ejecta deposits are
indicated. Curved lines indicate major rings. Panel (a) is the Moon’s nearside and (b) is its farside.
Blue indicates the deposits of the youngest (Imbrian) basins, yellow-orange Nectarian, dark brown
Pre-Nectarian. After Plates 3A and 3B in Wilhelms (1987).

EBSCOhost - printed on 11/3/2023 10:00 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
(b) NORTH POLE
80° ? 80°
70° 70°
60° 60°
IMB

?
RIU

?
M
50° HUMBOLDTIANUM 50°
?

? Coulomb

?
40° Sarton 40°
BIRKHOFF

30° 30°
MOSCOVIENSE
LORENTZ
CRISIUM
20°

V
20°

ONO
HAR

?
?

H-S
?
?

LF ?

LIC
10° ?
10°
UND
?

G
FRE
?

?
RU
EV

SP
0° AK

L
0°

Z
RO
ii

MENDELEEV ?
yth

?
KO
Sm

HE
ORIENTALE
Keeler-
?

?
Heaviside en 10°
?
10° Aitk
TS e-
Pol ?

?
20° 20°
h

MILNE
ut
So

Ingenii

?
30° 30°
GW APOLLO
?
Au
st
40° rale Poincaré 40°
Planck
RG
?

BE
° 50°160° 170°
180° 170° 160° 150 1 YD
50° 0°140 1
° 40 13 -R
?

SR 13 SCHRODINGER ?
° 0°1
20 DEL
0 °

° EN
12
0°

60° M
11
11

10°
0°

60°
0°
10

70° 70°
80°AG 80° FARSIDE
SOUTH POLE
EAST WEST
LONGITUDE LONGITUDE

Figure 1.3 (cont.)

EBSCOhost - printed on 11/3/2023 10:00 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
(a)

EBSCOhost - printed on 11/3/2023 10:00 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
(b)

Figure 1.6 (cont.)

EBSCOhost - printed on 11/3/2023 10:00 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
Figure 1.8 The Galilean satellites of Jupiter as imaged by the Galileo spacecraft. In order from left
to right are Io, Europa, Ganymede, and Callisto. Volcanoes dominate Io, Europa is covered with an
ice shell, Ganymede’s surface is a patchwork of bright young and dark old terrain, and Callisto is an
undifferentiated mixture of ice and rock. NASA/JPL/DLR image PIA01400.

Figure 1.9 Global view of Titan’s surface from the VIMS spectrometer aboard the Cassini spacecraft.
This false color composite is constructed from three wavelengths in the infrared that penetrate Titan’s
hazy atmosphere (1.3 μm is shown in blue, 2 in green and 5 in red). The dark region in the center of
the image is named Xanadu and may be the site of a large ancient impact. NASA/JPL/University of
Arizona image PIA09034.

EBSCOhost - printed on 11/3/2023 10:00 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
EBSCOhost - printed on 11/3/2023 10:00 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
Figure 1.10 Topographic map of the Earth from NOAA. ETOPO1 is a 1 arc-minute global relief model of Earth’s surface that integrates land topography
and ocean bathymetry. It was built from numerous global and regional data sets (Amante and Eakins, 2009).
Figure 6.17 Night-time IR thermal image of the 6.9 km diameter Martian crater Gratteri, located
at 17.8° S, 202° E. The dark streaks are created by secondary impact craters that extend up to 500
km from the crater center. In images of this type, dark regions are cold and emit little IR radiation
because they have low thermal inertia, indicating that the streaks are composed of fine-grained
material compared to their surroundings. The overall image measures 545 x 533 km across. THEMIS
image courtesy of Phil Christensen. NASA/JPL/ASU.

Figure 7.14 Cavernous weathering surface in sandstone. Image is about 1 m across, showing deep
pits and columns that have become detached from the mass of the rock behind them. This variety is
often called honeycomb weathering from its appearance. Canyonlands National Park, Utah, USA.
Photo courtesy of Ingrid Daubar-Spitale, 2010.

EBSCOhost - printed on 11/3/2023 10:00 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
(a)

(b)

EBSCOhost - printed on 11/3/2023 10:00 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
(a)

EBSCOhost - printed on 11/3/2023 10:00 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
Figure 8.17 (a) a landslide triggered by the 2002 Denali earthquake in Alaska, looking west toward the divide of the Black Rapids and Susitna glaciers.
Image courtesy Dennis Trabant and Rod Marsh, USGS. For panel (b), see p. 342 in text.
(b)

Figure 9.14 Yardangs on Mars. For panel (a), see p.376 in text. (b) Panoramic view of a soft rock
unit dissected into yardangs south of Olympus Mons. The three flat regions in the fore-, middle-, and
background measure about 17 x 9 km in this oblique view. Mars Express HRSC image, orbit 143,
ESA/DLR/FU Berlin.

Figure 10.9 Ares Vallis is one of the large outflow channels on Mars. This image shows the transition
between Iani Chaos region to the lower left and the plains of Xanthe Terra to the top (north). The
spurs between the individual channels have been shaped into crude streamlined shapes by massive
floods of water. 10 km scale bar is at lower right. Mars Express images by ESA/DLR/FU Berlin.

EBSCOhost - printed on 11/3/2023 10:00 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use
Figure 10.18 Liquid methane lakes near the North Pole of Titan imaged by the Cassini synthetic
aperture radar. Dark regions are smooth lake surfaces and brighter regions are the surface. Intermediate
brightness levels near the lake shores indicate some radar return from the lake bottoms. Image is
centered near 80° N and 35° W and the strip is 140 km wide. Smallest details are about 500 m across.
The radar strip was foreshortened to simulate an oblique view from the west. Image PIA09102.
NASA/JPL/USGS.

(b)

Figure 11.12 Ice-wedge polygons on Mars. For panel (a), see p. 461 in text. (b) Troughs are spaced
1.5 to 2.5 m apart near the Phoenix landing site at 68°N and 26° W. They are believed to represent
ice-wedge activity. On Earth, ice wedges may also be manifested by either ridges or troughs. NASA/
JPL/University of Arizona.

EBSCOhost - printed on 11/3/2023 10:00 AM via YDUQS. All use subject to https://www.ebsco.com/terms-of-use

The Geology of The Terrestrial Planets
100% (2)
The Geology of The Terrestrial Planets
322 pages
Introduction to Marine Geology
No ratings yet
Introduction to Marine Geology
31 pages
Volcanic Plumes PDF
100% (2)
Volcanic Plumes PDF
254 pages
San Andreas Fault
No ratings yet
San Andreas Fault
40 pages
Guide To Geophysical Equations
100% (1)
Guide To Geophysical Equations
296 pages
Hamblin EarthDynamics
100% (2)
Hamblin EarthDynamics
766 pages
Satellite Remote Sensing: Monitoring Ground Surface Movement & Application On Tailings Dams
No ratings yet
Satellite Remote Sensing: Monitoring Ground Surface Movement & Application On Tailings Dams
20 pages
Geodynamics
100% (2)
Geodynamics
863 pages
Stellar Interiors - Physical Principles, Structure, and Evolution
100% (2)
Stellar Interiors - Physical Principles, Structure, and Evolution
527 pages
2019 Book VolcanicUnrest
100% (2)
2019 Book VolcanicUnrest
313 pages
Turcotte D.L., Schubert G.-geodynamics-CUP (2014) GEODYNAMICS
100% (11)
Turcotte D.L., Schubert G.-geodynamics-CUP (2014) GEODYNAMICS
657 pages
The Trace Fossil MesozoicCenozoic PDF
100% (3)
The Trace Fossil MesozoicCenozoic PDF
497 pages
Bohmvitense e Introduction To Stellar Astrophysics Volume 1
100% (1)
Bohmvitense e Introduction To Stellar Astrophysics Volume 1
255 pages
Ammonitoid
100% (1)
Ammonitoid
615 pages
(Ebook) Planetary Surface Processes (Cambridge Planetary Science) by H. Jay Melosh ISBN 9780521514187, 0521514185 Instant Download Full Chapters
No ratings yet
(Ebook) Planetary Surface Processes (Cambridge Planetary Science) by H. Jay Melosh ISBN 9780521514187, 0521514185 Instant Download Full Chapters
166 pages
Planetary Surface Processes Cambridge Planetary Science 1st Edition H. Jay Melosh Download
100% (6)
Planetary Surface Processes Cambridge Planetary Science 1st Edition H. Jay Melosh Download
71 pages
Planetary Surface Processes Cambridge Planetary Science 1st Edition H. Jay Melosh
No ratings yet
Planetary Surface Processes Cambridge Planetary Science 1st Edition H. Jay Melosh
51 pages
Planetary Surface Processes Cambridge Planetary Science 1st Edition H. Jay Melosh Available All Format
100% (17)
Planetary Surface Processes Cambridge Planetary Science 1st Edition H. Jay Melosh Available All Format
92 pages
Remote Compositional Analysis Techniques For Understanding Spectroscopy Mineralogy and Geochemistry of Planetary Surfaces 1st Edition Janice L. Bishop Download
No ratings yet
Remote Compositional Analysis Techniques For Understanding Spectroscopy Mineralogy and Geochemistry of Planetary Surfaces 1st Edition Janice L. Bishop Download
111 pages
Dokumen - Pub - Remote Compositional Analysis Techniques For Understanding Spectroscopy Mineralogy and Geochemistry of Planetary Surfaces 978 1 107 18620 0
No ratings yet
Dokumen - Pub - Remote Compositional Analysis Techniques For Understanding Spectroscopy Mineralogy and Geochemistry of Planetary Surfaces 978 1 107 18620 0
633 pages
Planetary Landscapes
100% (3)
Planetary Landscapes
297 pages
Planetary Geology
No ratings yet
Planetary Geology
223 pages
Remote Compositional Analysis Techniques For Understanding Spectroscopy Mineralogy and Geochemistry of Planetary Surfaces 1st Edition Janice L. Bishop Full Chapters Instanly
100% (1)
Remote Compositional Analysis Techniques For Understanding Spectroscopy Mineralogy and Geochemistry of Planetary Surfaces 1st Edition Janice L. Bishop Full Chapters Instanly
113 pages
Remote Compositional Analysis Techniques For Understanding Spectroscopy Mineralogy and Geochemistry of Planetary Surfaces 1st Edition Janice L. Bishop Latest PDF 2025
No ratings yet
Remote Compositional Analysis Techniques For Understanding Spectroscopy Mineralogy and Geochemistry of Planetary Surfaces 1st Edition Janice L. Bishop Latest PDF 2025
159 pages
Planetary Tectonics Cambridge Planetary Science 1st Edition Thomas R. Watters PDF Available
100% (8)
Planetary Tectonics Cambridge Planetary Science 1st Edition Thomas R. Watters PDF Available
148 pages
NSO 2018 Solar System Overview
No ratings yet
NSO 2018 Solar System Overview
8 pages
Rossi Van Gasselt 2017 - Planetary Geology
No ratings yet
Rossi Van Gasselt 2017 - Planetary Geology
478 pages
Jupiter and Saturn 2nd Edition Linda T. Elkins-Tanton PDF Download
No ratings yet
Jupiter and Saturn 2nd Edition Linda T. Elkins-Tanton PDF Download
112 pages
Martian Surface Composition Mineralogy and Physical Properties Jim Bell PDF Version
100% (2)
Martian Surface Composition Mineralogy and Physical Properties Jim Bell PDF Version
176 pages
An Introduction To The Solar System 3 3nbsped 1108430848 9781108430845 - Compress
100% (1)
An Introduction To The Solar System 3 3nbsped 1108430848 9781108430845 - Compress
444 pages
Planetary Crusts Their Composition Origin and Evolution 1st Edition S. Ross Taylor PDF Download
No ratings yet
Planetary Crusts Their Composition Origin and Evolution 1st Edition S. Ross Taylor PDF Download
49 pages
Planetary Rings A Post Equinox View Esposito PDF Version
100% (18)
Planetary Rings A Post Equinox View Esposito PDF Version
196 pages
The Earth and The Moon, Revised Edition (2010)
100% (1)
The Earth and The Moon, Revised Edition (2010)
327 pages
Planetary Rings A Post Equinox View Esposito L. Online Reading
100% (1)
Planetary Rings A Post Equinox View Esposito L. Online Reading
159 pages
Planetary Rings A Post Equinox View Esposito L. Online Version
100% (1)
Planetary Rings A Post Equinox View Esposito L. Online Version
82 pages
Burbine - Asteroids - Astronomical and Geological Bodies - Cambridge (2017) 410pp
100% (1)
Burbine - Asteroids - Astronomical and Geological Bodies - Cambridge (2017) 410pp
410 pages
Jupiter The Planet Satellites and Magnetosphere Larry W. Esposito Latest PDF 2025
100% (1)
Jupiter The Planet Satellites and Magnetosphere Larry W. Esposito Latest PDF 2025
111 pages
Saturn in The 21st Century Cambridge Planetary Science Series Number 20 1st Edition Kevin H. Baines PDF Download
No ratings yet
Saturn in The 21st Century Cambridge Planetary Science Series Number 20 1st Edition Kevin H. Baines PDF Download
162 pages
Earth - Evolution of A Habitable World (2nd Ed) (Gnv64)
100% (2)
Earth - Evolution of A Habitable World (2nd Ed) (Gnv64)
344 pages
Martian Surface Composition Mineralogy and Physical Properties Jim Bell PDF Download
No ratings yet
Martian Surface Composition Mineralogy and Physical Properties Jim Bell PDF Download
52 pages
The Cambridge Guide To The Solar System 2nd Edition - Kenneth R. Lang PDF
100% (6)
The Cambridge Guide To The Solar System 2nd Edition - Kenneth R. Lang PDF
503 pages
Astronomy by OpenStax Hardcover Version Full Color 1 1st Edition Andrew Fraknoi: Foothill College Digital Version 2025
No ratings yet
Astronomy by OpenStax Hardcover Version Full Color 1 1st Edition Andrew Fraknoi: Foothill College Digital Version 2025
160 pages
Planetary Science
No ratings yet
Planetary Science
6 pages
Byrne2020 Planetary Geology
No ratings yet
Byrne2020 Planetary Geology
15 pages
Ebooks File Planetary Rings A Post Equinox View Esposito L. All Chapters
100% (8)
Ebooks File Planetary Rings A Post Equinox View Esposito L. All Chapters
51 pages
Interiors of The Planets Cambridge Planetary Science Old 1st Edition A. H. Cook Full Access
100% (4)
Interiors of The Planets Cambridge Planetary Science Old 1st Edition A. H. Cook Full Access
144 pages
The Solar System Saturn S Satellites X Ray and Gamma Ray Astronomy Appendixes Index 2nd Edition David G. Fisher Newest Edition 2025
100% (14)
The Solar System Saturn S Satellites X Ray and Gamma Ray Astronomy Appendixes Index 2nd Edition David G. Fisher Newest Edition 2025
136 pages
Martian Surface Composition Mineralogy and Physical Properties Jim Bell Download
No ratings yet
Martian Surface Composition Mineralogy and Physical Properties Jim Bell Download
83 pages
Saturn in The 21st Century 1st Edition Kevin H. Baines Online Reading
100% (4)
Saturn in The 21st Century 1st Edition Kevin H. Baines Online Reading
167 pages
1783earth: Portrait of A Planet 6th Edition (Ebook PDF) Available Instanly
100% (9)
1783earth: Portrait of A Planet 6th Edition (Ebook PDF) Available Instanly
79 pages
(Ebook PDF) Earth: Portrait of A Planet Sixth Edition by Stephen Marshak PDF Download
100% (2)
(Ebook PDF) Earth: Portrait of A Planet Sixth Edition by Stephen Marshak PDF Download
47 pages
Levin - Origin of The Earth and Planets - Foreign Languages Publishing House (1956)
No ratings yet
Levin - Origin of The Earth and Planets - Foreign Languages Publishing House (1956)
83 pages
Earth Evolution of A Habitable World 2nd Ed Edition Jonathan Irving Lunine Instant Download
No ratings yet
Earth Evolution of A Habitable World 2nd Ed Edition Jonathan Irving Lunine Instant Download
52 pages
Jupiter The Planet Satellites and Magnetosphere Larry W. Esposito PDF Download
No ratings yet
Jupiter The Planet Satellites and Magnetosphere Larry W. Esposito PDF Download
52 pages
Earth Surface Processes Landforms and Sediment Deposits 1st Edition John Bridge Download PDF
No ratings yet
Earth Surface Processes Landforms and Sediment Deposits 1st Edition John Bridge Download PDF
77 pages
Planetary System Report Suprabha Mukhopadhyay
No ratings yet
Planetary System Report Suprabha Mukhopadhyay
2 pages
Planetary Crusts Their Composition Origin and Evolution 1st Edition S. Ross Taylor Digital Version 2025
100% (4)
Planetary Crusts Their Composition Origin and Evolution 1st Edition S. Ross Taylor Digital Version 2025
157 pages
Earth As An Evolving Planetary System 1st Edition Kent Condie All Chapters Available
100% (1)
Earth As An Evolving Planetary System 1st Edition Kent Condie All Chapters Available
143 pages
Planetary Geology
No ratings yet
Planetary Geology
30 pages
(Encyclopedia of Earth Science) Richard A. Simpson, Jon M. Jenkins (Auth.) - Encyclopedia of Planetary Science-Springer Netherlands (1997)
No ratings yet
(Encyclopedia of Earth Science) Richard A. Simpson, Jon M. Jenkins (Auth.) - Encyclopedia of Planetary Science-Springer Netherlands (1997)
942 pages
A New Approach For Serving 1984
No ratings yet
A New Approach For Serving 1984
3 pages
A Practical Model For 1980
No ratings yet
A Practical Model For 1980
7 pages
Conception of Gifted 2003
No ratings yet
Conception of Gifted 2003
27 pages
Geographies of Outer Space
No ratings yet
Geographies of Outer Space
24 pages
An Introduction To The Data and Tools of Planetary Geomorphology
No ratings yet
An Introduction To The Data and Tools of Planetary Geomorphology
9 pages
Design Thinking Project On The Case of TVL Graduates
No ratings yet
Design Thinking Project On The Case of TVL Graduates
37 pages
The Little Big Loo: Rethinking Public Toilets
No ratings yet
The Little Big Loo: Rethinking Public Toilets
13 pages
Datasheet - 50kVA - 800-415V (1) (28-Mar-22)
No ratings yet
Datasheet - 50kVA - 800-415V (1) (28-Mar-22)
3 pages
Internship Student List 2023 24
No ratings yet
Internship Student List 2023 24
4 pages
Lab Commands
0% (1)
Lab Commands
9 pages
Singapore's Pro and Anti-Natalist Policies
No ratings yet
Singapore's Pro and Anti-Natalist Policies
12 pages
Acknowledgments
No ratings yet
Acknowledgments
2 pages
Cut Paper Drawing Lesson Plan
No ratings yet
Cut Paper Drawing Lesson Plan
4 pages
Climate Change Impact on Indus Delta
No ratings yet
Climate Change Impact on Indus Delta
53 pages
Brand Preferrence of Packaged Milk
82% (17)
Brand Preferrence of Packaged Milk
73 pages
Kunci Determinasi Alismataceae
No ratings yet
Kunci Determinasi Alismataceae
6 pages
Cash Management for Small Retailers
No ratings yet
Cash Management for Small Retailers
151 pages
Secondary Math Flowchart 2018
No ratings yet
Secondary Math Flowchart 2018
1 page
Buku Murid Bahasa Inggris - Work in Progress - Sports Events Buku Panduan Guru Bahasa Inggris SMA Kelas X Unit 2 - Fase E
No ratings yet
Buku Murid Bahasa Inggris - Work in Progress - Sports Events Buku Panduan Guru Bahasa Inggris SMA Kelas X Unit 2 - Fase E
24 pages
PRESIDENTIAL DECREE NO. 1706 and For Other Purpose.: Section 1. Section 2
No ratings yet
PRESIDENTIAL DECREE NO. 1706 and For Other Purpose.: Section 1. Section 2
9 pages
Halogen-Free Safety Cables Guide
No ratings yet
Halogen-Free Safety Cables Guide
2 pages
Harsha Khorana's Creative CV
No ratings yet
Harsha Khorana's Creative CV
34 pages
The Conversion of Constantine PDF
No ratings yet
The Conversion of Constantine PDF
18 pages
Amigurumi Lamb Pattern Guide
100% (4)
Amigurumi Lamb Pattern Guide
6 pages
Dawkins - 2003 - Regional Development Theory Conceptual Foundation
No ratings yet
Dawkins - 2003 - Regional Development Theory Conceptual Foundation
42 pages
Government: of Tamil Nadu
No ratings yet
Government: of Tamil Nadu
31 pages
VMW VCP DCV Certification Preparation Guide
No ratings yet
VMW VCP DCV Certification Preparation Guide
5 pages
Building Confidence
No ratings yet
Building Confidence
2 pages
Unit 4 Environment
No ratings yet
Unit 4 Environment
32 pages
Obstetric Complications in Osisioma-Ngwa
No ratings yet
Obstetric Complications in Osisioma-Ngwa
9 pages
Ac Performance
No ratings yet
Ac Performance
1 page
7605 - Sales Force Management
No ratings yet
7605 - Sales Force Management
144 pages
Impressions Clothesaccessories Ws 718635
No ratings yet
Impressions Clothesaccessories Ws 718635
3 pages
MAN01 01 001 Appendix3 v5
No ratings yet
MAN01 01 001 Appendix3 v5
22 pages
Untitled
No ratings yet
Untitled
10 pages

Planetary Surface Processes

Uploaded by

Planetary Surface Processes

Uploaded by

Copyright © 2011. Cambridge University Press. All rights reserved.

Planetary Surface Processes

h. jay melosh is Distinguished Professor of Earth and Atmospheric Science at Purdue

Francis Nimmo, Sara Russell

Books in the series

1. Jupiter: The Planet, Satellites and Magnetosphere†

12. Protoplanetary Dust: Astrophysical and Cosmochemical Perspectives

Plane tary Sur face P ro ce sse s

Cambridge, New York, Melbourne, Madrid, Cape Town,

Cambridge University Press

© H. Jay Melosh 2011

This publication is in copyright. Subject to statutory exception

First published 2011

Printed in the United Kingdom at the University Press, Cambridge

Library of Congress Cataloguing in Publication data

ISBN 978-0-521-51418-7 Hardback

Cambridge University Press has no responsibility for the persistence or

2.1.4 A scaling law for planetary figures? 34

2.2.3 Where are we? Latitude and longitude on the planets 41

2.3 Spectral representation of topography 44

Further reading 100

4.2 Sources of tectonic stress 108

4.2.1 External sources of tectonic stress 108

5.1.4 Segregation and ascent of magma 187

5.2.3 Volcanic surface features 204

5.3 Lava flows, domes, and plateaus 208

6.8.3 Leading/trailing asymmetry 266

6.9.5 Impact-induced volcanism? 270

6.9.6 Biological extinctions 271

8.2.2 Cohesive materials c > 0 331

9.1 Sand vs. dust 349

Further reading 431

11.2 Flow of glaciers 439

11.2.1 Glen’s law 440

The grand tour

1.1 Structure of the Solar System

of interest to us is the inclination of the ellipse to a reference plane, conventionally chosen

1.1.1 Major facts of the Solar System

Table 1.1 Planetary orbits

Semimajor Semimajor Orbital

Mercury 0.241 0.387 0.55 0.2056 7.004

1.1.2 Varieties of objects in the Solar System

1.2 Classification of the planets

1.2.1 Retention of planetary atmospheres

Table 1.2 Physical data on terrestrial planets and major moons

Mercury 58.6 2439 5430 2.78 2 700

1.2.2 Geologic processes on the terrestrial planets and moons

Geologic Process: Planetary Diameter*:

1 km 10 km 100 km 1000 km 10 000 km

Tectonics _ __ __ ___ _______________________

Volcanism __ ___ __________________________

Impact Cratering _________________________________________________ _

Mass Wasting _______________________________ _______________________

Interaction with Surface Fluids: ______

Surface Fluids: Airless Trace Dense

Internal Heat Transfer Conduction Convection

gest and most diverse planet in the Solar System, Earth.

1.3 Planetary surfaces and history

1.3.1 The Moon

BIRKHOFF Sarton 40°

20° MILNE 20°

30° GW APOLLO 30°

present) to the 1540 km diameter Caloris Basin.

loss of hydrogen to space.

unearthly degree of strength to the near-surface material.

1.3.5 Jupiter’s Galilean satellites

1.3.7 The Earth

is created at mid-ocean spreading centers. This crust is only 5 to 10 km thick, compared to

2008), as have comets (Festou et al., 2004).

The shapes of planets and moons

tures of planetary topography and its geometric properties.

Table 2.1 Physical characteristics of large bodies in the Solar System

Mean Equatorial Sidereal

Mercury 2 439 3.303 x 1023 5430 2.78 58.75 days

2.1 The overall shapes of planets

2.1.1 Non-rotating planets: spheres

2.1.4 A scaling law for planetary figures? 34

2.2.3 Where are we? Latitude and longitude on the planets 41

2.3 Spectral representation of topography 44

Further reading 100

4.2 Sources of tectonic stress 108

4.2.1 External sources of tectonic stress 108

5.1.4 Segregation and ascent of magma 187

5.2.3 Volcanic surface features 204

5.3 Lava flows, domes, and plateaus 208

6.8.3 Leading/trailing asymmetry 266

6.9.5 Impact-induced volcanism? 270

6.9.6 Biological extinctions 271

8.2.2 Cohesive materials c > 0 331

9.1 Sand vs. dust 349

Further reading 431

11.2 Flow of glaciers 439

11.2.1 Glen’s law 440

Tectonics _ _ _____________________

Volcanism _ __________________________

Mass Wasting _________ _