Elements of Physical Hydrology

2
Elements of Physical Hydrology

SECOI'ID EDlTtO'll

George M. Hornberger

Patricia L. Wiberg

Jeffrey P. Raffensperger

Paolo D'Odorico

Johns Hopkins University Press Baltimore

3
© 2014 Johns Hopkins University Press
All rights reserved. Published 2014
Printed in the United States of America on acid-free paper
98765432 1

Johns Hopkins University Press

2715 North Charles Street
Baltimore, Maryland 2121 8-4363
www.press.jhu.edu

Library of Congress Cataloging-in-Publication Data

Elements of physical hydrology I George M. Homberger . . . [et al.]. - 2nd
ed.
p. cm.
Includes bibliographical references and index.
ISBN 978-1 -42 14-1373-0 (hardcover : alk. paper) - ISBN 978-1-4214-
1396-9 (electronic) - ISBN 1-4214-1373-6 (hardcover: alk. paper)­
ISBN 1-4214-1396-5 (electronic) I. Hydrology. I. Hornberger, George
M.
GB661.2.E44 2014
551 .48-<1c23 201 3036530

A catalog record for this book is available from the British Library.

Special discounts are available for bulk purchases of this book. For more
il!formaliull. please colllact Special Sales at 4/0-5/6-6936 or
[email protected].

Johns Hopkins University Press uses environmentally friendly book

materials, including recycled text paper that is composed of at least 30
percent post-consumer waste, whenever possible.

4
Contents

Preface

[ The Science of Hydrology

2 Precipitation and Evapotranspiration
3 The Basis for Analysis in Physical Hydrology: Principles of Fluid
Dynamics
4 Open Channel Hydraulics
5 Catchment Hydrology: Streams and Floods
6 Groundwater Hydraulics
7 Groundwater Hydrology
8 Water in the Unsaturated Zone
9 Ecohydrology: imcractions between Hydrological Processes and the
Biota
10 Catchment Hydrology: The Hillslope-Stream Continuum
II Water, Climate, Energy, and Food

Appendixes
1 Units, Dimensions, and Conversions
2 Properties of Water
3 Basic Stalistics in Hydrology
Answers to Example Problems
Glossary
References
Index

5
Preface

The world population is expected to grow to 9 billion by 2050. Food

production will have to grow even faster than the population because, as
countries develop and achieve a higher standard of living, diets lend to
include more animal protein and thus there is a concomitant increased
consumption of grain by animals. Agriculture consumes the greatest
amount of fresh water of all sectors, so we anticipate increased stress on
water resources as a result of increased food production. The growth in
population will also result in increases in the use of energy, which entails
the use of water as well. Furthermore, these changes will occur amidst
large-scale changes in the use ofland resources (e.g., increased
urbanization) and changes in climate caused by the burning of fossil fucls.
11 is clear that dealing with water resources issues will be a key activity for
regions, nations, and the global village as these changes unfold. Arguably,
university students need to acquire knowledge about the hydrological
cycle just to bc informcd citizens of the world in the coming decades.
This book is based on the premise that students of environmental
science must learn the quantitative physical basis of hydrology if they are
to appreciate the scientific approach to understanding observed phenomena
and the basis for achieving solutions to water resources problems. The text
is not a catalog of observations nor is it a compendium of applied
tcchniqucs of engineering hydrology. Rathcr, it prescnts a basic coveragc
of physical principles and how these allow one to grasp the essential
elements of hydrological processes. One tenet that we hold is that an
understanding of fundamental fluid mechanics is essential to the study of
hydrology. Our aim is to provide an integrated coverage of flows of water
on and bcneath the Earth's surface with the underpinning ofa knowledgc
of basic fluid mechanics.
This book originated from lectures in a coursc on Physical Hydrology at
the University ofYirginia that began more than 25 years ago. Following
publication of the first edition of our book in 1998, many colleagues
elsewhere told us that they found it useful, but that there were a few topics
that were underreprescnted. Thus, in addition to updating material
throughout the book, in this second edition we include a good bit of new
material, including a new chapter on Ecohydrology and a completely new
final chapter on water-climate-energy-food interactions. The changes

6
notwithstanding, the book is still designed to accompany an undergraduate
course in physical hydrological science. It is aimed at upper-level
undergraduates majoring in environmental or Earth scicnces. The coverage
presupposes a modest background in calculus and physics.
We use several conventions in an attempt to make the book "user
friendly." Each chapter has introductory and concluding remarks that seek
to place the material presented in the chapter in the context of some
contemporary environmental issue. Tenns that appear in boldface at their
first occurrence are found in the Glossary. Terms that appear in italics
deserve emphasis. Supporting material is contained in three appendixes: a
review of units and dimensions (Appendix I ); a tabulation of certain
properties of water (Appendix 2); and a review of some elementary
statistical concepts (Appendix 3) .
Over the ycars we have benefited from ideas and data shared by many
colleagues and students. We are grateful to all who have contributed to our
education, but will refrain from attempting to produce a comprehensive
list. We do want to acknowledge specifically colleagues who either have
taught the course at UVA that stimulated the first edition of the book in the
past or who are currently in the teaching rotation for the course: John
Albertson, Keith Beven, Keith Eshleman, John Fisher, Peter German,
Aaron Mills, Matt Reidenbach, and Todd Scanlon. In addition, we
especially thank Margot Bjoring for the preparation of the figures for this
edition of the book.

7
Elements of Physical Hydrology

8
1 The Science of Hydrology

1.1 Introduction

1.2 Hydrology and Water Resources

1.3 The Hydrological Cycle

1.4 The Water Budget

1.4.1 The global water budget
1.4.2 The catchment water budget
1.5 Concluding Remarks
1.6 Key Points
1.7 Example Problems
1.8 Suggested Readings

1.1 Introduction

The role of waler is cCOIral to most natural processes. Water transports

sediment and solutes to takes and oceans, thereby shaping the landscape.
The global energy balance is influenced strongly by the high capacity of
waler for storing thennal energy and the large amount of heat required to
change water from liquid to vapor and vice versa. The abundance of water
in the atmosphere and oceans makes it an important regulator of climate.
Water vapor is the most important of the greenhouse gases. Life depends
on water.
Hydrology, literally "water science," encompasses the study of the
occurrence and movement of water on and beneath the surface of the
Earth, the properties of water, and its relationship with the living and
material components of the environment. Ultimately, many hydrological
questions involve the transport of dissolved nutrients, energy, sediment, or
contaminants. The starting point for investigations of transport must be the
physical processes of water movement.
Hydrological science has an important place in discussions of natural
resources. Water resources, especially freshwater resources, are the subject

9
of intense scmtiny and speculation. In arid regions, the fair allocation and
wise use of freshwater resources are signifieant challenges facing
governments and people, affecting relations between nations, states, cities,
and individual users. As a resource, water appears unlimited. However, the
twentieth century saw a tremendous growth in the use of water, as well as
an increase in the threat of its contamination, and the trends have
continued in the new millennium.
Hydrological science has aspects related to "curiosity-driven" questions
and to "problcm-driven" questions. The first aspect relates to questions
about how the Earth works, and specifically about the role of water in
natural processes. The second relates to using scientific knowledge to
provide a sound basis for the proper use and protection of water resources.

1.2 Hydrology and Water Resources

Hydrological science has both curiosity-driven and problem-driven origins

that stretch back to antiquity (Biswas, 1972). Many of the great ancient
philosophers of Greece and Rome speculated on hydrological phenomena.
VitTUvius, writing during the second halfofthe first century BC, often is
credited with first recognizing that groundwater is derived primarily from
infiltration of rain and snowmelt, rather than upwelling of subterranean
water from great depths. By applying hydrological and hydraulic
principles, the ancients constructed great hydraulic works such as the
ancient Arabian wells, the Persian kanats, the Egyptian and Mesopotamian
irrigation projects, the Roman aqueducts, the ancient water civilizations of
Sri Lanka, and the Chinese irrigation systems, canals, and flood control
works. During the Renaissance, Leonardo da Vinci wrote what is likely the
earliest complete statement of the hydrological cycle, including the notions
of condensation and evaporation and the dissolution of rock minerals (after
Eagleson, 1970):

Whence we may conclude that the water goes from the rivers to the sea and
from the sea 10 the rivers, thus constantly circulating and returning, and that
all the sea and rivers have passed through the mouth of the Nile an infinite
number of times. The conclusion is that the saltiness of the sea must
proceed from the many springs of water which, as they penetrate the earth,
find mines of salt, and these they dissolve in part and carry with them 10 the
ocean and other seas, whence the clouds, the begetters of rivers, never carry
it up.

During the eighteenth and nineteenth centuries, hydraulic experiments

10
flourished and, until about 1950, pragmatic considerations dominated
hydrology. Primarily due to the development and availability of digital
computing, thcoretical approaches in hydrology have incrcasingly allowed
hydrological theories to be subjected to rigorous mathematical analysis.
Freshwater resources are needed to meet the needs of humans,
livestock, commercial enterprises, agriculture, mining, industry, and
thermoelectric and hydroelectric power. Ln today's world, the necessity of
solving water-supply problems has become obvious, with many regions
exhibiting signs of looming water shortages (Figure 1 . 1 ).
Most of the human consumption of freshwater resources is associated
with food production. (We use cOl/slimp/ion to refer to water that is used in
ways that return it to the atmosphere rather than to a stream, river, or
groundwater. For example, thermoelectric power plants may withdraw
large quantities of water from a river for cooling, but then return the bulk
of that water to the stream, albeit at a higher temperature.) Relative to the
consumption of water for food production, drinking, household, and
industrial uses of water are overall smaller (Figure 1.2). We need much
more water to produce the food we eat than the amount of water we use for
drinking or other activities. The global water crisis is more likely causing
hunger than thirst. A report from the office of the U.S. Director of National
Intelligence, Global Water Security
(http://www.dni.gov/index.php/aboutlorganization/national-intelligence­
council-nie-publications) concludes: "Between now and 2040, fresh water
availability will not keep up with demand absent more effective
management of water resources. Water problems will hinder the ability of
key countries to produce food and generate energy, posing a risk to global
food markets and hobbling economic growth." The need for improved
water management, based on scientific knowledge, is of critical
importance in the coming decades (Jury and Vaux, 2007).

II
 r
·f ....
·

• •

o o

tt! ; ,/
asins:'"
;
Water stress indicator
�o""'!�'o'b
.. ., ......
Sou"*,, Smakht"" Ailvenga
.nd DOlI, 20001.

Figure 1.1 Water resources are highly stressed in many parts of the world.
The water stress indicator is the fraction of available water appropriated
for use by humans. Fractions greater than one indicate the use of fossil
groundwater.
http://www.grida.no/graphicslib/deta iI/water-scarcity-index_14f3

12
 o Agriculture

o Household
9.6%
o Industrial

85.8%

Figure 1.2 Relative contributions of agriculture. household, and industrial

consumption to the human appropriation of freshwater resources.
Data from Hoekstra and Chapaga;n 2008.

Although the rate of increase in water usc in some arcas has slackened
over the past several decades, the total consumption of water globally has
continued to increase (Figure 1.3 ). Population growth, depiction or
deterioration of freshwater resources, and changing demands (mosl
notably the tendency to adopt more water-demanding diets in emergent
countrics,-i.c., an increase in meat consumption) will tend to further stress
water resources in the future for many countries. Falkenmark et al. (2007)
indicate their view of the severity of the issue. "We are on the verge of a
new and more serious era of water scarcity, and it is clear that we will face
increasingly complex challenges. Water supply to different sectors will
become more challenging as supplies of blue water (e.g. water in rivers
and aquifers) become overstretched, while a scarcity of green water (e.g.
water in the soil) will limit food and biomass production."
World population is expected to increase in the next 50 years, stressing
water resources. Issues related 10 the qualifY of water supplies have
occupicd an increasingly important niche in hydrology. Ii is estimated that

13
80% of all diseases and over one-third of all deaths in developing countries
result from the consumption of contaminated water. Provision of basic
sanitation and water treatment in much of the world is still lacking. [n
1980, the United Nations launched the International Drinking Water
Supply and Sanitation Decade, with the goal of clean water and sanitation
services to those without them. Despite enonnous effort, expense, and
progress, at the close of the decade 1.8 billion people still had no access to
sanitation services, and nearly 1 .3 billion people still laeked access to
clean water. Population growth wiped out the progress achieved through
this effort (Gleick, 1993).

2000
Europe
�
North America
�
-

>-
,

Africa
� 1500
E Asia
..
-
South America
<:
0 Australia and Oceania
a.
'-

1000
-

E
"
fI)
<:
0
u
Q)
�
500
C\I
-

;:

o
1900 1920 1940 1960 1980 2000 2020
Figure 1.3 Water consumption grew during the twentieth eemury and is
projected to continue increasing in the future.
Data from Shiklomanov 1999.

In the developed world, the availability of safe drinking water and

adequate sewage treatment is taken for granted, but other water quality
issues abound. These include eutrophication of surface waters (delCierious
clTects duc to excess nutrient supply), contamination of groundwater with
a variety of organic compounds and metals, and the acidification of surface
waters from acid rain. The ability of streams, rivers, lakes, and estuaries to
dilute contaminants to safe levels and to purify the water through natural

14
processes (their assimilative capacity) depends on the quantity of water
flowing in them.
Understanding surface-water now is a requisite for water quality
studies. Similarly, knowlcdge ofsubsurfacc flow is nccessary for
understanding the movement of pollutants underground. To predict or
evaluate the effects of liquid-waste disposal in deep injection wells or the
use of solid waste as fill for reclamation of strip mines, one must
appreciate the mcchanics of water flow in rock and soil. The massive
cffort now undcr way in the Unitcd States to clean up sitcs wherc
groundwater has been contaminated represents another challenge for
hydrological science.
Major floods are the most dramatic and visible of hydrological
phenomena. The number of people afef cted by floods and the number of
lives lost havc increased in recent decades (Figure 1.4). With pressure for
increased usc of floodplains, the prediction and control of floods remain
among the most important applications of hydrology.
Addressing hydrological challenges, such as those related to floods or
groundwater contamination, requires a finn understanding of the basic
principles of the physics and chemistry of water. Hydrological science
uses the fundamentals of the basic sciences and mathematics to develop
explanations (modcls) of obselVed phenomena. One of the basic problems
in hydrology is a description of water motion. One goal of physical
hydrology is to identifY the paths of water movement on and beneath the
surface of the Earth. Using physical theory and associated mathematical
models, hydrologists seek to describe quantitatively the motion of water in
the nafUral environment. As Jury and Vaux (2005) put it, "There is little
question that science must play a critical role in fonning a succ·essful
solution to the world's emerging water problems." They go on to conclude
that "a new science of sustainability will be needed if the prospects for
managing and solving the world's emcrging watcr problcms are to bc
bright." Knowledge of physical hydrology will be a critical component of
this new science.

15
 1010
8 Number aHected
8 Number killed

"
10'
-

a.
0
"
a.
-
0 10'
"
�

.0
E
"
Z
10'

10'
1975 1980 1985 1990 1995 2000
Year
Figure 1.4 Losses during river flooding have trended upward over the past
several decades.
Redrawn from Jonkman 2005.

1.3 The Hydrological Cycle

The motion of water can be described at many different scales. The

fundamental concept of hydrology is the hydrological cyclc--the global­
scale, endless recirculatory process linking water in the atmosphere, on the
continents, and in the oceans. We can think of this recireulatory process in
tenns of reservoirs or compartments that store water (the oceans,
atmosphere, etc.) and the movement of water between them. Within the
various compartments of the hydrological cycle, water can be stored in any
one of three separate phases or states: gas (vapor), liquid, or solid. For
example, water in the atmosphere can exist as vapor (the concentration of
water vapor is expressed as humidity), in liquid (cloud droplets, rain
drops), or in solid phase (ice crystals, snowflakes). Similarly, all three
phases of water can be found on and below the land surface. Movement of
water from one compartment to another can occur in any of the three
phases. For example, the movement of water between the oceans and

16
atmosphere occurs in vapor phase (evaporation from the ocean surface),
liquid phase (rain onto the ocean surface), and solid phase (snowfall onto
the ocean surfacc).
Solar energy drives the hydrological cycle; gravity and other forces
also play important roles. The dynamic processes of water vapor fonnation
and transport of vapor and liquid in the atmosphere are driven largely by
solar energy. Precipitation and the now of water on and beneath the
Earth's surface are driven primarily by gravity. Within partially dry soil,
gravitational and other forces are responsible for the movement of water.
The hydrological cyclc can be considered to "start" anywhere, but let us
consider atmospheric water first (Figure 1.5). As hydrology is concemed
mainly with water at or near the Earth's surface, from our point of view
the dominant process involving atmospheric water is the precipitation of
water on the land surface. The portion of the precipitation that reaches the
land surface as solid precipitation (mostly snowfall) can be retained
temporarily on vegetation or ground surfaces, or accumulate in seasonal
snowpacks or in pennanent snowpacks known as glaciers. Considering
liquid precipitation (rain), a portion also can be retained temporarily on
vegctation surfaces or in surface depressions, a portion enters into the soil
(infiltration) and a portion nows over the land surface first into small
rivulets and ultimately into larger streams and rivers. This last process is
called sllIface rtlnoff; which can be augmented by runoff during periods of
snowmelt (snowmelt runofl). The portion of rainfall that infiltrates into the
soil can also follow one of scvcral paths. Somc of the watcr el'aporates
from the soil and somc is returned to thc atmosphere by plants
(frallspira/iol/). We often refer to the total evaporation and transpiration
from vegetated land surfaces as evapotranspiration. The remaining water
continues to move downward through the soil and recharges thc saturatcd
portion of the subsurface, becoming groundwater. At lower elevations,
groundwater discharges into streams and rivers or directly to the ocean
(groundwater rllnoj}). Water evaporates from the surface of the oceans and
thereby replcnishes the water in the atmosphere. Thus we have returned to
thc particular compartmcnt that we considered first, atmosphcric water.
Looking at the relative magnitude of thesc watcr nuxes (Figure 1.6), wc
observe that over the continents evapotranspiration is on averagc smallcr
than precipitation. The fraction of precipitation exceeding
evapotranspiration contributes to surface and groundwater runoff.
Conversely, over the oceans evaporation exceeds precipitation. Therefore,
precipitation alonc is not sufficient to rcmovc from the atmosphere above
the oceans all the water vapor coming from ocean evaporation. In fact, the
atmospheric circulation transports part of this water vapor on land masses,

I7
thereby allowing continental precipitation to exceed the evapotranspiration
ratc.

Snow Rain
• • • • • • I I f II
• • • • • •
• • • • • • I I II
• • • • • • I .. II
I' I I

11111 11111
Transpiration
Evaporation
I I I I I
I I I I I

• • •

Infiltration Surface runoff

11111
•

Evaporalion
· . . . .

I I I I I
-

• • • • • • • • •
Groundwater runoff
• • • • • • • • • • •

. .
• • • • • • • • • • • •
. . . . . . . ,
. . .
·
· . . . . . . . . , . , . . . .
. .

• • • • • • • • • • • • • • • • • •
• • • • • • • • • • • • • • • • • • • • • • • • • •
. , . .

.
, . . . . . . . . . . . - - -
. . . . . . . . . " . . . . . . . . , . " .
. .
. . . . . . . .
· . . . . . . . . . . . . . . . - . .

• • • • • • • • • • • • • • • • • • • • • • • • • • • • •

• • • • • • • • • • • • • • • • • • • • • • • • •
• • •

Figure 1.5 Mechanisms of water movement within the hydrological cycle.

Water movement from the atmosphere to the oceans and continents occurs
as precipitation, including rain, snow, sleet, and other fOnTIs. On the
contincnts, watcr may be stored temporarily but cvcntually returns to thc
oceans through surface and groundwater runoff or to the atmosphere
through evapotranspiration.

It is instructive to analyze the water cycle at different scales. We have

seen that at the global scale waler reaches Earth's surface as precipitation.
This water is transported over land and in the occans until it is returned to
the atmosphcre as watcr vapor in thc processcs of evaporation and
transpiration. Similarly, water vapor is transported by Ihe atmospheric
circulation until it contributes to precipitation (Figure 1 .7a). At a local
scale, it is unlikely that the same water molecules that evapotranspire from
an area of, say, a fcw hcctares will contribute to the precipitation on the
same area. Most of Ihe local precipitation is contributed by water brought
into the area by atmospheric transport; the excess of rainfall with respect to
evapotranspiration (locally) generates runoff (Figure I . 7b). At a regional
scale (e.g., a fcw thousand square kilometers) precipitation can bc
contributed both by moisture coming from regional evapotranspiration and
by water vapor brought into the region by atmospheric transport (Figure

18
1 .7c). Thus at the regional scale the water cycle may have an important
internal component, whereby part of regional evapotranspiration
contributes 10 a substantial part of the regional prccipitation. Known as
precipitation recycling, this phenomenon is usually quantitatively
expressed in tenns of the recycling ratio, which is the fraction of the total
regional precipitation that is contributed by water evapotranspiring from
the same region. The recycling ratio is overall greater during the growing
season, when evapotranspiration is more intense, and it increases with the
region size. Studies on precipitation recycling provide important insights
into the impact of terrestrial vegetation on the water cycle. This research
has shown how the removal of forest vegetation (e.g., deforestation) may
affect regional precipitation by changing (decreasing) transpiration and
precipitation recycling.

li
'''' Ocean
So 39 moisture
-0

li
0.­

�o
- 0 •
0

et
-

i�
-

• �
· . .
0

et°
• • •

Surface
38
• • • •

discharge
• • • •

• • • • •

• • • • • • •

• • • • • • • •
Groundwater 1
•

•
•

•
•

•
•

•
•

•
•

•
•

•
•

•
•

•
•

•
discharge
•
· . . . . .
• • • • • • • • • •
· . . . . . . . . . . . . . ,

• •
.

•
.

•
.

•
.

•
. .
• •
. .
• .�
.
- ��
• •
. .

• •
.
• • --c---:--,-
-
• �
.
. . . . . . . " . , . , " , . , " " , . , '
· . . . . , . . . . . . . . . . . . . . . . . , . .
. . . . . . . . . .

. . . .
, . . . , . . . . .,. . , . . . .

• •
.
• • • • • • • • • • • • • • • • • • • • • • • • •
. . . . . . . . , . , . . . . . . . . , . . . ,

Figure 1.6 Flows within the hydrological cycle. Units are relative to the
annual precipitation on the land surface ( 1 00 I 19,000 km3 yr-!). Black =

arrows depict flows to the atmosphere, gray arrows depict flows to land or
oceans, and blue arrows indicate lateral flows.
Data from Maidmcllt 1993.

1.4 The Water Budget

The hydrological cycle can be described quantitatively by applying the

19
principle of conservation of mass, which often is referred to as a water
balance or water budget when used in this way. A simple statement of
conservation of mass for any particular compartment (usually referred 10
as a COil/rot volume) is that the time rate of change of mass stored within
the compartment is equal to the difference between the inflow rate and the
outflow rate. For example, if we are adding two grams of water to a bucket
every minute and one gram of water is leaking out each minute, then the
mass stored within the bucket is increasing at the rate of one gram per
minute. Symbolically, we can write this as:

20
Atmospheric tr ;';:::
�.:---...,
sport

Precipitation
--1 ----- - -- ----
Evapotranspiration

,...._
.. _.=.:e:
o c�a:.:
n;;& land transport
(a)

Atmospheric transport
---- .. II __.
.... --
...... ./'
Evapotranspiration Precipitation

Runoff
(b)

Atmospheric transport
- --- .... ....
,... ... --
Recycled precipitatio

Precipitation

Runoff
Evapotranspiration

(e)

21
Figure 1.7 The water cycle at the global (a), local (b), and regional (c)
scales.
Modified after Eltahir and Bras 1996.
dAl
=1'-0'' (1.1)
d,
where M= mass within the control volume [M]; t = time [T] ; I' = mass
inflow rate [M il]; and 0' = mass outflow rate [M il]. The expressions
in square brackets are the mass-length-time dimensions associated with the
defined quantity; for example, the dimensions of I' are mass per time or M
,I. (See Appendix I for a discussion of units, dimensions, significant
figures, and unit conversions.)
In many instances, the density of water can be taken as approximately
constant and the conservation law expressed in tenns of volume. The lenns
involving mass in Equation 1 . 1 can be expressed in tenns of density times
volume and density can then be canceled from both sides of the equation.
Thus, Equation 1.1 can be rewritten:

dV =/-0
- ' (1.2)
d,
where V = volume of water within the control volume [L3]; I = volume
inflow rate [L3,1]; and 0= volume outflow rate [L3,1].

IA.I The global water budget

We can construct a global water budget by applying the principle of mass

conservation (Equation 1 .2), using the continents as our control volume.
The quantity V is then the volume of water stored on or within the
continental land masses. Inflow is precipitation and outflow consists of
evapotranspiration (evaporation and transpiration combined) and runoff
(both surface water and groundwater). Note that in addition to ignoring
density variations we must express the rate of evaporation or transpiration­
outflows of water vapor from the continents to the atmosphere-in "liquid
water equivalent," or LWE, units. Otherwise, density is varying (water
vapor is much less dense than liquid water), and mass, rather than volume,
is the conserved quantity.
If we consider only average (llIlIIwl conditions for this water budget,
the dVldt lenn in Equation 1 .2 becomes negligible. That is, over a period
of years the average amount of water stored as ice (ice caps and glaciers),
as surface water (rivers and lakes), and as subsurface (groundwater) does

22
not change significantly. Ovcr much longer time periods such as centuries
or millennia this may not be true if there is a dramatic shift in climatic
conditions. Ir thcre is no change in storage over time, we say that the
system is at steady slare. For any given control volume at steady state, a
completely general water budget equation can be written (using bars over
the terms to indicate that they are annual average quantities):

dV _
- = 0= p+r" +rj{l-r..... -rl:" -et. (1.3)
d,

where V = average volume of water stored, and assumed to be constant; p

= average precipitation rate; ;::. = average surface water inflow rate; '�oj =
average groundwater inflow rate; r..., = average surface water outflow rate;
'"xo, = average groundwater outflow rate; and el = average
evapotranspiration rate. All tenns in the equation have dimensions of
volume per time [L3•1]. For the continents, we will simplify Equation
1.3 by neglecting the inflows and outflows of groundwater, because they
tend to be very small. We also will neglect surface water inflows, because
surface water flows from the continents to the oceans, and will refer to
surface water outflows as runoff, "5' With these simplifications, Equation
1.3 becomes:

dV
p r
' O (104)
d,
�

- "" - - et""

0'

(1.;)

where p = average precipitation rate [L3•1]; ,:� = average surface runoff

rate [L3•1]; and eJ = average evapotranspiration rate [L3•1].
To quamify the global hydrological cycle, we can examine the relative
sizes of the various storage compartments and the magnitudes of the
various flows to and from these compartments (Figure 1.6). Nearly 97% of
all water on the Earth is stored in the oceans, while only about 0.001 % is
stored in the atmosphere (Table 1 . 1 ). Considering only freshwater (defined
as having a concentration of total dissolved solids less than 0.5 parts per
thousand and considered potable), which accounts for about 2.5% of the
total storage, 69.6% is contained in the polar icecaps and glaciers while
30.1 % is contained in groundwater. The freshwater contained in lakes,
streams, rivers, and marshes represents only 0.26% of all freshwater and
0.008% (80 drops in a million) of all water on Earth. Another useful
concept for thinking about the size of the various reservoirs in relation to

23
the flows of water into and out of them is the residence time. The
residence time, Tr [T], is a measure of how long, 011 average, a molecule
of water spends in that reservoir before moving on to another reservoir of
the hydrological cycle. The residence lime is easily calculated for systems
at steady state, when the inflow and outflow rates are idemical:
y
T, =-. (1.6)
I

The residence time has dimensions of time, because volume divided by

volume per time is lime. The residence time provides an indication of the
lime scales for flushing a solute out of that particular reservoir. Watcr in
the oceans has a residence time approaching 3000 years, while water in the
atmosphere has a residence time of only 0.02 years or about 8 days; the
residence time of water in rivers is 0.05 years or about 1 7 days (Table 1 . 1 ).
The water budget for all the land areas of the world is: p = 800 mm , ,
:
,

= 3 1 0 mm, and eJ = 490 mm (Figure 1 .6). (Note that we now are referring
to thc volumes divided by the areas being considercd. It is sometimes more
convenient to use depth rather than total volume, because the volumes can
be quite large; also, we are probably more familiar with the statement, "20
mm of precipitation was recorded at Smith Airport," than "Smith Airport
received 20,000 m3 of water".) On average, 39% of precipitation to the
continents runs ofT and 6 1 % is rcturncd to the atmosphere through
evapotranspiration. In other words, the runoff ratio V)m is equal to 0.39.
The balance is, of course, affected by many topographic and climatic
factors and the budgets for individual continents can be quite different
from the average (Table 1 .2). The budget for North America is p = 670
mm I� = 290 mm and el = 380 mm Thus, in North America 43% of
, , .

precipitation runs off and 57% evapotranspires on average.

Table I.t. Sizes and residence times for major reservoirs in the
hydrological cycle

24
 Pe rcenuge Percentage of
Volume (kml) Residence time (yr)
of total freshwater

Water in land areu 47,971,710 3.5

lakes

Fresh 91,000 0.007 0.26 (All surface water:1

Saline 85,400 0.006 •.,

Rivers 2,120 0.0002 0.006

Marshes 11,470 0.0008 0.03

Soil moisture 16,�00 0.0012 0.05

Groundwater (All subsurface water:)

fresh 10,530,000 0.76 30.1 20,000

saline 12,870,000 0.93

Biological water 1,120 0.0001 0.003

Icecaps and glaciers 24,364,100 1.76 69.6

Atmosphere 12,900 0.001 0.04 0.Q2

Oceans 1,338,000,000 96.5 2,6�0

Total 1,385,984,610 '" 100

Source: Maidment (1993).

1.4.2 The catchment water budget

A global view of the hydrological cycle is not appropriate for discussion of

details of hydrological processes. Just as the water budget for each
continent can differ from the global average, we expect the local water
budget to vary from place to place within a continent, and most problems
require the use of control volurnes much smal1er than continental scale.
Although any volume can be defined as the control volume for the
application of Equations 1 . 1 and 1. 2, hydrologists typically choose the
volume to be a catchment (other terms used in the United States are
watershed and drainage basin). The catchment is a fundamental
hydrological unit. A catchment is an area of land in which water flowing
across the land surface drains into a particular stream or river and
ultimately flows through a single point or outlet on that stream or river;
thus, the catchment is defined relative to a specific location along a water
course and the associated land area can be considered to "catch" the water
that flows past that point (Figure 1 .8) . Clearly, then, any number of
catchments can be defined for a particular river (corresponding to any
location along the river).

25
Table 1.2. Average annual water budget for the continents (excluding
Antarctica)

Runoff fatio,
Continent p(mm] r,(mmj rt (mm) r�/p

Africa 30.3x lOG 690 140 550 0.20

Asia 45x lOG 720 290 <30 0.40

Australia 8.7x 10' 740 230 510 0.31

Europe 9.8x 10' 730 320 <10 0.44

North America 20.7x 10' 670 290 380 0.43

South AmeriCa 17.8x 10' 1650 590 1060 0.36

Source: L'vovich (1979).

Note: Values for average annual preeipilation, runoff, and evapotranspiration arc
reported as depths of water over each land area. The total volume may be
calculated by converting the depths to km and multiplying by the land areas.
Several estimates of these quantities exist, all of which are uncertain to some
degree; Glciek (1993) presents a summary.

We can define a catchment by specifying when a point is part of the

catchment and when it is not. A point is within a catchment if surface
water hypothetically flowing from that point ultimately appears at the river
station defining the catchment outlet; the point is not within the catchment
if surface water flows from it into another river or into the same river
below the given river station. The boundary separating regions which do
and do not contribute water to that river station is called a divide. A
catchment is then defined as all points that potentially can contribute
surface water to a particular river station.
As an example, consider the Potomac River Basin near Washington,
D.C. A point on the western slope of the Blue Ridge Mountains near
Waynesboro, Virginia, drains into the South River, which is tributary to
the Shenandoah River, which is tributary to the Potomac. The selected
point is therefore within the Potomac River catchment. On the other hand,
a point on the eastern slope of the Blue Ridge near Charlottesville,
Virginia, ultimately drains into the James River and is, thus, nol within the
Potomac catchment. A portion of the crest of the Blue Ridge Mountains
forms the divide between the Potomac and James River catchments. (Note
that in the United Kingdom the tcnn watershed is synonymous with divide
rather than catchment. Also, in the United Stales the tenn catchmem is
sometimes used to refer to a small drainage basin, while the tenn river
basin refers to a large drainage basin.)

26
 Let us consider application of Equation 1.2 to a catchment. In this case,
V is the volume of water stored on or beneath the surface of the catchment.
One inflow is precipitation falling on the catchment. By definition, thcre is
no surface water inflow into the catchment (this is the primary reason for
using the catchment concept in hydrology, as discussed). There may be
both inflow and outflow of groundwater across catchment boundaries.
However, if the assumption is made that the groundwater divide coincides
with the surface water divide, then groundwater inflows can be neglected
in the formulation of the water budget equation. This assumption is often
valid but should be evaluated for each application. Outflows consist of loss
to the atmosphere (via evapotranspiration) and discharge at the river
station chosen in defining the basin. The conservation equation for the
catchment may be written:

Figure 1.8 The catchment. The boundary of the catchment is referred to as

a divide. If the catchment has been properly delineated, there should be no
surface-water inflows or outflows across the divide, except at the outlet. [n
this case, the major inflow is precipitation (P), and the major outflows are
evapotranspiration (et) and surface-water outflow through the catchment
outlet (1'8). The topography of the land surface controls where divides are
drawn. In the figure, two mountain peaks and their adjacent ridges
constitute the divide.
dV
-=p-r., -r� -el' (1.7)
dl

27
where V = volume of waler stored in Ihe catchment [L3]; p = precipitation
rate [L3 .1]; rs = rate of surface runoff [L3 . 1 ]; rg = net rate of
groundwater runoff [LJ .1]; el = rate of evapotranspiration [LJ .1].
We can illustrate the use of Equation 1.7 by considering long-term
average conditions. For questions involving long-range regional planning,
average quantities are often adequate. In this case, changes in volume of
water stored in the eatehmcnt can bc neglected. If, in addition, we assumc
net groundwater runoff to be negligibly small, the budget equation
becomes:

p-r, -el :0. (1.8)

As an example of the use of Equation 1 .8, let us estimate the average

annual evapotranspiration in the James River Basin above Scottsville,
Virginia. The groundwater systems in Ihis area are highly localized (i.e.,
controlled by the small-scale topography) so the assumption of negligible
groundwater flow is likely to be valid. According to the U.S. Geological
Survey, the James at Scottsville drains an area of 1 1,834 km2 and the
average discharge is 144.5 m3 S- I . The U.S. Weather Bureau
climatological records report data that indicate Ihat average precipitation
on the catchment is about 1080 mm yr- 1 • The terms in Equation 1 .8 must
be expressed in comparable units, so we will express the river discharge in
mm yr l distributed over the catchment; that is, the volume of flow will be
converted to the equivalent depth of water over the entire catchment by
dividing by catchment arca. Thus,

Ill' I 1000 lllll1

1 ., . b4 X IO X
lllll1 1 S

;= 44.5-;- X 1. l834 X lo1o m1 X m (1.9)

, _

I� .11' yr
Perfon11ing the indicated calculation we find that ,:� is 385 nun yr- I over
the basin area. Equation 1.8 indicates that el = p - ' = 695 mm yrl or that
:.

about 64% of the rain that falls in this humid region of the United States is
returned to the atmosphere through evapotranspiration. II is small wonder
that evaporation control is of major concern in arid regions of the world.
The use of a water budget for annual average conditions is
straightforward and provides useful infonnation about climatological mean
values. One reason that these applications of the conservation laws arc
simple and straightforward is that in dealing with annual averages we have
been able to neglect temporal changes in storage. We have not had to
consider how soils store and release water or how channel storage changes
with the passage of a flood wave. An understanding of hydrological

28
phenomena at time scales of hours, days, or even months requires that
changes in storage be described and that the relationship between the
forces applied to water "panicles" and the motion of these "particles" be
studied. Thus, the dYllamics of water flow must be studied.

1 . 5 Concluding Remarks

Hydrology may be broadly defined as "water science." Therefore, one's

background, interests, or motives are probably in some way responsible for
one's perception of hydrology. In truth, hydrology can be about a rather
large variety of subjects. Those whose background is in Earth science will
probably alrcady bc familiar with the imponant role of groundwater in a
variety of geological processes, or how river flow influences and is
influenced by Ihe topography. Others may be interested in issues of water
resources, protection, or water rights. Because water is essential to life,
biologists and ecologists also need to understand temporal and spatial
distributions of water ncar the land surface.
Increasingly, hydrologists arc collaborating with other Eanh scientists
to solve problems. Such collaboration is often difficult, because each
specialist will have his or her own unique language and methods, but
ultimately it is beneficial. Today, scientific and practical problems exist
that cross disciplinary boundaries. Examples include storage of radioactive
waste (involving geoscientists, hydrologists, physicists, and engineers) and
global climate change (involving atmospheric scientists, ecologists,
environmental chemists, and hydrologists). Scientists are working across
traditional disciplinary boundaries to mutual benefit. Throughout this
book, we discuss examples of how hydrological problems involve other
disciplines. Chapter I I concludes with a discussion of key challenges
facing hydrologists now and in the future, including the impacts of humans
and climate on the water cycle and problems related to ensuring the
quantity and quality of water needed for drinking, agriculture, and energy
as human populations grow and climate changes.
This book is organized around the unifying concepts of the hydrological
cycle and the catchment as a basic unit of study. The goal of physical
hydrology is to explain phenomena of water flow in the natural
environment by application of physical principles. Solutions to many
hydrological problems require an understanding of the dynamics of water
motion. Thus, much of the remainder of this book is devoted to the
quantitative description of components of the hydrological cycle based on
models that arise from fluid dynamics.

29
1.6 Key Points

• Hydrology is defined as the study of the occurrence and movement of

water on and beneath the surface of the Earth, the properties of water,
and water's relationship to the biotic and abiotic components of the
environment. {Section I . I }

• Hydrological science has both curiosity-driven and problem-driven

aspects that stretch back to antiquity. The importance of water as a
resource remains one of the central reasons for studying hydrological
processes. {Section 1 .2}

• The fundamental concept of hydrology is the hydrological cycle: the

global-scale, endless recirculatory process linking water in the
atmosphere, on the continents, and in the oceans. This cyclical process is
usually thought orin lemlS of reservoirs (e.g., oceans, atmosphere) and
thc volumctric nows of water betwccn thcm. {Scction 1.3 }

• Within the hydrological cycle, the dynamic processes of water vapor

fonnation and transport of vapor and liquid in the atmosphere are driven
by solar energy, while precipitation and many of the various flows of
water at or beneath the Earth's surface are driven primarily by
gravitational and capillary forces. {Section 1 . 3 }

• For a given control volume, a water budget may be constructed based on

the principle of conservation of mass: the time rate of change of mass
stored within the control volume is equal to the difference between the
mass inflow rate and the mass outflow rate. If density is constant, then
dVld, = / - O. { Section 1 .4}

• A catchment is defined as an area of land in which water flowing across

the land surface drains into a particular stream or river and ultimately
flows through a single point or outlet on that stream or river. {Section
1 .4.2 }

• Catchments are delineated on the basis ofland-surface topography. The

boundary of a catchment is called a divide. { Section 1.4.2}

• On an average annual basis, the water budget for a catchment (p - 1".. - eJ

= 0) indicates that precipitation is balanced by surface water runoff and
evapotranspiration, p = 1� + eJ. {Section 1.4.2 }

30
1.7 Example Problems

Problem I. Precipitation is typically measured as a volume [L3) per unit

area [L2], which has dimensions of length [L]. In the United Slates, the
average annual precipitation varies from minimum at Death Valley,
California (1.6 inches), to a maximum on Mt. Waialeate on the island of
Kauai in Hawaii (460 inches) What is Ihe average annual precipitation (in
millimeters, mm) at each of these locations?

Problem 2. In the United Slates, stream discharge is often measured in

units of cubic feet per second (ft3 S-l, or "cfs"). In most other countries,
discharge is measured in cubic meters per second (m3 sec- I ). What is the
equivalent now (in m3 5- 1 ) of \8.2 ft3 S- I ? (You might want to review
Appendix I on units, dimensions, and convcrsions.)

Problem 3. In an average year, 1 . 0 meter of precipitation falls on a

catchment with an area of 1 000 (or 1 03) km2•

A. What is the volume of water received during an average year in cubic

meters?

B. In gallons?

Problem 4. The polar ice caps (area = 1 .6 107 km2) are estimated to
x

contain a total equivalent volume of2.4 x 107 km) of liquid water. The
average annual precipitation over the ice caps is estimated to be 5 inches
per year. Estimate the residence time of water in the polar ice caps,
assuming their volume remains constant i n time.

Problem 5. In an average ycar, a small (area = 3.0 km2) agricultural

catchment receives 950 mm of precipitation. The catchment is drained by a
stream, and a continuous record of stream discharge is avai lable. The total
amount of surface-water nmoff for the year, determined from the stream
discharge record, is 1 . 1 x 106 m3.

A. What is the volume ofwaler (in m3) evapotranspired for the year
(assume no change in water stored in the catchment)?

B. What is the depth of water (in mm) evapotranspired for the year (again,
assuming no change in watcr storcd in the catchmcnt)?

C. What is the runoff ratio VIm for the catchment?

31
1.8 Suggested Readings

Oki, T., and S. Kanae. 2006. Global Hydrological Cycles and World Water
Resources. Science 3 1 3: I 068-1 072.
Univ. of Oregon. Global Water Balance Animations.
http://geography.uoregon.edulenvchange/clim animationsl
V6rosmarty. C. J., P. Green, J. Salisbury, and R. B. Lammers. 2000.
Global Water Resources: Vulnerability from Climate Change and
Population Growth. Science 289:284-288.

32
2 Precipitation and Evapotranspiration

2.1 Introduction

2.2 Precipitation

2.2.1 Point measurements

2.2.2 Spatial characteristics of precipitation and radar

estimation

2.2.3 Temporal characteristics of precipitation

2.2.4 Analysis of rainfall extremes

2.3 Interception
2.4 Evapotranspiration
2.4. J The water balance

2.4.2 Mass-transfer methods

2.4.3 The energy balance

2.4.4 Other methods

2.4.5 Soil moisture controls on evapotranspiration

2.5 Concluding Remarks

2.6 Key Points

2.7 Example Problems

2.8 Suggested Readings

2.1 Introduction

Chapler 1 introduced the concept of the hydrological cycle and bri efly
described some of the important processes involved in the motion of water
at or near the Earth's surface. This chapter explores two processes of great
importance to the fields of hydrology and meteorology: precipitation and
evapotranspiration. Precipitation is the primary input of watcr to a
catchment. Evapotranspiration is often the primary output of water from a
catchment. A detailed treatment of the precipitation process is most often a

33
subject of meteorology and is beyond the scope of this book. Our
discussion of precipitation will focus on methods for directly or indirectly
quantifying precipitation inputs to catchments and on describing thcse
inputs using statistical techniques. Evapotranspiration, on the other hand,
depends both on properties of the land surface and the state of the near­
surface air and is, therefore, well within the domain of hydrology. Our
concern with thesc processes is in knowing the rates, timing, and spatial
distribution ofthese water fluxes between the land and the atmosphere.
A significant portion of precipitation that falls on vcgetated catchmcnts
is intercepted by and temporarily stored on the surfaces of vegetation (for
example, within dense forest canopies). The remainder falls through to the
land surface, where it may infiltrate into the soil or run off. Water stored
on the surfaces of plants and at the surface of bare soil, as well as at the
surface of open water bodies, can be returned directly to the atmosphere by
evaporation. Water that infiltrates into the soil may be taken up by roots,
converted from liquid 10 vapor in micropores on leaf surfaces (stomata) of
plants, and released to the atmosphere in a process referred to as
transpiration. As it is very difficult to keep track of the fractions of water
lost by the various pathways, we adopt the convention of referring to the
combined processes of evaporation and transpiration as evapotranspiration.
Evapotranspiration represents a dominant outflow of water from most
catchments and accounts for approximately two-thirds of precipitation
over most contincntal land masses. Understanding the physical processes
and factors that govern the rale, timing, and distribution of this flux of
water from the Earth's surface is, therefore, an important goal of
hydrological science. In practice, direct measurements of evaporation from
surfaces (such as leaves, soils, or ponds) at the scale of the catchment are
unavailable, and we must rely on indirect measures of evapotranspiration
or inferences from empirical techniques and the use of proxy variables.
Later in this chapter we explore the process of evapotranspiration and
discuss some techniques to calculate or infer rates of this flux from the
catchment.
There arc a number of reasons for studying the processes of
precipitation and evapotranspiration. The most apparent reason derives
from the catchment annual water budget. Equation 1 .8 tells us that for a
known or specified value ofp, there are infinite combinations of el and '"s
that will satisry Equation 1 .8. What processes ultimately govern the way in
which these two annual outflows are proportioned in a particular
catchment? What controls the relative proportioning on a seasonal or
individual stonn basis? How might changes in land use and climate
influence the water budget for a particular catchment (or for an entire

34
continent)? Will clear-cutting a forest in a catchment cause an increase or
decrease in et or in rs?
The historical development of hydrological concepts indicates the
importance of understanding both precipitation (the ultimate supply of
freshwater) and evapotranspiration (which can reduce the supply of
freshwater and affect its quality by concentrating impurities). The earliest
measurements of precipitation, attributed to Kautilya, an Indian chancellor
of the exchequer during the fourth century BC (Biswas, 1972), were used
as a basis for taxation because agricultural production was presumed to be
proportional to rainfall amounts. Knowledge of precipitation and
evapotranspiration is crucial to water resources and agricultural questions,
especially in arid regions.
One historical example illustrates just how important it is to know the
magnitude of hydrological fluxes. In the early part oflast century, rapid
growth in the western and southwestern United States led to efforts to
"reclaim" the desert, mostly through management of the Colorado River.
To apportion the flow of the Colorado among the states that would use the
water (the Colorado River Compact of 1 922), it was necessary to
determine the amount of water available each year (i.e., p - el). This was
done by averaging the annual discharge measured at a single point on the
Colorado River over the available period of record ( 1 896-1921), which
turned out to be about 1,233 m3 eaeh year. Unfortunately, this period of
time turned out to be a particularly weI era (or, the following years were
particularly dly). From 1 922 to 1 976, the average annual discharge of the
Colorado River at the gaging station was 1 ,020 m3. When the budget was
calculated for the Colorado River Compact of 1922 and the water
apponioned among the states, there was not enough waler to go around! [t
should be clear that the temporal and spatial patterns of precipitation and
evapotranspiration within the Colorado River basin strongly influence
water availability and, hence, its use and management.

2.2 Precipitation

Precipitation is defined as the deposition of liquid water droplets and ice

particles that are fonned in the atmosphere and grow to a sufficient size so
that they are returned to the Earth's surface by gravitational settling.
Precipitation is ordinarily classified according to the phase it is in when it
reaches the surface of a collector or the ground: (a) solid (snow and ice
crystals, including sleet and hail) or (b) liquid (rain and freezing rain).
Some classifications also consider a class of mixed precipitation (liquid

35
and solid). Precipitation is the dominant dcposition mcchanism by which
atmospheric moisture is cycled from the atmosphere to both oceans and
continents. Other mcchanisms (i.e., dircct dcposition of dcw and fog) may
be important in some instances, especially in coastal mountains.
There are three primary steps in the generation of precipitable water in
the atmosphere: creation of saturated conditions in the atmosphere,
condensation of water vapor into liquid water, and growth of small
droplets by collision and coalescence until they become large enough to
precipitate. Saturated conditions occur whcn thc air (which is a mixturc of
gases, including water vapor) has the maximum water vapor content it can
hold without the emergence of condensation (e.g., dew or fog). Water
vapor content can be expressed in tenns of water vapor density (or
absolute humidity), which is the volume of water vapor per volume of air.
However, in mierometeorology and hydrology, the partial pressure of
water vapor, or vapor pressure, e [M L-1 J2], which is proportional to
the vapor density, is typically used as a measure of how much water vapor
is in the air. The saturation vapor pressure, e.m [M L-I J2], is the value
of e for saturated conditions. This value strongly depends on the air
temperature; that is, wann air can hold more water vapor than can cool air,
or, conversely, cool air cannot hold as much water vapor as warm air
(Figure 2. 1). Thus, cooling an air mass tends to produce saturated
conditions, because e.<a/ is reduced. Cooling typically occurs when an air
mass is lifted vertically. Uplift can happen in a number of ways, such as
when (a) air masses rise over mountains or other topographic features
(referred to as orographic uplift), (b) wann air masses rise above cooler air
masses at fronts (frontal uplift), or (c) when heating of the Earth's surface
(especially during the summer) makes air near the surface less dense so
that it rises (convcction) and cools, often producing thunderstonns.
Condensation is simply the phase change whereby water vapor becomes
liquid or solid water. It requires not only the creation of saturated
conditions within the air, but also the presence of condensation nuclei,
small particles (size of 0.00 1-10 jlm), such as dust, pollutants, smoke from
burning biomass, sea salts or prcviously fornled water droplets or iee
particles. Cloud condensation nuclei provide surfaces on which the
condensation of water molecules can start. Condensation may produce
such small particles that they remain stable in the atmosphere. The while
clouds observed on a fair day, for example, arc composed of water droplets
that arc too small to precipitate. It is the coalescence of small droplets into
larger drops, through collision of small droplets with each other or with
larger drops that gives rise to precipitable raindrops or ice crystals. To

36
generate substantial amounts of precipitation, clouds also need to draw
atmospheric moisture from the surrounding air.

�
-

"
II) Supersaturated
II)
Q)
�
Co

Undersaturated

Temperature, T •

Figure 2.1 Temperature-dependence of the saturation vapor pressure, eUJi'

Cooling of an undersaturated air mass (at a constant pressure) will produce
saturated or supersaturated conditions, the first step in the precipitation
process.

Close to the land surface, locations suitable for condensation arc

provided by grass and tree canopies, which-with adequate cooling­
allow for dew formation. Dew then either evaporates or drips down to the
ground surface. Similarly, plant canopies may favor the deposition of fog
droplets, a process known as occult precipitation. Canopy condensation
and fog deposition may provide an important input of moisture in some
landscapes, including areas of the west coast ofthe United States affected
by frequent fog during the rainless summer season, cloud forests, and
coastal deserts adjacent to cold ocean surfaces (e.g., the Namib and the
Atacama deserts).
Most of the precipitation falling on the continental United States
originates from the bordering oceans, even in the interior of the continent.
Some studies estimate that up 10 30 10 40% of the precipitation over large

37
land areas is derived from local evapotranspiration, a process known as
precipitation recycling (sec Chapter 1 ). Although this indicates that local
evapotranspiration docs influence local precipitation, much of the
precipitated water must be transported significant distances across the
continents from the oceans. It is the large-scale motions of the atmosphere
that are responsible for the broad patterns that are observed in annual
precipitation.
Average annual precipitation onto the continents is geographically
extremcly variable, reflecting the influencc of a number of important
physiographic factors. For example, in the United States, average annual
precipitation ranges from a minimum about 40 mm yr- 1 at Death Valley,
California (in the Mojave Desert), to a maximum of nearly 12,000 mm yr-1
at the summit of Mt. Waialeale on the island of Kauai in Hawaii. The
lowest average annual precipitarion that has been measured anywhere on
the Earth is less than I mm yr- 1 at Ariea, Chile (the 59-year record
includes a period of 1 4 consecutive years totally devoid of precipitation).
In general, average annual precipitation onto the continents is a function
of: (a) latitude (precipitation highest in latitudes of rising air-O° and 60°
north and soulh�and lowest in latitudes of descending air-30° and 90°
north and soulh); (b) elevation (precipitarion usually increases with
elevation, a phenomenon known as the orographic effect); (c) distance
from moisture sources (precipitation is usually lower at greater distances
from the ocean, a phenomenon known as continentality); (d) position
within the continental land mass; (e) prevailing wind dircction; (t) relation
10 mountain ranges (windward sides typically cloudy and rainy, with
leeward sides typically dry and sunny, a phenomenon known as rain
shadow); and (g) relative temperatures of land and bordering oceans
(Eagleson, 1970). Average annual precipitation onto the oceans is thought
to bc similarly variablc.
Ovcr much of the world precipitation is also extremely variable in lime,
with infrequent large stonns (e.g., hurricanes and monsoons) delivering
large portions of the total annual precipitation. As with the spatial
variability of precipitation, the temporal variability has both random and
persistent components. For example, although the arrival of individual
hurricanes is not predictable far into the future, the season in which they
arrive is consistent from year to year. Accurate forecasts of how many will
appear, the amount of rainfall they will produce, and the path they will
follow cannot be made for any given year.

2.2.1 Point measurements

38
Two problems arise in quantifying precipitation input to a given land area:
how to measure precipitation at one or more points in space and how to
extrapolate these point measurements to determine the total amount of
water delivered to a particular land area. A variety of instruments or gages
are used to measure point precipitation (amount of precipitation deposited
at a particular station representing a point in space). Point precipitation
usually is expressed in depth units (volume divided by collector cross­
sectional area). Point measurement devices arc generally of two types:
non-recording (storage) gages and recording gages. The first category
includes simple wedge- or funnel-shaped containers to collect precipitation
over a period of time between observations. There are two common types
of recording precipitation gages: weighing and tipping bucket gages.
Weighing gages collect precipitation, typically through a funnel and rccord
the weight of precipitation as a function of time. Tipping bucket gages
consist of a container with a funnel at the top leading to a pair of small
"buckets" attached to a fulcrum. When one bucket fills with water, it tips,
emptying the water and moving the other bucket bcncath the outlct of the
funnel. The device records the lime at which the bucket tips, and so the
time over which a certain amount of precipitation fell (e.g., 0.25 mm for a
bucket ofa certain size) is recorded. In regions that receive snowfall, point
gage data usually are expressed in liquid water equivalent (LWE), which is
the cquivalcnt amount of total precipitation if it had all fallen as rainfall.
Most precipitation gages are equipped with windbreak devices or shields
to minimize the measurement error caused by disruption of the airflow
pattern around the gage.
Rainfall rates measured at a point are tremendously variable in time
(Figure 2.2). The record of hourly precipitation over time (known as a
hyetograph) for a station in Virginia illustrates that precipitation
commonly is organized into discrete stonn events of varying intensity and
duration separated by interstonn periods of variable duration. Because we
may be interested in longer periods of time (perhaps an annual water
budget for a catchment), we can add hourly measurements together to
derivc daily or monthly hyctographs. Variability decreases as the period of
reporting increases, as illustrated by the daily and monthly hyetographs for
the Coweeta experimental watershed for the 1992 water year (Figure 2.3).
(In the United States, the U.S. Geological Survey water year begins
October I and runs through the end of the following September.) Both
annual prccipitation dcpths and temporal precipitation variability depend
on local climatology. For example, in monsoon-prone regions most of the
precipitation arrives in a certain season when large-scale atmospheric
flows drive moist air from oceanic areas onto terrestrial areas. In rain

39
forests and some coastal areas the precipitation is spread more evenly over
the year.

6 ,------,

5
E
-

E
�

c 4
0
-

CO
.

-
-
.

a. 3
.-

u
..
�
a.
,., 2
"
-

0
J:
1

O+rTT'"n
M m �
N M
• � � . �

01 0) 0) Ol
� �

� � � �
< < < <
Figure 2.2 A hyctograph for a site near Charlottesville, Virginia, showing
temporal variability in precipitation.
Data courtesy of Greg V. Jones.

40
 �
�-
E E
-
" E
'"
E 0 -

E
"
-

..
-
-

a.
-
.-

u
.,
.-

a.
�

,..
..
-
-

Cl

� N � � � � m � �
• • • • • • • • • • •

§ � �
� <
c .c � ""

8 � � � � � � • '

Figure 2.3 Daily and monthly (inset) hyetographs for Ihe Coweeta
experimental watershed, North Carolina, water year 1992.
Data courtesy of Wayne Swank.

The measurement of precipitation in the form of snow presents

additional difficulty. Gages must be heated, for example, to reduce the
precipitation to liquid water equivalent. [n mountainous areas, in
particular, it can be difficult to keep a standard gage running through the
winter. Snow pillows, devices that record the pressure at the base of a
snowpack, and snow surveys, in which teams periodically go into the field
and take cores of snow along transects to estimate depth of snow, are
among alternative measurement methods. The Soil Conservation Service
in the United States maintains approximately 500 snowpack telemetry sites
in remote mountainous areas of the western United States at which data
from snow pillows are collected automatically and continuously.
Nevertheless, point measurements of precipitation in the fonn of snow in
mountainous areas are collected on a much sparser network than are
rainfall data in lower-lying areas.

2,2.2 Spatial characteristics of p.'ecipitation and radar

estimation

41
Thus far, we have discussed temporal variations in precipitation at a point
or station in space. In hydrology we typically want to estimate the total
volumc of water delivered to a given area (e.g., a catchment) within some
period of time. That is, we want to know the average precipitation depth
over the area, i.e.,

p '" � If p(x,y)duly (2.1)

,

where P is the average depth of prccipitation over the catchment, A is thc

catchment area, and p(x, y) is the spatial distribUlion of cumulative
precipitation for the time period of interest. Once P is detennined, the total
volume of water delivered to the area is simply PA. In many situations,
hydrologists arc faced with using gage measurements 10 estimate P. Given
measurements made at discrete points within an area or catchment, how
can the average precipitation for an area be estimated?
In the simplest and most common case there is only one rain gage in (or
even near) the catchment of interest. Hence, by necessity the spatial
variability of precipitation within the catchment is ignored and the single
point measure is assumed to represent the mean value, such that the rate is
simply multiplied by the area to estimate thc total prccipitation over the
catchment. Given the great spatial variability in actual precipitation rates,
this approach is likely to yield large errors. Unfortunately, with limited
data there may be no other choice.
Over very large catchments or heavily instrumented research sites,
several precipitation gages may be available (Figure 2.4a). In this case we
might use a weighted average of these gage measurements to estimate P.
The precipitation values at all gaging stations (p,) are used, with each value
weighted in some reasonable way. The simplest example is an arithmetic
average that weights each gage equally. In some instances it may be more
appropriate to weight one gage more heavily than others. For example, one
gage might be located within the catchment that is the focus of a water
rcsources investigation and all other gages might be some distance from
the catchment. In this case we might prefer to assign weights on the basis
of how important we think the gaging station is or how representative of
precipitation over the catchment area that station is.

42
 .s, .s,

s,
s ,·
'----.I
S3
'-- __

_�
. S3 .87 • S,
•

(a) (b) p"

Figure 2.4 Depiction of gaging stations within a hypothetical catchment

(0) and the construction of isahyets (b) to estimate the spatial distribution
of precipitation.

We might approximate the true distribution of precipitation by drawing

"contour lines" of equal precipitation depths, called isohycts (Figure 2.4b).
The isohyetal method uses the arcas bounded by each pair of isohYCIS to
detennine P;

l '
p = - "" a/p," (2.2)
•

A L-
, ,
.

where PI (pj_ + 1'1+)12, the average p for each subregion; Pi- = value of
=

the lower magnitude isohyet bounding subregion i; PH = value orthe

higher magnitude isohyet bounding subregion i; aj = area of subregion i.
Again, the weights are fractional areas but this time the "contour" values
of precipitation are used in place of the gage values. Alternatively, spatial
intcrpolation algorithms may be used to dctcmlinc the spatial distribution
of precipitation based on point measurements from rain gages. These
algorithms are included in most geographic information system (GIS)
software packages.
Interpolation and weighting schemes that use rain gages are generally
inadequate for defining average precipitation to a catchment on a storm­
event basis. Even with fairly dense rain gage networks, the spatial
distribution and intennittency of rainfall rarely is captured in the point­
measurement data. For example, in a study involving a dense gage
network and a NEXRAD (NEXt generation RADar), numerous storms for
which radar estimates indicated hourly precipitation exceeding 50 mm
were missed completely by the gage network (Smith, Seo, ct aI., 1 996). As

43
radar data become available on a routine basis, they may supplant the
weighting schemes outlined above, or at least provide a rational basis for
constructing isohyctal lincs.
WCalher radar has become an increasingly important tool for cstimating
the spatial distribution of rainfall, p(x, y). Radar signals reflect from
raindrops in the atmosphere and the characteristics of the reflected signal
can be related to rainfall rates. Radar is far from an absolutely accurate
measurement method, but it provides detailed infonnation on the time and
spacc distribution of rain and can bc particularly valuable for heavy
rainfall. The spatial distribution of total precipitation in a storm may
depend in part on the path of the storm (Figure 2.5). We also see from
these data the potential for significant spatial variability in precipitation
over a catchment. Imagine the difTerent estimates oftola] precipitation
from a single gage located in the southcrn end of the catchment as
compared with one in the northern end.
Measurements from satellites also may be used to infer precipitation
rates and snow-accumulation patterns. Snow-cover maps are produced
from data collected with NOAA's (National Oceanic and Atmospheric
Administration) Advanced Very High Resolution Radiometer (AVHRR).
These satellite data are used in conjunction with ground-based point
measurements to detennine spatial distributions of water in the snowpack
(e.g., see DeWalle and Rango, 2008). Whereas the satellite observations
estimate amounts of snow present on the ground, techniques are being
developcd to cxtract snowfall ratcs from the NEXRAD data. NEXRAD
sites are being installed presently around the United States, providing
coverage of most regions. With the data from these instruments being
made available over the Internet, it is reasonable to anticipate that they will
soon be applied widely in hydrology to estimate rainfall and snowfall rates
distributed in space and time over catchments.

44
 a
./
10:17 a.m.

1 1 : 38 a.m.

(a) Storm track (b) Total rainfall accumulation

Figure 2.5 Radar observations of the June 27, 1995, stonn in the Rapidan
River basin, central Virginia. NEXRAD tracked the center of the stann
(a). Total rainfall accumulations (mm) were calculated based on the lime
series of radar images (b). This intense storm produced an average of344
mm (1 3.5 inches) over the catchment and caused extensive flooding
throughout the basin.
Smith ct al. ( 1996).

2.2.3 Tempo."al charactcri�tics of precipitation

Precipitation at any particular point on the Earth's surface is subject to

extreme temporal variation. One minute it may be raining heavily, while a
few minutes later it may not be raining at all. OUf ability to forecast this
temporal variation even a few hours in advance is limited and our ability to
forecast several days in advance is almost zero. The extreme uncertainty
associated with precipitation forecasts suggests substantial randomness in
the occurrence of precipitation at a point in space and implies the necessity
of a probabiIistic approach for characterizing the temporal variations in
precipitation.
As noted previously, if we examine a typical short-tenn hyetograph
(e.g., Figure 2.2), we see that precipitation is organized into discrete

45
(storm) events of varying amounts (storm depths). Average precipitation
intensity is the rate ofpreeipital'ion over a specified lime period, the
preeipilation depth divided by the time over which that depth is recorded.
For example, the data in Figure 2.2 are reported for each hour (i.e., the
hourly precipitation intensity). The hourly precipitation intensity for the
period shown varies from zero to about 22 mm hel. Average precipitation
intensities depend on the time period over which the computation is done.
That is, the variation in hourly precipitation intensity typically will be
much greater than the variation in 6-hour intensity (average over a longer
time), but less than that of 1 5-minute intensity (average over a shorter
time). Average precipitation intensity for a storm is the total depth of
precipitation for the stoml dividcd by the storm length. In general, the
longer the storm-event duration, the less the (average) storm intensity.
However, the greater the stoml duration, the greater the storm depth.
Hydrologists commonly employ a statistical technique known as
frequency analysis to describe systematically the temporal characteristics
of precipitation at a particular station. To illustrate the frequency analysis
technique, we will consider annual precipitation amounts for several
United States cities over a 64-year period (Table 2. 1). We assume that the
quantities in Table 2. 1 (annual precipitation) are samples of a random
variable. Ifwe look at how often (the frequency with which) values within
a certain range arc encountered, we find that a plot of magnitude versus
frequency displays a characteristic shape, a "bell-shaped" curve. In the
case of the data from Seattle (Figure 2.6), we find that the most frequent
values of annual precipitation are between 950 and 1000 mm. Higher and
lower values occur with lower frequency. We also could normalize these
data by dividing by the total number of observations (64) to determine the
relative frequency. For example, annual precipitation values between 950
and 1 000 mm occur with a relative frequency of 10/64, or 0.16.
Alternatively, we might say that values within that range occurred 16% of
the time. Note that, because annual precipitation is a continuous random
variable, we need to be concerned with the probability of a range of
values, rather than with one specific value. Also, note that these data
represent yearly precipitation amounts. As we will discuss shortly,
decreasing the averaging period (say, monthly or daily precipitation
amounts) often causes the shape of the curve to become more skewed and
look less like a bell.
Because annual precipitation amounts appear to follow a normal (or
"Gaussian") distribution, the mean and the standard deviation (Figure 2.6,
Table 2. 1 ) are the only parameters that are required to describe the
magnitude-frequency relationship. Once we have detennined the

46
parameters of the nom1al distribution (the mean and the standard
deviation), we can detcrmine the probability associated with a particular
range of annual precipitation values (Appcndix 3). As an example, we will
calculate the probability that annual precipitation in any given year will
exceed 1.0 m (this is referred to as the exceedance probability). The first
step is to calculate the z-value, which is a way of normalizing the data to a
distribution with zero mean and unit standard deviation. This is
accomplished by finding the difference between the sample mean and the
target value ( 1 .0 m in this example) and dividing this value by the sample
standard deviation:
x-x
== • (2.3)
'..

; = 392 x = 973
J.. _ 96 05.. _ 1 66

Denver
(1948-201 1 )

Seattle
(1 948-201 1 )

Annual precipitation (mm)

Figure 2.6 Absolute frequency of values of annual precipitation at SeanIe
(gray bars) and Denver (blue bars), with the calculated mean (x) and
standard deviation (s...) .

where x = the value of the variable (P) Ihat we are interested in (e.g., 1 . 1
m); x = the mean ofp; s = the standard deviation ofp. For the Seattle
..

47
data, the calculated z-value corresponding tox = 1 . 1 m is 0.77. For this z­
value, the cumulative probability (Table A3.2 in Appendix 3) is 0.7794.
This valuc is thc probability that thc annual prccipitation will bc less than
1 . 1 m, so we need to subtract 0.7794 from 1 . 1 (the total probability) to
arrive at a value of 0.2206. Therefore, there is a 22% chance that annual
precipitation will exceed 1 . 1 m in Seattle. Another way of looking at the
same result is to say that precipitation in excess of 1 . 1 m occurs roughly
evcry 4-5 years or with a return period (the inversc of the cxcecdance
probability) of approximately 4-5 years.
Magnitude-frequency relationships for many hydrological variables are
useful to hydrologists and water resources planners. For example, in the
design of a reservoir for irrigation or hydropower it may be important to
know how ortcn part of the reservoir's storage capacity would remain
unused. To that end, hydrologists usc the methods described in this
subsection to calculate the probability that the annual rainfall in a region is
smaller than the minimum value required to reach full storage conditions
in the course of the water year. Although annual precipitation often is
describcd wcll by a normal distribution (Figure 2.6), not all hydrological
parameters are (see Haan, 2002). In fact, exccpt for annual totals
precipitation intensities (or amounts) typically do not follow a nonnal
distribution. We have restricted our application of the nomlal distribution
to annual precipitation amounts. If the precipitation averaging time is
decreased to a much shorter period, the distribution begins to diverge from
nonna!. This is shown in Figure 2.7 for a 50-year record of precipitation
on the southern coast of Ireland. In this figure, probability density is
analogous to a relative frequency in that it is a relative measure of how
likely certain magnitudes of precipitation are. Notice the extreme skewness
in the daily precipitation distribution, the moderate skewncss in the
monthly amounts, and the rclative absence of skewness in the annual
amounts. Hence, it would be incorrect to use the nonnal distribution to
describe a variable such as daily precipitation at this location. It should be
stressed that the probabilistic approach requires a significant amount of
data (a long time series) to produce meaningful analyses.

Table 2.1. Total annual precipitation (mm), 1948-20 I I , for selected cities
in the United States

48
Year Seattle Santa Barbara Oenver Chariotte§vllie

1948 1164 230 321 1771

1949 '" 326 428 1116

1950 1401 288 355 1090

1951 1024 350 495 1106

1952 604 693 342 1307

1953 1256 153 362 971

1954 1050 416 192 1066

1955 1185 468 408 1087

1956 937 274 349 1078

1957 881 420 549 1084

1958 1084 557 479 1057

1959 1182 209 420 1205

1960 998 408 381 100'

1961 1081 143 483 1412

1962 90' 459 216 1248

1963 984 493 311 772

1964 1051 310 258 1076

1965 854 516 556 90'

1966 971 306 276 1009

1967 904 457 593 1134

1968 1275 304 308 1012

1969 857 '29 5<8 1402

1970 951 470 350 1140

1971 1098 265 279 1429

1972 1229 219 430 1678

1973 891 502 584 1249

1974 962 482 357 1106

1975 1130 315 395 1520

1976 '78 405 342 1267

1977 834 304 260 809

1978 864 892 298 1265

1979 820 457 518 1558

1980 905 459 348 883

1981 900 459 320 92'

1982 99' 527 367 1317

49
1983 1040 1035 514 1409

1984 '39 239 419 1249

1985 63' 306 415 1173

1986 975 494 308 844

1987 '" 373 510 1237

1988 838 318 381 672

1989 881 110 393 1400

1990 1137 139 424 1294

1991 900 565 516 1119

1992 833 456 399 1143

1993 732 574 377 1204

1994 885 321 280 1349

1995 1082 721 408 1198

1996 1287 708 318 1373

1997 1100 <70 606 1152

1998 1120 1024 453 1302

1999 1070 300 479 1154

2000 729 642 353 1095

2001 955 597 402 902

2002 798 234 216 1031

2003 1062 558 408 1895

2004 790 257 436 1133

2005 901 793 370 1151

2006 1231 467 321 1176

2007 991 186 443 880

2008 782 384 215 1159

2009 977 256 426 1328

2010 1195 636 299 1013

2011 925 718 480 1346

mean 973 438 392 1179

standard dev. 166 201 96 226

50
 '.2..-- ----- --,--, 0.01 -,---------
. ( b)
-;,-
,.,
-

(� •
� 0.15 •
<

� �
� �

:0�
;:
-
0.1 ;: 0.005

0.05
'\
� , �::::p-.,.�
o 10
.
ro � 40 SO
,-f-'---
o
100 150SO
-c--r-r-�.....j
250 300 350 200
Dalty precipitation (mm) Monthly precipitation (mm)

�'o:.----7---,=,-----(/;c'il)
:; 0.0015
•

o.OOO:-J-.c.:;:::::.-��_�_"'�>-�
1i 0.001

£
Jl

800 1000 1200 1400 1600 1800 2000

Annual precipitation (mm)
Figure 2.7 Probability densities for daily (a), monthly (b), and annual (c)
precipitation at Valentia, Ireland.
Data courtesy of Gcrard Kiely.

2.204 AnalYSis of rainfall e"Xtrcmes

•

A number of hydrological applications require an analysis of extreme

events and their frequency of occurrence. Urban planners, land developers,
and insurance companies need to know the nooding frequency of areas
subjected to dife f reOl land uses. Similarly, in the design of bridges, canals,
drainage systems or dams engineers need to refer to suitable hydrologic
design conditions. It is important to know with what probability these
conditions may be exceeded. Different exceedance probabilities
correspond to different risks of damage to people and property so design
decisions are made based on how much risk is tolerable in a given
situation.
To explain how the exceedance probability of extreme rainfall events
can be calculated, we need to revisit some of the concepts presented in the
previous section. If X is the rainfall depth measured with a given resolution
tJ.t (e.g., !J.t = 15 min, I h, or I day) and X* is a reference value ofX (e.g.,
the design conditions) the time interval, T, between two consecutive events
exceeding X* is known as the recurrence interval of events greater than X*
(see Figure 2.8). The average duration of the recurrence interval is defined

51
as the return period, Trc1llnp of X* and is typically measured in years. The
return period of an event, X*, tells us that X* is exceeded on average every
Trc tum years. Based on this definition, the return period increases with
increasing values of X*. In other words events exceeding X* become more
rare as X* increases. It should be stressed that the definition of return
period is given as the mean recurrence interval, t, which means that the
rainfall event with return period of, say, 100 years (sometimes called "the
1OO-year rainfall") is exceeded on average (but not exactly) once every
100 years. We can have some I OO-year intervals with more than one event
greater than the hundred-year rainfall and others with none. It can be
shown that if events with X> X* are statistically independent, l/T,.cmr M is
equal to the exceedance probability of X* (i.e., Prob[X> X*] = I/Trctum;
Appendix 3).
The study of extreme rainfall events faces two challenges: First,
extreme rainfall events are by definition rare and the rainfall records that
are typically available to detennine their return period are often very shon.
For example, we might need to study events with the return period of 100
or 200 years, while having access to records much shoner than these return
periods. Second, as noted in the previous section, event-scale precipitation
does not exhibit the "bell-shaped" Gaussian distribution that we have
considered in the ease of annual precipitation. The propenies of the
distribution ofX arc in gencral unknown and depend on thc timc scalc !:J.t
at which Xis measured (see differences between daily and monthly
precipitation in Figure 2.7). Statistical theories, however, have shown that
the extreme values of any random variable tend to follow only three
possible classes of distributions. In particular, extreme precipitation is
typically distributed according to the Gumbel (or "Type I") distribution.

52
 140

E 120
-

E
-
100
.c
T
a.
-
:
----- ,------ --- ! ! -- -"----
� BO
"C
-
-
�
- 60
c
�
.-

�
40
,.,
-

�
20
.-

0
1958 1962 1967 1972 1977
Year
Figure 2.8 Record of daily precipitation depth, X, for the city of Turin.
The return period of the extreme value X· = 88 mm d-' (dashedhorizollfai
line) is the mean of the recurrence intervals 'I' T2, and so on.

To use extreme value distributions to analyze a precipitation record we

need to extract from the "parent population," X, of precipitation the
extreme values, X'. This can be accomplished by considering only the
maximum annual value for each year of record. We then assume that X'
has a Gumbel distribution and dctcnnine the return period ofX'" (or its
exceedance probability) as

ProliX';?:X.]= I
T...,,,,,,
= l-CXP [ ( X'-f3)]
-cxp -
a
(2.4)

Equation 2.4 shows that the Gumbel distribution h�nly two parameters,
a and /3, that need to be estimated using the mean (X,) and the standard
deviation (ox) of X' as

.J6 (1\" and -

Ct == - fJ X'- O.5772Ct. (2.5)
,
==

We can use the precipitation record from the city of Turin to examine
this method. For every year we consider the maximum rainfall depth with
a duration of 1 , 3, 6, 12, and 24 hours (Table 2.2). To detennine the return
period of a rainfall depth, e.g., X· = 70 mm with the duration of 3 hours,
we first need to calculate the mean, X', and the standard deviation, 0x� of

53
the annual maxima of that duration (Table 2.2, third column). Using
equations (2.5) we detennine the corresponding values of a and P for the
3h duration (sec Table 2.2) and use Equation 2.4 to calculate the return
period of X*. We find that 70 mm of rainfall in 3 hours is exceeded on
average once every 40 years (Trelum = 40 yr). Alternatively, we might need
10calculate the rainfall depth corresponding to a given return period.
Solving Equation 2.4 for X* we obtain

(2.6)

where InO denotes the natural logarithm. Using Equation 2.6 we find that
the rainfall depth with a 3 hour duration and a return period of 100 years is
X* = 79.5 mm. As expected, the rainfall depth associated with the same
return period (Trelllrn = \00 yr) increases with increasing rainfall duration
(see Table 2.2).

Table 2.2. Maximum annual precipitation depths, X', for different

durations, lll, obtained from the rainfall record of the city of Turin (1 927-
2009)

54
Year lIt=1h lIt=3h lIt=6h at=12h lIt=24h

1927 31.0 33.0 33.0 38.0 56.0

1928 20.6 22.5 25.4 27.4 27.8

1929 21.0 33.5 36.5 61.5 98.0

1930 18.9 21.0 30.0 50.4 70.7

1931 62.2 64.2 65.2 65.2 7 2.6

1932 17.6 29.0 32.6 39.6 51.2

1933 32.4 36.0 36.0 40.0 60.0

1934 15.6 24.6 35.6 48.2 51.4

1935 31.0 52.0 54.8 68.2 68.'

1936 42.6 54.6 54.8 54.8 70.6

1937 31.0 33.0 33.0 35.4 51.6

1938 33.0 38.0 38.0 38.0 43.0

1939 24.2 24.2 25.4 32.2 44.6

1940 15.0 25.6 32.6 37.0 41.0

1941 15.6 23.0 25.0 41.0 66.0

1951 32.6 38.0 67.0 87.4 104.0

1952 30.4 30.4 30.4 44.0 65.6

1953 20.4 30.0 48.0 67.0 71.0

1954 53.6 69.0 70.6 70.6 72.8

1955 57.6 61.0 61.0 61.0 64.0

1956 30.8 45.6 45.6 64.6 81.6

1957 32.6 41.0 52.2 52.8 73.6

1958 23.8 23.8 27.8 46.0 73.0

1959 31.4 33.0 3 5.8 53.4 74.4

1960 36.0 48.0 70.0 110.0 140.0

1961 60.0 61.6 72.6 SO.O 82.8

1962 58.4 61.2 61.2 78.0 135.0

1963 23.4 25.4 SO.O 52.6 74.2

1964 27.2 52.0 64.0 73.6 93.0

1965 21.0 24.0 29.2 42.2 49.6

1966 28.0 28.0 28.0 28.0 28.0

1967 28.0 31.2 34.8 54.6 57.8

1968 25.0 43.0 43.0 43.0 43.0

1969 28.0 34.8 53.6 53.6 66.8

55
1970 18.0 26.4 27.6 28.2 46.8

1971 26.2 27.0 'll.O 40.2 43.0

1972 24.0 28.4 37.4 37.4 44.0

1973 36.0 40.0 48.6 57.0 66.0

1974 46.8 50.2 55.0 107.0 120.2

1975 19.8 30.8 31.0 42.2 74.4

1976 43.6 48.8 49.8 62.4 99.2

1977 22.6 47.2 66.8 85.0 85.0

1978 25.6 30.0 30.0 47.0 60.4

1979 51.0 55.2 55.2 55.2 55.2

1980 37.0 54.0 67.6 75.8 83.0

1981 20.8 26.2 39.6 52.0 99.0

1982 25.8 37.8 39.0 40.0 44.0

1983 28.2 31.2 34.2 47. 4 60.6

1984 16.0 22.0 28.6 42.4 61.4

1985 32.2 38.2 44.0 45.8 52.6

1986 30.4 31.0 37.0 45. 4 52 .0

1987 35.6 48.0 54.0 56.8 80.8

1988 30.6 42.0 42 4

. 47.4 59.2

1989 23.6 35.4 39 4

. 64.2 100.2

1990 45.2 46.2 46.2 46.4 52.6

1991 36.4 56.0 87.2 88.6 99.4

1992 21.8 34.2 44.0 78.8 109.4

1993 32.6 51.8 53.2 53.2 74.4

1994 23.6 43.8 63.0 109.0 149.4

1995 28.4 32.8 32.8 44.1 45.9

1996 3 4.3 39.5 46.9 47.7 50.7

1997 49.6 57.1 57.1 61.5 66.5

199 8 28.6 28.6 28.6 28.6 51.9

1999 25.8 41.2 41.6 41.6 46.8

2000 24.0 41.4 63.2 82.4 97.1

2001 21.2 24.6 26.1 26.9 28.1

2002 55.3 50.' 61.0 72.5 118.2

2003 29.5 29.5 31.2 37.8 51.3

2004 26.2 35.9 36.1 36.1 49.6

2005 65.2 74.2 77.7 79.5 80.5

56
2006 31.0 34.8 46.7 64.4 106.9

2007 52.2 56.4 56.8 56.8 65.6

2008 20.6 23.8 34.2 56.6 86.4

2009 29.6 37.0 52.2 69.8 69.8

X' 31.5 39.1 45.3 55.4 70.4

0, 12.1 12.9 14.7 19.0 25.7

a 9.' 10.0 11.4 14.8 20.0

P 26.1 33.3 38.7 46.8 58.9

,.=100yrs)
X· (T",.. 69.5 79.5 91.3 115.1 151.0

Note: These values are used to calculate the mean, standard deviation, and the
parameters, a and P, of the Gumbel distribution, along with the rainfall values, X*,
with lOO-year return period. All values are expressed in mm.

2.3 Interception

Not all precipitation reaching the land surface is available for streamflow
or replenishing groundwater. Rather, a portion is temporarily stored by
vegetation (interception), where it is subject to evaporation. If we
consider the total amount of precipitation (gross precipitation) delivered to
a point within a land area (p), such as might be measured by a tipping
bucket gage placed in a clearing or above a forest canopy, some of that
precipitation in a vegctated area will be intercepted by the plants, some
will fall between plants to land on bare ground or ground covered by lower
vegetation or leaflitter, and some will run down the stems and trunks of
plants to the ground surface (Figure 2.9). Precipitation stored by the
canopy or leaf litter is subject to evaporation. We can define the total
interception (IT) as the sum of the canopy interception (lJ and litter
interception (I,). It is important to stress that, according to this definition,
interception is associated with the evaporation of rainwater stored in
canopy and litter. Interception should not be confused with the temporary
retention of water in canopy or litter before it eventually drips to the
ground.
The capacity of the canopy and litter to store water is limited, and the
rain interception and snow interception capacities are not the same value.
After a period of time during a stonn, these "reservoirs" will begin to
approach their limits, such that no additional water can be stored.
Precipitation that is not intercepted by the canopy or which leaves the
temporary storage of interception may be classified as either stemnow or
throughfall (see Figure 2.9). The fonner is exactly what the name implies

57
and thc latter refcrs to watcr that rcachcs the ground dircctly or by dripping
ofT leaves. This partitioning of prccipitation between vegetated surfaces
and thc soil is critical in dcfining thc reservoirs of watcr availablc for
direct evaporation from surfaces and for transpiration by plants of water
taken up from the soil through roots.

Not all precipitation reaches the ground surface in vegetated areas, such
that the combination of throughfall and stcmnow is less than the total
precipitation. Water storcd in vcgctation (Ie) or lcaf litter (I,) may return to
the atmosphere through evaporation.

2.4 Evapotranspiration

The tenn evapotranspiration describes all the processes by which liquid

water at or ncar the land surface becomes atmospheric water vapor.

58
Evaporation is a physical process that occurs whenever a wet surface is
exposed to unsaturated air. Transpiration is the vaporization and transport
of plant water from leaf chloroplasts to the atmosphere through the
stomata, small cavities existing on leaf surfaces. It is difficult in practice
to separate evaporation (from wet surfaces) and transpiration (water
evaporating inside plants) from each other. Hence, we focus on the
combined quantity, evapotranspiration (et). As we saw in Chapter I,
looking at a global average, about two-thirds of the precipitation that fal1s
on the continents is evapotranspired. Of this amount, 97% is el from land
surfaces and 3% is open-water evaporation. It is through the process of
evapotranspiration that the sun's energy is introduced to drive the
hydrological cycle.
Both evaporation and transpiration require a diffusion mechanism (sce
2.4.2) to sustain a water vapor flux that removes vaporized water
molecules from the evapotranspiring surface. The two ingredients for
vaporizing water are energy and water. Hence, evapotranspiration is where
the surface-water balance and surface energy balance meet. The energy is
derived from the solar radiation and the water is typical1y provided by
local precipitation. Because both solar energy and available water are
necessary to cause evaporation (and transpiration), energy will limit the
rate of evapotranspiration at some times, and water availability will limit
the rate at other times. Considering water availability, we see that
evapotranspiration is a two-way street, where the amount of water present
affects the rate of evapotranspiration, which in tum affects the amount of
water present for subsequent allocation. When the near-surface soil is
saturated or nearly saturated with water, evapotranspiration may proceed at
a rate-known as the potential evapotranspiration rate-limited only by
the availability of energy. When the soil becomes drier the actual
evapotranspiration rate becomes reduced below that found for a wet
surface.
Most plants have openings (stomata) on their leaves to allow them to
take up carbon dioxide from the atmosphere. When the stomata are open,
plants transpire water. Unlike evaporation, transpiration is not control1ed
solely by physical conditions because plants regulate the rate at which
water is released in transpiration in a manner that varies by plant type. Of
the water taken up by plant roots about 95% is transpired through the
stomata. The remaining 5% or so is converted to biomass through
photosYnlhesis. Hence, to first order, the water taken up by the roots is
converted to vapor and lost to the atmosphere. When the availability of
soil water is limited, plants conserve it by restricting flow to the
atmosphere by contracting the stomata. However, the degree of restriction

59
varies considerably across plant species and even throughout the year for a
given species. This renders quantitative treatment possible only in an
average sense, say with an entire stand of trees trealed as a single "big
leaf' for which the resistance 10 water flow is handled as a simple function
of soil moisture content and time of year. Here hydrologic and ecologic
conditions must be explored simultaneously, to incorporate a quantitative
description of plant water use characteristics into the framework of
hydrology.
The direct measurement of evapotranspiration is a difficult task and in
most hydrological applications the rate of evapotranspiration is calculated
as a function of quantities that are easier to measure. To that end, we can
in principle use methods based on the water balance, the energy balance,
or the diffusion of water vapor (mass-transfer methods). As it will be
shown in the following subsection, water balance methods have poor
accuracy. Thus, energy balance and mass-transfer approaches are most
commonly used. To eliminate the dependence on some variables that are
difficult to measure, these two methods are often combined (combination
methods).

2A.) The \\ ater balance

Evapotranspiration may be estimated with a water balance approach, if the

change in storage and all the inputs and outputs except et are known (sec
Equation 1.3):

dV
P(/)+ r.,(t)+ 'gi(/) -,..,(1) - '.>"'0) - el(/). (2.7)
dt
0::

In this equation, the time rate of change in volume of water stored within
the control volume (left side) is balanced by the difference between the
inputs and the outputs. Each of the tenns on the right side of the equation
has units of volume per time [LJi1], and all are functions of time.
Equation 2.7 may be simplified for some problems, for example, by
neglecting rgi and rgo when considered unimportant. Writing Equation 2.7
for average annual quantities reduces it to Equation 1 .3. Once all the tenns
except et have been measured, then et may be computed as the residual to
force a balance that conserves mass in the system. An example of the use
of this equation for an entire calchment is presented in Chapter I. The
result is a very reasonable estimate of average annual evapotranspiration,
el, for the catchment. In many cases, however, the water balance approach
may suffer from the accumulation of errors in the measured variables. As

60
an example, consider a lake with a very rapid throughflow rate, which over
the course of a year does not see a net change in storage. Suppose that the
measurcd avcrage annual fluxes and associated errors for the lake arc:

p = 107+5x IO� 013 yr-1 (+5%)

r" = I09± 1.5 x 108 Ill] yr-I (± 15%)
,,,,=9.95x 108+ 1.5x 108 m3 yr- I (+ 15%).

If we neglect groundwater inflows and outflows, we can use these values

and Equation 2.7 to solve for el. The result, accumulating the errors as we
go, is 1 . 5 x 1 07 ± 3 x 108 m3 yrl. (Bear in mind that for a lake such as
this, thc el is a rclatively small part ofthc budget. You might want to
compare this example with the example presented in Chapter I, Section
1.3.2.)
It is unrealistic to expect to be able to quantifY accurately all the terms
in a water balance for a catchment to solve for el, especially over short
time periods where storage changes are both significant and very difficult
to measure or predict. Furthermore, this is a diagnostic rather than a
predictive approach. In many cases we need to predict the rate of water
loss to the atmosphere for certain anticipated conditions.

2....2 'lass-transfer methods

Water vapor fluxes from the soil surface or through the stomata typically
occur by diffusion, the same process that is responsible for the spread of
contaminants in streams and lakes or the transfer ofheat through rock
surrounding an igneous intrusion. In a diffusive process the flux takes
place in the direction of decreasing concentration of the transported
quantity and the rate of transport is proportional to the concentration
gradient (i.e., the difference in concentration between two points in the
flow direction divided by the distance between the two points). In the case
of evapotranspiration the transported quantity is water vapor and its
concentration (i.e., mass of water vapor per unit volume, or water vapor
density) is proportional to the vapor pressure (see section 2.2). Therefore,
evapotranspiration is a function of the vapor pressure difference between
the evapotranspiring surface(s) and the overlying air (a) at a height Za
above the surface (Figure 2.1 0)

61
 T,/ e
"

H,o

vapor stomata wet

ground surfaces to a measurement point a (e.g., top of the tower) at
temperature TO' vapor pressure eO' and height z". The evapotranspiring
surfaces are treated as a "big leaf' at temperature T$ and with saturation air
humidity conditions.

(2.8)

where eo and II" arc the vapor pressure and the wind speed at the height zO'
respectively, and es is the vapor pressure of the air in cont(lct with the
evapotranspiring surface. Because this air is saturated with water vapor
(i.e., eJ = esat), eSal can be calculated as a function of the surface
temperature, TJ (see Figure 2. 1 and Table 2.3).
The dependence of et on Jill reflects the fact that diffusion in a turbulent
atmosphere is morc effective with stronger wind speeds. (Turbulence is
discussed in Chapter 3.) In general the diffusion of water vapor from the
evaporating surface depends on the aerodynamic properties of the surface
and is enhanced by surface roughness. The dependence on roughness is
expressed through the K£ factor in Equation 2.8, which, in the case of
evaporation from a wet surface (i.e., without accounting for the effeet of
stomatal regulation) and under neutral conditions (i.e., in the absence of
convection or subsidence), can be expressed as

62
 (2.9)

Table 2.3. Values of air density, saturation vapor pressure, tatent heat of
vaporization, psychrometric constant, and slope of saturation vapor
pressure curve as a function of air temperature

63
T ['e) p� [kg m41 e"" [kPal l..[kJ krl y(kPa'C-<1 So (kPa 'C-<I

0 1.292 0.611 2.501x1()1 0.0654 0.044

I 1.284 0.657 2.499 x 101 0.0655 0.047

2 1.279 0.706 2.496xl0' 0.0656 0.051

3 1.274 0.758 2.494x10' 0.0656 0.054

4 1.270 0.814 2.492x 10' 0.0657 0.057

5 1.269 0.873 2.489x 10' 0.0658 0.061

6 1.261 0.935 2.487x10' 0.0659 0.065

7 1.256 1.002 2.484 X 1()1 0.0659 0.069

8 1.252 1.073 2.482 X 10' 0.0660 0.073

9 1.247 1.148 2.480 X 10l 0.0660 0.078

!O 1.243 1.228 2.478 X 10l 0.0661 0.082

II 1.238 1.313 2.475x 10' 0.0661 0.087

12 1.234 1.403 2.473 X 10' 0.0662 0.093

13 1.230 1.498 2.470x 101 0.0663 0.098

!4 1.225 1.599 2.468 X 10) 0.0663 0.104

15 1.221 1.706 2.466 X 10' 0.0664 0.110

16 1.217 1.819 2.463x 10' 0.0665 0.116

17 1.213 1.938 2.461 X 10l 0.0665 0.123

18 1.209 2.065 2.459 X 10l 0.0666 0.130

!9 1.205 2.198 2.456 x10' 0.0666 0.137

20 1.204 2.337 2.454 X 10' 0.0667 0.145

21 1.196 2.488 2.451 X 10' 0.0668 0.154

22 1.192 2.645 2.449x 10' 0.0668 0.161

" 1.188 2.810 2.447 X 10' 0.0669 0.170

24 1.186 2.985 2.444 X 10' 0.0670 0.179

25 1.184 3.169 2.442 X 10' 0.0670 0.189

26 1.180 3.363 2.440x10' 0.0671 0.199

27 1.176 3.567 2.437 x 10' 0.0672 0.209

18 1.172 3.781 2.435 X 10l 0.0672 0.220

" 1.169 4.007 2.433 X 10l 0.0673 0.232

30 1.165 4.243 2.430x101 0.0674 0.243

31 1.161 4.494 2.428x 10l 0.0674 0.256

32 1.157 4.756 2.425 x10l 0.0675 0.269

64
33 1. 153 5.032 2.423x 101 0.0676 .
0 282

34 1.149 5.321 2.421 x 10' 0.0676 0.296

3S 1.149 5.625 2.418>< 101 0.0677 .

0 311

36 1.142 5.943 2.416 x 101 0.0678 0. 326

37 1.138 6.277 2.414 )( 101 0.0678 0.342

38 1.135 6,627 2,411><101 0.0679 0.358

39 1.131 6.994 2.409x 10' 0.0680 0.375

where Po is air density (sec Table 2.3), Pw is liquid water density, 1nO
indicates the natural logarithm, k = 0.4 is the VOIl Karman's universal
constant (dimensionless), Po is the atmospheric pressure, Zu is the height at
which "a and e" arc measured (Figure 2. 1 0), Zo is the roughness height, a
parameter that accounts for the roughness of the surface (Table 2.4); and z"
is a parameter known as displacement height, which accounts for the fact
that in the presence of a forest canopy the surface over which the wind
blows is not at the ground level but displaced by a height that is a function
of the tree height, II (i.e., zd:::: 0.65h) (Campbell and Nonnan, 1998). In the
case of transpiration the diffusion of water vapor from the stomata is
controlled by the plant, which can reduce the opening of the stomata to
limit water vapor losses in conditions of limited soil water availability.
Stomatal regulation can, therefore, reduce the value of K£ with respect to
the values given by Equation 2.9.
As expected, looking at Equations 2.8 and 2.9, evapotranspiration rates
are greater in the presence of dry air (i.e., eo < eS(l/) and stronger winds.
Evapotranspiration is also enhanced by the roughness of the surface. This
representation of evapotranspiration as a diffusive flux is due to the
English scientist John Dalton ( 1802).

2A.3 The energy balance

According to the first law of thermodynamics, the net radiant energy

received at the land surface must be conserved. As in Chapter 1 for the
conservation of mass, we consider a control volume and seek to balance
energy inflows with outflows and net changes in energy stored within the
system. For convenience, we establish an imaginary control volume near
the land surface that includes a very thin top layer of soil (say, 10 mm
thick), the vegetation, and the immediate surrounding air (Figure 2.11).
Thermodynamic principles hold Ihat the net radianl energy arriving across
the boundary of this system must be exactly balanced by other energy

65
fluxes across the boundary and the net change in energy held within the
volume. The energy may change among its possible forms (radiant,
thermal, kinetic, and potential), but it must be conserved.
All matter has internal energy, known as heat Q [M L21"2], which is
due to the kinetic and potential energy associated with individual
molecules. Heat is an extensive property (it depends on the amount of
material) and is expressed in units of calories or joules. There are two
Iypes ofheal: sensible and latent heat.

Table 2.4. Some example of surface roughness for a few types of ground
surfaces and vegetation covers
•

Surface Zo (cm)

Ice 0.001
Dry lake bed 0.003
Calm, open sea O.QI
Closely mowed grass 0.1
Snow-covered farmland 0.2
Tilled bare soil 0.2-0.6
Thick grass (50 cm high) 9
Forest (on level ground) 70-120
•

Source: Data from Campbell and Nonnan (1998).

Sensible heat is proportional to temperature. As the name implies, this

is the heat you would "sense" by contact or touch. The specific heat
I
capacity cp [L2 E)- 1"2] provides a measure of how a substance' s sensible
heat changes with temperature. It is the amount of heat that is required for
a unit increase of temperature in a unit mass of that substance. For
instance, in the case of air cp is about I J kg-1 K-1 , which means that I J of
energy is required to increase by I K the temperature of I kg of air. Based
on this definition, cp can be expressed as

where m is mass and T(K) is absolute temperature (note that absolute

66
temperature is represented in degrees kelvin, which is indicated as "K").
We can use the specific heat capacity to determine how the temperature of
a given mass of water will change if we add energy (heat the water). The
heat capacity of water at 20°C is approximately 4.2 103 J kg-I K-l or 1.0
x

cal g-1 K-l. If we have 1.0 kg of water and add 12 kJ (kilojoules) of

energy, then the temperature change can be calculated as:

I!.Q = 12000)
I!.T= =2.9K
mc" 1.0kgx4.2xIOJJkg' K-'

or the water temperature will increase by approximately JOe. We have

assumed in this problem that we have not evaporated any of the water.
Latent heat is the portion of internal energy that canl/ot be sensed or
felt. Instead, the latent energy is the amount of internal energy that is
released or absorbed during a phase change, at a COllstallllemperature.
Evaporation involves a liquid to vapor conversion, which requires energy
to be added to the water. This energy is called the latent heat of
vaporization:

I
--------

I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
----- ------- ----
___ I

67
Figure 2.11 A control volume for energy conservation. Solar energy (R'l)
entering the control volume must be balanced by fluxes of energy out of
the volume and by the time rate of change in energy stored (dQldt).

(2.10)

This tells us that we need to add about 2.5 million joules of energy to
evaporate 1 kilogram of water. Other types of latent heat energy include

• latent heat of melting: A.m = 3.34 x 105 J kg-1 at O°C

• latent heat of sublimation: )' = 2.83

" x 106 J kg-1 at O°C

Note that evaporation of snow or ice involves a change in phase from solid
to vapor. This may occur in either two separate steps, melting and then
evaporation, or in one step as sublimation, which is the direct phase
change from ice to water vapor. In either case latent heat is added to
support the sum of the two unique phase changes, as evidenced by the fact
that A.s = A.,. + A",.
The net solar radiation (Rn, i.e., the fraction of solar radiation that is
received at the surface and is not reflected back to the atmosphere) warms
the exposed surfaces inside the control volume. When water is present
some of this heat energy is absorbed by the water to support the phase
change from liquid to vapor (evapotranspiration). Evapotranspiration will
not typically absorb all the energy, and so the surface continues to warm.
As the surface becomes warmer than the air and the underlying soil, heat
will be conducted from the hot surface to the air (H) and from the surface
down into the soil (G). During this process, both heat and water vapor
have been added to the air inside the control volume. The moist, warm air
becomes less dense than the surrounding air and tends to rise up and out of
the control volume. Therefore, we see that the main energy outlets for the
sun's energy that hits the Earth's surface are conduction into the soil,
conversion of liquid water to water vapor, and heating of the overlying air.
The energy used to evaporate the water is stored in the water vapor and is
removed from the system as the water vapor is removed. Finally, the latent
heat (evaporation energy) is converted back to thermal energy when and
where the water recondenses. From this we see that the flux of water vapor
induced by evapotranspiration is associated with a flux of latent heat, £,.
The latent heat flux is related to the rate of evapotranspiration through the
latent heat of vaporz
i ation:

68
 EI
el "" . (2.11)
p�A.\·

where et = the evapotranspiration rate [L 11]; p", = density of water [M L-

3]
; � = latent heat of vaporization at the temperature of interest [L 2,2].
The energy balance approach to evaporation was proposed by Bowen
( 1 926) and is based on the faci that evapotranspiration involves an energy
flux (i.e., of laten! heat energy to the atmosphere). The rate of
evapotranspiration can be calculated llsing Equation 2 .11 with the latent
heal flux detennined through the energy balance equation (Figure 2. 1 1 ):
dQ
-=R-G-H- £I' (2.12)
dl n

where Rn = the net (solar) radiation input, G = energy output through

conduction to the ground, H = net output of sensible heal to the
atmosphere, E, = output of latent heat to the atmosphere (the latent heat
flux) due to evaporation, and Q = the amount of heat energy stored in the
control volume per unit area of surface. Each of the inputs and outputs on
the right side of Equation 2 . 1 2 is in the form of an energyflux (i.e., energy
l
per unit area per unit time). In SI units this would be J m-2 S- , or W m- 2.
RIO is an input of energy to the control volume. Therefore, the convention
adopted here is to express RIO as a positive quantity. Conversely, in the
energy balance Equation 2.12, G, H, and E, are assumed to be positive
when directed outward from the control volume (i.e., when they are energy
outflows). We can rearrange Equation 2. 1 2 to solve for E" the latent heat
flux:
dQ
E=R-G-H-- . (2.U)
dl
I n

We can substitute Equation 2. 1 3 into Equation 2. 1 1 and rearrange to solve

for the evapotranspiration rate:
-G-H-dQldl
el = R� . (2.14)
P�,A.

Let's consider a simple example of how we might apply the energy

balance approach to calculate the daily evaporation from a forest on a
sunny day (RIO = 200 W m-2) in Virginia, USA. If we can neglect G and H
and assume that the heat (Q) stored within the forest remains
approximately constant, then Equation 2 . 1 4 becomes:

69
 � 200Wm-�
et: O'J I ",8.0xlO-i IUS-I :O.7anday-.1
PW./"'J "'(000
1 kgm-32-
X .)X 1 kg- )
�

In practice, it is common 10 neglect changes in heat storage because of the

relatively low heat capacity of air (the same assumption cannot be made in
the case of water or soits; therefore, we choose a control volume that does
not include a substantial mass of water or soil). However, typically we
cannot neglect the sensible heat flux H. Also, for averaging times
significantly less than a day we cannot safely neglect G either. Thus, to
estimate the evapotranspiration rate using Equation 2.14 we need to
dctcnnine H. Sensible heat is transported by diffusion across gradients of
temperature. This indicates that H should be proportional to the
temperature difference between the surface and the overlying air (Figure
2.10), and can be expressed using a formulation similar to the case of
water vapor Ouxes
(2.15)

with

(2.16)

2AA Other methods

In the previous sections we have considered two methods that can be used
to calculate the evapotranspiration rates. The mass-transfer method
outlined in Section 2.4.2 uses Equations 2.8 and 2.9 to detennine et as a
function of wind speed (uo)' air vapor pressure (eo), the saturation vapor
pressure at the surface (esol' which is a function of surface temperature Ts),
and a number of parameters that depend on air temperature and surface
characteristics (Tables 2.3 and 2.4). The energy balance method outlined
in Section 2.4.3 uses Equations 2.14 to 2.16 to calculate et as a function of
the net radiation (R), heat conduction to the ground (G), wind speed (110),
surface temperature (Ts), the air temperature (To) at the height (zo), and the
set of parameters reported in Tables 2.3 and 2.4.
It can be convenient to combine the two methods to eliminate one of
these variables, preferably a quantity that is hard to measure. The Bowen
ratio method eliminates the dependence on wind speed. It is based on the

70
definition of Bowen ratio (B)-the ratio between sensible and latent heat
fluxes-and can be expressed (using Equations 2.8, 2.9, 2.15, and 2.16) as

B=!i=r T,-T"
E,
( e_-e"
l. (2.17)

where y = cp pj(O.622 t..) is known as "psychrometric constant" and is

calculated as a function of the air temperature, T(l (Table 2.3). If we
express sensible heat in Equation 2.14 as H = B EI = P Av et and solve for
....

et we obtain
R,,-G
1.'1 = . (2.18)
p�A.,.(l + B)

The Bowen ratio method uses Equations 2.17 and 2.18 to calculate the
evapotranspiration rate. Because Equation 2.17 is obtained using
Equalions 2.8 and 2.9, it docs not account for the effect of stomatal
regulation and can only be used in the case of evaporation from a lake, or
other wet surfaces. We see from Equation 2.17 that as a surface becomes
wanner (Ts increases) and drier (eSat decreases), the Bowen ratio tends to
increase. Consequently, sensible heat flux increases relative to latent heat
flux. The Bowen ratio ranges from approximately 0.1 in very humid
regions (8 0.1 over the oceans) to values greater than I in arid regions.
-

In deserts, 8 may be greater than 10. With estimates or measures of Rn, G,

and 8, Equation 2.18 provides a simple and direct estimate of et.
The combination method (or Penman's method) combines mass
transfer and energy balance approaches to eliminate the dependence on the
surface temperature (�,), which is oftcn difficult to mcasure. In this case
the evaporation rate is expressed as

e/=
Sa [R" G J + KIIIIAc"" (T,,)
- -
e" J. (2.19)
p A,.(S" + r)
•. (S" +r)

where S" is the slope of the saturation water pressure curve (Figure 2.1); its
values are reported in Table 2.3. As in the case of the mass transfer method
presented in Section 2.4.2, Equations 2.18 and 2.19 do not account for the
effect of stomatal regulation. Thus, the Bowen ratio method and the
Penman method can be used to estimate evaporation but not transpiration.
The Penman Equation 2.19 can be modified to account for stomatal
regulation (known as the Penman-Montieth method), though this
introduces a number of additional parameters. Brutsaert (2005) provides a

71
good explanation of the Penman-Montieth method.

2AoS Soil moisture controls on e"apotranspintion

In general we need to consider the state (wetness) of a surface to

understand and quantify how the solar energy reaching the earth surface is
partitioncd. When watcr is in limitcd supply, thc surface bccomcs wanncr
than in the wet cases and more of the energy is removed from the control
volume through conduction into the soil and heating of the air. Hence, in
this case the surface properties rather than the atmospheric properties
control the rate of evapotranspiration. Under wet surface conditions, the
rate of evapotranspiration is governed by the supply of radiant energy, the
relative dryness of the air, and the efficiency of the wind in removing the
water vapor from the surface. This is evidenced by how quickly water
evaporates from wet clothes on a dry day, as compared to a humid day. We
know intuitively that for the same example, higher winds will increase the
evaporation rate, and reduced solar input, say from heavy cloud cover, will
reduce the evaporation rate.
Through field experiments we can explore the relationship between the
surface wetness and the partitioning of the received energy between
evaporation and heating of the air and soil. Figure 2.12 shows a set of
experimental measurements of the terms in the energy balance over an
irrigated bare soil field in Davis, California. After wetting the soil with
sprinkler irrigation the energy balance tenns were monitored and graphed
against time for a two-day period. The net radiation was very similar from
day to day, except for a few passing clouds on the afternoon of day 2. The
main difference between the two days was that the surface was wet on day
[ and partially dry on day 2. Notice how the measured latent heat nux
(from evapotranspiration) is nearly twice as large as the heating of the air
(If) on the wet day. On the drier day the two terms are nearly equal. (f we
were to plot additional days (following day 2), where the soil becomes
increasingly dry, we would see H begin to exceed £, as the
evapotranspiration steadily declines.

72
 600

1\ R"
500
G f\.
---
H
400
£,
-

300
"
,

E
=:
� .
-

><
200
,�
"
u..
-

(/ \
/ ,
-
100
' ,

',t
, ,

,
, ,
,
,
,

\.\\.
,

--'
,

1'"
c
�
, ,
o

�'"
,�-----
-
-
.-

.
-

-100 ..
, . , . .
, • .
, •
, .
, ..
,
. . .
E E E E E E E E E
• • • . . •

• • � � • • � � •
. . . .

0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 .. 0 0
N '" N '" N '" N iri N
� � � � �

Figure 2.12 Measured energy fluxes from an experimental field in

California. As the surface dries during the experiment, the latent heat flux
(EI) is reduced and the sensible heat flux (H) increased.
Dala courtesy of John D. Albertson.

The rate of el that occurs under the prevailing solar inputs and
atmospheric properties, if the surface is fully wet, is commonly referred to
as potential evapotranspiration (PET). By definition PET does not depend
on surface properties and is not affected by stomatal regulation. For a
catchment water balance, we arc interested in the actual et, because that is
the rate at which water is removed. The concept of PET is useful as a tool
if we have some means to estimate actual et from knowledge of the
potential rate and the nature of the surface conditions. By definition, when
the surface is wet the ralio of ei to PET will be unity. Conversely, when
the surface is completely dry, the ratio will go to zero. The transition

73
between these two extremes is shown in Figure 2.13 for the simple case of
evaporation from a bare soil field. The actual et was detemlined from
detailed measurements of wind velocity and vapor pressure in the
atmosphere; PET was calculated from measured atmospheric properties.
1 ,------,

o
I:
2 0.75 -
1ii
.

.
�

a.
-

on
I:
.. o
0.5 -
�

'0 o
a.

� o
.. o
-

o
i
o
.-

0.25-
(;
a.
-

O� __
-- -
--._
, -- -- -- - - --
----
- ,
._ -- -- --�
o 0.25 0.5 0.75 1
DRY SATURA TED
Degree of soil saturation
et to PET decreases as the available soil
Figure 2.13 The ratio of actual
moisture decreases.
Data courtesy ofJohn D. Albertson.

Currently there are no widely used methods for estimating areal

distributions of actual et, although satellite-derived surface data offer
promise of delivering useful estimates of components of the surface energy
balance from which actual ef might be inferred (Moran and Jackson,
1991). In hydrology, computations using empirical evapotranspiration
equations typically arc used to estimate actual et. For a complete
discussion of techniques to measure and estimate et, see Brutsacrt (1982).

74
2.5 Concluding Remarks

This chapter has presented the concepts of precipitation,

evapotranspiration, and interception, which are fundamental components
of the hydrological cycle. A number of issues were raised in the
introduction dealing with how we quantify these fluxes and stores. Our
conclusion that it is very difficult to predict precipitation led us to
investigate how hydrologists use statistical analyses to describe the spatial
and temporal patterns of precipitation. Because we typically use point
measurements of precipitation, there are significant problems with
extrapolating these values to land areas of interest (e.g., catchments).
Studies have shown that two precipitation gages (on a ridge exposed to the
wind), separated by only a small distance (3 m), can produce results that
differ consistently by as much as 50% (Court, 1 960). Because of the
temporal and spatial variability in precipitation fluxes, it is no wonder that
water management in a basin as large as that drained by the Colorado
River is a significant challenge. Despite problems with interpretation, it is
apparent that radar and satellite data will be essential for resolving the
measurement problem of areal precipitation.
It has been known for many years that changes in catchment vegetation
can alter the water budget. One dramatic example is deforestation.
Clearing land areas of trees typically decreases the amount of transpiration
and interception (and consequently evaporation from interception stores);
as a result, surface runoff and soil moisture increase. Data summarized in
Dunne and Leopold (1978) indicate a linear relationship between the
percentage of vegetation removed and the first-year increase in
streamflow. More recent reviews (Stednick, 1996; Brown et aI., 2005)
indicate that a simple relationship between deforestation and increased
streamflow is not so apparcnt, and requires further study (see Chaptcr 9).
As vegetation regenerates, interception and evapotranspiration increase,
and surface runoff declines. Other changes in catchment water dynamics
accompanying deforestation have also been observed. Evaporation is
increased from exposed soils. Infiltration is increased, enhancing leaching
of soiI solutes. The increased surface runoff accelerates soiI erosion. It
may take many years for reforestation to halt these effects (Stednick,
\996).
In introducing the concepts of the hydrological cycle and catchment
water budgets in Chapter I, several fluxes and reservoirs were discussed.
This chapter has dealt with the connections between the land and the
atmosphere. Ultimately, there is a close coupling between the atmosphere
and the land surface, which controls the flux of water across this interface.

75
Subsequent chapters will deal with the terrestrial portion of the
hydrological cycle. To proceed to a physical treatment of water flows in
streams and beneath the Earth's surface, we will need to gain a basic
understanding of fluid mechanics. The physics of fluids wi II be the subject
of Chapters 3 and 4.

2.6 Key Points

• The three primary steps in the generation of precipitable water in the

atmosphere are: creation of saturated conditions in the atmosphere,
condensation of water vapor into liquid water, and growth of small
droplets by collision and coalesccncc until thcy bccome largc enough to
precipitate.

• Precipitation onto the land surface is measured with relative ease at

individual points or stations by using precipitation gages. A record of
precipitation versus lime is called a hyelograph. Areal distribution of
precipitation can be estimated using weather radar. {Section 2.2.1 }

• Mean areal precipitation can be estimated using data from multiple

stations within a catchment using the isohyetal method. { Section 2.2.2}

• Hydrologists use frequency analysis to describe the temporal variability

of historical observations from individual stations. Annual precipitation
data often are described well by a normal distribution {Section 2.2.3} ,
while rainfall events are typically distributed according to a Gumbel
distribution. { Section 2.2.4}

• A portion of the gross or total precipitation may be temporarily storcd

above the land surface by vegetation and then evaporated. This is
referred to as interception. {Section 2.3 }

• Evapotranspiration from a catchment includes a variety of different

vaporization processes, including evaporation from open water, soils, and
vegetation surfaces, and transpiration from plants. {Section 2.4}

• By making appropriate assumptions and by collecting appropriate field

data, a long-tenn average rate of evapotranspiration can be estimated
using the mass balance approach. {Section 2.4. 1 }

• The mass transfer approach is based on a representation of the water

vapor fluxes associated with evapotranspiration as a diffusion process

76
 along gradients of decreasing water vapor pressure {Section 2.4.2}

• The energy balance approach is based on the principle of conservation of

energy in which the nct solar radiation (Rn) is partitioned between latent
heat flux (E,) to the atmosphere, sensible heat flux (N) to the atmosphere,
heat flux to the ground (G), and the change in energy stored in the
control volume (dQldt). {Section 2.4.3}

• Some of the most commonly used methods for the calculation of

evapotranspiration (e.g., the Bowen ratio and Penman methods) are based
on a combination of mass transfer and energy balance approaches.
{Section 2.4.4}

• Over land surfaces water may be limited in availability (e.g., in dry

soils), and rates of evapotranspiration are then reduced from those over a
fully wet surface. Actual evapotranspiration (et) is commonly much less
than potential evapotranspiration in catchments where moisture
limitations occur, but et can approach potential evapotranspiration in
extremely wet climates where energy may become limiting. {Section
2.4.5}

2.7 Example Problems

Problem 1. Two tipping bucket rain gages are used to collect the
following rainfall data:

77
 Cumulative precipitation (mm) Cumulative precipitation (mm)
Station #Z
Time
Station'1

4:00a.m. 0.0 0.0

6:00a.m. 0.0 0.0

8:00a.m. 1.0 1.0

10:00 a.m. '.0 3.0

12:00 noon 13 11

2:00 p.m. 1.7 15

4:00 p.m. 19 16

6:00 p.m. 19 17

8:00 p.m. 19 17

10:00 p.m. 19 17

12:00 midnight 19 1.7

2:00a.m. 19 17

4:00a.m. 19 17

A. Calculate the mean daily rainfall intensity for each station (mm hr-1 ).

B. Calculate the maximum 2·hour rainfall intensity for each station (mm
hr-1 ).

C. Calculate the maximum 6·hour rainfall intensity for each station (mm
hr-1).

D. Using the arithmetic average method and knowing that the drainage
basin area is 176 mi2, calculate the total volume of rainfall (m3)
delivered to the basin during the event.

Problem 2. Measurement of changes in volume of water in an evaporation

pan is a standard technique for estimating potential evapotranspiration.
United States Class A evaporation pans are cylindrical with the following
dimensions: depth = 10.0 inches and diameter = 47.5 inches. An
evaporation pan can be considered a hydrological system with an inflow,
outflow, and storage volume. Evaporation from pans is not the same as
evaporation from natural surfaces for a variety of reasons. For example,
water temperatures in shallow pans will be much more variable than
temperatures in a nearby lake. Evaporation measured in pans is adjusted by
a factor called a pan coefficient to convert to an estimate of potential
evapotranspiration (see Brutsaert, 1982).

78
A. Calculate the cross-sectional area (m2) of a United States Class A
evaporation pan through which inflows and outflows of water can pass.
Also, calculate the total storage volume of the pan (m3).

B. Initially, the pan contains 10.0 U.S. gallons of water. Calculate the
depth of water in the pan (mm).

C. Assuming a water density of 997.07 kg m-3 (25°C), calculate the mass

(kg) of water in the pan.

D. After 24 hours in an open field (no precipitation), the pan is checked

and the volume of water left in the pan is determined to be 9.25 gallons.
Calculate the average evaporation rate (mm hr-1) from the pan.

E. The pan is emptied and refilled with 10.0 gallons of water and left in an
open field for another 24 hours. During this period, rain fell for a 3-hour
period at a constant intensity of 2.5 mm hr-1; after 24 hours, the volume
of watcr in the pan was 11.50 gallons. Calculate thc average
evaporation rate (mm he l ) from the pan during this period

F. If the evaporation rate calculated in E remains constant and no

additional precipitation occurs, then estimate the time (days) for the pan
to empty as a result of evaporation.

Problem 3. For Problem 2 part 0, calculate the flux of latent heat from the
water in the pan to the atmosphere (W m-2). Use a water density, Pw =
1000.0 kg m-3.

Problem 4. A small (area = 300 ha) catchment in Iowa absorbs a mean Rn

= 330 W m-2 during the month of Junc. In this problem, apply the energy
balance approach to estimate evapotranspiration from the catchment
during the month of June.

A. Write a complete energy balance equation (i.e., including all tenns) for
the catchment for the month of June.

B. Neglecting conduction to the ground (G) and the change in energy

stored (dQldt = 0), simplify your energy balance equation for the
catchment so that it can be solved for the latent heat flux, E,. Also,
replace the term H (the sensible heal flux) with B x E, (where B is the
Bowen ratio).

79
C. Using a mean Bowen ratio ofD.2D for the catchment, calculate the mean
daily flux of latent heat to the atmosphere (W m-2) and the mean
evapotranspiration rate (mm day-I) from the catchment. Usc a water
density,pw= 1000.0 kg m-3.

D. Calculate the tolal evapotranspiration from the catchment during the

month of June (mm).

2.8 Suggested Readings

Campbell, G.S., and J. M. Norman. 1998. AI/ Introduction to

Environmental Biophysics, New York: Springer.
Oke, T.R. 1987. Boul/daly Layer Climates, 2nd ed. New York: Routledge.
Chapler I, pp. 3-32.

80
3 The Basis for Analysis in Physical
Hydrology: Principles of Fluid Dynamics

3 . 1 Introduction

3.2 Definitions and Properties

3 .3 Forces on Fluids

3.4 Fluid Statics

3.5 Fluid Dynamics

3 . 5 . 1 Fluid acceleration

3 .5.2 The Bernoulli equation

3.5.3 Applications of the Bernoulli equation

3.6 Energy Loss

3 .6. 1 Friction factor

3.7 Laminar and Turbulent Flows

3 . 7 . 1 Laminar pipe flow

3.7.2 Turbulent pipe flow

3 .8 Concluding Remarks

3.9 Key Points

3 . 1 0 Example Problems

3 . 1 1 Suggested Readings

3.1 Introduction

Hydrology is concerned with the occurrence and movement of waters on

and beneath the surface of the Earth. Water moves, or flows, in response to
forces arising from gravity and pressure, for example, that act on water
above and below the ground. Much can be learned about the movement of
water on and in the Earth by considering the forces causing l1uid motion
and the response of l1uids to those forces. The study of the physical

81
processes governing fluid motion is called fluid dynamics. Flow of water
in streams and subsurface aquifers, infiltration of precipitation into soils,
cvaporation, design of flood control measures, and transport of
groundwater contaminants are just some of the problems hydrologists
study that depend on a knowledge of fluid dynamics. In hydrology we are
obviously most concerned with understanding the behavior of water, but
the principles discussed in this chapter apply to a broad range of fluids.
The notions of force balance and Newton's second law (force equals
mass times acceleration), familiar from basic solid mechanics, can be
extended to fluids, including water. The forces that cause fluids to move
are due to gravity, pressure differences, and surface stresses. In this
chapter we analyze relatively simple cases such as flow through a pipe or
hose, but the same principles apply to flow in streams and through soil and
rocks. These applications are explored in subsequent chapters. The basic
principles developed in this chapter also apply to the flow of air, lava,
glaciers, and the Earth's mantle. While all of these can be described as
fluids, they differ dramatically in their viscosity. The viscosity of lava is
almost a billion times that of air. We will find that fluid viscosity plays an
important role in the nature of fluid flow. Flows in which viscous forccs
dictate the nature of flow are called laminar flows; flows in which viscous
forces are relatively unimportant are often turbulent. Most groundwater
flow is laminar, while most surface-water flow is turbulenl.

3.2 Definitions and Properties

We usually classify matter as either solid, liquid, or gas, based on

macroscopic properties. Typical classifications are made on the basis of
the most easily specified diffcrences among the three types of matter: a gas
takes on the shape and volume of a container, a liquid takes the shape of
the portion of the container that it fills but retains a fixed volume and a
solid has its own defined shape as well as volume. Liquids and gases are
called fluids.
When describing fluid motion, a more fundamental definition of fluid is
needed. The word fluid comes from the Latinflllere (to flow) and a
definition found in a dictionary might be "a substance capable of flowing."
The scientific use of the term is not much different: a fluid is a substance
that continuously deforms when subjected 10 a shear stress. Shear stress is
a tangential force (i.e., a force parallel 10 a surface) per unit area acting on
a surface. Our experience tells us that fluids move in the presence of a
shear stress; for example a puddle of water will be disturbed if a wind

82
exerts a tangential stress on the water surface.
The above definition is valid for any fluid. The rate at which
defonnation occurs, however, is not independent of the fluid itself. The
property of a fluid that describes the resistance to motion under an applied
shear stress is tenned viscosity. The viscosity of honey is greater than the
viscosity of water, which in tum is greater than the viscosity of air. This
common sense notion of viscosity corresponds with the scientific notion.
To fonnulate a precise definition of viscosity we will consider a simple
conceptual experimcnt. Imagine a very large, thin sheet of matcrial (area =
A) floating on a very shallow layer of water (depth = d) lying above a
horizontal boltom surface. Suppose further that the sheet is pulled (with
force F) across the surface of the water. This experiment can be
conceptualized in more concrete lenns as attaching a rope to the edge of a
square sheet of Styrofoam 2 meters on a side, floating this on a layer of
water that is 5 mm deep at the bottom of a large, flat-bottomed pool, and
pulling slowly on the rope to cause the Styrofoam to move tangentially
(parallel to the water surface) across the surface of the water. We would
find thai the fluid adjacent to the bottom of the pool would "stick" to the
bottom and not move and that the fluid adjacent to the underside of the
floating sheet of Styrofoam would stick to that surface and, therefore,
move with the same velocity as the sheet itself, say up/ale (Figure 3. 1).
After a short time we would find that the velocity of the water between the
sheet and the pool bottom is proportional to the distance above the bottom,
as long as the product lip/ale x d is less than about 0.002 m2 S-l; the reason
for this limit is explored in Example Problem 3 ( Section 3. 10). For
example, at a distance of d/2 units above the boltom-halfway between the
bottom surface and the floating plate-the water velocity would equal
IIp/O/e12. We would also find that the applied force divided by the area of
the plate is proportional to the surface (plate) velocity divided by the water
depth or

F 11[>1<",
-=JI (3.1)
A d
The constant of proportionality, 11, is the viscosity of the fluid.
Equation 3 . 1 is a specific case of an important relationship for
understanding the flow of water, air, and many other fluids. More
generally, FIA is the tangential force per unit area (shear stress) acting on
the fluid surface and lip/auld is the slope of the curve (a straight line in this
case) relating fluid velocity to distance above the bottom boundary. A
curve relating velocity to distance above the bottom boundary is called a

83
velocity profile; the slope of such a curve is called the velocity gradient.
We can rewrite Equation 3.1 to relate the shear stress to the velocity
gradient:
d"
=
r J1 d:: ,
(3.2)

where ! shear stress [M L- I 12] ; duldz = the velocity gradient [I I ]; /I =

=

velocity in the x-direction [L I I ]; and 11 = dynamic viscosity [M L- I II ].

(Note that equivalent dimensions for viscosity are [F L-2 T], where F is
force. Thus viscosity is often reported in SI units of pascal· seconds.)
Equation 3.2 is known as Newton's law of shear and provides a definition
for viscosity: the viscosity of a fluid is the ratio of the shear stress at a
point in the fluid to the velocity gradient at that point.

area of floating plate F,

applied
force

d,
z water depth

Figure 3.1 Schematic diagram of hypothetical experiment to detennine the

viscosity of water. Viscosity is given by the ratio of the applied force per
unit area (FIA) to the slope of the velocity profile (Uplate-'d).

The phrase "at a point" in the above definition deserves further

comment. We want to use the idea of a point in the mathematical sense
because the use of calculus is necessary for quantitative descriptions of
fluid systems. However, we know, because of the molecular nature of
matter, that if we picked a point at random in a fluid, that point might be
interior to an atomic particle or alternatively might be in the space between
particles. The fluid "properties" associated with such a point would
obviously depend on the location of the point. We avoid the problems
associated with an approach that recognizes the discrete nature of matter
by making the continuum assumption: the fluid is idealized
macroscopically as bcing continuous throughout its entirety-the

84
molecules are pictured as being "smeared" or "averaged" to eliminate
spaces between atomic particles. Thus, we will use the terminology "at a
point" so that we havc the mathematical tools that are necessary, but it
should be borne in mind that the "point" actually represents average
conditions over a small volume. For our purposes explanation in physical
hydrology-this assumption is valid. (There are situations involving
rarefied gasses important in other branches of science for which the
continuum assumption may not bc acceptablc.)
Oncc the continuum assumption has bcen made, we can define the
density of fluid at a point. Density is simply mass per unit volume. If
density varies spatially, the ratio of mass to volume will vary with the
sample size. Hence, the definition of density at a point is the limit of the
ratio of mass to volume, as the volume under consideration shrinks to an
infinitesimal:
,. aM
p= lin . (3.3)
<1.1'-0 AV

where p density [M V3] and 6. V is a small volume of fluid [L3] with a

= =

mass 11M [M]; the "11" before V and Mjust reminds us that we are
considcring small volumes and masses. Of course, the continuum
assumption means that density " at a point" is really the average density of
a volume that is small compared with the macroscopic scale of the fluid
motions being studied but large compared to intennolecular distances. If
the density of a fluid does not vary spatially, the fluid is said to be
homogeneous. A fluid property related to density is unit weight. The unit
weight (y) of fluid is its weight per unit volume and is equal to the product
of density and the acceleration of gravity, that is y pg.
=

The viscosity of water at 20°C is approximately 1.0 x 10-J

pascal'seconds (Pa . s) and the density of water at that temperature is
slightly less than 1000 kg m-J (or 1.0 g cm-3); a pascal is 1 newton m-2
(see Appendix 1). Both viscosity and density are temperature dependent.
Viscosity decreases with increasing temperature from 1.787 x 10-3 Pa . s
at O°C to 0.7975 x 10-3 Pa . s at 30°C (Table 3. I), a range that includes
much of the variation expected in the natural environment. Density varies
with temperature in a somewhat more complex fashion. Water exhibits a
density maximum at about 4°C; the density decreases slightly both above
and below that temperature. More extensive infonnation on the density
and viscosity of water is providcd in Appendix 2.
We should note that density also depends on pressure. Pressure will be
defined in the next section, but, roughly speaking, it is the force per unit

85
area that tends to compress a fluid. Values of density given in Table 3.1
are for a pressure of I atmosphere. The deviations in density caused by
changes in pressure are often ncgligible. In these cases, the fluid is said 10
be incompressible. However, there are cases of importance in
environmental sciences where the compressibility of water must be
considered and, thus, the variation of density with pressure taken into
account. These cases include the variation of water density with depth in
the oceans and with the variation of confining pressure in deep aquifers.

Table 3.1. Properties of water as functions of temperature

Temperature, ·C Viscosity, j.1. Pa·s Density, p, kg m-3

0 1.787 x 10-1 999.87

5 1.519 x 10-1 999.99

10 1.307x 10-1 999.73

15 1.139 x 10-3 999.13

20 1.002 x 10-) 998.23

25 8.904 x 10.... 997.07

30 7.975 x 10.... 995.67

3.3 Forces on Fluids

In discussing the flow of water (or any other fluid) it is necessary to

consider the forces acting on a fluid particle (a very small volume of fluid).
These forces are usually divided into two classes, body forces and surface
forces. Body forces are those that do not require direct contact with the
fluid; they "act at a distance." One body force that plays an important role
in hydrological systems is the force due to gravity, the weight of the fluid.
Another example of a body force is an electromagnetic force. Surface
forces are those caused by direct contact between two fluid particles or
between fluid and solid. The tangential force in the conceptual experiment
we employed in defining viscosity is an example of a surface force-the
dragging of the water along by the floating plate required the plate to be in
contact with the fluid surface. A normal force is one oriented
perpendicular to a surface.
Instead of speaking of surface forces directly, we usually employ the
concept of force per unit area or stress. There are two types ofstrcsses that

86
we consider in fluid dynamics-normal stresses and tangential stresses.
The latter, as we have already seen, are termed shear stresses. The inward­
directed (compressive) normal stress, when applied to a fluid medium, is
referred to as pressure.

3.4 Fluid Statics

A special case of fluid motion is that ofa fluid at rest, i.e., the case of no
motion. If a fluid is not moving, there are no shear stresses present (by
definition) and only pressure need be considered when analyzing surface
forces. The special case of no motion has important applications and the
study involving such problems is referred to as fluid statics.
The variation of pressure with depth in a fluid at rest is one of the
fundamental relationships in fluid dynamics. The equation describing this
variation in pressure is known as the hyd rostatic equation. We will derive
this equation by considering a thin slice of fluid at rest and considering the
forces acting on it (Figure 3.2). The slice has a surface area A and a
thickness tlz. The horizontal pressure forces acting on the sides on the slice
must be equal and opposite so that they sum to zero. Ifnot, there would be
a net pressure force causing the slice of fluid to move sideways, at odds
with our assumption that the fluid is not moving. The upward force on this
slice, pA, is due to the pressurep on the bottom face of the slice. There are
two downward forces, the wcight of the slice itself and the force due to
pressure on the top face of the slice. The weight of the slice is given by the
product of density p times gravitational acceleration g times the volume of
the slice V (= Atlz). The pressure on the top face of the slice will differ
from the pressure on the bottom by some small amount I1p, so that the
downward-directed pressure force on the top face is (p + I1p)A. Because
the fluid is not moving, the upward and downward forces acting on the
slice must balance each other, giving:

(p + I1p)A + pgAtlz = pA,

which we can solve for the change in pressure !!.p from the top to the
bottom of the slice:

or

87
 /;. P = _
pg
/;.=

(p+/;.p)A

pA pg6zA

Figure 3.2 A thin slice within a static fluid. Because the fluid is at rest, the
upward and downward forces acting on the slice must balance.

As the thickness of the slice shrinks to an infinitesimal,

dp
-� -pg. (3.4)
do
Equation 3.4 is the hydrostatic equation. It indicates that the rate of
decrease of pressure with distance as we proceed upward (relative to
gravity) is pg, the unit weight of the fluid. The pressure is caused by the
weight of the overlying fluid. In many instances involving water we can
consider density to be constant and we can then integrate Equation 3.4 to
give pressure as a function of depth. Suppose the slice in Figure 3.2 is at a
level d units below the surface oflhe fluid and thaI z = 0 at the surface. We
will integrate Equation 3.4 from z = -d to z = 0, i.e., from the slice to the
surface. Rewriting Equation 3.4 in differential fonn,

dp = -pgd= (3.5)

we can integrate to obtain:

JP• dp = -pg f' d=.

P -d

where Ps is the pressure at the surface, typically atmospheric pressure.

Evaluating the integral gives

88
 p, -p � - pg[O - (-d)]

0'

p - p. = pgd. (3.6)

Equation 3.6 is a form of the hydrostatic equation that can be used to

determine the absolute pressure at any point in a static fluid of constant
density. Commonly, we take the pressure at the surface to be zero gage
pressure (Ps 0) so that what we actually are calculating is the pressure
=

relative to atmospheric pressure. To convert gage pressure to absolute

pressure, we would add the atmospheric pressure to gage pressure.
Throughout the remainder of this text, we will follow the convention that p
is the gage pressure. In this case, Equation 3.6 becomes:

p= pgd. (3.7)

Thus, the pressure in a constant density, stalic fluid is the product of the
unit weight and the depth. Pressure increases linearly with depth.
As an example, let us calculate the pressure on the bottom of the 1-
meter high pond wall shown in Figure 3.3a. The unit weight of water is 9.8
kN m-3 and the depth is 1 meter. Therefore, the pressure at the bottom of
the pond, and hence at the base of the wall, is 9.8 kN m-2. The pressure on
the wall varies linearly with depth such that the pressure at 0.5 m is 4.9 kN
m-2 and the (gage) pressure at the surface is 0 kN m-2 (Figure 3.3b) . The
lotal force on the wall is the integral of pressure over area. Ifwe consider a
I -meter width ofthc wall (perpcndicular to the plane of the page in Figurc
3.3), then the force on the wall is detennined by:

(a) (b)

1 m

Pressure at bottom
= 9.8 kN m-2

89
Figure 3.3 Water pressure on a vertical wall (a) increases linearly with
depth at a rate of9.8 kN m-2 per meter (b).

kN kN
Force per meter of width =
J'I -9.8
m
3 :d: = 4.9
In
.

One application of the hydrostatic equation is in calculations for drilling

fluid when an oil well is drilled into a pressurized formation (Figure 3.4).
If the pressure in the oil-bearing formation is not controlled, then a
blowout can occur. These occurrences are now rare, although the 20 I 0
Deepwater Horizon event in the Gulf of Mexico is a reminder of the
consequences of blowouts. One method for preventing blowouts as a well
is drilled is to fill the borehole with a drilling fluid (or drilling mud). The
hydrostatic pressure created by the column of mud counteracts the
formation pressure of the oil reservoir. Ifa well is drilled into an oil
2
reservoir with fomlation pressure equal to 1.8 x 1 04 kN m- at a depth of
2000 m, then the height of a column of drilling mud required to balance
the formation pressure ean be calculated using the hydrostatic equation

Prequired = form atioll pressure = Pmu£d,

where d is the height of the mud column above the formation. For a
drilling mud with unit weight (Pm,,,' g) = 10.5 kN m- ,
3

Prequired = 1.8 x 10 kN m-2 = (10.5 kN m-3) x d.

4

90
 Dril li ng mud fills the
pipe stem

Formation ===, Back pressure of

pressure -= drilling mud

Figure 3.4 Drilling mud, with a density typically 10 to 20% greater than
that of water, is used to stabilize wells by providing a hydrostatic pressure
at depth to counterbalance the formation fluid pressure.

So the required height of the drilling mud in the well is about 1 700 m. The
depth of the formation is 2000 m so there the balancing pressure of the
drilling mud is more than adequate to prevent a blowout.

3,5 Fluid Dynamics

As indicalcd previously, a description of fluid dynamics requires

specification of the relationship between forces and acceleration. Fluid
motions (other than the special case of no motion that we have just
discussed) are described using Newton's second law, F = mao The force F
is the sum of all the forces that cause fluid motion. Acceleration a is the
ratc of changc of vclocity with timc. In this scction, wc will invcstigatc thc
nature of fluid acceleration. Acceleration will be related to the forces
acting on a fluid to derive the Bernoulli equation, one of the fundamental
equations of fluid dynamics. Several examples of the application of the

91
Bernoulli equation are also considered.

3.5.1 Fluid accele..atiolJ

Acceleration of a solid object, such as an automobile, can be found by

measuring its velocity as a function of time u(t) and calculating a = duidt.
It is difficult to take the same approach to determining fluid acceleration
because a fluid does not move as a body with a single velocity that can
directly be measured. Instead it is a collection of many small fluid
"elements," not all of which are moving with the same velocity. Rather
than attempt to measure the velocity or acceleration of each fluid element
as it moves downstrcam, we typically use mcasurements of velocity at
fixed points to calculate the acceleration.
The velocity of a flow can vary in space and time. The change in
velocity with time at a single location is the local acceleration. A change
in velocity from point to point in a fluid at a single time gives rise to a
convective acceleration. The local and convective components of
acccleration in a rcgion of a flow can be determined from velocity
measurements made over a period of time at several locations along the
flow path in the region of interest. The total acceleration experienced by
individual fluid elements moving through the region is the sum of these
two components.
An examplc may serve to elarify this concept. Suppose you leavc
Washington, D.C., at 8 a.m. on a summer day and drive due south. That
morning, the temperature throughout the mid-Atlantic region of the U.S.
increases at a rate of 2°C per hour. This is the local rate of change of
temperature. The temperature in Washington, D.C., for example, may
increase from 20°C at 8 a.m. on a summer morning to 26°C at 1 1 a.m. In
addition to the local change we might expect a warming trend in the
southerly direction-the 8 a.m. temperature at Raleigh, North Carolina,
may be 23°C-so that one would encounter higher temperatures on
traveling south. Suppose this warming occurs at a rate ofO.8°C per
hundred kilometers. Now, if you mcasurc thc tcmperature as you drivc
south from Washington at an avcrage speed of 75 km hc l , what will bc thc
total rate of change of temperature with time that you record? It will be the
sum of the local rate of change (2°Clhr) and the convective rate of change
(velocity times rate of change with distance = 75 km hr- I x 0.8°C per 100
km = 0.6°C hr-I) or 2.6°C hc l .
To express mathematically the local and convective components of the
rate of change of temperature we employ the notation for partial

92
derivatives. We write the local rate of change in temperature as the partial
derivativc of tcmperature with time, oTlot. This is just a notational
convenience for expressing the change of temperature with time at a fixcd
location (e.g., Washington). Likewise, aT/Oy is the partial derivative of
temperature with distance in the y direction or the rate of change of
temperature with distance at a fixed time (e.g., 8 a.m.). The total rate of
change oftcmperature with time is written as dTldt and is equal to aTlot +
v (oT/Oy) where v is the velocity in the y direction.
Acceleration is treated in an entirely analogous manner to temperature
in the above example. Total acceleration is the sum of local acceleration
and convective acceleration, and acceleration in the x-direction can be
written:

rill au au
= + 11 .
(3.8)
dl al ax
- - -

where u = velocity in the x-direction.

In this text we will often restrict our discussion to steady flow. The
term steady applied to fluid flow means that conditions at a point are not
changing in time. For example, if the velocity of water at some point in a
stream does not change over a certain time period, then the flow has been
steady over that period. Steady flow implies that the local acceleration, oul
at, is zero. Velocity can still vary from point to point in the fluid, however,
and if it does, fluid acceleration will be non-zero due to the convective
tenn. For example, an upland stream may be characterized by a sequence
of pools and riffles, mpids following a slow-flowing pool following
another set of rapids, and so forth. Although the average velocity at any
fixed point in the stream is not changing with time, water accelerates as it
flows from a pool to a riffle, or vice versa. In contrast, the flow in a
straight canal with a constant cross-sectional geometry would not vary
from point to point along the flow path. Such a flow is said to be uniform,
whereas the flow through a series of rimes and pools is nonuniform. A
flow that is both steady and uniform has no acceleration; the local
acceleration is zero and the convective acceleration is zero.

3.5.2 The Bernoulli equation

To be able to predict the behavior of a moving fluid, for example water

flow in a river or through an irrigated field, it is necessary to develop a
model, or more specifically, a mathematical equation. This equation must
express the correct relationship between the basic physical laws and our
observations of fluid motion. We want a relatively simple equation, easily

93
solvable, but one that nevertheless is general enough to describe practical
flow situations. An aim of hydrology is to determine what simplifying
assumptions are acceptable for giving useful predictions of the behavior of
real hydrological systems.
We will eonsider the problem of the flow of water through a garden
hose as the starting point for developing an equation for fluid motion. The
results, as we will see, apply to a far wider range of problems than just a
garden hose. The hose is connected to a house spigot with sufficient
pressure to cause an ample stream to leave the open end, perhaps 1 0 m
away. Clearly the pressure at the spigot will affect the flow in the hose. We
must also consider what other factors might be influencing the water in the
hose. As in the hydrostatic case (i.e., no motion), we approach the problem
by making an assessment of the body and surface forces acting on the
fluid.
We will consider an arbitrary short section of the hose of length ds
oriented in the s (an arbitrary) direction to begin constructing the force
balance for the water movement through this small volume (Figure 3.5).
Water is flowing into this small piece from the house, or upstream, end
and out of it at the downstream end.
Pressure causes a surface force to act on the small volume. There is
both an upstream pressure, PI' and a downstream pressure, p2' Thus there
is a force i n the downstream direction, PIA, and a force in the upstream
direction, py4. . Another force that acts on the volume of water in the short
section of hose is the weight of the water itself, that is, its mass multiplied
by the acceleration of gravity:
F;r (pdV)g.
: (3.9)

where Fg = body force or weight [M L .2] and dV = volume = Ads [L3].

94
 z

x
u.. 3.5 A section of a garden hose used to analyze the pressure and
gravitational force acting on a small volume of water.

The forces from pressure and the weight of the water are the only two
forces we will include in our simple model. It is easy to imagine a more
complex situation where other forces would be important. If, for example,
the fluid were beryllium, a relatively magnetic fluid, instead of water, then
it might be appropriate to consider magnetic forces. A morc obvious force
we have nOI listed is the frictional force between the water and the hose or
within the flowing water. Frictional forces are tangemial surface forces
due to the viscosity of the water and the fact that fluids in contact with a
surface "stick" to that surface. Although friction is certainly a real surface
force, we wish to keep our model simple, and so will neglect frictional
forces for the moment. We must keep i n mind, however, that this
assumption of negligible friction is not correct. We recognize from the
start that the final result will be useful only in cases when this assumption
does not cause too much error. In many cases in hydrology, the equation

95
resulting from this simple, frictionless case has to be modified, as we will
see.
To develop our flow model, knowledge of physical laws and
observations of the flow itself arc used. According to Newton ' s second
law, the product of the mass and acceleration of the fluid within the hose
must be equal to the sum of the forces acting on the fluid volume:

(3.10)

where Fg is the body force due to gravity and Fp is the surface force due to
pressure, the net force resulting from the sum of the force due to upstream
pressure, PIA, acting in the +s-direction (positive force) and the force due
to downstream pressure, P;0, acting in the -s-direction (negative force).

(3. 1 1 )

Over the small section of hose that we are considering, the pressure at the
downstream end,P2' differs from the upstream pressure, P ' by a small
I
amount dp,

dp
P2 = PI +-ds.
ds
The tenn dplds is the pressure gradient in the s direction. Substituting this
into Equation 3. 1 1 ,

or

dp
Fp = -- dsA . (3.12)
d,
The body force Fg acts in the vertical direction. We are interested only in
the component of this force acting along the axis of the hose, the s
direction. Fgs can be computed from a simple trigonometric relationship
(see Figure 3.5):
�, = F, sine. (3.13)

where e is the angle the hose is inclined from the horizontal. We can
further simplify this expression if we note that

96
 do
sin O = --. (3.14)
d,

Equation 3. 1 3 can be written

F = (pgd V)sin 8 = -pgd V!5.

d...
g1

do
F.
" =-g-pAds.
d,
(3.15)

We now have all of the elements of the force balance in useful terms.
The mass of the fluid in the control volume is the density times the
volume. The forces are as indicated in Equations 3. 12 and 3. 1 5.
Assembling these in the fonn of Newton's second law (F= ma) gives:

de dp
(pAds)a = -g-=-pAds - - Ads,
d... d...

where a is the acceleration in the s-direction.

If we assume that the flow is steady in time, then BulBI is zero, and
Equation 3.8 becomes

d"
a = u-, (3. 16)
d,

where II is the velocity in the s-direction. (Because II is now a function of s

only, we write the convective acceleration term as a regular rather than a
partial derivative.) This assumption would imply that the supply pressure
at the house is steady in time. Finally, we can divide through by pAds,
du d: l dp
u-+g-+--=O. (3.17)
ds ds p dv

To review briefly how Equation 3 . 1 7 was developed, recall that it is a

simple statement ofNewton's second law of mechanics. The acceleration
ofa fluid element was equated to the sum of the forces per unit mass
exerted on the element. We have assumed that the frictional force is small
enough to neglect and that the flow is steady in time.
Now that we have derived an equation describing flow in a garden hose,
we can use it to predict some characteristics or properties of this flow. For
example, if we know what the pressure gradient is, can we compute the
water velocity or the volume discharge rate? How is pressure related to

97
slope and velocity? To answer these questions, it is first necessary to solve
Equation 3.1 7. Our solution will be simplified if we recall from the rules
for differentiation that

(3.18)

Using Equation 3 . 1 8, Equation 3. 1 7 can be written:

�(�+:+...E....l =o.
ds 2g pg
(3.19)

Equation 3. 1 9 can be solved by integration along the s axis. However, to

do so we must make another assumption: that the fluid is both
homogeneous and incompressible. Making this assumption, we can
integrate with respect to s, and write,

1/2 P
- +: + - '" const:lnt. (3.20)
2g pg

Equation 3.20 is known as the Bernoulli equation, after Daniel

Bernoulli (1 700-1 782), a Swiss physician, mathematician, and physicist,
who was one of the principal founders of the field of fluid mechanics. The
constant on the right side of Equation 3.20 is total head, H.
The terms in the Bernoulli equation derived above contain elevation,
pressure, and velocity as variables. When written as in Equation 3.20, each
term in the Bernoulli equation has units of length, and can be thought of as
a component of the total head:

112 L2 T2
Velocity head: -dlln -- = L
•

2g T2 L

Elevation head: : dim L

I'ressure head: P dim ML £J' L

pg T'L' M L
�

The notarion "dim" is used to denote dimensional equality, for example, z

dim L means that elevation has dimensions of length. Each of the total
head components can also be thought of as having units of energy per unit
weight. Multiplying the Bernoulli equation by the weight of a volume of

98
fluid (pgV) gives the equivalent equation in units of energy:

111112
p V + mgz + 2 = eonstant.

We see that pV is the "flow work" or the work due to pressure, mgz is the
potential energy, and nlu2/2 is the kinetic encrgy. Thus, the Bcrnoulli
equation can also be thought of as a statement of the conservation of
energy.
At this point it is desirable to recall all of the assumptions needed to
derive the Bernoulli equation, namely:

a. no friction,
b. incompressible fluid,
c. homogeneous fluid,
d. flow steady with time.

In addition, the solution is valid only along the path of integration, the s
axis, a line everywhere parallel to the flow field that is referred to as a
streamline.

3.5.3 Applications of the Bernoulli equation

The use ofthe Bernoulli equation can best be illustrated by considering a

simplc example. A tank is filled with watcr and drains at a small spigot at
the bottom (Figure 3.6). By allowing for overflow at the top, and a
continual supply of water, the level in the tank remains constant. Our
problem is to compute the velocity of the water flowing out of the spigot.
To apply the Bernoulli equation, all of our assumptions must be valid, and
we must choosc an appropriate streamline. The fluid is water, and we can
reasonably assume it is both homogeneous and incompressible. Since the
level in the tank does not change with time, our problem satisfies the
steady assumption. We must assume that friction is unimportant, an
assumption whose validity can be tested only by experiment. The
streamline to be followed is a question of convenience. A possible
streamline from point I to point 2 is postulated and the Bernoulli equation
is applied. Restating the Bernoulli equation,

u' P
-+=+ constant H.
2g
= =

pg

99
Because the total head is constant, the sum of the velocity, elevation, and
pressure heads will be the same at every point on a streamline. Using
subscripts to identify points I and 2 on the streamline, the Bernoulli
equation can therefore be written in the useful fonn:
,
IIi PI IIi Pl
-+- =-+- 1 +-
,

1+- _ _

(3.21)
2g pg 2g pg

If a spot of dye were placed carefully on the surface of the tank, its
speed of motion would be seen to be very small with respect to the speed
at which the water exits the tank because the surface area of the tank is so
much larger than the cross-sectional area of the spigot at the outlet.
Because II I is so small relative to 112 (1/1 « 112), we can assume that 111 is
equal 10 zero in our analysis. The pressure on the surface would be the
local atmospheric pressure, as would be the pressure at the outlet of the
spigot, point 2. The small difference in atmospheric pressure due to the
elevation difference between I and 2 can be ignored. With these
considerations, we can now write,

_I
_ + Put",.,, = -
2
"2 + Z, + !P�m�"��"
pg 2g - pg
or

100
 ===
9 1"1
=� Inflow keeps water
II level constant
PI = ['2
-
'"
-
-
111 ", 0 1) :::r =

=
PaUfIOS
0 (gage)

-,

z �"
172
,

"
-

Datum (z 0)
=
-

Figure 3.6 Tank with steady flow of water. The Bernoulli equation
provides a relationship between the depth of water in the tank and the
velocity of water exiting the tank at point 2.

If we let d = zl-Z2 (the depth of water betwcen points I and 2), then,

(3.22)

Thus, the velocity of the water leaving the tank depends on the depth of
the waler in the tank.
If a plug were placed in the spigot, shutting off the flow, we could use
this same approach to compute the pressure at point 2. In this case, both II I
and //2 are zero but P2 is no longer equal to atmospheric pressure:

- -
-1 - + p,
- -2
pg
0'
P2 = pgd. (3.23)
This equation, for thes tatic situation, is identical to Equation 3.7, the
hydrostatic equation. Thus, we have shown that the hydrostatic equation is
a special case of the Bernoulli equation.
Let us return to our general fonn of the Bernoulli equation (Equation
3 .20),

101
 u2
+=+ L = constant,
2g pg
-

and considcr the relationship betwcen pressure and velocity for thc garden
hose. If the velocity along a horizontal segment of a streamline (i.e., one
for which z did not change) were to increase, say because of a constriction
in the hose, we would conclude that the pressure would decrease. That is,
pressure and velocity are inversely related, assuming no change in
elevation, z.
Often we are more interested in the average or mean velocity of flow
through a hose or pipe than we are in the velocity along any particular
streamline. The mean velocity U at any cross section is the discharge Q
divided by the cross-sectional area A, that is, U = QIA. The mean velocity
also can be thought of as the avcrage value of the velocities at each point
in the cross-section. For the frictionless flow described by the Bernoulli
equation, the velocity at any point in a cross section is equal to the mean
velocity, so u = U. This is not true for flows in which friction is important.
One of the fundamental constraints on flow through pipes or channels is
conservation of mass: rate of inflow minus rate of outflow equals rate of
change of storagc (Eq. 1.1). In the case of steady flow in a full hose, there
are no changes in the amount of water in any segment of the hose at any
time (no change in storage) so inflow rate must equal outflow rate.
Because we are taking density to be constant, the volumetric inflow rate
must equal the volumetric outflow rate. The conservation of mass (or
volume, in this case) equation is oftcn referred to as the continuity
equation, simply written,
Q = UA = constant. (3.24)

The continuity equation and the Bernoulli equation are two of the
fundamental relationships of fluid mechanics.
As an example of the way conservation of mass can be used to provide
infonnation about flow, consider a section of hose where the diameter
changes from D1 to D2 , where D2 = 2D1 (Figurc 3.7). Becausc we have
assumed the flow to be steady, the same quantity of water flowing through
the narrow scction must be flowing through the wide section:
(3.25)
and thus,
(3.26)

102
where the cross-sectional area of the hose is A = rrD 2/4 and D is the hose
diameter. Substituting the expression for cross-sectional area into Equation
3 .26 gives,

lfDI2 = U, lfDi
UI 4 - 4

"\
•
•
•

VI
� �
•
•
•
• °2
•
•

"l-
•

/
I

2
A 1 - ( rrO I )/4
-

2
A2=(rr02)/4
Figure 3.7 An expansion joint in a hose. Discharge is the same at points I
and 2 (upstream and downstream ends of expansion joint) for steady flow
through the hose. As a result, if cross-sectional area increases, velocity
must decrease.

01

Substituting the relationship between D I and D2, we find

which may be rearranged to give:

UI 4.
= (3.27)
U,

Thus, the velocity in the narrow section of the hose is four times the
velocity in the wider section.
If the hose is level (zl = z2), the Bernoulli equation is given by:

103
 Ur + PI _ ULl + P2
::c
2g pg 2g pg

(Recall that for frictionless flow, V u, and therefore the Bernoulli

=

equation can be written in tenns of mean velocity, u.) Because V1 is

greater than U2 (Equation 3.27), we see that the pressure at point I must be
smaller than the pressure at point 2. Thus, we have shown that an increase
in hose diameter results in a decrease in mean velocity and a
corresponding increase in pressure.

3.6 Energy Loss

In deriving the Bernoulli equation, we assumed that fluid friction was

negligible. Laboratory experiments can be conducted to test the validity of
this hypothesis for simple flow situations involving, for example, flow
through glass tubes. There are many flow situations in which fluid friction
can be neglected, especially if we arc concerned with flows over short
distances. For this reason, the Bernoulli equation has been used as the
basis of many flow measurement devices used in pipes and streams. An
accurate description of flow, however, often requires that we account for
the energy loss due to friction, which is associated with the viscous
properties of fluids. Reconsider the example of the garden hose. If we were
to punch very small holes in the hose every meter along its length, with the
water running, it would cause a number of small fountains, as in a lawn
soaker (Figure 3.8). The height of each fountain is a measure ofthe
internal pressure at that point in the hose,
h, = PI ,
(3.28)
pg

where the subscript i refers to position along the hose ( I , 2, and 3 in Figure
3.8).

104
 •

•
I

hi
•

Figure 3.8 The height ofwaler spouting from holes punched in a hose
decreases along the length of the hose because offrictional losses of
energy (head loss) as the flow travels downstream in the hose.

For a horizontal section of the hose with no friction, the Bernoulli

equation would predict that pressure and velocity along the hose would be
constant provided the hose has a constant diameter. This implies that the
h,.'s in Figure 3.8 would all be the same. However, observations of the
heights of sprinklers in a real hose with punctures show that hi > 112 > 113;
that is, adjacent fountain elevations become successively smaller as the
open cnd of the hose is approached. Because the hose is level in the
section we are interested in and the discharge and diameter are constant
(and therefore U is not changing along the hose), the pressure must be
continuously decreasing along the hose. But, this conclusion contradicts
the conservation of energy equation (Bernoulli) that we obtained using the
assumption of zero friction:

u'
P +=+ - = H = constant.
PI!. 21!.

In fact, we observe a loss in total head along the hose. hL• which if added
to the other tenns in our equation would give a total head equal to the

105
initial sum H. That is, the Bernoulli equation must be modified to account
for the energy loss to friction:

U'-
J!.... + = + - + "t constant, (3.29)
pg 2g
=

where hL is called the head loss. Head loss is simply an empirical way of
dealing with the fluid friction that is dissipating energy (converting
mechanical energy to thennal energy) inside the hose over a specified
length of hose. For a fluid with a non·zero viscosity, like water, friction
causes fluid adjacent to a solid surface to "stick" to that surface. This
means the velocity of the water in contact with (stuck to) the wall of the
hose mllst be zero because the hose is not moving. As a result, there will
always be a velocity gradient away from the wall within a frictional flow
(Figure 3.9). Energy dissipation is a result of this velocity gradienl.

/ /
'.
•
•
•
•
•
,
,
,

,
,
,
,
,
,

I----�/
/'
,
,
,
\
Figure 3.9 Water "sticks" to the walls of the hose and vIscosity causes a
frictional energy (or head) loss. Velocity is zero at the walls of the hose
and a maximum at the centerline.

3.6.1 Friction factor

For flow through a circular pipe or hose, head loss between the two ends
of the hose depends on the viscosity of the fluid, the velocity oflhe flow,
hose diameter and length. Because head loss also is a function of the
roughness of the wall material itself, we must include a friction factor in
its definition:

L u�
"/. - f --, (3.30)
D 2g

106
wheref= friction factor [dimensionless], L= length [L], and D = diameter
[L).
lf we consider the special case of a horizontal hose of constant diameter
(and therefore constant cross-sectional area), velocity and elevation don't
change along the hose. We can write the Bernoulli equation including head
loss as

I' v' 1' '- u2

I +:1+ ' =_
' _' +:,+ 2 +111 , (3.3t)
pg 2g pg - 2g '
_ _ _

where ilL is the head loss between cross-sections I and 2 of the hose. If
velocity and elevation don't change along the hose, then VI V2, ZI Z2 '
= =

and thcrefore hI. (Pl-P2)/pg. If we cxpress head loss as head loss per unit
=

length, then

!!L = , PI - Pl , tJ.p , dp
(3.32)
_
_ _
_
_

=
__
_

I. pg I. pg L pg d'C

where Lis the length of the hose between cross-sections I and 2. Ifwe
combine Equations 3.30 and 3.32, we can obtain the following expressions
for velocity:

(3.33)

Note that dpldx is negative for flow in the positive x-direction, In other
words, there is a pressure drop in the direction of flow. Equation 3.33
suggests that if we can measure the pressure drop between two points
along a hose or pipe a distance Lapan and we know the friction factor, we
can calculate the velocity or discharge of flow through the pipe.
The problem of calculating or predicting the velocity at which water or
other fluids would flow through a given pipe, channel, or other conduit is a
problem that has great practical importance in the design and management
of water supply and sewer systems, waterways, and industrial pipelines.
Calculating velocity using Equation 3.33 requires knowledge of the correct
friction factor. Friction factor is a quantity that must be determincd
experimentally. Because of its importance, many measurements of friction
factorfhave been made under controlled laboratory conditions. A number
of these measurements of friction factor are ploned in Figure 3.10 against
the term pUD/�. This dimensionless term, the Reynolds number R, is
discussed in the next section. When plotted in this manner, the data for all
pipe diameters, discharges, and fluids describe a well-defined relationship

'07
for pipes that are smooth inside. We refer to this as a friction factor
diagram.
2X l0 ',-
---,C- __________________________________ --,

k,lD
----
----- --- ------- 4.0><10"

2.0><10" ilr
-- ----------- !!
------------·4.0Ml0..
�.
,::- .:::- -- -
,;: --- -
g�
Turbulent �
'0
. � '0'
- --
- - - - - -

,
- _

pipe flow -'1.0>< 10�

1.0><'0� �
10' 1.0><10"

5xl 0-3 +-
_____ ....,
______-r ,-
_______ _____ ...,
'-'- �
__

10' 103 10' 100 10' 5xlO"

Reynolds number, R pUD/1!=

Figure 3.10 Friction factor diagram for pipes. Laboratory measurements

of friction factor fversus Reynolds number R for smooth pipes over a
range of Reynolds numbers arc shown by the blue dots. The break in the
measurements at Reynolds numbers between 2000 and 4000 marks the
transition from laminar to turbulent flow. The Blasius approximation
(Equation 3.39) is a linear approximation to the turbulent flow
measurements valid for 4000 < R < 100,000. Friction factors for turbulent
flow in rough pipes are indicated by the gray dashed curves, where relative
roughness, k,JD, is the ratio of the roughness length, kr (see Figure 3.12),
to pipe diameter.

3.7 Laminar and Turbulent Flows

It is obvious looking at the friction factor diagram (Figure 3.10) that

something happens to the flow when the Reynolds number exceeds about
2000. It turns OUI that the frictional losses in a fluid flow depend on
whether the flow is laminar or turbulent. In order to understand the
difference between these flows, let us review an experiment first

108
conducted by Sir Osborne Reynolds (1843-1912). Reynolds was an
English civil engineer noted for both his theoretical and applied studies in
fluid mechanics. Rcynolds's expcriment used a simple apparatus to
visualize a flow field (Figure 3.11). In this apparatus, dye is injected along
the centerline of a pipe in which fluid is flowing. The dye serves to show
the nature of the flow as the fluid moves along the tube. At low velocities
the dye rcmains approximately on the centerline, traveling downstream
with, and at the velocity of, the flow. Because of molecular diffusion there
will be a small flux of dye away from the centerline, but this will be so
small that no departure from the line will be observed with the naked eye.
Reynolds characterized this flow situation by describing the fluid as
moving in dislinellamilJae (Latin for layers), and today we refer to such
flow as laminar flow.
Now consider what happcns as the velocity of fluid flow in thc tube is
increased. At some value of increased velocity, the dye is observed to
become mixed within small, randomly located spots within the flow; with
higher velocities the dye rapidly becomes completely mixed with the
surrounding fluid in a short distance of flow. Rcynolds referred to this
flow as sinuous or disturbed flow. Today we refer to flows exhibiting such
violent mixing as turbulent flows. In this case the mixing takes place
much more rapidly than realized by molecular diffusion, and in fact we
call this mixing tllrblllellf diffllsion, We can conceptualize the turbulent
mixing as small parcels of the fluid "jumping" away from a streamline to a
different portion of the flow field. Most flows we observe in nature, for
example, clouds, rivers, ocean waves, are turbulent flows. In this regard,
we note that "laminar" and "turbulent" refer to properties of the flow, and
not properties of the fluid, For example, the water in the garden hose could
be in either laminar or turbulent flow without changing its propenies, that
is, density or viscosity.

109
 '.
•
•
•

�--.:;-
•

1
•
---------

•
•
,
•
·

(a) .'

(b) .'

Figure 3.11 Reynolds 's experiment. In laminar flow (a), the dye remains a
thin line, moving with the fluid. The dye trace becomes convoluted in
turbulent flow (b).
To explain the last statement, we return to Reynolds's experiment. By
carefully changing the properties orthe fluids in his apparatus, as well as
the tube diameter and flow velocity, Reynolds empirically defined a
dimensionless number R which describes the flow properties:

=ULp
R (3.34)
I'

where U = IIJ; L a characteristic length [L];

a characteristic velocity [L =

p = fluid density [M L-3]; 11 = viscosity fM L-I .1]. R is called the

Reynolds number. According to Figure 3.10, if R is small (with Utaken
as the average velocity and Ltaken as the diameter of the tube), say less
than 2000, then the flow in a pipe will be laminar. IfR is large, say greater
than 4000, then the flow will be turbulent. At values between these limits,
called the transition region, the flow is not easily characterized as one or
the other. The choice of the velocity and length is a matter of accepted
convention. For circular pipes such as the garden hose, the characteristic
velocity is taken as the mean value over the cross section (the discharge
divided by the cross-sectional area), and the characteristic length as the
hose diameter. If water flows at [5°C at 2 m S-l, through a 30-mlll

110
diameter hose, the Reynolds number would be calculated as follows:
1
u= 2 m 5- (mean velocity);

L = 0.03 m ( pipe d iametcr);

/1 = 1.139 X 10-3 Pa' 5;

(2XO.03XiO')
53 000
=

R �
• •

(1.139xI0-')
which indicates that the flow is turbulent. The flow could be made laminar
by somehow reducing R to some value less than 2000. We could
accomplish this by:

a. reducing the velocity,

b. reducing the diameter,
c. reducing the density of the fluid. or
d. increasing the viscosity.

In order to accomplish (c) or (d), we would have to change the fluid

temperature, or perhaps the fluid itself. This latter altemative suggests
another important conclusion from Reynolds's experiments. Two different
fluids with the same Reynolds number will have similar flows, that is,
laminar or turbulcnt. Thus, we can expect the same degree of mixing or
turbulence inside the hose for both water and glycerin, if they both have
the same value for R. Because the viscosity of glycerin is much greater
than that of water, it would be difficult to achieve this similarity in
practice.
Now that we have some feeling for the meaning of the Reynolds
number, let us reconsider the differences between laminar and turbulent
flow as evidenced by Reynolds's experiment. For the laminar flow, any
time the dye began to stray from the centerline of the tube, it was hindered,
or its motion was restored to the centerline. The attempted deviation or
acce1t:ration is an apparent inertial force, and the restoring rorce is due to
the viscous forces within the fluid. Thus, for laminar flow, the mixing is
restrained as the viscous force must dominate over the inertial force. If the
inertial force becomes greater than the viscous force, then the dye would

III
have a tendency to deviate from the centerline, and the flow could become
turbulent. We can therefore take the ratio of the inertial to viscous force as
an index of whcn flow would be expcctcd to be laminar or turbulcnt. If the
viscous forces» inertial forces, the tlow is laminar, and if the inertial
forces» viscous forces, the flow is turbulent. We know that the inertial
force is proportional to mass x acceleration, which could be expressed as
(p)(L2)(velocity2). Similarly, the viscous force is proportional to the shear
stress x area, which could be expressed as (1l)(L)(velocity). Thus,

; ," �n�'; �I fO�'" ��' « p_L' _v '

"� � _ _ LV = .
= p___

R (3.35)
viscous force JlLV Jl

The Rcynolds number is equal to the ratio of the inertial to viscous forces
within the fluid.

3.7.1 Laminar pipe flm"

Let us return to our example of water flowing through the garden hose . If
the hose is narrow enough, the flow slow enough, or the viscosity high
enough, the flow could be laminar. In this case, the experimental
relationship between friction factor and Reynolds number indicated in
Figure 3.10 is linear and is given by:

1 =64,. 64JI.
(3.36)
R pUD

If we substitute this into Equation 3.33, we obtain:

1) I2 dp
v= (_ D pUD (_ VD')1 2
dp 2
dx P 64p
=

de 32p

or, if we square both sides and divide by U:

D2
U=_dp . (3.37)
dr 32J1

The velocity given by Equation 3.37 is the mean velocity for laminar
flow through the pipe; this velocity times the pipe cross-sectional area
gives the discharge, Q UA. Steady laminar flow through a uniform pipe
=

is referred to as Poiseuille flow and Equation 3.37 for the mean velocity is
sometimes referred to as Poiscuillc's law. As depicted in Figure 3.9, the
actual velocity at any point in the pipe varies with distance between the
wall and the center of the pipe. !fwe integrated this velocity profile over

112
the pipe cross section, wc would obtain thc discharge Q; the mean velocity
Uis this discharge divided by pipe cross-sectional area A.
In deriving Poiseuilte's law from the Bernoulli equation we assumed
that the elevation along the hose or pipe is constant, that is, ZI Z2' We can
=:
generalize the equation by relaxing this assumption and allowing the hose
10 be inclined at an angle as in Figure 3.5. In this casc a term relatcd to thc
clevation difference is added to Poiseuille's law giving:

( dP dZ D' )--
U=- -+pg-
ds ds 32p

or

(.3.38)

Thc tcrm in parcntheses in Equation 3.38 is onc that is uscd frequently in

hydrology and is referred to as hydraulic head h.

3.7.2 Turbulent pipe flo"

If the flow velocity, pipe (or hose) diameter, and fluid viscosity combine to
produce a Reynolds number in excess of 4000, as is often the case, then
the flow will be turbulent and Poiseuilte's law does not apply. Instead, we
must return to Equation 3.33 and use Figure 3.10 to detennine the friction
factor. For values of Reynolds number bctween 4000 and 100,000, thc
friction factor for smooth pipes can be approximated by the expression:

f=0.316R-'I� (3.39)

This approximation, proposed by Blasius, is indicatcd in Figure 3.10.

When the inside of thc pipc is rough, thc friction factor for turbulent
flow increases. In this case flow experiences rcsistancc not only from the
pipe walls themselves (termed skin friction) but also from the roughness
elements protruding up into the flow as suggested in Figure 3.12. This
roughness resistance is due to drag (called fonn drag) produced as flow is
forccd to go around roughness elcmcnts on the flow boundary. Examples
of roughness elements that contribute to bottom friction in streams include
rocks, boulders, and other bed fonns such as sand ripples and dunes. The
rougher the surface is, the greater the fonn drag, and the higher the friction
factor for turbulent Ilow, as shown in Figure 3.10. Thus, in turbulent flow,
unlike laminar flow, it is neccssary to characterize the roughncss of the

113
walls bounding thc flow in addition to thc othcr factors. Wc will scc in thc
next chapter on channel flow that characterization of channel roughness is
an imponant but difficult pan of cstimating flow velocity in streams and
other turbulent surface flows.

Velocity 1------<--1
gradient
implies shear
stress or skin-
friction drag

Flow separation
at small scales
implies form drag
Figure 3.12 Flow through a rough-walled hose. The roughness length, kf '
is a measure of the size of the irregularities or unevenness of the surface.

3.8 Concluding Remarks

The material in this chapter was meant to introduce you to the basics of
fluid mechanics. We should emphasize again that our objective is to apply
these fundamental principles of fluid flow to hydrological problems. It
may not yet be clear to you how a description of pressure distribution, flow
and fluid drag in a garden hose is relevant to our objective. The
applications to the "real world" are discussed in the succeeding chapters
and we will see that the garden hose example will assist us in problems
ranging from the relationship between river velocity and bed roughness to
the interpretation of water levels in wells.
"Real world" applications of the principles of fluid mechanics are not
restriclCd to the flow of water. The winds that blow over the surface of the
Eanh are flows of air in response to differences in atmospheric pressure or
other forces. rlows of air near the Earth's surface are characterized by
high Reynolds numbers because air has a relatively low viscosity and the
flows arc always "deep." In contrast, lava has a viscosity more than a
thousand times that of water (the value depends strongly on temperature)

114
while its density is only 2 to 3 times larger. As a result, flows of lava are
low Reynolds number, laminar flows that can be described by a fonn of
Poiseuille's equation. One of the many things that makes water interesting
is that there are commonly occurring examples of both laminar and
turbulent flow.

3.9 Key Points

• A fluid is a substance that defonns, or flows, when subjected to a shear

stress, no matter how small. {Section 3.2}

• Viscosity is a temperature-dependent property of a fluid that

characterizes its ability to resist defonnation. The viscosity fl of a fluid is
given by the ratio of the shear stress at a point in the fluid to the velocity
gradient at that point (Newton's law of shear): fl t/(duldz). The
=

viscosity of water at 200e is 1.0 x 10-3 Pa . s. {Section 3.2}

• When defining properties of a fluid at a point we make the continuum

assumption that, at the macroscopic level, the fluid is continuoliS
throughout its entirety, that is, that the molecular structure of the fluid is
smeared out to eliminate spaces between atomic particles. {Section 3.2}

• The density of a fluid is the mass per unit volume at each point in the
fluid. A homogeneous fluid is one in which density is constant
throughout the fluid. The density ofwater is approximately 1000 kg m- 3.
The unit weight of a fluid is its weight per unit volume, pg [N m-3].
{Section 3.2}

• Two classes of forces act on fluids: body forces and surface forces. Body
forces, such as gravity, act unifonnly on each fluid clement. Surface
forces, such as pressure and friction, act on the surfaces of fluid
elements. In fluid mechanics we typically represent surface forces as
forces per unit area, or stresses. There are two types of stresses: nonnal
stresses and tangential stresses tenned shear stresses. The inward­
directed (compressive) nonnal stress, when applied to a fluid medium, is
referred to as pressure. {Section 3.3}

• In a fluid at rest, pressure increases with depth at a rate of pg, the unit
weight of the fluid. This is expressed by the hydrostatic equation, dpldz =

-pg. The integrated form of the hydrostatic equation for a fluid of

constant density indicates that pressure increases linearly with depth

115
 below the surface, p = pgd. The pressure given by this relationship is
gage pressure, the pressure relative to atmospheric pressure. To obtain
absolute pressure, the atmospheric pressure acting on the fluid surface
must be added to p. {Section 3.4}

• The acceleration of a fluid can be divided into two components. The

local acceleration is the change in velocity with time at a point, alliat.
The convective acceleration is related to changes in velocity from one
place to another in the flow, c.g., II (all/ax) for flow in the x-direction.
The total acceleration is the sum ofthesc two components. A flow is
steady if the local acceleration is zero. A flow is unifonn if the
convective acceleration is zero. {Section 3.5.1}

• The Bernoulli equation states that, for a frictionless flow, the sum of the
velocity head 112 /(2g) [L], elevation head z [L], and pressure head p/(pg)
[L] along a streamline is a constant tenned total head, H [L]: [1l2j(2g)] + z
+ [P/(pg)] = H. Thc assumptions made in deriving the Bernoulli equation
are I) no friction; 2) incompressible fluid; 3) homogeneous fluid; and 4)
steady flow. For frictionless flow, the velocity II on any streamline in a
cross section is equal to the mean velocity U, so that U can be substituted
for II in the Bernoulli equation to obtain an equation in temlS of mean
velocity. {Section 3.5.2}

• The continuity (conservation of mass) equation for steady flow through a

pipe or channel states that the rate of inflow is equal to the rate of
outflow, as long as there is no water entering or leaving the flow through
the sides. If the density of the fluid is constant, this means that the
discharge must be the same at each cross section of the flow: Q VA = =

cOllslallt. The continuity equation is often combined with the Bernoulli

equation when solving flow problems. {Section 3.5.3}

• Frictional losses of energy in a flow result in a loss in total head with

distance downstream. When friction is important, head loss hL must be
added to the tenns in the Bernoulli equation to keep the sum of the tenns
constan t [1I1/(2g)] +z, + [PApg)] [1I1/(2g)] +=,+ [P,l(pg)] + ",: Foe
=

flow through a horizontal pipe, head loss is related to the pressure drop
along the pipe: ilL = (PI - P2)/(pg). {Section 3.6}

• Head loss for flow through a pipe or tube is related to mean velocity V,
pipe diameter D, and pipe length L by the empirical equation ilL =

j[(LV2)/(2Dg)], wherefis the friction factor [dimensionless]. Values off

116
 are determined experimentally and are typically represented in friction
factor diagrams offas a function of pipe Reynolds number R pUDfl1=

(Figure 3.10). {Section 3.6.1}

• The Reynolds number R = pUD/J.I is a measure of the relative importance

of viscous forces in a flow. When the Reynolds number is less than 2000
for pipe flow, viscous forces are large enough to damp any disturbance in
the flow resulting in laminar flow. Viscous forces are less effective at
higher Reynolds numbers (above about 4000 for pipes) so that
disturbances to the flow can grow, causing the flow to become turbulent.
Two different fluids with the same Reynolds number will have similar
flows. {Section 3.7}

• The friction factor for laminar pipe flow is given by f= 64fR. Combining
this with the head loss equation gives an equation for the mean velocity
for laminar flow in a pipe (Poiseuille's law): U -{dpidx)([j1f32J.1).
=

{Section 3.7.1}

• For turbulent pipe flow, friction factor must be obtained from the friction
factor diagram or determined experimentally. The friction factor for
turbulent flow increases with increa3ing roughness of the pipe walls.
{Section 3.7.2}

3.10 Example Problems

Problem 1. The following questions make use of the hydrostatic equation.

A. What is the gage pressure (Pa) at a depth of 10.0 m in a lake with a

water temperature of 15°C?

B. Would the pressure change significantly if the water temperature was

22°C instead?

C. At what depth (m) is the gage pressure 300 kPa?

D. What depth (m) of mercury, with a unit weight of 133 kN m-3 , would
be required to produce a pressure of300 kPa?

Problelll 2. A plali: is pulkJ UVi:r a huriwntallayt:r uf walt:r thal is 10.0

mm deep (Figure 3.1). The temperature of the water is 20°C. !f the plate
exerts a shear stress of 0.0 J N m-2 on the upper surface of the water, what
is the speed (m 5-1) of the plate?

117
Problem 3. Observations show that flow in a circular pipe of diameter D
remains laminar up to a Reynolds number Rpipe pUD/� 2000 (Figure
= =

3.10). What about flows in other flow in channels or pipes with different
geometries? Consider the example of flow between two flat plates shown
in Figure 3.1. In this case, the appropriate length scale for the Reynolds
number is L = d. I t also makes sense to use plate speed rather than mean
flow velocity speed as the characteristic velocity in R, giving Rp1me =
Pllp/ale d/�. Changes in length and velocity scales can alter the upper limit
for laminar flow, but for the case of flow between two parallel platcs, flow
is again laminar up to Rp1ate = 2000. (A parameter called the hydraulic
radius [Chapter 4.5] can be used to find values of R corresponding \0 the
laminar-turbulent transition for different flow cross-sections. This is
explored further in an example problem in Chapter 4.)

A. For20°C water between2 plates separated by a distance of 4.0 mm

(i.e., d = 4.0 mm; Figure 3.1), what is the maximum speed that the
upper plate can move and still maintain laminar flow?

B. Is the flow in Problem2 laminar? (Note: if not, Equation 3.1 is no

longer correct.)

Problem 4. Surface temperaturc in a rivcr is mcasurcd by a thermometer

drifting with the water at a rate of I km hr- I. The water in the river as a
whole is warming al a rate of 0.2°C hr-1 , and the temperature along the
stream increases by 0.1 °C every kilometer in the downstream direction.
What change in temperature eC) does the thermometer record in 6 hours?

Problem 5. A tank like the one pictured in Figure 3.6 is filled to a constant
level of 0.70 m. The center of the outflow opening near the bottom is 0.10
m above the bottom of the tank. What is the velocity (m S-I) of flow
exiting from the outflow opening?

Problem 6. The pressure drop through a well-designed constriction can be

used to measure the velocity of flow through a pipe. If the pressure drop
from a O.I-m diameter cross section to a 0.05-m diameter cross section is
7.5 kPa, what is the velocity (m S-I ) in the O.I-m diameter section of the
pipe? Hint: Use the conservation of mass equation to relate the velocity at
the smaller cross section to that at the larger cross section.

Problem 7. A steady discharge of2.0 x 10-4 m 3 S-I is flowing through a

20-mm diamcter hose. The viscosity ofthc watcr is 1.0 x 10-3 Pa . s, and

118
the density of the water is 1000.0 kg m-3.

A. Calculate the Reynolds number. Is the now laminar or turbulent?

B. What is the friction factor and the head loss per unit length for this now
(both are dimensionless)?

C. What is the change in pressure (Pa) over a 10-m length of the hose?

Problem 8. Lava, with a density of 2700 kg m-3 and viscosity of 1.0 x 103
Pa-s, nows through a conduit that is circular in cross section. The diameter
of the conduit is 1.0 m. Flow of lava through the conduit is driven by a
pressure gradient of -2.0 kPa m-I. Assuming the now is laminar, what is
the discharge of lava (m) S-I)? Is the assumption of laminar flow valid?

3.11 Suggested Readings

Fox, R. W., A.T. McDonald, and P.J. Pritchard. 2011. II/troductiol/ to Fillid
Mechanics, 8th ed. New York: John Wiley & Sons.
Middleton, G.V., and P.R. Wilcock. 1994. Mechanics in the Earth and
Environmental Sciences. Cambridge: Cambridge University Press.
Chapters 9 and II, pp. 296--336, 365-394.

119
4 Open Channel Hydraulics

4.1 Introduction
4.2 Specific Energy
4.2.1 Flow over a vertical step
4.2.2 Flow criticality
4.3 Discharge Measurements Using Control Structures
4.4 The Effect of Bed Roughness
4.5 Channel Flow Equations
4.5.1 Velocity distribution in open channels
4.6 Measuring Flow in Natural Channels
4.7 Concluding Remarks
4.8 Key Points
4.9 Example Problems
4.10 Suggested Readings

4.1 Introduction

In this chapter, the principles of fluid mechanics are applied to surface­

water flow in channels, for example, rivers, streams, and canals.
Observations of flow in channels suggest many interesting and important
questions. Why does the flow in rivers vary from deep and tranquil to
shallow and torrential? What controls the depth of watcr in a river? How
are water depth and discharge in a stream related? How does thc velocity
in a stream change as the amount of water carried by the stream increases?
Consider the last question: How are velocity and discharge in a stream
related? It turns out that the answer to this question is a key to determining
how fast a flood will move down a river valley. We also need to know
about thc speed of watcr movemcnt if we want 10 know how fast a
contaminant that is spilled into a river moves toward the water intake of a
city's water supply. An instance of the latter occurred in the early 1990s

120
when a train car containing a hazardous chemical derailed and spilled into
the river upstream of a city in California. It was imperative in this case to
know when the city water intakc had to be shut down to avoid drawing the
contaminant into the city's water-supply system. It turns out that, for river
channels, there is generally a useful relationship between mean velocity of
flow in a channel and the discharge in the channel (Figure 4.1).
The relationship between water depth and discharge is one of very
broad importance in hydrology. We often need to know how the depth of
water in a channel is related to thc quantity of watcr the channel is
carrying. Unlike the pipes and hoses that we considered in the last chapter,
channels do not have "lids." The water depth varies along with the other
hydraulic quantities. In the following discussion of hydraulics in stream
channels, the rclationship between watcr dcpth and discharge will be seen
to be important for flow mcasurement. Our primary goal in this chapter is
to explore the physical basis for relationships among the hydraulic
variables that we can measure in channels. In Chapter 5 these principles
will be applied to stream hydrology.

10 ,-------�--__,

-
�

,
.,

E
�

:;, 0

>-
""
0 0
0
Qi
>
c:
co
.,
:0

0.1 �--�����o�--�------�----------�
0.1 1 10 100
Discharge, Q (m' S-')
Figure 4.1 Velocity and discharge in rivers usually are related by a power

121
function, which plots as a straight line on logarithmic axes. The equation
of the straight line shown in the graph is U= 0.28(11-62.
Dala from Miller el a1. (1971).

Obviously, all of the important questions relating to flow in channels

cannot be answered fully in one brief discussion on channel hydraulics.
Many important flow phenomena can be described in a reasonably
straightforward manner, however, by considering the equation of
continuity, the Bcrnoulli equation, and the friction factor relationships
developed in the previous chapter.
In developing some of the principles of fluid mechanics in Chapter 3,
we started with the assumption of frietionless flow. Of course, there are no
fluids that have zero viscosity or any pipes that have no roughness at their
surfaces. Nevertheless, the Bernoulli equation assuming no friction is a
uscful starting poim for a discussion of the physics of fluid flow in pipes
and hoses and can be applied directly in many problems. A similar
approach is useful for channel flow. Some concepts about how water
depths vary with discharge are appreciated best by considering frictionless
flow. It again turns out that the concepts derived from an assumption of
frictionless conditions are useful for measuring flow in channels.
Therefore, we will start our discussion by exploring specific energy in a
frictionless channel and will then proceed to consider flow in "real"
channels where friction has a dominant effect.

4.2 Specific Energy

To begin, consider how velocity and water depth are related in open
channel flow. An open channel flow differs from flow in a pipe in that the
channel flow is only partially enclosed by a solid boundary. The upper
"boundary" of an open channel flow, between the water and thc
atmosphere, is called a free surface. To keep thc analysis simplc we will
consider the flow to be steady and frictionless so that the Bernoulli
equation can be used.
Our first step will be to adapt this equation to a foml particularly suited
for application to channel flow. First consider flow in a short section of
channel (Figure 4.2). In a very short channel section, or reach, frictional
losses often c(ln be disregarded, (IS can any chrmge in Ihe elevation oflhe
bed. The Bernoulli equation for a streamline in this short segment of
channel is:

122
 II' 1
+_
1 _+: = H (4.1)
2g pg
_

1/2 P
-+-+(=.+"-d)� fI,
2g pg

f
", water depth d
Streamline

FIOW --...
I

I
Zb' elevation z

t� atum (z = 0)
j of bed

Figure 4.2 Horizontal flow in a short channel section. The flow can be
considered uniform and frictionless over this section. The total energy per
unit weight of the water along the streamline is lJ2J2g + Ii + Zb = E + Zh,
where E is the specific energy.

where zb is the elevation of the stream bed above datum [l], It is total
water depth [L], d is the depth of the streamline below the free surface [L],
and H is the total energy per unit weight or total head [ll The "datum" is
just a convcnient rcference elevation; often mean sca level is used. As
discussed in thc derivation of the Bernoulli equation in Chapter 3, H is a
constant. The tenn zb + II d is the elevation head. If the flow is essentially
-

horizontal, then the fluid acceleration in the vertical direction is negligible

and the pressure will vary with depth beneath the free surface according to
the hydrostatic equation. Thus, p = pgd and Equation 4. 1 can be written as:
'
1/
-+d+(:b+h-d)= H.
2}l

At this point a useful modification can be introduced: d + (Zb + II - d) is

123
equal to Zb + h, where h is the total water depth and zb is the elevation of
the stream bed above datum. This would be tme for any streamline
because it is independent of d (e.g.. the gage pressure at the water surface
Cd = 0) is zero and the total elevation head is zb + h).
For frictionless flow the velocity is assumed constant over the depth. In
this case, the velocity on each streamline is the same and is equal to the
channel mean velocity V (L .1], so that Equation 4.1 can be written:
u'
-+,,+=� = H. (4.2)
2g

The first two terms of this equation, V2/2g + II, can be interpreted as the
energy per unit weight of the flowing water relative to the stream bottom.
This combined term is defined as the s pecific energy, E [L]:
u'
£=-+11. (0)
2g

Equation 4.2 can be rcwritten:

£+:b= fl. (4.4)

From Equation 4.4 we see that irthe elevation of the channel bottom
remains constalll, then because H is constant, E must also be constant.
The continuity equation for the steady, uniform flow of a constant­
density fluid states that the discharge Q [L3•1] remains constant so that
VI A I = V2A2, where A is the cross-sectional area of flow [L 2]. To simplify
the analysis even further, leI the channel have a reclangular cross section.
Then the cross-sectional area is equal 10 the channel width, IV [L], times
the water depth, II:

Q '" Vwl!.

If, for the moment, we also assume that the width of the channel is
constant, Equation 4.5 implies that:

Q
q.., = = VII= constant. (4 .6)
w

where qw is called the specific discharge [L2.1]. From Equation 4.6, V=

q,/II, which can be substituted into Equation 4.3 to give:

(4.7)

124
 Equation 4.7 provides a means for detennining the depth of a
frictionless channel flow of known discharge and specific energy. That is,
given numerical values for E and for qw' h is the single unknown quantity
in Equation 4.7. Although it is possible to solvc Equation 4.7 directly, a
graphical solution is more convenient. A specific energy diagram is a
graph of specific energy, E, versus depth, II, for a given value of specific
discharge, qw (Figure 4.3). From the specific energy diagram, we see that
for any specific discharge there are three depths for each value of specific
energy. (Equation 4.7 is a cubic equation in II, so there are three roots).
Only two of these possible depths, hi and hz, are positive; the third depth,
113, is negative and can be disregarded in that it has no physical meaning.

The two positive depths that are possible for a single value of the specific
energy and the given specific discharge are called alternate depths.
The specific energy diagram (Figure 4.3) exhibits a minimum in E,
which corresponds to the condition (called the critical condition) in which
the solutions hi and 112 of Equation 4.7 coincide. Thus, under critical
conditions, a flow with specific discharge, qw' has the minimum specific
energy. The corresponding depth is known as the critical depth, he.
Critical flow is a unique condition in which, for any given value of specific
discharge, thcre is only one possible water dcpth (i.c., the critical dcpth).
For values of E with two possible (alternate) depths, one will be larger
than the critical depth and the other will be smaller (Figure 4.3). The flow
depth hi corresponds to II flow that is deeper and slower than the critical
flow and is called subcritical flow. llle flow with depth il2 is shallower
and fastcr than the critical flow, and is termcd supercritical flow.
The original problem poscd at the beginning of this discussion was to
detennine the water velocity and depth for a given discharge and stream
cross section. Application of the Bernoulli equation and the continuity
equation led to the specific energy diagram. This suggests that an answer
to the question dcmands additional knowledge (the specific cncrgy of the
flow) and that two combinations of depth and velocity are possible. Thus,
we will have to examine the question further to delennine how specific
energy and depth are related.

125
 1.5 -,------"71

E
-
'"
£0.5
a.
�
c

-O.5+�----_r---��_,-----___l
o 0.5 1 1.5
Specific Energy, E (m)
I
Figure 4.3 A specific energy diagram for q", = 0.5 m 2 S- . The three
depths, "1 ' "2' and 113, are those allowed by theory for a given specific
discharge and a given specific energy (0.75 m in this figure). The three
depths arc the solutions to the cubic equation relating E, II, and q",.
Because negative depths have no physical meaning, the upper pan of the
diagram, with positive values of It, is used in practice.

4.2.1 FIO\\ OHr a yertical step

A comparison between now through a hose with a contraction and the free
surface analog, a stream with a shallow risc, or step, in the bottom
illustrates the role of specific energy in understanding problems of open
channel flow (Figure 4.4 ). From the equation of continuity and Bernoulli
equation, it is easy to compute the velocity or pressure in the narrow
section of the hose if we know these values for the larger section. As found
in Chapter 3, V2> VI and P < PI for a steady, homogeneous, frictionless
2
now through a pipe contraction such as that illustrated in Figure 4.4.
Unlike the now through the hose contraction, the stream would seem to
have a "choice" as it nows over the step. To constrain the now we will

126
stipulate that the width efthe stream is eenstant and that flew remains
within the ehannel. What will the surfaee an as the flew eresses the step:
rise.. fall. er remain unehanged? Assuming the transitien is essentially
frietienless. the Berneulli equatien {4.2} ean be written:

A \ -
._ . fl- ______ ? ————————

_...
U1
_....
U:
_...
U]
.........
{_f1

U /

(a) Pipe with eenstrietien (in) Channel with step

Figure 4.4 Fer the ease efa pipe (at) er an epen ehannel (b). ean UL] and p;
er a: he ealeulated if U] and pl er in are knewn‘? The answer is a
straightferward “yes” fer (a). hut fer (b) the ealeulatien efh2 is net quite
se simple. The answer is gained by eensidering the relatienship ameng
speeitie energy. speeifie diseharge. and water depth.

if: if;
j"+Jn: 1 ' +hg+hm [+8]
"fg "' I?

where the stream hettem at. the upstream leeatien is ehesen as the datum
and he is the height ef the step. Restating Equatien 4.3 in tenns e'fspeeifie
energyt

El = E3+.-i't:= H. {4.1};
Heeause this is a frietienless Ilew. the tetal energy H must be eenstant.
Hewett-er. Equatien 4}} shews that speeifie energy will net he eenstant if
the elevatinn et" the ehannel hettem ehanges.
We begin by eensidering the ease in whieh the flew upstream efthe
step is suheritieal. Presenting that its flew yeleeity and depth are knewn,
we earl eempule the upstream speeifie energy.
1'.
a; = at +gi~ teen
as
and we ean lee-ate El en a speeifie energy diagram heeause the speeifie
diseharge alse is knewn (Figure 4.5]:
 (4.11)

From Equation 4.9 we next can find £2 ' the specific energy over the step;
it is equal to EI minus the height of the step, Liz. As the flow moves over
the step it gains potential energy at the expense of specific energy. We can
make Ihis subtraction directly on the energy axis of the diagram. We now
have the specific energy over the step, £2 ' but we still have to determine
which of the two depths possible for the single values of £2' h2 or hi. is
corree!.

1 ,----0
E,
0.9

0.8

0.7
Subcriticallimb
0.6
-

E
-

""
£ 0.5
C.
� 0.4
-.6.z-
0.3

0.2

0.1
Supercriticallimb
o
o 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Specific Energy, E (m)
2
Figure 4.5 Specific energy diagram for now over a step; q", = 0.5 m 5-1,
The specific energy at station 1 , 0.8 tn, is represented by the vertical line to
the right. The specific energy at station 2 is 0. 1 m less than that at station I
because a step up i n the bottom has increased the potential energy, thereby
decreasing the specific energy. £2 is shown as a vertical line on the
diagram, 0.1 m to the left of (less than) £1.

The proper choice of depth is best argued from a consideration of the

specific energy diagram, Figure 4.5. Regardless of the depth the flow

128
establishes over the step, the change must proceed along the same specific
discharge curve, because q", is constant. Thus, as the flow moves up the
step with decreasing specific energy, the depth decreases, first to h2, and
then perhaps to hi.
To go from hi to 112 requires only a rise in the bed of elevation z. For the
flow finally to reach II; it must pass through the minimum specific energy
at depth 11o, and then procced to hi. But to go from ho to hi would require
an upward step of a certain critical height followed by a downward step
such that the final height of the step is less than the initial height (Figure
4.6). It is obvious then, because our step does not include this hump, that
the ollly depth accessible to Ihe flow is the first choice, "2'

.. .. ......................
:: '
"

" '.
" "

.
'.
•

.
"'" 112
'

1
............... ----.--
t,
",
6.Zaitiwl

[
Figure 4.6 A critical step is rcquircd to reach an alternate depth. If the
upstream flow (shawl/to file left ill the diagram) is suberitieal, the
hi
alternate depth can be reached only if the flow follows the specific
energy diagram around the "nose" of the curve. This means that the flow
must be critical at some point. The only way that critical flow can be
created under the conditions we are considering is by having a step
sufficiently high to cause the specific energy to drop to Eo.

We now know the water depth over the step, "2' and the height of the
step, z, but we still don't know if 112 + D.z is larger or smaller than hi; that
is, whether the water surface elevalion increases or decrcases over the step
(see Figure 4.4). To resolve this question, Equation 4.8 can be rearranged
in the fonn:

129
We already have found from Figure 4.5 that 112 is less than hi; therefore,
U2 must be grea\er than VI if the discharge is constam, and so,

U,' U'
1 >0.
2g 2g
-'-'-_

Thus, III - ("2 + 6z) > 0, or

(4.12)

So, in a subcritical flow the water surface drops and velocity increases as
water flows over the step. This conclusion explains why canoeists are wary
of sudden changes in stream elevation, as they often mean that large rocks
are below.
Notc that if the upstream flow is supcrcritical with specific cncrgy E1
and depth 11; the specific energy will still decrease as the flow crosses the
step. However, in this case the decrease in E is accompanied by an
increase in depth from 11; to II;, and, therefore, a decrease in velocity
because the discharge remains constant (see Figure 4.5).
It is also important to stress that if the step height, 6z, is greater than the
differcncc bctwccn E 1 and the spccific cnergy of the critical flow (Figure
4.5), EI would not be sufficient to allow for the flow of the specific
discharge qw across the step. [n these conditions it is observed that the step
modifies the flow and induces an increase in E in the surroundings of the
step, thereby allowing the flow to occur across the step with minimum
specific energy (i.e., in critical conditions). [n this case the step functions
as a broad crested weir, as explained in Section 4.3.

4.2.2 Flo" criticalit)

From the example of the flow over the step we have shown that for a
single value of the specific discharge, q,,� and specific energy, E, there
generally are two possible depths. It is also clear from the specific energy
diagram, however, that there is a minimum specific energy, Eo, which has
only a single corresponding depth, 110 (Figure 4.5). This implies that for
any value of specific discharge, there is a depth at which specific energy E
'=II + t.J2/(2g) is a minimum. Note, however, that this does not mean that

130
either the depth or the velocity is at a minimum value. We can derive a
relationship between the velocity and depth at this condition of minimum
specific energy as follows:

(4.\3)

At E equal to a minimum,

(4.14)

or

(4.14)

and

(4.15)

This last equation states that when the mean velocity is equal to JgJ'o' the
specific energy is a minimum.
The line defining critical flow conditions divides the specific energy
diagram into two regimes (Figure 4.5) . As noted in the previous section,
for any value of specific discharge, the regime above this line has slow,
deep flow relative to the flow below the line. This upper regime is called
subcritical flow and the lower regime is referred to as supercritical flow.
Critical flow occurs along the dividing line of these two regimes. For any
value of C/o. the specific energy will be at a minimum when the flow is
..

critical, that is, when,

(4.16)

The dimensionless parameter F is called the Froude number. IfF> I, the

flow is supercritical, and if F < I, the flow is subcritica1. When F = I, the
flow is critical.

4.3 Discharge Measurements Using Control

Structures

I3 I
How does a flow come to be either subcritical or supercritical? One of the
obvious ways that flow can be controlled is by using some type of
structure. Wc intuitivcly expect that a dam will cause a river to back up
and that the resulting flow within the rescrvoir will be deep and suberitieal.
On the other hand, the rapid, shallow flow of water down the face of a
steep spillway must be supercritical. For the case of frictionless flow
considcred so far, subcritical flow can be thought of as controlled by a
structure downstream (e.g., the dam is downstream of the suberi{ical
region) and supercritical flow controlled by a structure upstream.
The control that a structure, such as a dam, has on flow through a
channel can be exploited for making measurements of channel discharge.
Upstream of the dam the flow is subcritical. The flow over the spillway is
supercritical. Therefore, somewhere over the dam crest the flow must
reach the critical condition where U = J
gh, that is, the Froude number
equals 1 . In fact, from the discussion of critical flow in the previous
section, we know that we can force any channel flow to become critical by
placing in the flow a properly designed step across the channel. An
artificial obstruction like a step or dam over which all the water in a
channel must flow is referred to as a weir. There are various types of
weirs, ineluding sharp-crested, broad-crested, and notched weirs. Here we
will focus on broad-crested weirs.
Consider the flow over the broad-crested weir shown in Figure 4.7.
Because we are interested in only a short reach of flow we can expect that
Frictional energy losses will be small enough that we can use the Bernoulli
equation. We also will assume that the flow is steady on the time scale
required for making a measurement. Using the Bernoulli equation to relate
the flow over the weir crest (position 0) to the flow upstream oFthe weir
(position I), we obtain:

(4.t7)

The upstream flow is subcritical, so UlIghl < I. In fact, generally the flow

just upstream of the weir will be relatively slow and deep, in which case
we can make the assumption that the Froude number is small enough that
UlI2K« II. In this case, we can neglect the velocity tenn on the left side
of Equation 4. 1 7. lfwe choose our datum to be at zl, the level of the
channel bottom upstream of the wcir, Equation 4 . 1 7 reduces to

132
 h lI'eir

"0
I ,

/ 1\,
Zo

.
Figure 4.7 A broad-crested weIr. The structure causes water In the channel to
• .

"back up" until the flow over the weir is critical. Because we know how velocity
and water depth are related for critical flow, measurement of the height of the
water in the pool behind the weir can be used to detennine the discharge in the
channel.

U6
IIt==o+lIo+-=/-I
2g

or, if we define h eir as the height of water upstream of the weir relative to
...

the weir crest, i.e., h"'eir h, -zo, then

=:

(4.18)

At some point over the weir crest, the flow must be critical. At this point
U6 gila or UJl2g
= hr/2
Substituting this into Equation 4. 1 8 gives:
=

"0 3 o
110+"2
h..
·nr=
"2h .
=

or

]fwe then substitute Ihis expression for 110 into Equation 4. 1 8, we find that:

133
Rearranging to obtain an expression for velocity,

(4.19)

Equation 4. 1 9 tells us the velocity at a particular point over the weir

crest, but what we really want to know is the discharge Q at that point
because, unlike the velocity, the discharge over the weir will be the same
as that upstream and downstream of the weir owing to conservation of
mass. To obtain discharge from velocity, we need to multiply the velocity
by the cross-sectional area of the flow. Over the weir where the flow is
critical, thc dcpth is ho and the across-channel width of the weir crest is a
constant we will call IV . Thus,
..

(4.20)

Equation 4.20 is the discharge equation or weir fonnula for a broad-crested

weir. In general the flow through a weir can be expressed as in Equation
4.20 in the f0n11

where C is a (dimensionless) discharge coefficient. Using Equation 4.20

q
we find that in the case of the broad-crested weir C = 0.385. Other kinds
q
of weirs have similar discharge equations but with a different discharge
coefficient. For example, in the case of a sharp-crested weir C is bigger
q
(0041), and for thc typical dam spillway C can bc as high as 0049-0.50.
q
The important thing to note is that we can use a measurement of "weir (a
simple task) to obtain a measure ofQ (an otherwise relatively difficult
task).
A broad-crested weir is one example of how a control stmcture can be
used to measure discharge. Other measuring devices that work on a similar
principle include the Parshall flumc in which the flow is forced 10 go
through a width constriction that causes the flow to become critical.
Downstream ofa control structure, such as the spillway ofa dam (or
waterfall), the flow in a channel will once again become subcritical (if the
channel bed has a relatively gentle slope), this time without the benefit of a
control structure. To explain how the channel itself exerts control requires
that the influence of friction be considered.

134
4.4 The Effect of Bed Roughness

Until now, we have considered specific energy in channels with no slope.

In "real" channels, the bottom elevation drops in the downstream direction.
First consider the impl ication of assuming frictionless flow in a channel
with a sloping bottom (Figure 4.8). If the total head, H, is constant and the
elevation of the stream bed above datum, z/>. is decreasing, then the
specific energy must increase continually since H = E + z/>. This cannot be
a general conclusion because cases of very nearly unifonn flows (E, in
particular, not changing in the downstream direction) are observed in
channels with sloping beds. The altemative is to consider that H is nol
constant but decreases in the downstream direction. That is:

dN to
"/ '
_

(4.21)
_

dx -

where Sj is the slope of the total energy line as indicated in Figure 4.8 . Sjis
known as the friction slope becausc it is frictional head loss that causcs H
to decrease. Note that Equation 4.21 can be writtcn as:

S friction slope
f
=

- / =
"L/ L

Ui,{I, 2g -
-L
h,
- I•

".
H,
1
S = slope of bed
I H,
1-
-t-
L
-

"
-
,
-

Datum (: 0)
=

. . .
•
.
Figure 4.8 Fnctlon along the bed causes head loss. The schematic diagram
•

shows the case of uniform now in a channel, a condition often

approximated in streams. If water depth and water velocity remain

1 35
constant in the downstream direction and the elevation of the bed
decreases, the Bernoulli equation indicates that energy is lost to friction as
the watcr flows along thc channel.
dE + d:=_S.
(4.22)
d... dt' J

As dzldx is the negative of the bottom slope, S, Equation 4.22 can be

wntten:

(4.23)

For the case ofunifonn flow (II, h, and hence E unchanging

downstream), Equation 4.23 implies that the friction slope is equal to the
bottom slope. Does the constant specific energy of unifonn flows mean
that these flows do not dissipate energy? Even though E is constant, the
total energy H E + z dccreases in the flow direction. The fact that E is
=

constant and that friction slope is equal to the bed slope means that
uniform flows dissipate as much energy as is provided by the bed slope. In
other words, if the flow is unifon11 the energy dissipation in a channel is
thc same as thc encrgy added by the changc in bed elevation between the
upstream and downstream sections of the channel. Another way to look at
uniform flows is that they exhibit a balance between friction forces and the
component of gravity in the flow direction.
Equation 4.23 can be used to infer how flow is controlled in a river
channel. To picture this, imagine a laboratory experiment in which water is
caused to flow ovcr a bed of sand and creatc a channel. Thc inflow might
be controlled by a reservoir of water and the upstream specific energy
would then be detennined by the height of the water in that reservoir. We
might suppose that initially the friction slope would exceed the bed slope
and thus that dEldx < O. This means that the specific energy would
decrease in thc downstream direction through bottom friction. If the flow
possesses sufficient energy to move the sand, we might expect the bed to
erode and, therefore, increase the bottom slope, S. As S approaches Sf'
dEldx approaches zero and the flow becomes unifonn, the slope and the
roughness detenn ining whether the flow is subcritical for a given value of
specific energy.
The above analysis is really just an intuitive examination of the
meaning of Equation 4.23. In actual practice, this equation can be used to
dctcrmine how depth and velocity vary in channels but the proccdurcs arc
rather involved (see Chanson, 2004, for a more complete discussion). A

136
simpler approach is to investigate the relationship between velocity and
head loss in a manner similar to that used for flow through pipes.

4.5 Channel Flow Equations

Friction slope is the head loss per unit length and can be represented as
II,JL (sec Figure 4.8). In Chapter 3 , an equation for the head loss in a pipe
was presented:

L u'
111-[--,
D2g
(.3.30)

where D is the pipe diameter andfis the friction factor, a (dimensionless)

function of the Reynolds number and surface roughness. A similar head
loss equation can be written for a free surface flow, but because there is no
"diameter" of the channel, an alternate measure of size must be used. This
measure is the hydraulic radius and the head loss equation becomes:

III. _ [ 1 U2
' (·U4)
L - 4R1/2g
where RI/ is the hydraulic radius and is defined as the ratio of the cross­
sectional area of the flow to the channel wetted perimeter. The numerical
factor 4 is introduced because for a pipe with a circular cross section, the
hydraulic radius is equal to the diameter divided by 4. For a rectangular
channel (Figure 4.9) the cross-sectional area is wh, and the wetted
perimeter is 211 + w. Thus,

wh
RI, = 0;;"'-- (4.25)
211+ 11'
If the width is much greater than the depth, w» II, then
wll
RI/ .. -=11. (4.26)
W

and so for wide streams and rivers it often is appropriate to approximate

the hydraulic radius by average channel depth. (For example, if IV = 4011 ,
the error in using RI/:::: It is 5%.) Note that if we use hydraulic radius to
define the flow Reynolds number R = pURHlop, the critical Reynolds
numbers below which the flow is laminar and above which the flow is
turbulent are now 500 and 1000, respectively. It follows that almost all
flows in channels are turbuleOl! For example, a "shallow, slow" flow

1 37
intuitively might be thought to be "laminar" when, in fact, it is not, a
mistake made quite onen. As a concrete example, suppose water at 20°C
nows in a wide channel with a depth of 0.20 m and a mean velocity of
0.30 m S-I . This now has a Reynolds number of about 6000, well into the
turbulent range.

"

I
. . . .
FIgure 4.9 Cross sectIOn of a rectangular channel. The hydraulic rad1Us IS
•

the ratio of the area of flow (wh) to the wetted perimeter (w+ 2h). For a
very wide rectangular channel, the wetted perimeter is very close to the
width of the channel so the hydraulic radius approximately equals the
water depth.

Rearranging Equation 4.24, we can solve for the velocity as a function

of the slope, hydraulic radius and friction factor,

(4.27)

where we have assumed unifonn now so that flLIL = S. However, if the

flow is not uniform, this equation and its subsequent variations (e.g.,
Equation 4.3 1 ) still can be applied provided that the friction slope rather
than the bottom slope is used. Equation 4.27 was originally proposed by
Antoine Chezy in the late 1 700s and the equation is referred to as the
Chezy equation. The constant, C, called the CMzy number, has the same
units as gll2.
Assuming that the value of C i s known, the Chezy equation permits a
very simple computation of the mean velocity (average over the cross
section) for steady, unifonn now. Unfortunately, the "constant" C depends
on the hydraulic radius and the bottom roughness. Several equations were
proposed after Chezy's initial effort, many of which sought to replace the

13 8
C with a parameter that is a function of roughness only. In 1 891, another
Frenchman, A. Flamant, proposed,

kRI16
C = 1/ . (.US)
"

where 11 is a function of only roughness and is today called the Manning

coefficient; the same expression was separately proposed by others and, in
fact, in Europe II is oftcn known as the Gauckler-Manning cocfficient.
Here we adopt the convention that II is dimensionless and introduce k, a
1
dimensional constant (k = 1 m 1/3 S- in SI units) to maintain dimensional
homogeneity. SubstinJting this expression for C into the Chezy equation
we have,

U=�RJPS'I2. (4.29)
"

Typical values of 1/ range from 0.02 for smooth-bottom streams to

0.075 for coarse, overgrown beds (Table 4. 1 ). For a wide range of natural
stream and river channels in the United States, Barnes ( 1 967) reports
values of II from 0.024 to 0.075. In upland streams, the effective roughness
is controlled by the riffle-and-pool structure and not by the size of the bed
material. [n these cases Manning's 1/ can be very large (Beven et al.,
1979).

Table 4. t. Roughness coefficients for channels

• •

Channel form Manning's II

Laboratory flume (glass) 0.01
Smooth concrete 0.012
Unlined earth canal in
0.02
good repair
Natural streams 0.024-D.075
Very steep upland
0.075-77 (>1)
channels
• •

Note: For natural streams, the report by Barnes (1967) is an excellent reference.

As an example of how bottom roughness affects stream velocity,

consider two streams that are identical except that for one II equals 0.025,

139
and for the other /I equals 0.075. Assume that the slope is 0.0006 mlm, and
the hydraulic radius is 3.0 rn. In this example,

V = 1.0 (3.0)lf3(0.0006)112 =2.35 rn S- I

I
0.025

0
(.4030)
I
V,. 1. (3.0)!Il(0.OOO6)"2 =0.79 rn S-
1 oms

This variation in roughness, equivalent to the difference between a stream

with a sand bed and a stream with a coarse, weedy bed, can have a
significant effect on velocity, and thus discharge. As noted above, for a
wide rectangular channel RH;:;; h, and Equation 4.29 can be written:

V=�,,2J3SI12.
(4.31)
"

Multiplying both sides of Equation 4.3 1 by depth gives an expression for

specific discharge:

q•. = VI! = � ,,!l1SI12.

(4.32)
"

This equation suggests that channel characteristics such as bed roughness

and channel slope control the flow in the sense that the relationship
between qll' and h is established once II and S are specified.

4.5.1 VcJodt)" distribution in open channels

Manning's equation provides a method for estimating channel mean

velocity U given depth or hydraulic radius, channel slope, and an estimate
of Manning's roughness coefficient. As we found for flow through pipes
(Section 3.6), once we include friction in the description of the flow, the
velocity is not uniform throughout the cross section. Friction with the
channel bottom requires that the fluid velocity adj acent to the bottom be
zero. Clearly the velocity is non-zero away from the bottom, so velocity
must vary with distance from the channel bottom. We found that for
laminar pipe flow, the shape of the velocity profile is parabolic (sec Figure
3 .9) . The velocity profile for a laminar channel flow also would be
parabolic, with the maximum velocity at the channel surface (rather like
cutting the pipe in half). The presence of turbulence in the flow greatly
enhances the mixing of fluid in a flow compared to the mixing resulting
from molecular viscosity. As a result, the velocity distribution in a

140
turbulent channel flow is logarithmic rather than parabolic (Figure 4. 1 0) .
The logarithmic velocity distribution of a turbulent channel flow is
given by the Kam1an-Prandtl equation:

(4.33)

where kr is a channel roughness parameter. In writing the equation in this

fonn we have assumed that the channel bottom is hydraulically rough, a
good assumption for most stream beds. For channel beds of nearly
unifonnly-sized sand or gravel, k,. is approximately equal to the diameter
of the bed material. For more complicated stream beds the problem of
detennining kr is a difficult one and often requires measurements of flow
velocity in the stream reach of interest.

4.6 Measuring Flow in Natural Channels

Often we don't know enough about a stream to make even good estimates
of flow velocity or discharge without making some direct measurements.
Channel cross section and slope can be measured using meter sticks and
surveying equipment. Channel roughness cannot be measured directly.
Instead, it must be detennincd by measuring all of the other variables in
Manning's equation or the Karman-Prandtl equation and rearranging the
equations to provide an expression for 11 or k,.. To do this requires
measurements of velocity in the channel using a current meter. Current
meters measure velocity at a point. How can we most efficiently use point
measurements to estimate discharge?
Mean velocity in a channel is the sum, or integral, of the velocities at
each point in a cross section divided by the cross-sectional area:

u= -
[ f·f'u(y.:)d:dy.
A • • (4.3")

The Kannan-Prandtl equation tells us how velocity varies in the vertical

direction (z), but there are no generalizations for the variation of velocity
in the cross-stream direction (y). If the channel is wide and rectangular,
then we can reasonably assume that the velocity is the same at each point
across the channel. In this case integrating across the channel is equivalent
to multiplying by channel width, so that Equation 4.34 simplifies to:

141
 1 ,-------,-

0.9 ,
,
Turbulent ,

flow ,

0.8-
,
,
,
,

'" 0.7-
,

.
,

."
,

I
,
,
,

;: 0.6-
-

,
,

e.
,
,
"

J//
'tJ 0.5-
"
>
:; 0.4
"
-

a: 0.3-
-
-

-
-
-

0.2- -­

Laminar /"
0.1-

o ¥ --
�/ ��
�
�........
-
..

----r "'" ::::;�----r------1-l

..
� -

o 0.5 1 1.5
Relative velocity, it/V
Figure 4.10 Velocity distribution in a channel. For laminar flow (almost
never observed in natural channels), the velocity profile is parabolic,
similar to the case examined for pipe flow in Chapter 3. For turbulent
flow, the velocity near the bed changes much morc rapidly than it does in
laminar flow. The velocity profile for turbulent flow generally is described
using a logarithmic equation.

1 1
V= -

11'11 I'
\II
0
lI(z)dz =

110
-

I'u(z)dz.
If we substitute the Karman-Prandtl equation for lI(z) and carry out the
integration, we obtain:

(4.35)

where, because we are assuming the channel is wide and rectangular, we

can take RH � h. We can deduce an interesting and useful result by
comparing Equations 4.33 and 4.35. The velocity in a channel varies from
zero at the channel bed to a maximum value at the channel surface (Figure

142
4.10). The mean velocity is some intennediate value and thus there will, in
general, be some level Zu at which the velocity u(zu) is equal to U. Setting
Equations 4.33 and 4.35 equal to each other and solving for the value ofZu
gives zu= 0.37h (Figure 4. 1 0). This means that all we have to do if we
want to estimate mean velocity is to measure the velocity at a level
roughly 0.411 above the bottom or 0.611 below the surface.
Of course, most natural channels are not rectangular in cross section so
that estimating U with Equation 4.33 letting II = average depth cannot be
expected to be very accurate. However, we still can make use of the results
for rectangular cross scctions by dividing a more complicated channel
cross section into a number of rectangular subsections as illustrated in
Figure 4. 1 1 . For each subsection, the width and depth can be measured
with a tape and meter stick and the mean velocity can be approximated as
the measured velocity at a depth 0.611 below the surface. Taking Ihis
approach, we can re-writc Equation 4.34 as:

U = �L N

,. ,
II',II,U" (4.36)

where N is the number of subsections, Ui = u(z = O.4l1i), and

Q UA = L 1Il,II,U,. (4.37)
,.,
=

Once we know mean velocity, slope, and depth (or hydraulic radius),
we can determine the value of Manning's 11 using Manning's equation. If
we did this for a number of different discharges in a given stream reach we
could estimate a characteristic value of 1/ for the re<lch that could then be
used to predict the mean velocity and discharge for depths other than those
for which direct mcasurements wcrc madc. Leopold and Wolman (\957)
derived an empirical relationship between the friction factor, water depth,
and DS4' the size of bed material of which 84% of bed material is finer (see
also Leopold, 1 994, for other examples of data supporting the empirical
relationship). In terms of Manning's 11, the empirical relationship is
(assuming that RII = II)

143
 I W'I
B 9
7
,

2 --- -'

3 6
h, 5 --- ,
- --

-- -- -

Figure 4.1 1 Method for determining discharge in a stream. The stream is

subdivided into 11 number of rectangular elements. A current meter is used
to measure the speed of the flow at a distance 0.411 from the stream bottom
in each rectangle. The water velocity at this depth is approximately the
average velocity for that segment, assuming that the logarithmic velocity
profile for turbulent flow is valid. Discharge is calculated by multiplying
the average velocity for each rectangle by the area ortha! rectangle and
summing across the stream. (For additional detail on this method, see
www.rcamnl.wr.usgs.gov/sws/SWTfaining/WRIR004036NAMethodlVelo

�-
II
- :EL[, (-"-) 1
"
116 _.Olog. lo
DS4
+ 1.0 . (4.38)

Measured and calculated (using Equation 4.38) values of 11 are shown in

Figure 4. 1 2 ror a site on Brandywine Creek in southeast Pennsylvania.
Manning's 11 generally decreases with increasing discharge at a site, at
least for flows in which the river stays within its banks. This is at least in
part because the bottom roughness, indicated by some measure of size of
the bottom material, e.g., DS4' protrudes farther up into the flow for
shallow flows than for deep flows, and thereby increases frictional energy
loss.

144
"
'"
• 0.2 i---- --- -;======;---
+ Measured
o Predicted

o
0.1
+ -

0.06 o
o

+
0.04

Measured discharge, Q (m3 5")

Figure 4.12 The variability of Manning's II with discharge for Brandywine
Creek at Comog, Pennsylvania, can be determined from the values of
measured flow, water depth, and other factors reported by Wolman ( 1 955).
Values calculated with Equation 4.38 used D84 = 0. 1 5 m and assumed that
RH=h.

Manning's equation, in conjunction with an estimated value of II, can be

used to infer stream velocity and discharge in the absence of direct
hydraulic measurements. For example, Manning's equation provides a
means of estimating velocity and discharge under very high flow
conditions when it is not possible to make flow measurements directly.

4.7 Concluding Remarks

There are many important applications of the theory developed above and
some oflhese will be explored in subsequent chapters. Here we briefly
consider examples of how the relationships derived above can be useful in
interpreting infonnation about flows in channels.

145
 Recall that our consideration of Manning's equation suggested that
discharge should be related \0 the 5/3 power of stream depth for a stream
with a rectangular cross section (Equation 4.32). Streams rarely have
strictly rectangular cross sections, of course, but our analysis could be used
to suggest that a power relationship, perhaps with an exponent different
from 5/3, might be useful for describing stream data. In fact this often is
the case. For Brandywine Creek near Comog, an equation Q = 28.26h2.26
fits the observations of discharge and depth reasonably well (Figure 4. 1 3).
As we will see in the next chapter, such power relationships are used
routinely in making continuous measurements of stream discharge,
measurements that are essential for solving many practical and theoretical
hydrological problems.

10

�
r

"'
,

"

E
�

Ol
Q) 1
"

'"
�
co
.c
"
"'
C
,-

0,1
0.1 1
Water depth, " (m)
Figure 4.13 Stream discharge is related to a power of stream-water depth
for many streams and rivers. For Brandywine Creek at Comog, the
exponent in the empirical relationship between Q and h is 2.26.

Fluvial geomorphologists use power relationships to describe the

interrelationships among many streamflow variables of interest to them,
not only Q and h. Consider some general observations on hydraulic

146
variations along the longitudinal profile ofa river, that is, on the changes
in slope, width, depth, and velocity in the downstream direction. Leopold
et al. ( 1 964) and Lcopold ( 1 994) present data of this type for a number of
rivers. Relationships tend to be straight lines on logarithmic axes (i.e., to
be power relationships) for most river systems; the Brandywine Creek
basin is an example (Figure 4.13). Of course, slope decreases as a river
progresses from the headwaters to the sea and discharge generally
increases bccause of the steady increasc in drainage basin area. The
observed increase in width and depth also seems intuitively plausible.
Velocity also increases in the downstream direction.
There are many interrelated changes that occur among hydraulic
variables in streams and these must be considered collectively in
explaining the observed changcs in mean velocity of a river system. Smith
(1 974) assumed that Manning's equation held for a hypothetical river
system with constant roughness (Manning's /I constant in the downstream
direction), that discharge increased directly with drainage area, and that the
basin was eroding at a constant rate. He used bank filII discharge, the
discharge in the stream when it is filled right to the top of the channel, in
his analysis. Using empirical equations for sediment transport in the river,
he then showed that slope should be proportional to discharge to the
negative 0. 1 8 power, that velocity should be proportional 10 the 0.09
power of discharge, that depth should be proportional 10 Ihe 0.27 power of
discharge, and that channel width should be proportional to the 0.64 power
of discharge. The corresponding powers for the Brandywine basin (the
slopes of the lines in Figure 4. 1 4) are -0.69, 0.30, 0.20, and 0.43. The
magnitudes of the powers for this particular basin do nOI correspond all
that closely with Smith's theory, but again the fonn of the relationship is
similar and the hydraulic interpretations help provide a consistent
conceptual basis for explaining natural phenomena. These ideas also are
useful in assessing the potential impact of modifications on a stream
channel (e.g., how stream channelization affects all of the pertinent
variables in a channel as a result of changes in slope).

147
 1 100
8 0

E E
0
- -

0
- -

�
00
•
"
� � 10
"

c- 0
-
-
"
•
c �
0.1 1
1 10 100 1 10 100
Bankfull discharge (rna 5-1) Bankfull discharge (m3 s-')
0.1 10
"
-

>-
-

E
,
-

0 'i3 .::--
E 0'.
- 00 -
0
'" 0.01 0 � E 1,
"
• 0 c- o
c- o;,
0 0 •
-
<J) "

0.001 0.1
1 10 100 1 10 100
Bankfull discharge (m3 5-') Bankfull discharge (m3 5-')
Figure 4.14 Hydraulic parameters in the downstream direction for the
Brandywine Creek basin.
Data from Miller ct a1. (1971).

4.8 Key Points

• Many important flow phenomena in rivers can be described in a

reasonably straightforward manner by considering the equation of
continuity, the Bernoulli equation, and the friction factor relationships
developed in Chapter 3. {Section 4.1}

• The Bernoulli equation for flow in a frictionless channel can be written

Lf2/2g + II + zb = H, where U is mean velocity [L r-I], II is water depth
[L], and zb is the elevation of the channel bottom above datum [L].
{Section 4.2}.

• Specific energy E [L] is the energy per unit weight of water in the stream
relative to the stream bottom: E= U2/2g + h. Total energy, H [L], is
therefore E + zb' {Section 4.2}

• For channels with a rectangular cross section, a useful relationship

148
 among spccific energy, specific dischargc, and water depth follows from
the definition of specific energy: E q�.!2 gh2 + II, where qw is the
=

specific dischargc, the discharge per unit width of channel [L211]. For
given valucs of E and qw' this equation is cubic in II giving three
mathematically allowable depths of water for the given conditions, only
two of which are positive. These two physically meaningful depths are
callcd thc altcmatc dcpths. {Section 4.2}

• For any specific discharge, thcre is somc minimum valuc of specific

energy possible. This minimum occurs at critical flow, with the defining
condition being that the Froude number F [dimensionless] is one:
Uo/�g ho I· For Froude numbcrs below one, the flow is subcritical,
=

slow and tranquil. For Froude numbers greater than one, the flow is
supercritical, rapid and torrential. {Section 4.2.2}

• Flow measurement devices for water in channels can be devised on the

basis ofthc known relationship bctwccn velocity and watcr dcpth at
critical flow. An example is a broad-crcsted weir for which the principlcs
developed in this chapter yield a way to estimate discharge from a
measurement of the height of water above the weir crest:
Q (8g 127)lf2 h���r we' where "weir is the distance from the top of the
=

weir to the water surface in the pool behind the weir and we is the width
of the weir crest. {Section 4.3}

• For steady, unifonn open channel flows, the friction slopc is cquallo the
bottom slope, and depth, velocity, and specific energy are constant.
{Section 4.4}

• The hydraulic radius R" [L] of a stream channel is defined as the ralio of
the cross-sectional area of flow to the wetted perimeter. {Section 4.5}

• Using the friction factor discussed in Chapter 3, the Chezy equation can
be derived: U �8g /f JSR" CJSRI!' where C is Ihe Chezy number
= =

[LI/211], S is channel slope [dimensionless], and RII is the hydraulic

radius. {Section 4.5}

• Manning's equation, U kR1PSlf2/11, is an expression relating mean

=

velocity to slope and hydraulic radius that is very widely used. Values of
the roughness parameter, II, have been tabulated for a range of channel
conditions. {Scction 4.5}

149
• The vertical profile of velocity in a river channel often is assumed to be
logarithmic by virtue of a derivation due to Prandtl and von Karman. The
mean velocity for flows with a logarithmic velocity profile occurs at a
distance of 0.411 above the bottom. The common method of gaging
streams using current meter measurement relies on this approximation.
{Sections 4.5. 1 and 4.6}

• Manning's II generally decreases as a function of discharge in streams, at

least for flows within the stream banks. This decreasing roughness can be
thought of intuitively as due to the relatively larger size of the bottom
roughness (e.g., pebbles in an upland stream) with respect to stream
depth at low flows versus the relatively smaller size at high flows.
{Section 4.6}

4.9 Example Problems

Problem I. Water flows in a 3.00-m wide rectangular channel. The water

depth is 1.50 m and the discharge is 1.50 m3 S-I . The channel bottom drops
smoothly by 0.100 m over a short distance (a step down in the bottom)
with no head loss or change in the width of the channel.

A. Calculate the specific discharge (m2 S- I ) and specific energy (m) at the
upstream station.

B. Calculate the specific discharge (m2 S-I) and specific energy (m) at the
downstream station.

C. Calculate the water depth (m) at the downstream station.

150
 w

w/2.S

.
Figure 4.15 Canal cross secllon for Problem 2.

Problem 2. A discharge of 2.0 mJ S- l is carried in a canal with the cross

section shown in Figure 4. 15. The canal is 1400 m long and drops 0.50
min elevation over thaI distance. Manning's /I for the channel is estimated
to be 0.020. What is the value of IV (m) for this canal?

Problem 3. As noted in Problem 3 of Chapter 3, channels and pipes of

varying geometry will be characterized by different formulations orthe
Reynolds number. For example, Rpip" =- pUD/�, where D is pipe diameter,
whereas a more appropriate rOm1 for channels might be R = pUR,/Il,
where RII is hydraulic radius. Use of a different length scale in R will
generally alter the upper limit of R for laminar flow relative that that found
for pipes. In this problem, we refonnulate R using hydraulic radius as the
length scale to obtain a more general fonn of R that we can use to relate
the laminar·turbulent transition in pipes to other flow geometries.

A. Find an expression for the hydraulic radius RH in temlS of pipe diameter

D for a pipe with a circular cross·section.

B. Rearrange the expression found in 3A to get a relationship for D in

terms of RH. Substitute this into Rpipe = pUD/)l to get an expression for
R in terms of R".

C. Set the expression found in 38 equal to 2000, the critical Reynolds

number for pipes. Rearrange this to find the equivalent critical value for
RRII = p UR,/�l .

15 1
D. Use the equation you developed in 3C to find the critical Reynolds
number for the transition from laminar to turbulent flow in a relatively
wide, rectangular channel (such that RH :::; II).

Problem 4. Use Equation 4.32 to estimate the depth and mean velocity of
a flow in a channel with a slope S 0.003, width )II = 15 m, discharge Q =
=:

1.0 m3 S- l , and Manning's 11 = 0.075. Does the assumption that RH :::; h

seem reasonable for this flow?

4.10 Suggested Readings

Dunne, T., and L.B. Leopold. 1978. Water in environmental planning. San
Francisco: W. H. Freeman. Chapter 16, pp. 590-660.
Leopold, L.B. 1994. A view ofthe river. Cambridge: Harvard University
Press. Chapter 9, pp. 148-167.

152
5 Catchment Hydrology: Streams and Floods

5. 1 Introduction

5.2 The Hydrograph

5.3 Movement ofFload Waves

5.4 Flood Routing

5.4. 1 Reservoir routing
5.4.2 Flood routing in riv ers
5.5 Flood Frequency Analysis
5.6 Concluding Remarks

5.7 Key Points

5.8 Example Problems

5.9 Suggested Readings

5.1 Introduction

The principles governing flow in channels, discussed Chapter 4, provide

the basis for describing the flow of water in rivers and streams. [n Ihis
chapter we emphasize the response of a river to precipitation and the
nature and dynamics of the resulting flood. The problems associated with
floods arc many and varied, and solving them is an important application
of hydrology.
Flood control is an issue that is important throughout the world. Every
year we read in the newspapers about damage and death caused by river
flooding. Flood insurance, levee construction, and future land-use zoning
are common topics of discussion in the wake of floods. Such discussions
point out the range of options that must be considered in flood protcction
schemes. Structural approaches to flood control, such as levees, dams, and
channel widening, and non-structural approaches, such as floodplain
zoning and preservation of natural wetlands, must be weighed in terms of
costs and benefits. The issues often cause contentious debate and rarely are
settled to cveryone ' s satisfaction.

153
 As an example, consider the situation for Sacramento, California. The
American River basin lies to the east of Sacramento. The basin area is
about 5000 km2 with headwaters in the Sierra Nevada. Three forks of the
river flow through scenic canyons in the mountains, valued by white-water
sports enthusiasts and hikers, and merge just above Folsom Dam. Below
the dam, the American River flows between levees through downtown
Sacramento. More than one million people live in the Sacramento
metropolitan area, within the floodplains of the American and Sacramento
rivers.
.

Extreme precipitation and snowmelt cause high discharges in the

American River from time to time, and the city of Sacramento
occasionally has suffered flood damage. After a major flood in 1 850, the
city of Sacramento decided to rebuild on thc floodplain. A series of flood­
control measures have been implemented, but the city nevertheless has
remained susceptible to devastating floods (NRC, 1 995a).
One of the chief protection measures for Sacramento is Folsom Dam,
on the American River. Folsom Reservoir is a mUltipurpose facility; it is
regulated for water supply, recreation, and flood control. Different
purposes often dictate conflicting strategies for reservoir opcration. To
provide an adequate water supply through a dry period, one would want to
keep a reservoir nearly full; however, to protect against floods one would
want to keep a reservoir empty so floodwaters could be stored. When
Folsom Dam was planned, it was supposed to protect against a flood that
would be expected to occur only once in 500 years on averagc. A scrics of
floods from 1955 through 1986 caused a reevaluation of the protection
levels offered by the dam, reducing the expected protection offered by
Folsom Dam to floods that occur once in 70 years on average. New
construction begun in 2007 and scheduled to be complete in 20]7 is
designed to double thc flood safety rating of Folsom Dam and to increase
the level of protection for the Sacramento area to a 200 year flood.
The question of what options should be considered in planning for
flood protection for Sacramento is not a purely technical one.
Socioeconomic factors are very important and ultimately the question is
one that must be answered as part of the political process. Ideally, thc
political decision is informcd by solid tcchnical knowledgc. Thus,
hydrological questions will arise. For example, the flows of water into and
out of a reservoir must be understood when considering the construction of
a new dam or when evaluating the operating policy of an existing dam.
How can an estimate be made of the level of protection offered by a
dam? A flood-control rescrvoir works by storing watcr during high inflows

154
to the reservoir and later releasing this water at a rate much lower than the
highest flood rates. The problcm of deducing how outflow from a reservoir
is relatcd to its inflows is known as flood routing. As will bc described in
this chapter, flood routing depends on some of the principles of fluid flow
that were covered in earlier chapters.
In this chapter we discuss some selected elements of surface hydrology
and show how the fom1Ulation and solution of surface-water problems are
developed from fundamental hydraulics. We consider the propagation of
flood waves through reservoirs, flood routing, and the frequency with
which a flood of a given magnitude would be expected to occur. There are
many facets of surface-water hydrology that are not included in this
chapter. The material presented here is intended to introduce the subject,
particularly with regard 10 river floods.

5.2 The Hydrograpb

In the analysis of floods, water supply, and other subjects included in

surface-water hydrology, the basic quantity to be dealt with is river
discharge, the rate of volume transport of water. Both river discharge and
depth (or stage) change with time, and an understanding of this temporal
variability is a prerequisite for approaching hydrological analyses. A graph
of river stage or discharge versus time at a point is called a hydrograph
(Figure 5. 1 ). The former is referred to as a stage hydrograph and the latter
a discharge hydrograph.

0.4 (a) Stage hydrograph >'2 -F===

1
03 1
0.8
E E
-

8,0.2 8,0.6
-

Sl � 0.4
li:
0.1 -

o
0.2

o • • O+---�-
0
-�r----r--
o 6 12 18 24 6 12 18 24
Time (hr) Time (hr)
Figure 5.1 The hydrograph. River stage (a) and discharge (b) as functions
of time.

155
 The discharge hydrograph is not measured directly, but is inferred from
the stage hydrograph. In the United States, the United States Geological
Survey is responsible for the measurement of river stage at over 7000
gaging stations. At many of these locations, data are continuously recorded
by a float gage installed inside a stilling well (Figure 5.2). As the river
rises and falls, the float moves with the water level and the motion of the
float is recorded. The stilling well serves both to protect the float
mechanism and to dampen the rapid fluctuations of the water surface due
to local disturbances and turbulence. In recent years, the float mechanism
has been replaced by a pressure transducer, a device that records the
pressure at the bottom of the stilling well. The hydrostatic Equation 3.6 is
used to calculate water depth from the measured pressure. Radar is also
used to measure stage in some rivers.
The record of water depth versus lime measured at a gaging station is
the stage hydrograph (Figure 5 . 1 a). To construct a discharge hydrograph
from these data, determinations must be made of the river discharge for
various values of flow depth or stage. By measuring stream velocity at
numerous positions across the width of the stream, the discharge at a given
stage can be detennined as described in Section 4.6 . These measurements
are generally made with a current meter either attached to a wading rod or
suspended from a cable, depending upon the depth and discharge of tile
fiver.
.

After discharge has been measured for a range of flow depths, a rating
curve can be constructed that relates stage to discharge. Rating curves
typically are nonlinear and often are approximated using a power function
(Figure 5.3). To be useful, a stream gage should be located at a point
where the stage-discharge relationship will not change from year to year;
such a location is called a control section. An example of a control section
might be a place where the flow is constricted by a natural rock fonnation
or by bridge abutments. A station where the cross section changes almost
continually because of streambed erosion or deposition would not provide
good control.

156
 Stilling well
Recorder

---' - ;....-----,

Inta ke
pipe

Figure 5.2 Schcmatic diagram of a stream gaging slalion used 10 record

flow depth (stage) as a function oftime. The resulting stage hydrograph
can be converted to a discharge hydrograph using a rating curve.

After the rating curve for a gaging station is detennined, it can be used
to convert each value of stage to discharge. Consider, for example, the
stage hydrograph shown in Figure 5.1 a. The stage peaks at 0.35 m (at time
= 6hr). Using the rating curve shown in Figure 5.3 , the corresponding peak

discharge is Q = 76.5(0.35 m)4.1 1.0 mJ S-1 (Figure 5. l b). In this way, a

=

continuous measurement of river stage is used, in conjunction with an

established rating curve, to detemline discharge as a function of time.
Almost all discharge hydrographs are dctcnnined in this manner.
A typical stream discharge hydrograph is shown in Figure 5.4. This
hydrograph of Holiday Creek near Andersonville, Virginia, for a period in
1 974, indicates that discharge is highly variable. During and after rainfall
and snowmelt events, water moves through the catchment to the stream
channel and the discharge increases. The resulting peak in the hydrograph
is tenned a flood, regardless of whether the river aClUally leaves its banks
and causes damage. Background discharge between floods is lenned
baseflow and is supplied by inflow of groundwater. There were a number
of floods on Holiday Creek within the 3-month period of the hydrograph
shown in Figure 5.4. Each ofthcse floods has a characteristic hydrograph,
with a steeply rising segment followed by a less steeply falling segment as
the flood passes the station. The exact nature of each hydrograph depends
on watershed and storm characteristics.
The difference among hydrographs is well illustrated by the examples
in Figure 5.5. Poplar Creek is a perennial stream in Tennessee. Floods
occur throughout the year, but the magnitude of the discharge peaks varies

157
seasonally in response to changing weather conditions. In contrast, the Frio
River is an ephemeral stream in Texas that is dry for long periods with
occasional large floods. Little Blackfoot River lies in a snowmclt4
dominated catchment. The hydrograph from a strcam like this is
characterized by a single, extended period of high discharge in the late
spring and early summer when the snow melts. Even within a single
climatic zone, differences in catchmcnt shape, geology, and vegetation will
produce differing hydrographs in response to similar precipitation events.

, ,5
(a) Arithmetic axes (b) Logarithmic axes
"----"
lO-i- -- ------r----,
, ,
- -
•
• ,
,
, ..
• • "

E E
.'
" "

....
•
- -

CI 0.5 CI 0.1
"

o 0,01 +':...-----�
o 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.1 0.5 1
Stage (m) Stage (m)
Figure 5.3 Rating curve (stage-discharge curve) for the Snake River above
Montezuma, Colorado. Thc equation for the lines is Q =
76.5 (stage)4. 1 .
Data from Boyer (1994).

E
- -

•
•

E
,

E
" -

0
c

•
-

,
-

'"
0.
-

,
•
-
•

u u
"

•
,.

•
is 0.
,

May 1. 74 Jun 1 . 74 Jul l . 74 Aug 1 . 74

Figure 5.4 Hyetograph (black bars, scale 01 righ/) and hydrograph (blue
lille, scale al left) for Holiday Creek near Andersonville, Virginia.

5,3 Movement of Flood Waves

158
The floods shown in Figure 5.4 may be thought of as waves that propagate
downstream. The way in which flood waves propagate-their speed, peak
height, and peak discharge-is an important area of study in hydrology.
Planning for the response of rivers to both normal and extreme rainfall
depends on this understanding.

(a) Poplar Creek near Oak Ridge, Tennessee

� 1 00
,
•

E 75
•

•
'" 50
"
•
"" 25
0
.•
-

0 0
OcI ' , St Jan l , S2 Apr t , 82 Jul l , S2 Sep 30, 82

(b) Frio River below Dry Frio River near Uvalde, Texas
� 1 00
:.,
E 75

•
-

l!'
50
•
""

��-----1
25

� 0
5 J,. , . ro . , . ro J." . ro Sep 30. 76
� '.7
� (c) Little Blackfoot River near Garrison, Montana
50
•
E
-

•
25
l!'
•
""
o
.•
-

o 0
Oct 1 , 90 Jan l , 91 Apr l , 91 Jul l , 91 Sep 30,91
Figure 5.5 Annual hydrographs for three rivers in catchments in three
climate zones illustrating the differences among streamflow in perennial
(a), ephemeral (b), and snowmelt (c) streams.

In an "ideal" channel with a frictionless fluid, we might conceive of a

flood wave traveling with no change in form from its point of origin.
However, mechanical encrgy is lost (dissipated) in a natural channel
because of the frictional resistance of the rough channel bed. Water also
becomes stored in pools and backwaters and is subsequently released. As a
result of friction and storage, we might expect the magnitude of a flood

159
wave to be reduced, or attenuated, as it travels downstream. This
attenuation will depend on several factors: the ability of the channel
system to store and release walcr, the channel roughncss, the amount of
vegetation in the channel and its floodplain, the relative straightness of the
channel, and the number of bridges and other obstructions in the path of
the flood wave.
The picture is complicated further because the lateral inflow of water
from tributaries and groundwater will increase the volume of water in the
channel as onc moves downstrcam. In gcncral, dischargc increascs
downstream as drainage area increases. An example may serve to illustrate
the relative effects of drainage area and flood wave attenuation in a river
or stream network. Hydrographs for the South Fork of the South River at
Waynesboro, the Shenandoah River at Front Royal, and the Potomac River
at Washington, D.C., for the same event are quite different (Figurc 5.6).
The South River is a tributary of the Shenandoah and the Shenandoah
flows into the Potomac weU above Washington. Perhaps the most
noticeable difference among the hydrographs is the highest peak flow at
Washington, the lowest peak at Waynesboro, and the intennediate peak at
Front Royal. This seems to contradict the "attenuation by friction and
storage" hypothesis stated above, but as we have already indicated, the
behavior is complicated by the increasing drainage area in the downstream
direction. The effect of increasing drainage area can be taken into account
by dividing discharge by drainage area to "normalize" the discharges
before they are compared. The nonnalized hydrographs indicate thai the
flood wave is indeed attenuated as it moves through the channel (Figure
5 .6). The normalized hydrographs show clearly the time delay between the
flood peaks and the reduction in peak discharge per unit area in the
downstream direction.

160
 5 (aJ Discharge hydrograph
000

-
4000
/ Potomac
1n 3000
•

§. 2000
'"
1000 � '/" �enandoah
0
/" South
0 2 4 6 8 10 12
Time (days from April 13, 1987)

normalized to catchment area

-
•
E
,

�
-

:.

E
•

•
•
• 0.1
,;
'"

o 2 4 6 8 10 12
Time (days from April 13, 1987)
Figure 5.6 Hydrogmphs for three nested basins. Discharge increases
downstream in the stream network due to increased drainage area (a).
Dividing discharge by catchment area (b) reveals both attenuation and
delay of the flood wave as it moves downstream.

5.4 Flood Routing

Flood warning and flood mitigation depend on understanding how quickly

a flood crest travels downstream and how the height of the crest changes
as it does so. Essentially, the problem is to be able to predict what a
downstream hydrograph would be if we know the upstream hydrograph.
This is known as flood routing. [n addition to the upstream hydrograph, the
prediction of the downstream hydrograph requires an estimate of how fast
the flood wave is moving, how much water is being added by inflow, the
influence of friction, and in general a complete understanding of the
hydrology and hydraulics of the drainagc basin. Wc can simplify the
problem by identifying depth and velocity of the flood wave as the two
most important variables. As in the channel transition problem in Section
4.3, the equations of continuity and Newton's second law provide the two

161
equations we need to solve for the two unknowns, depth and velocity.
If a section of river channel, or reach, is taken as the control volume,
then the inflow is given by the upstream hydrograph and the outflow by
the downstream hydrograph. We will assume that the lateral inflow of
water is small relative to the change in storage associated with the flood
wave and, therefore, can be neglected. This assumption may not be valid if
the reach is too long. Analyzing the movement of a flood wave is different
from analyzing he t steady flows considered in Chapter 4. Depth and
velocity change quickly with time during a flood and, therefore, we must
solve this as an unsteady flow problem.
In simplest tenns, neglecting any lateral inflow, the continuity equation
for unsteady flow can be wrinen as

dV
/ - 0.. (5.1)
d,
-
=

where I is the known upstream (inflow) hydrograph, 0 is the downstream

(outflow) hydrograph that we want to predict, and Vis the volume of water
stored in the reach. Under steady-state conditions, the storage volume is
constant and 1= O. During a flood, inflow increases and stage increases.
As a result, the volume of water stored in the reach increases. The greater
the change in storage in a given interval of time (clVlclf), the greater the
difference between the inflow and outflow hydrographs.
Solving for the two unknowns V and 0 in Equation 5. 1 requires a
second equation. Equation 5. 1 is a statement of conservation of mass.
Therefore we expect the second equation to follow from Newton's second
law (conservation of momentum). All hydrological problems must obey
these physical laws. An example of this second typc of equation is the weir
equation that relates discharge to depth. Depth, in tum, is related to the
volume of water in a stream reach or reservoir. Thus we see that an
equation such as a weir equation can also provide a relationship between
volume Vand outflow 0, that is, 0 j{V). Substituting this expression for
==

o into Equation 5 . 1 produces an equation in which the rate of change of

volume (dVldl) depends on volume. Such an equation is called a
differential equation.
Typically, we cannot solve the differential equation for V directly.
Instead, a numerical method is used to solve the equation. Many
problems in hydrology must be solved numerically. Basically, this
involves transforming the differential equation into one or more algebraic
equations that can be solved more easily. We will use this approach to
calculate outflow hydrographs for a couple of interesting and infonnative

162
flood routing problems.
The first step in obtaining a numerical solution to the flood routing
problem is to rewrite Equation 5 . 1 in tenns of finite differences. To do
this, we choose a computational time interval, !J.I, separating two instants
in time, I" and f" + I ' Lei the subscripts /I and 11 + 1 refer to values of each

!-v. Iw + 1n;!
variable at these times. Equation 5.1 can then be approximated as:

v••
(5.2)
III 2
-

At each time step, the terms at time 1/ are known; the terms at time /I + 1
are the unknowns we are solving for. At the initial time step (1/ = I),
volume and outflow must be specified. We refer to the values of VI and 01
as initial conditions. In addition to the initial conditions, we have to
specify the inflow hydrograph during the interval of interest. That is, In
and lIP + I are known at each time step. The unknown terms being solved for
at each time step are V,, + I and 0,, + I ' Rearranging Equation 5.2 to put the
unknowns on the left side of the equation gives:

(5.3)

Given conditions at time I", the quantity on the left-hand side of Equation
5.3, (2 V" + I /!J./) + 0" + I ' can be calculated. We need a second equation, as
described below, to separate this into values of V" .. I and 0" .. I ' Once these
are detennined, all of the quantities on the right-hand side of Equation 5.3
are known for time t" I ' The procedure is then repeated to solve for the
..

values on the left side at the next time step.

Two flood routing problems are considered to illustrate the effect of
storage on outflow hydrographs and the use of a numerical method to
solve hydrological problems. The first problem is the routing of a flood
through a reservoir. This problem has important applications to flood
management and the maintenance of water supplies for public and
agricultural use. The second is an empirical approach to the routing of
flood waves through a river channel. River routing is important for flood
prediction.

5..1.1 Rcseryoir routing

Reservoirs are an effective tool in flood control because of their large

storage capacity. Hydrograph peaks downstream of a reservoir are smaller

163
in magnitude and delayed in time compared to those on the same stream
upstream of the reservoir. Understanding how a flood wave is modified as
it passes through a reservoir makes it possibJc (0 design and manage
reservoirs to protect downstream areas.
Reservoirs are formed when a river is dammed. The simplest dam is
one designed so that when reservoir depth exceeds a certain value, water
will flow over a spillway (Figure 5.7). The spillway is like a weir (see
Section 4.3). There is a fixed relationship between the depth of water
above a weir crest (h ei�) and discharge over it. This relationship provides
....

the second equation we need to complete the reservoir routing problem.

Discharge over a weir is given by

Inflow - known for

routing problem

Reservoir

Outflow over the

spillway - to be
Flood-control compu ted
dam

Figure 5.7 An example flood-control dam and reservoir.

Q = Cq �2g"�t�r 11',. ,

where lVe is the width of the weir crest and Cq is an empirical coefficient.
With Cq = 0.5, typical for a dam spillway, the weir equation gives us an
expression for reservoir outflow

(5.4)

with 0 expressed in mJ S-I and both h ej� and We in m. Water depth in the
....

reservoir h is given by the sum of hh"tir and the height or the weir

164
(spillway) crest, hspilll<'lIY" Therefore, Equation 5.4 allows us to relate water
depth in the reservoir to the outflow over the spillway.
Water depth in a reservoir is also related to the volume of water in the
reservoir V in a manner dictated by the geometry or topography of the
reservoir basin. For example, a 20 I I bathymetric survey was used to
establish the relationship between depth and volume in Totten Reservoir, a
small reservoir in southwest Colorado (Kohn 2012; Figure 5.8). This
reservoir has a spillway that is roughly 20 III long at a height of 1 0 . 1 III
above the reservoir bottom. Reservoir depth and volume above the
spillway arc provided in Table 5. 1 .
The application of these relationships to the reservoir routing problem
is best illustrated by example. The reservoir in the example has the
geometry indicated in Figure 5.8. The computational time step D.( is
dictated by the resolution of the input hydrograph. For this example we
will use the input hydrograph given in Table 5.2, which has a time step D.I
= 6hr or 2 . 1 6 x 1 04 seconds. We also must specify the initial values of 0 1
and VI before we begin the calculation. The spillway height h''Pi/hl'll), = 10 . 1
m (Figure 5.8) and we will assume h""ei� = 0.0 m initially, corresponding to
a reservoir volume VI = 3.7 x 1 06 m3 and initial outnow 0 1 = 0 m3s-1
(Table 5. 1 ). In other words, the water level of the reservoir is just al lhe top
of the spillway at the initial time I I '

Table 5. t. Depth and volume for Totten Reservoir in Montezuma County,

CO (Kohn 2012), and related quantities needed for reservoir routing
calculation

Oepth, Volume, 12V/lI.t) +

h_Im) o ImJ/sl
h im) VlmJ) o ImJ/s)

10.1 3.69E+06 0.00 0.00 '"

10.4 3.98E + 06 0.30 7. 4 '76

10.7 4.30E+06 0.61 21 '"

11.0 4.61E+06 0.91 38 466

11.3 4.93E+06 1.22 59 516

11.6 S.27E + 06 1.52 " 570

11.9 S.60E+06 1.83 107 627

12 2
. S.9SE+06 2.13 137 687

12.5 6.29E+06 2.44 167 750

12.8 6.64E+06 2.74 200 '"

NOle: Assuming III = 2 1 6005 (= 0.25 days), outflow 0 is given by Equation 5.4,

165
 14 --
--

12

10
E
_
------------ - - - - - - - \- - - - - - - - - - -
- -

\
Spillway height
li
• B
."
-
--

�
•
6
•
•
a:
4

�----�,----�2�--�3�--�4�--�5�--�6�--�7
Reservoir volume (m3) x 10e
Figure 5.8 Relationship between reservoir volume and depth for Totten
Reservoir in Montezuma County, CO. Such relationships are typically
obtained from topographic and bathymetric surveys of the reservoir area.
Data from Kohn (2012).

Table 5.2. Inflow hydrograph for reservoir routing example

Time (days) Inflow 1m] s.....) Time (days) Inflow 1m] s.....j

0.00 0.70 2.00 2.80

0.25 0.90 2.25 2.30
0.50 2.00 2.50 1.80
0.75 5.00 2.75 1.50
1.00 9.00 3.00 1.20
1.25 6.70 3.25 1.00
1.50 5.00 3.50 0.85
1.75 3.80 3.75 0.70

166
 The solution to Equation 5.3 can be found by constructing Table 5.3.
The quantities in each column are identified by the expressions at the lOpS
or thc columns. To begin, values for In and In + ,'' + 1 from the inflow
hydrograph (Table 5.2) arc entered into columns A and B, respectively, for
each time I". Based on the initial values VI and 01 given above, the first
entry in column C of Table 5.3 is [(2 * 3 .7 x 106 m3)/(2. 1 6 x 1 04 s)] - 0 m3
S-I = 342 6 m3 S- I . The entry in column D is computed using Equation 5.3:
.

2 V, 211:
=--
c
· + 02 = 'I + '2 + 1 01 = 344.2 m ) S- I .
!>J !>J
-

That is, 0 = B + C.
Now we need to figure out how to separate the teon in 01 (column 0,
row 1: 2 V2//::..1 + O2) into values of V2 and O2, From Table 5. I (and Figure
5.8), we know reservoir volume V for a range or values of water depth, h.
We can convert the values of II to corresponding values of 0 using
Equation 5.4 with Iiwe'r = II - Ii'1'''/lI"I1)" This gives us values of V and 0 for
each value of h. From these we can calculate 2V/61 + 0 (column D) for a
given /::..r. The resulting values of "wei,., V, 0, and 2 V//::..r + 0 are listed in
Table 5. 1 for !:J.r = 2. 1 6104 s and Ii.,pill"''')' = 10.1 m. A graph of2 V/!:J.r + 0
x

versus 0 (Figure 5.9) allows us to detenlline the value of 0 for any value
of2V/!:J.t + O. In practice, it is often more convenient-particularly when
doing the calculation in a spreadsheet-to fit a curve to the relationship
shown in Figure 5.9. Since we don't expect outflows thai are larger than
inflows, we focus on getting a good fit to small values of outflow, 0
(Figure 5.9):

Table 5.3. Reservoir routing computation

167
 Time t. lV
e 2V_1
'. 1. +'.�1 - 0. 0.�1
+0"'1
Step n f.�1
Idays) '" M

A • C D ,

1 0.00 0.70 1.60 342.' 344.2 0.06 0.25

, 0.25 0.90 2.90 344.1 347.0 0.60 0.50

• 0.50 2.00 7.00 345.8 352.8 1.79 0.75

4 0.75 5.00 14.00 349.1 363.1 4.15 1.00

5 1.00 9.00 15.10 354.9 370.6 .5.98 1.25

6 1.25 6.70 11.10 358.6 370.3 5.91 1.50

7 1.50 5.00 '.80 358.5 361.3 5.lS 1.15

• 1.75 3.80 6.60 357.0 363.6 4.15 2.00

9 2.00 2.80 5.10 355.1 3602 3.45 2.25

10 2.25 2.30 4.10 353.3 351.4 1.81 2.50

11 2.50 1.80 '.30 351.8 355.1 1.19 2.75

12 2.75 I.SO 2.10 350.5 353.1 1.89 3.00

" 3.00 1.20 2.20 349.4 351.6 1.55 3.25

14 3.25 1.00 1.85 348.5 350.4 1.19 3.50

IS 3.50 0.85 1.55 341.8 349.4 1.08 3.75

0 = -2.64 >< I 0-f>(2/f/61 + 0)1 + 0.00408(2V161 +0)2

- 1.68(21'/61 + 0)+ 202.6. (5.S)

Note that Equation 5.5 may not be a good fit for values of 0 > 50 mJ S-I.
Returning to Table 5.3, we find from Equation 5 .5 (or Figure 5.9) that
the value of O2 (E 1) is 0.06 mJ S- I for a Dl value of 344.2 m3 S-I . We
enter 12 in the last column to keep track of the proper time for the outnow
hydrograph.
At time step 11 = 2, we can calculate the tenn C2 in Table 5 .3 as the
term 01 - 2 )( the tcrm E l . That is,

2V., 2V., J 1
�-
.:c...c 02 = � + °2 - 2°2 = 344.1 m s- .
AI M

This is added to 12 + IJ (B2) to obtain the value for column entry 02 =

347.0 m3 S- I , the value of the left side of Equation 5.3 for /I + I 3. We =

repeat these steps for each 11 until the table is complete. The calculated
outflow hydrograph is tabulated in the last two columns of Table 5 .3. To
summarize, the series of steps for completing Table 5.3 (using i to
represem the row number and letters to represent thc variables in each

168
column) is:
50

45
::::
I �
-
::: �
Values
��l-----/
from
- - - Equation 5.5
Table 5. 1

40

35
'"
-

;:;- 30
E
-

� 25
0
'"
-

" 20
0

15

10

gOO
� -
�3�
20
--�
�-�
0��
����
�0--
�4OO��
�2�
4 -
0 �
� 0--
�4�
60
--��
48�
0 �
50a
2V/dl + 0 [col O] (m'/s)
Figure 5.9 Outflow versus (2 Vldt) + 0 for the reservoir routing example
bascd on low outflow valucs in Table 5. 1 . Equation 5.5 providcs a fit to
the observcd relationship useful for completing the calculations in the
reservoir routing example.

I . Bi + Ci = Di.
2. Di + Equation 5.5 or Figure 5.9 produces Ei.
3. Di - 2 x Ei = C(i + 1).
4. return to step 1 for next i.

The predicted outtlow hydrograph shows how the reservoir has worked 10
reducc thc efTect ofthc inflow flood (Figure 5. 1 0) . The rcservoir stores a
portion of the inflow during high inflow periods and releases the water
when the inflow subsides. Again, note that the effect of this storage is to
delay the peak discharge and to reduce its magnitude.
It is instmctive to consider how the outflow hydrograph for the
reservoir routing problem changes with reservoir characteristics. First, if
the reservoir werc larger, it would store morc water. For the same inflow,

169
the rise in water level in the reservoir would be smaller and, therefore, the
increase in outflow would be smaller. The outflow hydrograph would have
a lower pcak valuc and would be spread over a longer period of time.
Second, we might conceive of an outflow mechanism that released water
more slowly than the spillway (or ofa spillway with a lower weir
coefficient). This also would have the effect of decreasing the peak
discharge because the water would have to build up to a higher level to get
thc same outflow. If thc watcr level in the reservoir were initially below
the level of the spillway, then there would be no outflow over the spillway
until the inflow could raise the water level in the reservoir to the level of
the spillway. In the example shown in Figure 5 . 10, 9.6 x 105 m3 of water
enters the reservoir over thc eoursc of 3.5 days, an amount roughly
equivalent to the volume of a 1 .0 m-thiek slab of water in the reservoir just
below the height of the spillway. Therefore, if water depth in the reservoir
were initially > 1.0 m below the spillway, the water surface would never
reach the level ofthc spillway during this flood.

10

0.5 1 1.5 2 2.5 3 3.5 4

Time (days)
Figure 5.10 Hydrographs for the reservoir routing example.

This example of a reservoir with a spillway is reasonable for reservoirs

170
located on relatively small rivers or streams. However, large dams, such as
the Grand Coulee Dam on the Columbia River, Glen Canyon Dam on the
Colorado River, and Folsom Dam on the American River, have outlets that
are more complex than simple spillways. Water is released through
carefully controlled outlets and over spillways. This allows the volume of
water in the reservoirs behind the dams to be varied seasonally and the
discharge of water through the dams to be adjusted according to the
operational policy for the dam. By gaging inflow and considering the
trade-offs between volume of water in the reservoir and outflow rates,
reservoirs can be managed for maximum benefit. For example, during
periods of drought, extnt water can be stored. However, if water levels in a
reservoir are kept high in amicipation of drought, the reservoir capacity for
storing floodwaters is decreased.
Some of the difficulties in optimally managing a reservoir are
illustrated by an example from the American River. In 1986, before the
flood on the American River, the operating procedures for Folsom Dam
called for a flood-storage capacity of 4.93 x 1 08 m3 beginning on
November 1 7 and extending until February 8. Subsequcm to February 8,
the allowable storage was varied according to how much the accumulated
precipitation for that season had been. The operating policy also called for
a maximum controlled release from the dam of3.26 x 1 03 m3 S- I up until
the reservoir level reached full pool. Subsequently releases were dictated
by emergency spillway procedures designed to protect the structure from
failure.
When the flood began in February 1986, Folsom Reservoir's flood­
storage capacity was about three-fourths of that called for in the operating
policy; that is, the amount of water stored in the reservoir exceeded the
permissible storage (Figure 5. 1 1 ). Outflows reached 3.68 x 1 03 m3 S-I ,
some 13% above the maximum controlled release. Fortunately, disastrous
nooding was avoided because the storm abated and because the levee
system downstream of Folsom Reservoir stood up to the "excess" releases
from the reservoir. The effect of the storage reservoir on the flood is clear
-the peak inflow discharge, for example, was 5.44 x 1 03mJ S- I , compared
3 3 l
10 the outflow peak of 3.68 x 1 0 m S- (Figure 5. 1 1 ).

I7 I
 2.5>:10'

Actual storage
E 2.0>:10'
,
-

g
8, 1.5>:10'

� 1.0>:10"
----
� ---=-
-c -
Permissable storage
, , , , -�"""1,---f
12 14 16 18 20 22 24 26
6000
,
-

"
•

E 4000
.
J£ Inflow

o
-

•
�
'"
-

2000
\ '\Outllow
u
.
• "-
0
-

/
0 • • •

12 14 16 18 20 22 24 26
Oay of the month, February, 1986
Figure 5.11 Flows and storage in Folsom Reservoir for the storm of
February 1986.
Data from NRC ( I 995a).

5... .2 Flood routing in ri\:ers

A comparison of Figure 5 . 1 0 with Figure 5.6b suggests that a flood wave

is affected by passage through a segment of a river in much the same way
as it is affected by passage through a reservoir. A detailed analysis of the
changes to a flood wavc as it moves through a river section is somewhat
more involved than reservoir routing, but the principles used are the same
in each case. The conservation of mass equation (Equation 5 1 ), or its
.

finite difference equivalent (Equation 5 .3), serves as the starting point for
river routing. The difference is that rather than using a weir equation to
relate storagc and outflow, we need an cquation relating the outflow from
a river rcach to the volume of water stored in thc reach. A complcte
analysis of the relationship between storage volume and outflow in a river
reach is beyond the scope of this book. Inste<ld, we adopt a simpler
approach in which observed inllow and outflow hydrographs are used to
develop an empirical relationship between storage and outflow.

172
 The Muskingum method of flood routing in rivers assumes that the
volume of waler stored in a river reach can be related to the inflow and
outflow for thc rcach by thc equation
v= K,[xI + (1- x)O], (5.6)

where K, is a travel time constant and x is a constant that weights the

contributions of inflow and outflow on storage in the reach. Although
these two constants are not entirely independent, K, primarily determines
the rate at which the flood peak moves through a reach and x primarily
determines (he attenuation of the peak flow. The best values of K, and x for
a specific reach are obtained using measured inflow and outflow flood
hydrographs. We can get reasonable estimates of the parameters, howcver,
if we take x 0.2 and set K, equal to the length ofa reach, L, divided by
""

mean velocity through the reach, U. Note that a rapid flow through a long
reach could have the same K, as a slower flow through a shorter reach.
Substituting Equation 5 .6 into Equation 5 .3 gives

(5.7)

where

Co = (-K,x+ 0.5/l/)/(K, - K,x + 0.5/l/).

C, = ( K,x +0.56t)/( K,- K,x + 0.56/).

C2 = (K, - K,x - O.5111)/(K, - K,x+ 0.5111).

The values of the coefficients Co, CI , and C2 must be positive (for x = 0.2,
0.75M < K, < 2.0.11) and the sum Co + C1 + C2 = I . Like the reservoir
problem, once the initial outflow 01 and the inflow hydrograph are
specified, the calculations can be stepped through to obtain the outflow
hydrograph.
The flood routing proccdure presented above allows one to compute an
outflow hydrograph given a known inflow hydrograph. To illustrate the
method, the inflow hydrograph used in the reservoir routing problem
(Table 5 .2) is routed through a river reach using the Muskingum method.
The value of x is taken to be 0.2. Outflow hydrographs for K, = 6hr and
12hr are shown with the inflow hydrograph in Figure 5. 1 2. For a given
flow rate, larger values ofK, correspond to increasingly long reach lengths.
For example, if U = I m S- I , the reach length for K, = 6 hr is approximately

173
22 km; the reach length is 44 km for K, = 1 2 hr. Alternatively, for a single
value of reach length, say 40 km, the two values ofK, correspond to flow
rates of approximalcly 1 .9 and 0.9 m S-l, respectively. Because we are
disregarding lateral inflow to the reach, the choice of reach length for the
river routing calculation would depend on the distance between major
tributaries and the limitations on values of tJ.t and K, necessary to keep the
coefficients in Equation 5.7 positive.
The hydrographs pictured in Figure 5 . 1 2 show that the lag in the timing
of the outflow peak relative to peak inflow is approximately equal to the
travel time constant. As the travel time constant increases, the flood wave
aucnuation incrcases. For K, 12 hr, the peak outflow is about 70% of the
=

peak inflow. It is interesting to compare the attenuation of the flood wave

by friction and storagc in a rivcr channel (Figure 5. 12) to the attenuation of
the same flood wave by storagc in a reservoir (Figure 5 .1 0). A flood wave
must move through a long stretch of a river before peak discharge is
reduced as much as a moderate-sized reservoir can accomplish in a
relatively short distance.
10
e Inflow
9 e Outflow (K,= 6 hr)

B OuHlow It<, 1 2 hr)

B
=

7
0
-

0;-
E S
-

�
'" 5
"
�
"
u 4
0
C
.-

o
o 0.5 1 1.5 2 2.5 3 3.5 4
Time since start of flood (days)
Figure 5.12 Outflow hydrographs calculated using the Muskingum

174
method for the same inflow hydrograph used in the reservoir routing
example (Figure 5 . 10). As the travel time constant Kt incrcascs, the time
lag and attenuation increase.

Because the peak in the outflow hydrograph is delayed relative to the

inflow hydrograph, therc is a direct operational use for river routing
techniques such as the Muskingum method. For example, a gaging station
on a river can be set up to transmit data automatically to the office of a
person responsible for flood warning. These data can be augmented with
forecasts of river inflow for the subsequent \0 days (based on weather
forecasts) and then fed continuously to a routing program that gives the
corresponding forecast for the outflow. Warnings about likely flood crests
can be issued on the basis of such forecasts. During early parts of a flood
event, forecasts of crest height downstream will be uncertain because the
peak inflow value is not yet known. After the inflow flood has crested and
the river has begun to subside, however, the forecasts of outflow crests
will become more accurate because the inflow hydrograph up through the
time of peak discharge is now available to the routing program.

5.5 Flood Frequency Analysis

Flood routing can be used to determine the effectiveness of a given flood­

control structure. There will always be floods that exceed the design
specifications of a structure. The question is, what is the probability of
those floods occurring in any given year, and is that probability acceptably
low? There are a variety of issues that demand the use of probabilities in
analyzing floods. For example, suppose a person erects a building in a
river valley and wishes to insure it against damage from floods. The
insurance company is faced with the problem of setting the premiums so
that it will have a reasonable hope of securing a profit. If the future
sequence of storms over the life of the structure could be predicted with
certainty and the resulting precipitation routed through the watershed with
certainty, then the company could determine the exact time that the house
would be damaged in a flood and calculate the premiums based on that
prediction. or course, this cannot be done, so an alternate solution must be
found.
As with any insurance, the calculation of premiums involves the
probability of an event occurring. That is, if a determination was made that
the structure would be flooded once every 50 years (on the average), then
the premiums would be set 10 cover the cost of damage this frequently.

175
This is another way of saying that the probability of the house being
flooded is 1/50 in any year. This situation docs not guarantee that the
company will make money on each individual policy. The structure may
be flooded in the first year of the policy or it may be flooded three times in
the first 1 0 years. On the other hand, it may not be flooded at all in 75
years. The probability statement does not specify that floods will occur
exactly 50 years apart but only that one is expected to occur an average of
once in 50 years (scc also Section 2.2.4.)
Thc same notion of probability is needcd in the analysis of flood·
protection measures. At the beginning of this chapter, the problem of
flooding on the American River system at Sacramento was introduced.
Mention was made that Folsom Dam was originally thought to offer flood
protection from a I-in-SOO-year stonn. Analysis of the level of protection
in this case required: an estimate of the I-in-SOO-year inflow hydrograph
and a routing procedure for this inflow hydrograph through the reservoir.
The probability of a flood equaling or exceeding a given magnitude can
be approximated using discharge records from a gaging station. For
example, if there were 20 years of record for a certain station we might
intuitively expect that the largest recorded flood is an approximation to the
h20-year flood"-a flood that occurs, on average, once in 20 years. The
fonnal procedure is summarized in the following steps:

I . The highest discharges recorded in each year for the II years of record
are listed, similar to the case ofrainf:111 extremes discussed in Chapter
2.2.4. These peak discharges are called floods and the series of the
largest flood in each year is tenned thc allllual series.
2. These floods are ranked according to magnitude. The largest flood is
assigned rank I , the second largest rank 2, and so on.
3. An initial detennination of the flood statistics can be found by plotting
the logarithm of discharge for each flood in the annual series against the
fraction of floods greater than or equal to that flood; this fraction,
tem1ed the exceedance probability, is given by r/(II + I), where r is the
rank of the particular flood.
4. If the data conform to a lognonnal distribution (i.e., the logarithm of the
data are nonnally distributed), they will plot along a straight line if a
nonnal probability scale is used on the y-axis as in Figure 5 . 1 3 .
5. If a log-normal distribution does not adequately fit the data, another
distribution, such as a log Pearson type III extreme value distribution,
might provide a better fit to the data (see Haan, 2002.) By fitting a
suitable distribution to the data, the exceedance probability and rctum

176
 period for a flood of any magnitude can be estimated.
6. The best fit to the data, whether lognormal or othenvise, defines the
exceedal1ce probabilities for floods of any given discharge. The retllm
period, the average span of time between any flood and one equaling or
exceeding it, is calculated as Trcruril = I/(exceedance probability).

As an example, we apply this procedure to the Powell River at Big

Stone Gap, VA, for the period from 1945 through 1 994. (There is a data
gap between 1994 and 2002, so we confine our analysis to the 50-year
period from 1945 to 1994.) The annual flood peaks range from 48 10 680
m3 S- I (Table 5.4). A line fit to the natural logarithm of annual peak
discharge versus exceedanee probability, plotted using a normal
probability scale (Figure 5. 1 3), gives a reasonably good fil to the data. We
can estimate the 100-year flood by finding the value of discharge indicated
by the straight line fit to the data for an exeeedanee probability o f 0.0 I . For
the example shown in Figure 5. 1 3, we estimate a discharge of
approximately 520 m3 S- I . (The log discharge from the graph is about 6.25;
e raised to the 6.25 power is about 520.)
The extrapolated value for the I DO-year flood probably is nol realistic
given that the highest few discharge values fall below the line fit 10 the
data. Because o f this, we have to be cautious about estimating return
periods of floods much larger than those in the record. This is a typical
problem that arises when fitting a lognonnal distribution to extreme flood
records. A better fit and, therefore, more reliable estimates of return period
are often obtained using an extreme value distribution (e.g., Hahn, 2002).
In addition, the length of the record strongly affects the accuracy of return
period estimates. As records become longer, estimates of extreme events
become better. Unfortunately, we are often faced with the exact problem
of wanting 10 estimate a 100- or 20D-year event from a relatively short
record.

Table 5A. Annual flood peaks for Powell River, Big Stone Gap, VA,
1945-1994

177
 rln + 1,
Waler year, 0_" ma.imum
r, flood
Oct 1 to discharlle for In(O_,)
rank year'
Sep 30 the year, m] s-'

1945 138.75 4.93 " 0.55

1946 260.51 5.56 7 0,14

1947 150.08 5.01 19 0.37

1948 240.98 5.48 9 0.18

1949 135.92 4.91 29 0,57

1950 140.45 4.94 15 0.49

1951 135.92 4.91 30 0.59

1952 148.95 5.00 10 0,39

1953 151.50 5.02 18 0,35

1954 81.55 4.40 45 0.88

1955 151.78 5.02 17 0.33

1956 185.19 5.22 " 0.24

1957 254.85 5.54 8 0.16

1958 106.75 4.67 37 0.73

1959 140.17 4.94 26 0.51

1960 115.53 4.75 35 0.69

1961 125.16 4.83 33 0.65

1962 146.11 4.98 22 0.43

1963 475.72 6.16 2 0.04

1964 105.91 4.66 38 0.75

1965 148.10 5.00 21 0.41

1966 223.42 5.41 10 0.20

1967 302.99 5.71 • 0.08

1968 90.61 4.51 43 0.84

1969 47.57 3.86 50 0.98

1970 286.00 5.66 6 0.12

1971 124.31 4.82 34 0.67

1972 107.89 4.68 36 0.7\

1973 179.53 5.19 1. 0.27

1974 181.79 5.20 13 0.25

1975 \44.42 4.97 23 0.45

1976 105.91 4.66 39 0.76

1977 679.60 6.52 1 0.02

1978 84.95 4.44 44 0.86

1979 139.89 4.94 27 0.53

178
 1980 93.73 <1.54 42 0.82

1981 57.77 4.06 49 0.96

1982 168.49 5.13 16 0.31

1983 62.86 4.14 " 0.94

1984 319.98 5.77 3 0.06

1985 63.43 <1.15 47 0.92

1986 10<1.77 <1.65 '0 0.78

1987 99.96 4.60 " 0.80

1988 10.79 4.26 " 0.90

1989 172.17 5.15 1S 0.29

1 990 131.39 4.88 31 0.61

1991 1<13.57 <1.97 " 0.<17

1992 197.09 5.28 11 0.22

1993 129.12 4.86 32 0.63

1994 288.83 5. 67 S 0.10

0.01
0.02 o

0.05

0.10
.-
.-
L>
� 0.25

5.
cu 0.50
u
o
1l
� 0.75
u
•
W
0.90

0.95

0.98
0.99

4 4.5 5 5.5 6 6.5

InIQ_)
Figure 5.13 Flood frequency curve for Powell River, Big Stone Gap,
Virginia, 1 945-1994.

179
5.6 Concluding Remarks

Dams, such as Folsom Dam on the American River, offer important flood
protection to people and property located on floodplains. Evaluation of
protection measures requires the routing of hypothetical floods with
specified exceedance probabilities. The U.S. Anny Corps of Engineers
(USACE) is responsible for the analysis of flood-protection altematives
for the American River system and many others. The USACE derives the
"design hydrographs" for various exceedance probabilities by using a
flood-frequency analysis similar to that described in this chapter.
Technical evaluation of dams or alternative flood mitigation measures,
such as natural wetland restoration, demands that we know how to "route"
rainfall through thc appropriatc storage zonc, be it a rcscrvoir or a wctland.
A comprehensive analysis of these routing problems requires that all
hydrological processes that can affect streamflow be understood, including
the movement ofwater in the subsurface. Chapters 6, 7, and 8 explore flow
in the saturated and unsaturated ponions of the subsurface. Wc rcturn to
consider thc dynamics of watcr movemcnt in catchments in Chaptcr 10.
Dams servc many valuablc functions, including watcr storagc and flood
protection, but they are not without problems. A dam regulates the
downstream flow of water in such a way that the discharge is more
constant than it would be naturally. Damming of streams can alter water
tcmperaturc and strcam chcmistry and fragment thc strcam itsclf. While
regulation of flow in strcams and rivers is bencficial in many respects for
the human populations they affect, it may have far-reaching consequences
for other pans of the river and riparian ecosystems. For example,
fragmentation and reduced flow rates in dammed streams impede the
upstrcam spawning migrations of salmon. Thcy also affect the transpon of
sediment in the stream. One consequence of this is accumulation of fine
sediment in reservoirs and increased water clarity downstream of dams.
The latter effect is pronounced in the Colorado River downstream of the
Glen Canyon Dam. Where the water was once warm and muddy, it is now
clcar and cold. This has transfonncd this part ofthc Colorado Rivcr into an
excellent trout fishing stream, to the detriment of the native fish
populations. Stream regulation also affects the riparian zones. For
example, reduced overbank flooding allows time for less flood-tolerant
plants to become established and for greater stabilization of the stream
banks by vcgetation.
Analysis of the world's largest river systems (292 rivcrs) revealed that
over half ( 1 72) are strongly to moderately affected by fragmentation by
dams and flow regulation by reservoirs (Nilsson et al., 2005). The large

180
impacts of such fragmentation and regulation on river hydrology and
ecosystems have prompted efforts to understand the scope of the changes
and, in some cases, to mitigatc thc changes by introducing controlled
floods (periods of high release intended to mimic naturally occurring
floods) or even removing dams entirely. For example, the Elwha and
Glines Canyon Dams on the Elwha River in the Olympic Peninsula,
western Washington, were removed in 201 1-2013, representing the largest
dam removal proj ect in U.S. history. A primary goal for removing the
dams was habitat restoration, particularly spawning grounds for salmon
and trout that used to be abundant in the Elwha River, but had become
reduced in numbers or even locally extinct. Other goats included
increasing the supply of sediment to the coastal area ncar the mouth of the
Elwha River-sediment that had been trapped behind the dams-and
renewing cultural traditions of the Lower Elwha Klallam Tribe, whose
native home is along the Elwha River.

5.7 Key Points

• Hydrographs record the time history of flow in streams. Hydrographs

typically are measured by continuously recording stream depth (stage) at
a gaging station and combining it with a rating curve relating depth and
discharge to obtain a continuous record of discharge. {Section 5.2}

• Peaks in discharge are tenned floods, regardless of whether the water

overtops the banks. Flood hydrographs are characterized by a steeply
rising section followed by a less steeply falling segment as the flood
passes. Background low flow conditions are tenned baseflow. {Section
5.2)

• As a flood wave propagates downstream, its peak is attenuated owing to

friction and storage of water in the channel. {Section 5.3 }

• Storage of water in a reservoir increases as inflow increases. The

conservation of mass equation, dVldf = 1 - 0, tells us that the greater the
rate of change of storage, the greater the difference between inflow at the
upstream end of a reservoir and outflow at the downstream end. {Section
5.4)

• Outflow from a rescrvoir or a river reach can be calculated from an

inflow hydrograph using the conservation of mass equation and a second
equation relating outflow to storage volume. The outflow hydrograph can

I8 1
 be found by numerically solving the finite difference fonn of the
conservation of mass equation. {Section 5.4 }

• Reservoirs are effective for flood control bccausc they can store a large
volume of water, thereby producing a significant decrease in the
magnitude of the outflow from the reservoir relative to the inflow at peak
flood conditions. { Section 5.4. I }

• The Muskingum method of flood routing can be used to estimate the

transformation of a flood wave as it moves through a river channel. The
resulting outflow hydrograph depends on the inflow hydrograph and two
constants: I) a weighting coefficient that detennines attenuation, x ::::: 0.2;
and 2) a travel time constant that detennines the time lag between inflow
and outflow peak discharge, K, (reach length)l(mean velocity).
-

{ Section 5.4.2 }

• The exceedance probability associated with a given peak annual flood

discharge in an n-year-Iong record can be estimated graphically by
ploning the logarithm of discharge against r/(n + I) on a nonnal
probability axis (see Figure 5 . 1 3), where ,. is the rank of each year's peak
discharge from I (largest) to 1/ (smallest). If the logarithms of the floods
are normally distributed, then they will fal l on a straight line. This line
defines the exceedance probabilities for a flood of any magnitude. The
return period is the inverse of the exceedanee probability. { Section 5.5 }

• Estimates of return periods for time spans longer than the existing record
or probabilities of floods exceeding the largest recorded value can be
obtained by fitting a suitable probability distribution to the data, which
can then be used to estimate the return period of discharges larger than
those found in the record. The farther the data are extrapolated (e.g., for
re!"Urn periods much longer than the record length), the larger the
potential error will be in the resulting values ofretum period or flood
magnitude. Errors also result if the data are not well fit by the chosen
probability distribution. While a lognormal distribution is relatively easy
to fit 10 an annual series of peak discharge values, an extreme value
distribution is often needed to provide accurate estimates of floods with
return periods much longer than the record. { Section 5.5 }

5.8 Example Problems

Problem I. If the flood used in the reservoir example delivered the same

182
volume of water in a shorter amount of lime (shorter duration with higher
peak discharge), as given by the inflow hydrograph in the table below,
how would the outflow hydrograph changc? Complcte the tablc below,
assuming the initial conditions and other reservoir parameters remain the
same.

Time t. 2V� _ 0 2V
/hJ+ 0
Step N '. 1. +1..., 1
'" 6t
...
(days)
.

A • C 0 ,

1 0.00 0.7 4.7 342.6 0.25

2 0.25 4.0 22.0 0.50

3 0.50 18.0 28.0 0.75

4 0.75 10.0 15.6 1.00

, 100 '.6 '.6 1.25

6 1.25 3.0 4.5 1.50

7 1.50 1.' 2.2 1.75

• 1.75 0.7 1.4 2.0

Problem 2. The Muskingum routing coefficients for a slream reach are

delennined to be: Co = 0.26, C1 = 0.55, C2 = 0.19. For the inflow
hydrograph given in the table below, complete the calculation of the
predicted outflow hydrograph.

Time (hr) Inflow (mJ s- ') Outflow (m3 s-')

0000 10 10

0600 SO

1200 130

1800 110

2400 70
Problem 3. In Figure 5 . 1 4, a 40-year annual series (1950- 1989) of floods
on the Eel River, California, is plotted against the fraction [r/(1/ + 1)J of
floods with discharges greater than or equal to each value.

A. Fit a line through the data and detemline the return period of an 8000
1
m3s- flood.

B. Estimate the magnitude of the 1 00-year flood.

183
 0.Q1
0.02 o
0.05 o
�
.­
-
0.10
.0
.-

.. 0.20
.0
o 0.30
�
Q. 0.40
., 0.50
u
c 0.60
..
" 0.70
� 0.80
u
><
W 0.90
0.95 o
0.98 o
0.99 ..'I . .. .. • ••• • •• •

1 0' 1 02 1 03 1 04 1 0'
Discharge, Q (m3 S-' )
Figure 5.14 Probability plot for Eel River peak annual discharge, 1 950--
1989.

5.9 Suggested Readings

Dunne, T., and L.B. Leopold. 1 978. Water ill environmental planning. San
Francisco: W. H. Freeman. Chapter 1 0, pp. 279-39J.
Haan, C.T. 2002. Statistical methods ill hydrology. New York: Wiley.
Mays, L.W. 20 I I . WaleI' resources engineering, 2nd ed. Danvers, MA:
John Wiley & Sons. Chapter 9, pp. 3 3 1-358.
Nilsson, c., C.A. Reidy, M. Dynesius, and C. Revenga. 2005.
Fragmentation and now regulation of the world's large river systems.
Science 308:405-408.

184
6 Groundwater Hydraulics

6. 1 Introduction

6.2 A Conceptual Model

6.3 Darcy's Law

6.3.1 Hydraulic conductivity, intrinsic penneability, and

porosity

6.3.2 Restrictions on Darcy's law

6.4 Water in Natural Fonnations
6.4.1 Construction of wells and piezometers
6.4.2 Water-level measurements
6.4.3 Geophysical techniques

6.5 Steady Groundwater Flow

6.5.1 Quantifying groundwater flow using flow nets

6.5.2 Heterogeneity and anisotropy

6.6 Concluding Remarks

6.7 Key Points

6.8 Example Problems
6.9 Suggested Reading

6.1 Introduction

Of the 10tal amount of freshwater on this planet, about 30% is contained

beneath the surface of the Earth. This figure is even morc impressive when
one considers that all but a few tenths of a percent of the remaining
freshwater is held in ice caps and glaciers (Table 1 . 1 ). Most subsurface
water is found in rocks and soils that arc saturated with water; that is,
materials in which water occupies all pores, openings, and fractures. We
refer to water in the saturated region of the subsurface as groundwater.
Groundwater is an important resource across the globe. Irrigated

185
agriculture in many areas is based on withdrawal of groundwater. For
example, Shah et al. (2000) estimate that 60% of irrigated grain production
in India depends on the usc ofgroundwatcr. Halfofthc world's mcga
cities (population of 1 0 million or more) are dependent on groundwater
(Giordano 2009). Even in areas where surface water is available,
groundwater often is a preferable source because of water temperature,
quality, or accessibility. For example, in Dhaka, a city of 1 0 million in
Bangladesh, groundwater is the predominant source of freshwater even
though the city is located in the delta region of the Ganges and
Brahmaputra rivers and is bordered by major rivers.
The study of groundwater is motivated partly by practical
considerations of water supply. In addition, an understanding of the
hydrological cycle for a catchment requires a quantitativc description of
how the groundwater reservoir functions. For example, recall that
streamflow (baseflow) is maintained in perennial streams between
precipitation events. We can hypothesize that this is in large part the result
of the discharge of groundwater into stream channels. In this chapter and
in Chapter 7 we discuss subsurface pathways through which precipitation
eventually may reach a surface water body. We retum to these ideas again
in Chapter 1 0, when we examine mechanisms of runoff generation.
A number of environmental issues involve groundwater, especially the
remediation of sites that have been contaminated by poorly controlled
dumping practices and the identification of and planning for sites to safely
dispose of hazardous wastes. An example is planning for the disposal of
radioactive wastes.
The Waste Isolation PilO( Plant (WIPP) in New Mexico, United States,
is an underground repository built for certain radioactive wastes that were
generated during the construction of nuclear weapons. These wastes,
including work gloves and laboratory glassware, have been stored in 55-
gallon drums at facilities of the Department of Energy. The wastes are not
high-level, but they contain isotopes that remain radioactive for very long
periods of time (tens of thousands of years). These wastes must be isolated
for millennia to ensure that they do not pose risks to human health. In the
U.S. these wastes are being buricd in a repository in a 600-m thick salt
formation (thc Salado Formation) some 700 m below the ground surface.
Rock salt is thought to be a good host rock for radioactive wastes because
it flows and over time will seal the wastes off from the environment. The
only path that would lead to the release of radioactivity to parts of the
cnvironmcnt wherc it might adverscly affect people or ceosystcms is
through dissolution of the waste by water and transport of the dissolved
constituents by groundwater. The Culebra dolomite, a regional aquifer,

186
overlies the Salado Formation. Assessment of the suitability of the WIPP
site for disposal of radioactive waste requires knowledge of how
groundwater flows in the aquifer above the repository. Several questions
might be raised. What causes or drives the movement of groundwater?
What physical characteristics of subsurface fluids and porous media
detemline the rate of fluid movement?
Because we are again interested in the flow of water in relation to
imposed forces, the equations of fluid mechanics provide the basis for the
quantitative description of groundwater flow. At thc scale of the porcs in
rocks and soils, however, the paths along which groundwatcr flows are
complex, with many twists, turns, contractions, and expansions. So we
recognize immediately that simplifications will have to be made to enable
useful equations to be derivcd.

6.2 A Conceptual Model

Let us try to picture the flow of water in a porous medium, for example,
likc sand. Thc path a "parccl" of watcr might follow in moving through a
material containing pores or void spaces is convoluted (Figure 6. I ). Not
only does the water follow tortuous paths, but the geometry of the
channels of now is extremely complex and cannot be specified completely
(i.e., the position, size, and shape of all of the sand grains cannot be
known). Finally, we recognize thai the opcnings in which water flows arc
very small. Therefore, we might expect that frictionless flow is totally
meaningless in this situation and that head losses will play the
predominant role.

187
Figure 6.1 Schematic of a thin section of a porous medium and the
tortuous flow path of two water "parcels." (Note that a real medium is
three-dimensional with flow through the open spaces within the three­
dimensional matrix.)

This picture of groundwater flow is exceedingly complex. To try to

understand the fluid mechanics of the flow, we must resort to a conceptual
model. We first consider the tortuous flow path (Figure 6. 1 ) to be
"straightened" by somehow stretching the path. The opening through
which flow occurs then might be depicted as a pipe with a continually
varying eross section (Figure 6.2). Of course, there are actually many flow
paths and the entire system would consist of many different variable-radius
tubes. Next we replace the variable-radius tubes with an equivalent set of
constant-radius tubes. Thus, our conceptual model of flow through a
porous medium is flow through a bundle of very small (capillary) tubes of
different diameters. To be sure, the model is almost unacceptably
oversimplified. Nevertheless, we can derive certain insights into flow in
rocks and soils by examining this conceptual model.
Consider flow through a capillary tube. Because the rube is very narrow
and the velocities are relatively small, it is reasonable to assume that the

188
flow in the capillary tube is laminar. Moreover, because the velocities are
very small, we can neglect the effect of changes in velocity and velocity
head through the capillary tube, and assume that the diameter is constant.
In Section 3.7. 1 , we found an expression for the average velocity,

Figure 6.2 Straightened flow "tubes" rcprescnting flow in a porous

_!!....(..E.... + _]
medium.

u D2pg
(3.38)
-

-
dl pg - 32jJ

of laminar flow through a pipe or tube of circular cross section in the

direction I; D is the diameter of the tube and 1 is the flow direction. The
quantity between parenthescs is the sum of pressure head and gravitational
or elevation head and is also known as the hydraulic head [L]

(6.1)

The hydraulic head i s a quantity, measurable at every point in a

groundwater flow system. Because in these slow flows the velocity head is
negligible, h represents the fluid mechanical energy per unit weight.
The negative sign in Equation 3.38 indicates that the flow is down the
hydraulic head gradient from high to low hydraulic head. The discharge
through the tube is given by the product of average velocity and the cross­
sectional area of the tube:

(6.2)

189
Equation 6.2 is a form of Poiseuille's law for the flow of a viscous fluid
through a capillary tube. Discharge is directly proportional to the hydraulic
head gradient (or hydraulic gradient), inversely proportional to the fluid
viscosity, and directly proportional to the fourth power of the radius of the
tube.
The equations derived from this conceptual model have several
important implications. First, Equation 6.2 indicates that for a given fluid
and given hydraulic gradient (dhldl), the discharge varies as the fourth
power of the radius of the tube. For example, the discharge from a tube of
radius 1 0 mm will be 104 times greater than that from a tube of radius 1
mm, if all other conditions are the same. Because the size of the capillary
tubes in our conceptual model is related (on an intuitive basis) to the
texture or grain size of the porous medium, flow rates will depend on the
texture. In addition, Equation 3.3 8 specifies that the average velocity, V, is
proportional to the hydraulic gradient, dhldl,

U
=
_ D2 pg!!.!!. (6.3)
32 Jl dl

Equation 6.3 is the fonn ofPoiseuille's law that provides the most useful
analogy for flow through a porous medium. As we will see, an equation
almost identical to Equation 6.3 is the basis for studies of groundwater
flow. Because Equation 6.3 was derived in Chapter 3 using well-known
principles and stated assumptions, we can usc the analogy between the
conceptual model and the actual porous medium to great advantage in the
sense that we can apply, in qualitative lenns at least, knowledge about the
factors that control laminar flow in tubes directly to the flow of
groundwater.

6.3 Darcy's Law

In 1 856, a French hydraulic engineer named Henry Darcy published an

equation for flow through a porous medium that today bears his name. In
designing a watcr treatment system for the city of Dijon, Darcy found that
no formulas existed for detennining the capacity of a sand filtration
system. Consequently, Darcy performed a series of experiments on water
flow through columns of sand.
Darcy packed sand into iron pipes and systematically measured
parameters that he expected to influence the flow. Consider flow through a
cylindrical volume of sand, with cross-sectional area A and length L
== ==

(Figure 6.3). The sum of elevation head (z) and pressure head (plpg) is

190
represented as II and varies from hi to h2 along the column. The hydraulic
head at each end of the column is measured with an open tube, a simple
manometer, as shown in the diagram. By varying L and the hydraulic head
difference across the column (hi - h2 = I1h), Darcy found that the total
discharge Q varies in direct proportion to A and to I1h and inversely with
L . That is,

(6A)

where K is a constant of proportionality called the hydraulic conductivity

[L ,..-1 ]. Equation 6.4 can be rewritten as:
Q == _K "l - hl __ K "l - /� . (6.5)
A I. /)-/1

This can be written more generally as:

== _ K dh
q dl .
(6.6)

where q =QIA is the specific discharge. Equation 6.6 is the ronn of

Darcy's law that we will use in our studies. Although q has dimensions of
velocity [L ,..-1], keep in mind that we obtained this tenn by dividing the
discharge by the lola! area and that water flows only through a fraction of
the area, the spaces between the solid grains of the medium.
In Equation 6.6, dhldl is the hydraulic gradient and the negative sign
indicates that positive specific discharge (indicating direction of flow)
corresponds with a negative hydraulic gradient. Thus, Darcy's law states
that specific discharge in a porous medium is in the direction of decreasing
head and directly proportional to the hydraulic gradient (Figure 6.4).

191
 •
t,},
I

Cross-sectional
area, A

z,

Datum "
- = 0
Figure 6.3 Schematic diagram of an apparatus for illustrating Darcy's law .

• 92
 200 .-------�
-
-

.'"
-

E
150
E
E
-

"
� 100
..
.t:
U S lope = K
'"
.-

"
U
;;:
. 50
u
-

1i
'"

5 10 15
Hydraulic gradient, -dhldi
Figure 6.4 Darcy's ( 1 856) original data showing a linear relationship
between specific discharge and hydraulic gradient for two different sands.

6.3.1 H)'draulic conductiyity, int.-insic pcrmeabilit)" and

porosity

We can envision the hydraulic conductivity as the slope ofa line relating
specific discharge, q, with hydraulic gradient, dhldJ (Equation 6.6) .
Imagine a sel of experiments using a given sample of material and a fluid
of constant density and viscosity. By varying the hydraulic gradient and
measuring the discharge, q can be plotted against dhldl. According to
Darcy's law a straight line should be the result, the slope of which will be
K (Figure 6.4). Using a fluid with a constant density and viscosity, the
slope of the relationship between q and dh/dl will depend only on the
material and generally will increase with the coarseness of the material.
Now imagine repeating the experiments with another fluid having different
properties (e.g., a greater viscosity). We would expect the more viscous
fluid to move more slowly if everything else remains constant.
The example above suggests that hydraulic conductivity depends on the
nature of both the fluid and the porous material. The way in which K
depends on these properties can be inferred by reference to the conceptual

193
model discussed earlier. We already have remarked on the similarity
between Equations 6.3 and 6.6, which are reproduced below (for one­
dircctional flow in the I dircction) to facilitate dircct comparison.

V = _ D! pf!.!!!!. (Poiseuille's law) (6.J)

32 11 dl

dll .
q : _K ( Oarcy's law) (6.6)
dl

This analogy suggests that pglJI should describe the variation of K with
fluid density and viscosity, and that D2/32 should describe the variation of
K with pore diameter. Of course, the "diameter" of the pores is not
measurable (or even well defined), so the analogy is not perfect. However,
qualitatively we might expect the grain size of the material to give an
indication of the size of the openings (i.e., we would expect larger pores in
a boulder field than in a silt deposit) and we can define and measure the
average grain size of a granular material such as sand. Experimental
evidence supports this analogy, at least for simple granular porous media.
Based on experimenlal results, the following empirical relationship for
hydraulic conductivity can be written:

(6.7)

where N = a factor to account for shape of the passages [dimensionless]; d

= the mean grain diametcr IL]; pg = the unit weight ofthc fluid [M L-2 1

2];/1 = the viscosity of the fluid 1M L- 1 t1]. From this we see that K is
composed of two factors, one represcnling fluid properties and thc other
representing properties of the medium. Darcy's law can be written in a
manner that clearly separates these two influences:

q : -k(�J( �:;) (6.8)

where k is referred to as the intrinsic pcrmcabiHty (L2] of the porous

medium. The factor Nd2 in Equation 6.7 is equivalent to k. Onee the
intrinsic permeability ofa ccnain rock formation is known, Equation 6.8
can bc used to describc the flow of any fluid (oil, gas, or water) through
that formation.
In addressing issues such as the movement of contaminants in the
subsurface, it should be apparent that the hydraulic conductivity or
intrinsic permeability of natural materials plays a major role, with higher

(94
values of K or k resulting in faster transport. Measuring or estimating these
properties is a fundamental step in applying Darcy's law to a natural
setting. There arc a variety of techniques and methods for either directly or
indirectly detennining the penneability of a sample of porous material. A
small sample can be placed in a device called a permeametcr, not unlike
Darcy's original column. The flow rate through the sample can be
measured for a known hydraulic gradient and the permeability can be
calculated directly using Darcy's law. [n some situations, for example in
looking at flow deep underground, samples may be difficult to obtain.
There are a variety of methods, known as aquifer tests or "pump" tests, for
detemlining penneability in such cases. By withdrawing or adding water
to a well, and measuring the water level in that well or other wells nearby
as a function of time, we can ealcu[ate the penneability. Finally, there are
indirect methods that are based on measuring some other parameter, such
as grain size, that is related to the permeability.
Literally thousands ofpenneability measurements have been made in
different materials. The results show that the range of permeability of
natural materials is quite large (Figure 6.5). Recalling the discussion of the
WIPP site in the introduction, one reported range of hydraulic
conductivities for salt deposits, such as the Salado Formation, is 10-12 to
1 0-10 m S-1 (Domenico and Schwartz, 1 990). Relative to the other types of
materials shown in Figure 6.5, this range is near the low end and is similar
to shales and unfractured crystalline rocks. The resulting slow rate of
groundwatcr flow is one rcason that salt fonnations have bcen considercd
for waste repositories. For example, a typical hydraulic gradient of 1/100
in a salt fonnation with a hydraulic conductivity of 10-10 m S-I will
produce a specific discharge of 10-12 III S-I, or less than I mm per 30
years !
Darcy's law indicates that for a given value of the hydraulic gradient
(dhldf), the specific discharge will be greater for a penneable material,
such as a sand or gravel, than for a granite. The difference can be several
orders of magnitude (Figure 6.5) . As mentioned earlier, the sped fic
discharge, despite having dimensions of velocity [L JI], is not a velocity.
We could not use specific discharge to detennine how long it will take a
parcel of water to move from one point to another. The cross-sectional
area available to the water is smaller than the actual cross-sectional area,
such that the solid portion of the porous medium acts as a constriction.
This constriction means that a tagged parcel of water, or, better, many
tagged parcels that are averaged together, will appear to move through a
porous medium at a speed that is faster than the specific discharge. The

195
effect is similar to the constriction in a pipe discussed in Chapter 3; the
constricted Ilow has a greater mean velocity for the same value of
dischargc.
Intrinsic permeability, Ii.
, 0-" 10-19 H)-17 H)-15 10-13
, , I t !

Gh&1 till

'"'"

I i i i i i i
10-1' 10-'2 10-'0 10-11 10-11 1 0.... 10-2
Hydraulic conductivity, K (m s-t)
Figure 6.5 Ranges of intrinsic permeability and hydraulic conductivity for
a variety of rocks (gray bars) and sediments (blue bars).
Data from Freeze and Cherry (1979).

We can detcrminc thc mcan pore water velocity ifwc havc some idea of
the amount or "constriction." For porous materials, this property is the
porosity [dimensionless], which is simply the fraction ofa porous material
that is void space:

v,. .
¢= (6.9)
V,

where V" is the volume of void space (L3] and VI is the total volume [L3].
lfwe have measured the porosity, which by definition must be between 0
and 1 , we can detennine the mean pore water velocity, or average linear
velocity, v [L I I]:
- q
I' =-. (6.10)
�

6.3.2 Restrictions on Darq's Jaw

196
Darcy's law is widely used for almost all situations involving motion of a
fluid through soil or rock in the natural environment. Although the
conceptual modcl we used to introduce our discussion of Darcy's law can
apply (to a limited extent) only to granular materials such as sand,
Equation 6.6 is applied in materials ranging from clay to limestone to
fractured crystalline and metamorphic rocks. The openings through which
fluid flows in most of these materials cannot be envisioned as capillary
tubes. Nevertheless, Darcy's law usually can be applied with success.
There arc some general limitations on the usc of Equation 6.6, and
because of the analogy we developed we can infer how these restrictions
arise by examining the assumptions implicit in the derivation of Equation
6.3. First, Darcy's law has been found to be invalid for high values of
Reynolds number. Experiments using very high flow rates in very
penneable materials have found that when values of the Reynolds number
exceed the range 1 - 10, Darcy's law does not describe the relationship
between flow rate and hydraulic gradient. For flow through porous
materials, the characteristic length in the Reynolds number is the mean
grain diameter, d:

(6. 1 1 )

where q, the specific discharge, is used in place of the mean velocity, U.

This restriction can be explained by virtue of the fact that Poiseuille's law
neglects inertial forces (accelerations). The simplest evidence of this
statement is that a velocity head term does not appear in Poiseuille's law
(or in the definition of hydraulic head) and that this tenn derives from
consideration of acceleration.
Darcy's law also may fail to hold at very low values of hydraulic
gradient in some very low-permeability materials, such as clays. Equation
6.6 implies that any hydraulic gradient, no matter how small, will cause
some motion of water. In certain clay materials it is observed that below
some threshold value, a small hydraulic gradient will not cause motion;
that is, the force due to pressure and gravity must exceed some critical
value before motion ofwater ensues. Poiseuil1e's law, which is our
analogy for Darcy's law, is an expression of Newton's second law, and is
based on the idea of balanced forces. However, the only forces considered
are pressure and shear. Deviation from the law can be the result of the
presence of other forces that we failed to consider (e.g., the inertial tenns
discussed in the previous paragraph). In the case of clays, there may be
important electrostatic forces resulting from charge imbalances within the

197
mineral structure and at low hydraulic gradients these may contribute to a
force balance.

6.4 Water in Natural Formations

The subsurface is a complicated assortment of different materials, some of

high penneability and some of low permeability. Hydrologists studying
groundwater flow arc interested in how the distribution of different
materials in the subsurface influences patterns and rates of groundwater
movement, and how the water interacts chemically with natural materials.
The first lopic will be addressed more fully in Chapter 7, but for now we
introduce some terms that are applied to different soil or rock units, based
on whether they are relatively penneable or not, and where they reside in
the subsurface. An aquifer is a saturated geological formation that
contains and transmits "significant" quantities of water under normal field
conditions. "Significant" is a vague term but the implication is that
aquifers are formations that can be used for water supply. Obviously,
whether the supply is significant or not depends upon whether one is
referring to a supply for a single rural dwelling or a large municipality.
Many aquifers are unconsolidated materials, mainly gravel and sand.
Examples of this type of aquifer include those in coastal plain settings and
intemlOntane valleys. Limestones, partially cemented sandstones and
conglomerates, and permeable volcanic and igneous rocks are also
important as aquifers. The limestone aquifers in the Paris basin in Western
Europe and those underlying most of the Florida peninsula in the United
States are critical resources as are the Deccan basalt groundwater system
in India and the basalts of the Hawaiian Islands. Similarly, unconsolidated
materials fonn important aquifers in the North China Plains in China and
in the High Plains in the central United States.
Of course, not all formations are aquifers. An aquiclude is a fonnation
that may contain water but does not transmit significant quantities. Clays
and shales are examples of aquicludes. An aquifuge is a formation that
neither contains nor transmits significant quantities of water. Unfractured
crystalline rocks would fall into this category. The more general term
aquitard is often used to denote formations that are of relatively low
permeability and that may include both aquicludes and aquifuges.
Aquifers are classified according to hydraulic conditions and type of
material. One type of aquifer is an unconfined or water-table aquifer. If
an excavation is made in soil, then the near-surface material is usually not
saturated (the unsaturated or vadose zone, Chapter 8). Deeper in the soil

198
profile, saturated conditions prevail (saturated zone); groundwater by
definition refers to water in the saturated zone of the subsurface. The
water table or phreatic surface is dcfined as a surfacc ofzcro gage
pressure within Ihe subsurface, and separates the saturated and unsaturated
zones. Water will flow inlo an excavation or well up to this level; the
water table is equivalent to a free surface. An aquifer with the water table
as thc bounding surface of the lOp of the aquifer is an unconfined aquifer.
The second type of aquifer is called confined or artesian. This type of
aquifer is found when permeable material (the aquifer) is overlain by
relatively impermeable material (an aquiclude; Figure 6.6). The water in a
confined aquifer is under pressure and, in a well penetrating the aquifer,
will rise above the top of the aquifer (see Figure 6.6). The height to which
water rises i n a well defines the piezometric or potentiometric surface.
Note that a well penetrating a confined aquifer can be thought of as a
piezometer, or single tube manometer. The water level in a piezometer is
a measure of waler pressure in the aquifer. Thus, the elevation of the
potentiometric surface above an arbitrary (horizontal) datum is the sum of
pressure head and elevation head, or the hydraulic head, h, of Darcy's law.

Recharge area for

confined aquifer

Potentiometric surface
-
-
____
_
__
l in confined aquifer
'.
- -
'.
' .

or water-table

Confined aquifer

Figure 6.6 Hydrogeological units. Three piezometers are depicted, which

are open in either the confined or unconfined aquifer, as indicated by the
short horizontal lincs. Notc that in unconfincd aquifers, the watcr level in

199
the piezometer (jar right) indicates the height of the water table; in
confined aquifers, the water level in the piezometers (left and cell1er) rises
abovc thc top ofthc aquifcr and indicatcs thc position ofthc potcntiomctric
surface.

6.4. t Construction of wells and piezometers

We dcrive infonnation about the subsurface in large part by constructing

wells and piezometers. A well is an opening, generally a cylindrical
opening, from the ground surface down to a geological formation
containing water. The formation is typically an aquifer, at Icast for wells
constructed to supply watcr or to monitor a water source. In
unconsolidated materials such as sand and gravel aquifers, a well casing is
installed as part of the construction. A casing is made of solid material
intended to support the opening and keep the well open. The well is
screened over some interval to allow water to enter the otherwise
impcrmcablc casing. A well screen can bc as simple as slots cut into a
polyvinylchloride (PVC) pipe that is used as a casing. Wells typically are
screened over a considerable depth of the aquifer to allow easy entry of
water. A piezometer, by contrast, typically is screened over a narrow
interval to reflect groundwater head at one location in an aquifer. If there is
little vertical flow of groundwater, the head is sensibly constant over the
depth of the aquifer and water levels recorded by wells screened over large
vertical intervals will provide the same measurements as piezometers.
There are many methods for constructing wells including hand digging,
augering, hammering, and jelling (using water under pressure to "erode"
material and excavate a hole). The maximum depth that can be reachcd in
the excavation ofa well strongly depends on which of these techniques is
used (only a few meters for hand dug wells, several tens of meters for
jetted wells). One common method for constructing wells and piezometers
in unconsolidated material involves the use of a hollow-stem auger. A
hoHow-stem auger isjust what it sounds like (Figure 6.7a): auger blades
surround an open core. There is a removable plug at the bottom and once
the desired depth is reached, the plug is removed and a casing is inserted
into the hollow stem. The auger is withdrawn, leaving the casing in place.
The screened area of the well outside of the casing is backfilled with
coarse material to allow rcady flow of water to the well through the screen
but 10 prevent fine material from clogging the screen. The upper part of the
hole outside the casing is backfilled with a relatively impermeable material
such as clay to prevent downward leakage of surface water into the well
(or upward flow in the case of a confined aquifer). Near the surface a

200
concrete cap is generally installed to complctc thc scal ofthc casing and
prevent surface water and contaminants from entering the well (Figure
6.7b).

6.4.2 Water-Ienl measurements

Kcy mcasuremcnts that wc derivc from wells and piezomcters arc thc
elevations of the potentiometric surface or the water table. It is these
measurements that allow us to establish head gradients that can be used
with estimates of hydraulic conductivity to calculate groundwater flow
rates. Once the elevation of the top of the well casing is established by
survcying or the use of satellite positioning, we can dctermine thc
elevation of the water surface in a well by measuring the distance from the
top of the well casing. One instrument that we routinely use to make these
measurements is a pressure transducer. This is a device, as the name
suggests, that measures water pressure. The transducer is lowered a known
distance from the top of the well casing and below the watcr level in the
well and the pressure is measured. The hydrostatic Equation 3.5 can be
used to compute the pressure at the depth of the transducer. Pressure
transducers can be left in place for extended periods and connected to a
dalR logger to record well hydrographs (Figure 6.8).

Rod inside ----!" (a) (b)

hollow slem

""-
• Clay seal

• Gravel pack
Screen

Figure 6.7 Schematic of a hollow-stem auger (a). one technique used to

drill wells. Once wells are drilled (b), construction Iypically is completcd
as follows: coarse material is placed around the well screen to facilitate
water inflow to the well, backfill is packed around the rest of the casing to
seal off the screened region from possible surface contamination, and a
grout and concrete cap is constructed to provide surface security.

201
 6.4.3 Geophysical techniques

There are other, indirect techniques that are used to estimate subsurface
conditions that cannot be "seen" in the conventional sense of the word.
Geophysics is a field that uses measurements of physical processes and
properties to infer information about the structure of the Earth, and in the
case of hydrogeology, the relatively shallow solid Earth in particular. That
is, we usc measurements involving the propagation of sound (seismic
vibrations), electricity, or radar, for example, to infer the distribution of
Earth properties that affect transmission. These include density for seismic
vibrations, electrical resistivity for electricity, and the dielectric constant
for radio waves (radar). These physical properties often can be correlated
with hydraulic properties because the differences in rock type and structure
that affect the physical properties also affect the hydraulic properties such
as porosity and hydraulic conductivity. In a sense, use of geophysical
techniques can be considered to be a way of "seeing into the Earth" (NRC,
2000). The array of quite sophisticated techniques used in geophysical
investigations is extensive and a description of methods is beyond the
scope of this book. Suffice it to say that lhe visualization of subsurface
structure can provide significant information for understanding
hydrogeology. Even a subtle underground feature may have important
implications for the flow of water in soils and rocks (Figure 6.9).

202
 2

�
�
Measured in feel from ground surface
0
=;
"
; j>-2
" ..
,,
o -4
-

C!J
� --------�--�-
Feb Mar Apr May
Time. 2012
Figure 6.8 Well hydrograph for USGS Well 351428085003600, Hamilton
County, Tennessee, USA.
Provisional data from USGS.

6.5 Steady Groundwater Flow

The concepts developed in Section 6.3 can be applied readily to flow in

natural fonnations. Consider first a simple example in which a confined
aquifer is bounded by two channels (Figure 6.10). The height of the water
in the channel on the left defines the hydraulic head at that boundary of the
aquifer and the same is true on the right boundary. Irthe flow is steady and
the confined aquifer is of constant thickness, b, then the specific discharge
through the aquifer from x = 0 to x = L is constant and we can write
Darcy's law for flow in the x-direction as:

, dl1
q �-I\. - . (6.12)
d,

We can rearrange this expression and integrate over the length of the
confined aquifer to obtain an expression for hydraulic head, II:

203
 "I"" .. I.... "I'" HI""
.'."..., .'&

�� OO� E.� ��
." '- . ....,
•

Figure 6.9 Trace a transect

buried channel. Such features often may be important for guiding
groundwater flow.
Image courtesy of Alan D. Howard.

__ Potentiometric surface
� •

in confined aquifer
•

•
-

I Confined
h, � b h,
Flow
aquifer

'-'

x=o x= L
Datum, ::. = 0
Figure 6.10 Horizontal flow in a confined aquifer.

204
 q
dh::--dr
K

(6.13)

Equation 6.13 specifics that head decreases linearly with distance from the
left boundary; that is, the potentiometric surface is a plane sloping from
left to right. This is shown as a solid gray line in Figure 6.10. Water flows
in the downhill direction of this surface.
On a plan view of the channel-aquifer system considered above, the
contour lines of the potentiometric surface arc parallel to the channels
(Figure 6.11). The spacing of the contours indicates the slope oflhis
surface, just as do the contour lines of a standard topographic map. Once
these lines of equal hydraulic head, or equipotentials, have been
established, lines that indicate the direction of flow can be sketched in by
constructing perpendiculars to the cquipotemials because this reprcsents
the downhill direction of the potcntiomctric surfacc, the direction of flow
specified by Darcy's law. These lines are called streamlines. Together, the
equipotentials and the streamlines constitute a flow net. Flow nets can be
applicd to great advantagc in actual field problems whcre groundwater
flow patterns are to be cstablished based on the measured watcr levcls in a
series of wells (see Fetter, 2000, for examples).
We will discuss the use of flow nets in greater detail in a later section of
this chapter, but at this point we need to consider the physical reasoning
behind the idea that streamlines must be perpendicular to equipotentials.
Darcy's law indicates that flow should always be from high valucs of thc
hydraulic head to low values. In fact, flow must follow the path ofsteepest
descent. Picture the potentiometric surface or the water table as a
topographic surface, with hills (high II) and valleys (low II). Now imagine
a drop of water moving along this surface in response to the pull of
gravity. We havc to imagine this watcr drop moving rather slowly, without
the momentum we might expect a ball bearing or other object to have on a
steep surface. The water drop moves down the slope and perpendicular to
the "topographic" contours rather than following some arbitrary path down
the slope; that is, it follows the path of steepest descent. The rate of
desccnt is proponional to the slope, just as the specific discharge is

205
is not horizontal (as one might expect intuitively) but has a significant
upward component Groundwater flow is not concentrated near the water
table with a large volume of stagnam water at depth-the flow patterns are
actually such that flow occurs throughout the saturated zone. The fact that
water flows along "U-shaped" paths is of practical and theoretical
importance. The U.S. Geological Survey conducted a study on the
presence and the movement of agricultural chemicals in shallow
groundwater on the Delmarva (Delaware-Maryland-Virginia) Peninsula
(Hamilton and Shedlock, 1 992). They found nitrate (partly from applied
fertilizers) at almost all depths sampled in the groundwater and attributed
the patterns of contamination to land-use practices and to groundwater
flow paths.
The construction of accurate flow nets by hand is a task requiring
considerable practice. More commonly today, flow nets arc constructed
using numerical solutions to the equations governing groundwater flow.
Regardless of how a flow net is constructed, it provides not only a
visualization of the groundwater flow paths, but also infon11ation on the
rate of groundwaler flow in a particular region.

net I a

207
hillslope.
Redrawn from Hubben, 1940, fig. 45.

6.5.1 Quantif):ing groundwater flow using flow nels

Returning to the example of horizontal flow through an aquifer bounded

by two channels scparated by distance L (Figure 6.10), we find that
q= K ill -IIH
(6.14)
L
Because the hydraulic gradient, hydraulic conductivity, and aquifer
thickness are constant, the value calculated using Equation 6.14 represents
the specific discharge at any point in the aquifer. Quite often the total
dischargc in an aquifer, rather than specific discharge at a point, is of
primal)' concern. Discharge per unit length of stream can be calculated
from Equation 6.14 by multiplying by the thickness of the aquifer, b:

Discharge per length =

It-II
Kb I' H = T h-II
I' H, (6.15)
L L
where T = Kb [L2.1] is called the transmissivity of the aquifer. This is
an important parameter when considering the development of a water
supply from an aquifer. For example, a highly permeable formation 10 mm
thick may not provide a usable supply of water but a formation 100 m
thick with only a moderate value of hydraulic conductivity vel)' likely will
be usable.
We can use the geometl)' of a flow net like that in Figure 6.11 to
calculate the discharge through the aquifer as well. This may not seem
terribly important for this simple example, but when we consider more
complicated flow patterns (e.g., Figure 6.12) we see that some simple
calculations can be performed that allow us to quantify flow even in these
settings.
In Figure 6.11, the simple flow net was constructed such that a series of
squares was created, each bounded by a pair of equipotentials and a pair of
streamlines. Within the flow net, water moves from high to low hydraulic
head and cannot cross a streamline. The area between a pair of streamlines
is referred to as a streamtube. In more complicated flow nets, these
squares might become "curvilinear squares," as can be seen in Figure 6.12.
If we isolate one of these squares (Figure 6.13) and make use of Darcy's
law, then we can calculate the discharge through the square and extend this
to determine the discharge through the aquifer. The square has sides of
length ds (in the direction of flow) by dm (perpendicular to flow).
Knowing that the aquifer thickness is b, we can apply Darcy's law to

208
detennine the total discharge through this box:

Q\, = qA K(dm b)!!!!. (6.'6)

'h
=

Equipotentials

;v �f
'. 't
-j-
· ..
·

, ,
•

..
•

,
•
.

.
·
• •
·

,
·

,
.
.

-+ · ,· ,
, tis
•
.' .

..
. ·

,, ,
'

,
· ·

·
,
.
.. ..
tim
" "
·
·
. .

·
..,.
.

•
.

",
·

·
',
, .
71· ..
S ream-
t
lines
,

·
"< >
•

.,
, • · ·

.
'.
•
•
•
· •
· . ·

' • ,
·

·.
..

. · · ··,
·

,
·
"

·
..
•
·

I"
•
' ,

I·. ·
·
· ' •
· ·
..

".
.
•
..
.. ,
·
· .

. .. ,
·

· ·
·
·
· ,
•
•

I I
·

h,=hL 1t2 1t3 It. It,=".

Figure 6.13 A simple now net showing terms used to quantify flow in the
aquifer. The gray square of dimension ds by dm is analyzed in the text.

where Q, refers to the total discharge [L3r-1J through a streamtube and dll
refers to the head difference across the box, or in this case "4 113, Because -

the domain is square, dm = ds, and equation (6.16) becomes:

Qs Kbdh.
=
(6.17)

In other words, if we know the hydraulic conductivity (or the

209
transmissivity, T= Kb), we can simply look at our flow net to see what
contour interval is used for hydraulic head (dh) and multiply that value by
thc transmissivity to detcnninc thc amount of water moving through cach
streamtube. If the aquifer is bounded at upper and lower ends, then we can
count the number of streamtubes and multiply by Qs to detennine the total
amount of water flowing through the aquifer.
We now look at a somewhat more complicated example. One effect of
placing a dam in a stream or river is that a hydraulic gradient is created
beneath the dam. At the upstream end, the hydraulic head (relative to the
basc of the dam) at thc bottom of the reservoir is equal to the depth of
water in the reservoir (Figure 6.14). On the downstream side, the hydraulic
head is equal to the height of water in the river. A flow net can be
constructed, as shown in Figure 6.1 4, to detennine the panem and rate of
steady groundwater flow beneath the dam. Note that a low-penneability
laycr exists at depth, which rcpresents the bottom of the aquifer beneath
the dam. We can envision this boundary, as well as the base of the dam
itself, as streamlines, because there will be relatively little flow across
them. For this example, we assume that the dam is 100 m wide (in the
direction into the page), and that the hydraulic conductivity of the material
beneath the dam is 10-10 m S-I. We can use Equation 6.17 to calculate the
total discharge beneath the dam. We use the length of the dam (100 m) in
place of the aquifer thickness (b). The contour interval for hydraulic head,
dh, is 2 m. Then,

Qs= Kbdh= 10-10 m 5-1 x 100 mx2 m = 2x 10-8 mJ 5-1. (6.18)

We can multiply this value by the number of streamtubes (3) to obtain the
total discharge, 6 x 10-8 mJ S- I, which is approximately equal to 1.9 mJ yr­
I. In this case, the flow beneath the dam is fairly small, which is what we

would hope for; the dam would be fairly inefficient if water was constantly
leaking around it! A larger value would result if the hydraulic conductivity
was higher, or the differcncc in head across the dam was greater.
Before we conclude this chapter with a discussion of some of the
complexities that enter into our evaluation of groundwater flow, we
mention several other points regarding the use of flow nets as we have
described it. First, we have considered only steady groulldwaterjlow. It is
not possible to construct a flow net for unsteady or transient groundwater
flow. Also, we can usc a flow net to describe thc variation in specific
discharge, and, therefore, flow velocity, in an aquifer. You can imagine
each streamtube as a "pipe," because water cannot cross a streamline.
Therefore, from the principle of conservation of mass described in

210
Chapters 3 and 4, we can state that within a streamtube, the specific
discharge will be greatest where the streamtube is narrowest. The total
dischargc through the streamtube must be the same at any cross section.
Referring to Figure 6.14, the flow velocity will be greater at point A than
at point B. Finally, we have considered only homogeneous porous media,
that is, materials with a spatially constant k. In the next section, we discuss
the implicat'ions for flow through materials thaI are not homogeneous.

-
It= 26m

11=10 m
Concrete dam -

>- � 12
B , \
, '4
•

16

20
24 22

I
Figure 6.14 Flow net for groundwater movement beneath a concrete dam.
The equipotentials (gray lilies) are labeled with values of hydraulic head.

6.5.2 Hctcrogcncit) and anisotropy

Virtually all natural materials through which groundwater flows display

variations in intrinsic penneability from point to point. This is referred to
as heterogeneity. The natural processes that create and modify rocks,
sediments, and soils give rise to heterogeneity at all scales-from minor
variations in grain size and the small holes called vugs in carbonate rocks
(mm scale) to sedimentological and soil features (m scale) to variations in
fracture spacing and lithological layering as depicted in Figure 6.6 (m to
km scale). These variations in permeability from point to point complicate
flow net construction. Penneable zones tend to focus groundwater flow,
while, conversely, flow tends to avoid less penneable zones. It is possible
to construct flow nets for some simple cases, such as layered aquifers and
aquitards. However, when the penneability distribution is more
complicated, hydrogeologists often rely on numerical simulation to depict

21 1
the pattern and rate of groundwater flow.
Another complication often arises when measuring the penneability of
natural materials. Rocks, sediments, and soils often have textural features
that cause the penneability at a point to vat)' with the direction of
measurement. Materials that display this trait are referred to as
anisotropic, whereas isotropic refers to the condition in which the
penneability does not depend on the direction of measurement. Consider a
fractured rock aquifer, in which the fractures are predominantly horizontal.
In this case the penneability will be higher when measured in the
horizontal direction than in the vertical direction. You might picture this
situation by referring to Figure 6.2 and imagining that the capillary tubes
are aligned in the horizontal direction with relatively little hydraulic
communication between them. Other features that produce anisotropy
include (but are not limited to) the orientation of: platy minerals and small­
scale layering in sedimentary rocks, such as clays; cooling cracks and lava
flow tubes in basalis; large pores due to animal burrowing and plant roots
in soils; and schistosity and fractures in metamorphic and igneous rocks.
Anisotropy, like heterogeneity, makes the job of constructing flow nets
difficult, so that once again we must rely on numerical models that are
capable of including this aspect of natural porous media. In the case of
anisotropy, because water will tend to flow in a "preferred" direction, that
is, in the direction of maximum penneability, Darcy's law must be
modified to include this preference. As a result, in an anisotropic medium,
the streamlines and equipotentials may not be perpendicular to one another
at all points.
In introducing the concept of a flow net in Section 6.5, we described
streamlines as paths of steepest descent in the "topographic landscape" of
hydraulic head. We can extend this analogy to anisotropic media if we
now imagine a series of ridges running at some angle with respect to the
equipotentials. These ridges cause the water drop to move at some angle
other than straight "downhill," since the water wants to follow the ruts
between the ridges at least part of the time. (The flow will be straight
downhill if the ridges are either straight downhill or follow the
equipotentials.) These ridges in the hydraulic "topography" have the same
effect as anisotropy in porous media. The flow direction will be altered
from the nonnal direction parallel to the hydraulic gradient toward the
direction of maximum intrinsic penneability.

6.6 Concluding Remarks

212
Darcy's law (and the law of conservation of mass implied in the
construction and use of flow nets) provides the basis for computing steady
groundwater flow patterns and rates. In some cases a simplc "back-of-the­
envelope" calculation is sufficient to gain a rough estimate of flow rates.
In other cases, the aquifer may have a complex geometry or be
heterogeneous and anisotropic and simple calculations will not suffice.
Today hydrologists routinely use groundwater models (by which the
groundwater flow equations are solved with the assistance of computers)
to address environmental issues (e.g., Konikow et aI., 2006). The
computational techniques used in groundwater models are much more
sophisticated than those we have used in this chapter. However, the ideas
behind the computation are essentially the same. The geometry, conditions
at the boundaries, and hydraulic parameters (intrinsic penneability,
porosity) of the aquifer must be specified before predictions can be made
with the models.
One use of groundwater models is in making assessments of
compliance with environmental regulations. For example, before the WIPP
site was licensed to begin operation, a set of regulations issued by the U.S.
Environmental Protection Agency (EPA) had to be met. For WIPP, the
U.S. Department of Energy had to show that the probability of significant
releases to the accessible environment over 10,000 years into the future
will be very small. The "accessible environment" for WIPP means
groundwater in the Culehra fonnal'ion "down gradient" of WIPP, that is, in
the direction of decreasing hydraulic head. The demonstration of
compliance with the EPA standard is done using a perfonnance assessment
analysis (Helton et aI., 1997). In short, performance assessment uses a set
of scenarios-sequences of hypothetical events that might occur in the
future-and models to predict the impact of these scenarios. For WLPP,
one scenario that is of concern involves someone in the distant future
drilling a well into the repository and allowing some of the radioactive
wastes to flow up the well and into the Culebra dolomite. A groundwater
model must be used to "route" the hypothetical contaminant through the
aquifer to decide whether significant quantities of contaminant might reach
a human population in this scenario. These and many other environmental
problems require the kind of knowledge of groundwater hydraulics that we
have introduced in this chapter.

6.7 Key Points

• The tenn "groundwater" refers to subsurface water found in completely

213
 saturated porous media. {Section 6. 1 }

• Flow through porous media may be approached using the analogy of

laminar frictional now through a bundle of small capillary tubes, where
the flow through an individual tube is described using Poiseuille's law.
{Section 6.2}

• Hydraulic head is a measurable quantity that can be used to describe flow

in porous media. Groundwater always flows from regions of high
hydraulic head to regions of low hydraulic head: h = (P/pg) + z. {Section
6.2 }

• Henry Darcy published a report in 1856 describing a set of experiments

in which he measured flow rates through sand-filled columns. His
experiments resulted in an empirical equation relating the specific
discharge in porous media to the hydraulic gradient, referred to as
Darcy's law: q = -K(dh/dl). {Section 6.3}

• The proportionality constant K in Darcy's law is called the hydraulic

conductivity, and is a measure of the ability of the medium to transmit
fluid. Hydraulic conductivity depends on both material and fluid
properties, and is often separated into two groups oftenns, K = k (Pg/JI),
where k is the intrinsic penneability. {Section 6.3.1 }

• The properties of porous materials need to be known or eSlimated to

solve groundwater flow problems: the porosity (cp), which is the fraction
oftolal medium volume occupied by pore space; the hydraulic
conductivity (K), which varies by at least 13 orders of magnitude for
natural materials; and the intrinsic penneability (k), which varies
similarly to hydraulic conductivity. {Section 6.3.1 }

• Darcy's law may be used for many situations of flow through porous
media, except when values of the Reynolds number are high or other
forces (accelerations, electrostatic forces) are significant. {Section 6.3.2}

• Geological fonnations may be classified as: aquifers, saturated

geological fonnations that contain and transmit "significant" quantities of
water under normal field conditions; aquicludes, fonnations that may
contain water but do not transmit significant quantities; aquifuges,
fonnations that neither contain nor transmit significant quantities of
water; or aquitards, a general tenn for fonnations that are of relatively
low penneability, which may include both aquic1udes and aquifuges.

214
 {Section 6.4}

• Aquifers may be either unconfined (bound at the top by the water table)
or conlined (bound at the top by an overlying aquitard). {Section 6.4}

• The water table is a surface separming saturated and unsaturated zones in

the subsurface. The fluid pressure at the water table is zero (gage).
{Section 6.4}

• Wells and piezometers are constructed to measure water levels that

represent the elevation of the potentiometric surface at a location.
{Section 6.4.1 and 6.4.2}

• Geophysicallechniques are useful for "seeing into the Eanh" and

learning about possible controls on water flow through soils and rocks.
{Section 6.4.3}

• Flow nets are used to depict patterns of steady two-dimensional

groundwater flow, and are made up of two families of onhogonal lines:
equipotentials (lines of constant hydraulic head) and streamlines
(representing the path of flowing groundwater). Flow nets may be used to
calculate the rates of groundwater movement. {Sections 6.5 and 6.5. 1 }

• Most natural porous media are heterogeneous (permeability varies from

point 10 point) and anisotropic (the permeability measured at a point
depends on the direction of measurement). {Section 6.5.2}

6.8 Example Problems

Problem 1. You are charged with designing a very simple filtration system
for a community water supply, using cylindrical sand columns (K = 5.0 m
day-I ). The filter needs to be 3.0 m long to adequately trap particulates in
the water, and since the system will be driven by gravity, the pressure
heads at the top and bottom of the (venically-oriented) filter will be zero.

A. What diameter filter is required to treat 4.0 x 1 03 gallons of water per

day? Is this value feasible (anything larger than about 1 m is not
feasible)?

B. Consider each of the alternatives and how you might modify your
design:

215
 i. Lengthen the sand filter (how long?)
ii. Raise the hydraulic head at the inflow (how high?)
iii. Use several filters (how many? what size?)

Problem 2. A penneameter is used to measure the hydraulic conductivity

of a porous medium (see Figure 6.3). The penneameter is perfectly round
in cross section, with a diamcter of 50.0 mm. The following parameters are
measured: Z I = 220.0 mm, Z2 = ISO.Omm, PI/pg = 230.0 mm, P2/pg = 280.0
mm, with L = 200.0 mm, and a discharge at the lower end ofthc column Q
= 500.0 mm3 min- I .

A. Which way is water flowing in the eolumn? [s water flowing from high
to low hydraulic head? From high to low pressure?

B. Calculate the specific discharge, q (mm min- I).

C. Calculate the hydraulic conductivity, K, of the material (m S- I ).

D. Calculate the intrinsic penneability, k (m l).

Problem 3. Consider two piezometers placed side by side but open in

different aquifers at depth (i.e., the two piezometers on the right side of
Figure 6.6). The following measurements are made:
Piezometer Ifl Piezometer #2

Elevation of piezometer 1m above mean sea level) 200 200

Oepth of piezometer 1m) 60 20
Depth to water in pielDmetE!r 1m) 20 18
A. Calculatc the elevation and hydraulic heads (rclative to mean sea lcvcl),
pressure head, and fluid pressure in the two piezometers, and fill in the
table below.

Piezometer n Piezometer #2

Elevation head, z 1m)

Pressure head, p/(pg) (m)

Hydraulic head, h (m)

Pressure, p (Pa)

B. Calculate the vertical hydraulic gradient between the two piezometers.

Is the flow ofwatcr upward or downward?

216
Problcm 4. Consider the flow net for a drainage problem shown in Figure
6. 1 5. Drains such as pipes and culverts placed in a wet field may be used
to remove groundwater by creating a "sink" or area of low hydraulic head.
In the figure, a cross section through such a field is shown. The hydraulic
conductivity, K, of the surficial material is 1 . 0 x 10-5 m S-I. The thick
black lines represent impermeable boundaries; a constant head is assigned
to the top and lower left side. The cross section is 20 m long by 10m deep.
The gray lines are cquipotentials and the blue lines arc streamlines.

,,= 1 m
, . ' ' .' .
' ..
.. :. ' . " C,
' '
.. . .
. .
.'. .
.
..

'
.
. :' . .
.. .
.

. . ' . . , . . . '
. .

: :
.

'. . .
.
. ,
."

, " , ' : ' ::

. ," ,

. . ., .
. . .

. '
"

. " , ':. :: .. . :
:. .... . : . : ." .';:� ::' . : '.': .<.:.:- .
'
. , : :
. "
.

. . . .
. '
' .
. .

. ... .. . .. . . . . ' . c"

",�., .
.

.... . . . .
..
. ' .
. '
. . . . . .. .
.. , . . . : .
.
•:
..,
. .
. . .
.
. .
. .
.' ' ',
.
.
"

:..
. . . . . • . . . . . . .
.

'. '
.

. .
.
. .
.
" . .

.
.
.
. .

::..:: : . :::<.: :. . . '

. .

.
. .
. .

. . '
. .

. . .: .
.
.

"; ' :::'

.

," . ; .... . '

, c',:"
.

,
.

'
.

, ':'.» :::';:.'
"
'
, , ,

. . :'.
" "
,

:
" "

: : < : .: . .
' .. . ... .. .'
:
.
. . .
.

. .
.
.

'
.
. .

'
" . .

. .' .
, . ' - ..

. :. ' .:: :::

. .:::. : : : .
.
. '

.. . ' . . .
.

'
: " ,
: .
. . .

Figurc 6.15 Flow net for drainage problem (Problem 4).

A. Place labels on the equipotentials, indicating the value of the hydraulic

head along the line.

B. Calculate the discharge through each streamtube, and the total discharge
or rate at which the field is drained (m3 day-l per width of material).

6.9 Suggested Reading

Fetter C.W. 2000. Applied hydrogeology, 4th ed. Upper Saddle River, NJ:
Prentice Hall, Chapters 3 and 4.

217
7 Groundwater Hydrology

7. 1 Introduction

7.2 Flow Nets and Natural Basin Yield

7.3 Regional Groundwater Flow

7.3.1 The effect of basin aspect ratio

7.3.2 The effect of water-table topography

7.3.3 The effect of geological heterogeneity

7.4 Well Hydrographs

7.4. 1 Unconfined aquifers

7.4.2 Confined aquifers

7.5 Well Tests to Estimate Aquifer Properties

7.5 . 1 Slug tests

7. 5.2 The Coopcr-Jacob method

7.6 Con tamina nt Hydrogeology

7.7 Land Subsidence

7.8 Groundwater Recession

7.9 Concluding Remarks

7. 1 0 Key Points
7. 1 1 Example Problems

7. 1 2 Suggested Readings

7.1 Introduction

Concerns about groundwater arise in many settings, including, for

example, developing plans for sustainable water uses in urban areas with
limited surface water. The question of how to deal with the resource
problems associated with urbanization-including water resources-is one
that taxes responsible governments in almost every part of the world. In

218
the United States, for example, more than 40% of the water used by people
in Orange County south of Los Angeles comes from groundwater. Because
precipitation and surface runoff vary with thc seasons, Orange County uses
spreading basins (gravel-lined areas) over which they spread water so it
can infiltrate to recharge the groundwater that is in high demand. During
most years there is not enough natural flow in the rivers to accommodate
the demand for recharge. In those years, reclaimed wastewater may be
used for recharge. For example, in the city of Anaheim the Santa Ana
River is diverted into spreading basins to allow the water to infiltrate and
recharge the underlying aquifer. In the summer over 90% of the flow in the
river is treated wastewater that is returned to the river upstream of the
spreading basins (NRC, 2012c). Trade-ofTs between amount and quality of
water are inevitable and careful plans must be made to sustain the
resource.
One example of a difficult problem in the use and protection of
groundwater resources can be found in Mexico City, which is now one of
the world's "megacities." Water must be supplied-and wastewater treated
-for about 20 million pcoplc. Usable surface water is scarce in thc
Mexico City Basin, so groundwater from the Mexico City Aquifer is a
primary source of freshwater. Near Mexico City, the aquifer is a sequence
of alluvial fill sediments interstratified with basalt deposits and overlain by
clays. The principal aquifer is from 100 to 500 m thick and the overlying
clay aquitard is approximately 100 m thick. The water table bclow Mcxico
City has been declining at a rate of I to 1.5 m per year; the rate at which
water is being withdrawn by pumping wells is much greater than the
natural rate of replenishment. The practice of withdrawing groundwater
faster than it can be replenished, referred to as "overdraft," has been
common since about 1900. Furthennore, poor waste-disposal practices
have adversely affected water quality in the aquifer. Clearly, if the water
resource for Mexico City is to be made sustainable, changes in the pattern
of use based on an understanding of the basin hydrology will be required.
In the United States, the High Plains Aquifer (often referred to as the
Ogallala Aquifer) has experienced similar problems (McGuire, 2011). The
High Plains Aquifer is an important source of water for much of the
central United States, including parts of Colorado, Kansas, Nebraska, New
Mexico, Oklahoma, South Dakota, Texas, and Wyoming. About 20
percent of the irrigated land in the United States is i n the High Plains, and
about 30 percent of the groundwatcr uscd for irrigation eomcs from thc
High Plains Aquifer. Overdrafts of groundwater have resulted in a decline
in water levels throughout much of the aquifer. Between 1950 and 2009,
the decline was more than 4 m (on average) and exceeded 45 m in some

219
places. Declining water levels (which are expected to continue into the
future) increase the cost of water, because pumps require more energy to
lift the water a greater distance.
In places such as the High Plains, Mexico City, northern India, the
North China Plain, and the southeast of Spain, groundwater is a limited
resource and is being depleted (Wada et al., 20 I 0). In other words, the rate
of groundwater withdrawal exceeds the rate at which it is naturally
replenished. As a result there is a net reduction in groundwater storage. In
some cases, modem societies are using groundwaler that accumulated in
aquifers over geological time scales and under different climate conditions.
This unsustainable use of non-renewable groundwater resources is
sometimes called groundwater mining. The water removed from the
ground is ultimately released to the ocean, and contributes-in addition to
the effects of climate change-to sea-level rise (Wada et aI., 2010;
Konikow, 201 1). Global estimates of groundwater depletion vary between
27 km3 yr-I (Margat et al., 2006) and 283 km3 yr-1 (Wada et al., 2010).
Table 7.1 shows an intennediate estimate (145 km3 yr-I) along with the
major hotspots of groundwater depletion in the world. As a tenn of
comparison, globally, the irrigation water used by crops is about 545 km3
yr- 1 (Siebert et aI., 2010).
To understand limitations on the sustainable use of groundwater, we
must examine the water balance for the subsurface. How and at what rate
is groundwater replenished or recharged? Hydrological basins (the rocks,
sediments, and soil underlying catchments) exhibit natural rates of
groundwater flow as a result of the infiltration of precipital'ion, exfiltralion
or discharge of groundwater, and evapotranspiration. Flow nets provide a
tool for evaluating the natural rate of groundwater flow. How is that flow
altered by perturbations such as pumping? Groundwater pumping in the
Mexico City Basin has changed the panern of flow as well as the water
balance. Because the flow pattern detennines how pollutants move within
the subsurface, flow nets are also important in evaluating the risk of
contamination.

Table 7. t. Total net groundwater depletion and average groundwater

depletion in 2001-2008

220
 Net depletion in Averase annual
Aquifer
2001-200S{kml) rate (kml yr")

In the USA

Atlantic Coastal Plain 2.8 0.3

Gull Coastal Plain 67.4 8.'
High Plains (Ogallala) Aquifer 94.7 11.8
Central Valley, California 31.4 3.'
Western Alluvial Basins 2.1 0.3

Western Volcanic Systems 2.' 0.'

Deep Confined Bedrock Aquifers 2.6 0.3

Agricultural and Land Drainage 0 0

Total (all USA systemsl 203.9 25.5

Non-USA Aquifer Systems

Nubian Aquifer System 18.9 2"

North Western Sahara Aquifer System 17.6 2.2

Saudi Arabia Aquifers 109.1 13.6
North China Plain 40.0 5.0
Northern India and Adjacent Areas 423.5 52.9

Indirect Estimates for Other Areas 350.0 43.7

Total Globill 1163.0 145.4

Source: According to Konikow (20 J J).

Our best infonnation about the water balance of an aquifer system,

infonnation that is essential for water resources planning, usually comes
from records of water levels and groundwater pumping. This is one of the
reasons why groundwater hydrology, as interpreted from flow nets and
water-level records, is an important area of study.

7.2 Flow Nets and Natural Basin Yield

Flow nets can be constructed for any selling where the approximations of
steady and two-dimensional flow are valid, as long as the conditions at the
boundaries and the distribution of hydraulic conductivity are known. Later
in this chapter we explore cases where flow is not steady. For the present
discussion, we consider steady flow to be an approximation to the real
system, such that temporal variations in the height of the water table are
small relative to the thickness of the flow system.

221
 I I
I I
I I
I I •
I I
I I
I I
I I
I I
Q;n I I QOfll
II dx I
I
I I
I I
I I

hi h2
I I
I I II
I I
I I
I I
Figure 7.1 Definition sketch for the derivation of the steady groundwater
flow equation.

Flow net construction is based on a mathematical model of groundwater

flow. The fonn oflhis model is a differential equation that describes the
physical process of groundwater motion. We can develop such an equation
using conservation of mass and Darcy's law. Consider a single curvilinear
square within a flow net as a control volume (Figure 7.1). This volume is
bounded by two equipotentials and two streamlines. The streamlines can
be thought of as impenneable boundaries, because no water crosses them.
For steady flow, the inflow (Qin) must equal the outflow (QolII)' or the
change in storage must be zero. For the streamtube (the area bounded by a
pair of streamlines), this may be written:

QOut - Q", = dQ =
0
d\· dr
,

where Q [L3r-l] refers to the total discharge at any section through the
streamtube. For uniform now in the x-direction (both streamlines are
horizontal), the specific discharge, q�, also must be a constant:

dq,
=0.
dx
(7.1)

Darcy's law may be written for the specific discharge as:

222
 dh
q� dx '
K (7.2)
_
_

where dh is equal to 112 " 1 ' Substituting Equation 7.2 into Equation 7. 1
-

and assuming that K does not val)' in space (i.e., the material is
homogeneous):

dq,
= K.!!... :!!!. =
0
dt dt
_

dt

or

d2h
=0. (7.3)
d,'
-

In two spatial dimensions, this would become:

(7.4)

Equal'ion 7.4 is referred to as the Laplace equation and may be solved

to detennine the distribution of hydraulic head (the equipotentials) in two
dimensions, provided that the region is homogeneous (constant 10. Of
course in the two-dimensional case, the equipotentials (and streamlines)
are not straight parallel lines as in the one-dimensional case in Figure 7. 1 .
The conditions at the boundaries of a domain are required to solve this
equation. A flow net may be thought of as a graphical solution to the
Laplace equation. The Laplace equation is simply a mathematical
statement of the law of conservation of mass (Equation 7. 1 ) combined with
Darcy's law (Equation 7.2). It is now common for the Laplace equation to
be solved using computers. Computer-based groundwater models have
become important tools for solving problems of groundwater hydrology
and contaminant transport.
For steady flow in two dimensions through a hillslope, conservation of
mass requires the inflow of water to be balanced by the outflow, because
there can be no change in storage. In other words, the same number of
streamtubes must leave the hi11s10pe as enter it. This idea gives rise to
naturally occurring recharge and discharge areas. A recharge area occurs
where water is crossing the water table downward, hence, recharging the
groundwater system. A discharge area occurs where groundwater is
moving upward across the water table, thereby discharging into the
unsaturated zone above, or to the land surface or a surface-water body
such as a lake or stream. Consider a hills lope bounded by an impenneable

223
bottom (for example, the bedrock beneath an unconfined aquifer) and two
impermeable sides, called groundwater divides. The upper boundary is
the water table. Often, the water table is a subdued replica of the land
surface topography, such that the water table is higher beneath hilltops
than beneath a valley. A flow net shows a balance between recharge in the
upland portion and discharge in the valley (Figure 7.2a).
Flow nets must obey certain rules at the boundaries. For example, it is
probably apparent that streamlines cannot cross impcnneable, or "no­
flow," boundaries. Equipotentials must be perpendicular to impenneable
boundaries, because otherwise some flow across the boundary would be
implied. A constant-head boundary is represented by an equipotential, and
streamlines must be perpendicular to a constant-head boundary. At the
water table, the hydraulic head is everywhere equal to the elevation,
because the gage pressure is zero. Equipotentials and streamlines may both
intersect a non-horizontal water table, and the orientation of these lines
provides a means of delineating recharge and discharge areas (Figure
7.2b).

Groundwater divides -------_�01

,.
,

Discharge area -1
,
,­
,
-·----- Recharge area ------ :
,

Recharge area
(b)
Discharge area � UphiU

Uphill angle > 900

Figure 7.2 The natural pattern of groundwater flow in a simple basin (a).
Equipotentials are gray and streamlines are blue. Recharge and discharge
areas may be distinguished by looking at the angle that an equipotential

224
makes with the water table (b).

The hydraulic head along any equipotential is a constant by definition.

One way of labeling an equipotential is by noting where it intersects the
water table (Figure 7.3). Water in a piezometer that is open at a particular
depth will rise to an elevation equal to the hydraulic head at that point. For
example, in Figure 7.3 the piezometer is open at the depth depicted. The
elevation head (z) at thai point is 20 m. Because the open interval
intersects the 40-m equipotential, the hydraulic head (h) at that point is 40
m. The pressure head (P/pg), therefore, is equal to II - z, or 20 m.
Flow nets have a "dimensionless" quality. They are a picture of the
paltern of steady groundwater flow and depend only on the physical
features of the basin. For example, the pattern shown in Figure 7.2a does
not depend on the actual size of the basin or the (homogencous)
penneability of the basin rocks or sediments. However, quantities
detennined from the flow net do have particular dimensions (and units).
Recalling the discussion in Chapter 6 on quantifying flow using flow nets,
the discharge through each streamtube may be calculated knowing only the
hydraulic conductivity and the difference bctween adjacent equipotentials
(dh):

Piezometer 60

'----- 55 50

50 ---1
30
45
-1
20
__

40 _
__ _
_

Open portion 10
of piezometer
o
Figure 7.3 The hydraulic head along any equipotential is equal to the
elevation of its intersection with the water table.

Qs = Kbdh. (6.17)

Equation 6.17 is valid as long as the flow net has been constructed
using curvilinear squares. For a horizontal flow net, b is the aquifer

225
thickness (see Chapter 6). The total volume of flow through a vertical
cross section depends on the cross section width, however, and so b refers
to that width. If we let b I unit [L], the discharge calculated from
Equation 6. 1 7 is simply the discharge per unit basin width.
=

Consider how we might use the flow net shown in Figure 7.2a. We
would have to know the length of the basin, for example L = 100 m, in
which case the hill is 30 m (0 . 3L) high on the right side. The difference in
hydraulic head between adjacent equipotentials is equal to 0.005L, or 0.5
m. If the hydraulic conductivity is 0.5 m day- I (approximately 5.8 )( 10-6
m S- I ), the discharge through each streamtube is:

Qs = Kbdh = (0.5 rn day-IXI rnX0.5 rn)= 0.25 m3 day-I.

Thcre arc four strcamtubcs in thc flow net. Thercforc, the lotal
discharge through the section is 4 x 0.25 or 1.0 m3 day-I . This is thc
discharge per meter basin width (b = 1 m). If the basin is 500 m wide, then
we could simply multiply 1.0 m3 day- I by the 500-m width to find the total
discharge from the basin (500 mJ day-I ).
The above calculation shows that the basin discharge is proportional to
the hydraulic conductivity of the basin material . Knowledge of the
permeability of basin materials is essential to detemlining the water
balance for a groundwater system. Freeze and Witherspoon ( 1 968) referred
to a calculation such as that above as the natural basin yield. Under
natural or undisturbed conditions (i .e., in the absence of anthropogenic
groundwater withdrawals or changes in climate or vegetation), this is the
average rate of discharge from a hillslope or basin .

7.3 Regional Groundwater Flow

The simple flow net above (Figure 7 .2a) provides a template for
understanding regional groundwater flow. However, flow patterns are
conditioned by variation in the shapc ofthc basin and the watcr table, and
spatial patterns of hydraulic conductivity. This section explores the
primary controls on the pattern and rate (natural basin yield) of steady
groundwater flow in a basin bounded by divides at the sides, the water
table at the top, and a low-permeability unit at the base . The terms
introduced in thc previous section (recharge and discharge areas, natural
basin yield) provide some keys to exploring the importance of each of
these controls.

226
 7.3.1 The effect of basin aspect ratio

Hydrological basins occur in a variety of shapes, and we would expect the

basin shapc to excrt some influence over the groundwater flow pattern.
Toth (1 962, 1963) examined the influence of basin aspect ratio (length to
depth) on the panern of groundwater flow in a homogeneous two­
dimensional basin with a gradual water-table slope. Imagine a basin of
constant length (L), but with a depth that might vary, depending on the
dcpth to a low-permeability unit, such as crystalline bedrock. The pattern
of flow will be similar to that shown in Figure 7.2a, with single recharge
and discharge areas. Our experience with Darcy's law should tell us that
the natural basin yield will be large for deeper basins (smaller aspect
ratio), because the volume through which groundwater is flowing is larger.
This assumes that we are comparing basins with identical waler-table
profiles. An analogy might be constructed using columns of different
cross-sectional area. For a given hydraulic gradient (analogous to the slope
of the water table in a basin), the discharge will be greater through a larger
column. Recall that Darcy found that, for the simple case of one­
dimensional flow, the total discharge was proportional to the cross­
sectional area.
Compare Figure 7.4a and Figure 7.4b. The natural basin yield in each
case is directly proportional 10 the number of streamtubes and to the
hydraulic conductivity of the basin materials. For the shallow basin (large
aspect ratio, Figure 7.4a), there arc only two complete streamtubes and a
small fraction of a third. For the deep basin (small aspect ralio, Figure
7.4b), almost eight complete streamtubes occur. This observation malches
our expectation, although the total discharge is not perfectly proportional
to the basin depth. In a catchment selling, we might expect the thickness of
permeable material ncar the surface to have a big influence on the rate of
groundwater flow. Where thin soils are found, the rate of groundwater
flow might be relatively small.

227
 (a) Large basin aspect ratio (shallow basin)
o. ,

I"I< I 0.0
-

I
0.0 o. , 0.2 0.3
I
0.4 0.5 0.6 0.7 0.8
= O.003L

0.9 , .0
x/[

(b) Small basin aspect ratio (deep basin)

-
0.4
7-
.--tB
\ 0.3

I- >\
I -'( 0.2::::
"

X \ o. ,
'f-
till = O.OO3L
0.0
0.0 o., 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 ' 0
xlL
Figure 7.4 The effect of basin aspect ratio (length to depth) on natural
patterns of groundwater flow. The water-table profile is the same for the
shallow (a) and deep (b) basins.

There is some difference between the patterns of flow in the two basins
in Figure 7.4. For the deep basin (Figure 7.4b), vertical hydraulic gradients
(and, therefore, vertical flow) exisl over a large portion of the basin. In the
shallow basin (Figure 7.4a), the flow is essentially horizontal over most of
the basin.

7.3.2 The effect of water-table topography

Variation in hydraulic head resulting from topographic relief of the water

table is in most instances the driving force for groundwater flow. Complex
land-surface topography should produce similarly complex water-table
topography, and so we need to consider the flow systems produced by
such topography. Although real topography can be quite complex, we
might consider a general picture in which "local" topography (small-scale
undulations or hills and valleys) is superimposed on a "regional" slope.
The end-member cases are shown in Figures 7.4 and 7.5, for regional and
local topography, respectively. These patterns are similar in some respects,
because flow occurs from highs to adjacent lows. A single hill-valley flow
system (dashed boxes in Figure 7.5) resembles the larger flow systems

228
produced by regional water-table topography (Figure 7.4). These flow
systems appear to have a lower depth limit, below which flow is very slow
or non-existent, as may be seen particularly well in Figure 7.5b. However,
the flow doesn'l actually become zero, but only slows as slreamlubes
become wider and the hydraulic gradient along the streamlines decreases.
It could also be shown that the apparent "depth" of these local flow
systems depends on the local hydraulic gradient.

(a) Local watertable topography, large basin aspect ratio

(b) Local watertable topography, small basin aspect ratio

0.4

0.3

0.2::::
..
,
,
0.1

_�
It:"
__ ���-;�� �����
I&�I�O�.O�O�3�
LI 0.0
__

0.0 0. 1 0.2 0.3 0.4 0.5 0.6 0.7 0.6 0.9 1 .0

xlL
Figure 7.5 Regional groundwater flow panems for the case of local water­
table topography. The dashed boxes indicate an individual local flow
system. The same equipotential spacing (d") of O.003L is used for both the
large (a) and small (b) basin aspect ratios.

T6th, in the same studies cited earlier, referred to these flow cells
(dashed boxes in Figure 7.5) in which water flows from topographic high
to adjacent low as local flow systems. Larger flow systems, from regional
high to low, h e referred to as regional flow systems. The similarity in
appearance between thcse types of flow systems suggests that the use of
different tenns to describe them depends only on one's definition of
"regional." However, the distinction becomes clearer when we
superimpose local hill-and-valley topography on top of a regional slope
(Figure 7.6). For the case of a deep basin (Figure 7.6b), both local and

229
regional systems develop (streamtubes labeled "L" and "R" in the figure),
as well as what Toth referred to as an intermediate flow system
(SlreamlUbe labeled "I" in the figure). If the basin is relatively shallow
(Figure 7.6a), a regional system may still exist, but may be attenuated as a
result of the dominant influence of the local flow systems.
The conclusions that we can draw from these flow nets are as follows:

I . If local relief is negligible, but a regional water-table slope exists, only a

regional flow system will develop.
2. Conversely, if local hill-and-valley topography exists, but no regional
slope, only local flow systems will develop.
3. Ifboth local and regional topography exists in a basin, all three types of
flow systems (local, intermediate, and regional) will develop. As a
result, precipitation infiltrating on a hilltop may eventually discharge at
an adjacent low or follow a longer flow path toward the regional low
point.

(a) Regional and local watertable topography, large basin aspect ratio

(b) Regional and local watertable topography, small basin aspect ratio

xlL
Figure 7.6 Groundwater flow patterns for the case of combined regional
and local water-table topography. Streamtubes labeled "L," "I," and "R"
indicate the local, intermediate, and regional flow systems, respectively.
The same equipOlential spacing (dh) ofO.003L is used for both the large
(a) and small (b) basin aspect ratios.

230
 Complex water-table topography will necessarily produce complex
patterns of groundwater flow, but the general observations discussed above
can be applied to other settings. Referring to Figure 7.7, can you identify
local, intermediate, and regional flow systems?

7.3.3 The effect of geological heterogeneity

Geological materials are always heterogeneous, a fact that complicates

analysis. Aquifers and other penneable zones in the subsurface are capable
of capturing and focusing groundwater flow. In some cases, recharge and
discharge areas may develop in locations that are not predicted from the
configuration of the water table. Hillside springs are one good example.
Freeze and Witherspoon (1 966, 1 967) performed a series of numerical
"experiments" in which they calculated flow nets for a variety of
configurations of the water table and subsurface geology. These
calculations demonstrated the ability of subsurface aquifers to alter the
flow system (Figure 7.8).

Figure 7.7 A flow net for a sha with complex water-table

topography.

A highly permeable unit such as an aquifer causes flow to become more

nearly venical within the overlying unit, and may narrow the discharge
area (Figure 7.8b). Aquifers act as conduits that are capable of rapidly
transmitting water to the principal discharge area. Because the mean
hydraulic conductivity of the basin is greater, the natural basin yield is
increased (compare Figure 7.8a with Figure 7.8b). A large portion of this
yield is transmitted through the aquifer. An aquitard at depth effectively
reduces the basin depth and natural basin yield (Figure 7.8c).
Discontinuous zones of high permeability, such as the truncated aquifer
shown in Figure 7.8d, may alter the areas of recharge and discharge. The
discharge produced at the end of the aquifer may appear as a surface
spnng or seep.
Hsieh (2001 ) produced a computer model that is useful for
demonstrating the influence of the topography of the water table and

231
heterogeneity on flow nets. Onee the model is downloaded
(http://waler.usgs.gov/nrp/gwsoftware/tdp£ltdpf.html), it ean easily be used
to explore a variety of diffcrent flow conditions to gain insights about
recharge and discharge areas of groundwater flow systems.
Mexico City lies within the discharge area for the Basin of Mexico,
which is a "closed basin." This means that there are no natural surface
water outflows for the basin, and all outflow occurs through
evapotranspiration. The Basin is ringcd by volcanic mountains that fonn
the recharge areas for the Basin. Because of the heterogeneity of the
materials filling the basin, an aquifer-aquitard system exists, and the
pattern of flow from the mountains to the center of the basin looks
something like Figure 7.8b. Prior to heavy pumping during the past 100
years, groundwater flowed from thc rccharge areas (mountains) into
discharge areas in the Basin center through the aquifers, and discharged
upward across the overlying aquitards. Surface springs and lakes resulted
from this natural pattent of groundwater flow.
Potable groundwater under artesian pressures (i.e., heads that caused
water flow to the surface in wells without pumping) was discovered in the
Basin of Mexico in 1846, promoting the rapid development of this
resource (NRC, I 995b). Since that time, the large overdrafts of
groundwater have reduced the pressures within the confined aquifers and
altered the natural pattern of groundwater flow. Now most of the water
within the Basin is moving downward toward the heavily pumped
aquifcrs; surfacc springs and lakes have dried up. In 1983, systematic
monitoring of the water levels in wells began. These records of water
levels in wells have provided important information on the water balance
in the basin.

232
 1<111 = o.oo2L1
(a) Homogeneous
O. 1

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
xlL

(b) Aquifer at depth

(e) Aquitard at depth

0.1

� ,-t
-

, ,
:!: , tt,
0.0
0.0 0.1 0.2 0.3 0.4 0.5 06 07 08 0.9 1.0
xlL

(d) Discontinuous aquifer at depth

� ...
,
"

xlL
Figure 7.8 The influence of geological heterogeneity on patterns of
groundwater flow. The homogeneous case (a). An aquifer at depth (b;
hydraulic conductivity of the aquifer is an order of magnitude greater than
the overlying unil). An aquitard at depth (c; hydraulic conductivity an
order of magnitude less than the overlying unit). A discontinuous aquifer
at depth (d). The equipotential spacing (dh) is the same for all, O.002L.

7.4 Well Hydrographs

The three major controls on patterns of groundwater flow described in the

previous section concern natural groundwater flow. The natural system is
assumed to be at steady state, such that recharge and discharge balance one
another and the water table is approximately constant. Groundwater
systems, however, are both dynamic (although sometimes sluggish) and
subject to human alteration, as we have seen for the Mexico City region.
The task of depicting patterns of flow in these circumstances is a difficult
one. We often have observations, in the form of well hydrographs, that

233
allow us to gain some understanding of these effects. As we will see in the
next sections, confined and unconfined aquifers behave somewhat
differently, and we consider each scparately.

7.4.1 llnconfined aquirers

A well hydrograph shows the variation in water level in a well through

time. In an unconfined aquifer, the water level in a well generally indicates
the position of the water table. Well hydrographs may show variations
over different time scales. For example, daily fluctuations may be
observed in carefully monitored wells, and very gradual changes over
several years of observation may also be evident. The most obvious cause
of water-level variations is fluctuation in the inputs and outputs. In Coffee
County, Tennessee, the water-table elevation is observed to vary with the
season (Figure 7.9). Precipitation in Tennessee is not very seasonal
(monthly precipitation averages are relatively constant). However,
recharge to groundwater does vary seasonally because it is influenced by
factors other than the amount of rainfall, the most important of these being
evapotranspiration. The variation in the water table in Coffee County can
be explained by noting that evapotranspiration is at a peak in the summer
months and greatly reduced in the winter months. Consequently, because
the ground remains unfrozen during the winter in southern Tennessee,
more water is available for recharge in the winter and groundwater levels
rise. The converse is true in the summer-little recharge occurs and levels
decline as the groundwater discharges to provide the baseflow in streams.

234
 1
>­
-

..

�
E
-

u
.-

f!
Q.

01110 04/1 0 07/10 10/10

0 r------,
-5

E
-

01/10 04/1 0 07/10 10/10

Time
Figure 7.9 Daily precipitation (a) for a weather station in Coffee County,
Tennessee, USA and water levels (b) for water year 201 0 in a nearby well
in an unconfined aquifer.

In areas where the water table is close to the ground surface,

groundwater levels are influenced directly by the transpirationai demands
of plants. During the day when transpiration is high, water movement is
upward from the water table and the level declines. At night, transpiration
is reduced, groundwater flows laterally from locations upslope that are
relatively unaffected by direct transpiration effects, and the water table
recovers (Figure 7.1 0).
Recharge 10 aquifers does not occur solely from direct infiltration of
rainfall. For example, water can seep from surfacc·watcr bodies, such as
rivers, ponds and lakes, into the ground. Artificial recharge (recharge
induced by activities of people as opposed to that which occurs naturally)

235
can be implemented by introducing water into wells (recharge wells) or by
routing water into infiltration basins in permeable material (the idea
mentioned in regard to Orange County in thc introduction).

-0.76
-0.77

E -0.78
�

� +-
"
.c -0.79
-

l!!
"
�
-0.80
'"
-

;: -0.81
0

-0.82
-

.t:
Co
-

"
C -0.83
-0.84
-0.85
'" '" '" '" 0> 0> 0> 0> 0> 0>
0 0 0 0 0 0 0 0 0 0
.., 0 '" "' "- 0>
- - -
<0 '"
- - - - - -

0 0 0 0 0
� �
N N N '" �

Time (day/month)
Figure 7.10 Water levels for a latc summcr period in 2010 in a shallow
well adjacent to White Clay Creek at the Stroud Water Center near
Avondale, Pennsylvania, USA.

Unconfined aquifers store water; a change in the elevation of the water

table indicates a change in the amount of water stored in the aquifer.
Consider a portion of an unconfined aquifer of (in plan view) area I m2
(Figure 7.11). A I-m drop in the water table would result from the removal
of a certain volume, V, of water. The volumc of water removed depends on
the porosity of the aquifer and how much water would remain behind after
gravity has drained the upper eubic meter of aquifer material. (The water
that remains is held by forces that will be described in the next chapter.)
Hydrologists rcfcr to this volume, V, pcr square meter of aquifcr per mctcr
drop in the groundwater table as the specific yield (Sy) of an unconfined
aquifer. Because it is the volume of water produced per unit aquifer area
per unit dccline in hydraulic hcad, it is dimensionless. Thc specific yield is

236
characteristic of a given aquifer and allows determinations to be made of
the change in the volume of water stored over time. Values of specific
yield typically are less than the porosity, and most range from 0.0 I to 0.30.
Not all of the water in an aquifer volume will drain under the influence of
gravity. Some will be retained by forces causing adhesion of water to
particle surfaces. This is why the specific yield is less than the porosity.

Area =

1 m2
.
I

_ Original
. " . ..
,
, .
water table
· - : · 1 ·.. ·

I
I
. .1
1 m r-,-
-
I _ Water table after
I
I water removal
=---
-+-=
I
I
I

Aquifer
) ,
- - -

,
/

Volume removed = V m3
Figure 7.11 The specific yield of an unconfined aquifer. For a I-m decline
in the water table, the volume of water produced per unit aquifer area is
the specitic yield, Sr

The specific yield is a hydrological parameter that detennines the

response of the water table to changes in inputs and outputs. In the case of
an increase i n evapotranspiration (Figures 7.9 and 7.10), the change in
water-table level may be fairly unifonn over a given area, although
variations will occur due to the lateral movement of groundwater and
spatial variations in evaporation rate and vegetation. Pumping a well has a
different effect. Pumping produces a decrease in hydraulic head at a point,

237
which increases the hydraulic gradient toward the well. The change in
water level in the pumping well, or in observation wells nearby, is referred
10 as a drawdown. The amount of this drawdown will decrease as one
moves away from the pumping well, and the pattern that is produced is
called a cone of depression because of its characteristic shape (see Figure
7.14). The shape and extent of the cone of depression within an unconfined
aquifer depend on the pumping rale, the transmissivity and specific yield
of the aquifer, and lime. From the definition of the specific yield, however,
we know that the volume of the cone of depression is equal to the volume
of water removed (by pumping) divided by the specific yield:

I""'f"'''
V
v.
-
=

S,
(7.6)

7.4.2 Confined aquifers

In an unconfined aquifer, the response to pumping a well is a change in the

water table. Water is drained out of the aquifer as the water table declines,
and the specific yield provides a measure of the volume of water released
from storage. However, the upper boundary of a confined aquifer, the
overlying aquitard, docs not move substantially in response to withdrawing
water from a well within the aquifer. Instead, a cone of depression is
created within the potentiomefric sUlface. The aquifer material is not being
drained, and the aquifer remains saturated. However, water is withdrawn
from storage within the aquifer (otherwise, the well would become a dry
hole almost immediately). Analogous to the specific yield of an
unconfined aquifer, we can define the storativity (S) of a confined aquifer
as the volume of water produced per unit aquifer area per unit decline in
the potentiometric surface (Figure 7.12). The question becomes, how is
water removed from a confined aquifer without de-watering (draining) thc
aquifer material?
Consider the forces acting on a horizontal plane within a confined
aquifer (Figure 7.13). A downward force due to the weight of the
overlying material exists, which when divided by the area of the plane (A)
is called the total stress (crr). Because the plane is not undergoing an
acceleration, Newton's second law tells us that this downward force must
be balanced by an equivalent opposing force or forces. In other words, the
weight of the overlying material must be supported or held up by
something. In this case the opposing forces per unit area (stresses) arc the
pressure (P) of the water and the upward stress exerted by the aquifer

238
solids, called the effective stress (ae):

Clr = p+CI,. (7.7)

Area =

1 m2 "
1-
/1 _
_ _

Original
// 1 //
/ 1 / potentiometric
.:: J
1 surface
_ _ _ _ _ ....

1
1
1
1 m
1 '-_ Potentiometric
1
1 surface after
1
•
water removal
Aquitard
1
,

1
1
,
Aquifer
1-
•

./

//
_ _

.
//

Volume removed = V m3
Figure 7.12 The storativity of a confined aquifer. For a l�m decline in the
potentiometric surface, the volume of water produced per unit aquifer area
is the storalivity, S. The aquifer material is not drained and remains
saturated.

This idea may be slated in another way: the weight of lhe overlying
material on a horizontal plane is supported in part by the fluid and in part
by the solid. The total stress won't val)' over time:

da,. = dp + da, = 0.

239
Therefore, any change i n pressure must be offset by a change in effective
stress:

dp = -d(J,. (7.8)

If water is being withdrawn from a well within a confined aquifer, the

hydraulic head, and, hence, the fluid pressure, is being reduced at that
point (dp is negative). As a result, the fluid will expand slightly. This is
because water is (slightly) compressible. This is one mechanism by which
water is released from storage in a confined aquifer. The additional l'o/lime
of water produced may flow to the well and be withdrawn by the pump.

Confining unit aTA

and Force down due to
overburden weight of overtylng
water and rocks

plane
Aquifer of area A

Force up due to Force up due to

effective stress water pressure
pA
ae A
Figure 7.13 Force due to water pressure and effective stress balance the
downward force due to the weight of the overburden.

Equation 7.8 indicates that the decrease in fluid pressure must be

accompanied by an increase in the effective stress (positive do..). A part of
the weight of the overlying material is being transferred from the fluid to
the solid. This results in the compression of the aquifer material, just as
decreasing the fluid pressure resulted in expansion of the fluid.
Compressing the aquifer material is similar to squeezing a sponge, and
produces water that may be pumped from the well. This is the second
mechanism by which water is removed from storage in a confined aquifer.
Water and most aquifer materials are not very compressible. As a result,
stomtivity values tend to be lower than values of specific yield. Storativity
generally ranges from 0.005 to 0.00005, as compared with 0.0 I 100.30 for
specific yield. In this discussion of storage in confined aquifers, we have
assumed that the aquifer is perfectly confined, or that all the flow occurs

240
within the aquifer and none within the confining layers. If an aquifer is
only semi-confined, with some water coming from surrounding aquitards,
then it is referred to as a leaky aquifer. Because of this additional source
of water, the storativity will often be higher in leaky aquifers.
The extent of the eone of depression depends, among other things, on
the value of the storage parameter (either specific yield or storativity).
Consider two aquifers, one unconfined and one confined, that are being
pumped at the same rate. Because the specific yield of the unconfined
aquifer will be greater than the storativity of the confined aquifer, the
drawdown in the unconfined aquifer will be less than that in the confined
aquifer (Figure 7.14). That is, the volume of the cone of depression in a
confined aquifer is equal to the volume of water pumped divided by the
storativity [change Sy to S in Equation 7.6], a very small number. Thus
removal of water from confined aquifers produces substantial drawdown
of the potentiometric surfacc. In confined aquifers that arc heavily pumped
over long time periods, the cone of depression ean be quite extensive,
approaching tens of kilometers in lateral extent, with tens of meters of
drawdown in the vicinity of the well (Figure 7. 1 5) and a cone of
depression that extends a long distance from the major pumping center
(Figure 7.16).

241
(a) Unconfined aquifer

r
•

Water table

Unconfined
aquifer

(b) Confined aquifer

- - - - - - -
' "
�'-" - -
"

/' -- -
--
-

Potentiometric '\,

surface

Aquitard

Confined
b
aquifer

242
Figure 7.14 The cone of depression in an unconfined (a) and confined (b)
aquifer. The wells are being pumped at a constant rate (Qp), and hydraulic
heads are being monitored at two piezometers at radial distances 1"1 and 1"2
from the pumping well.

0
"
> -5
E
-

.-

.. -
-

-10
" "
"
-

..
�

"O 't: -15

\
-

.. "
" '"
-20
"" "0
" "
. " -25
" 0
-
-

.. �
�
'" -30
"g, o
,,

\.
-
-35
-40
a � a � a � a � a � a
� � <D <D � � <D <D m m a
m m m m m m m m m m a
� � � � � � � � � � '"
Year
Figure 7.15 Water levels in a well in a confined aquifer in Calvert County,
Maryland, USA.
Data from DePaul et a1. (2008).

Other things can affect the balance between water pressure and
effective stress in aquifers and so affect the water levels in wells.
Atmospheric pressure varies as low-pressure and high-pressure systems
move across the landscape. An increase in atmospheric pressure creates an
additional force on the top of a confined aquifer that is distributed between
water pressure and effective stress. That is, only a portion ofthe force
from the increased atmospheric pressure will be transmitted to the water in
the aquifer. The increased atmospheric pressure also acts on the water
standing in the well. In the well, the full increase in the atmospheric
pressure is transmitted to the water. Thus, for a confined aquifer, an
increase in atmospheric pressure causes a decrease in the water level in a
well and, conversely, a decrease in atmospheric pressure results in an
increase in water level. For an unconfined aquifer, changes in atmospheric
pressure are transmitted equally to the water table and the water in a well,
and so do not produce changes in water levels.

243
7.5 Well Tests to Estimate Aquifer Properties

The rate at which water levels change in wells depends in part on the
aquifer properties. For example, if a quantity of recharge water is suddenly
added to an unconfined aquifer, the water levels will decline as the water
drains out to streams. The rate of decline will depend on the hydraulic
conductivity and the storativiry of the aquifer. Thus, if the water table or
potentiometric surface in an aquifer is purposefully disturbed, by removing
or adding water for example, measurements of the time variation of water
levels in wells can be used to back-calculate the aquifer properties. Below
we briefly describe two well tests that are llsed in the field.

244
 --
�-50
� --
--
�� - --
-30
-­
--
- -

�
-- -
�

'"
----_tIl: --,-6() -
-
--
-
-
- -
- --

-70

()

Figure 7.16 Map of the potentiometric surface around a groundwater

pumping center in Maryland, United Slates. Contours arc feet relative to
mean sea level. (Original level was approximately 1 0 feet above mean sea
level.)
Figure modified from Soeder ct lIl. (2007)

7.5.1 Slug tests

Slug tests use measurements made on a single well. The water level in the
well is suddenly perturbed by removing a "slug" ofwaler. This lowers the

245
water level in the well, creating a head gradient toward the well.
According to Darcy's law, water will then flow radially into the well until
the water lcvel in thc well is restored. The level of the watcr in the wcll
relative to the steady-state water table, H, is measured as a function of
time.
The initial perturbation inside the well is a depression of the water level
by Ho and subsequently the level varies with time, H(l). The well casing
has radius rc whereas the radius of the borehole is R. The length of the well
screen (the non-impervious part of well casing through which water flows)
is L (Figure 7. 17).
The rate of water flow into the well, Q, can be calculated using Darcy's
law.

dh
Q � 21CrLK- .
dr

where r is the distance from the center of the well and II the saturated
thickness ofthc unconfincd aquifer at distance r (Figurc 7. 1 7) .
This equation can bc rearranged as:

Q dr = ,111.
2trKL
(7.9)
r

246
 I n it ial
H yd raulic co n dition s
head at t ime h
T. after slug
is removed 2R

R. ---�)

Figure 7.17. Schematic for a slug test. Re is the radial distance beyond
which the perturbation by the slug has no effect. Note that the right side of
the figure depicts conditions initially, before the slug is removed, and the
left side depicts conditions that evolve through time after the slug test is
initiated.

Equation 7.9 can be integrated across the step change in H caused by the
removal of the slug, lelting r vary from R to Re, the effective radius beyond
which there is no effect.

(7.10)

Because after the initial abrupt withdrawal no other water is extracted from
the well, a mass balance in the well itself gives the following relationship.

dH Q.
dl
- : -

(7. 1 1 )
JCrl
•

Equations 7. 10 and 7. 1 1 can be combined to get

247
 2KL
- =- ,
dH
dl. (7.12)
H r,,- ln( R"I R)

which can be integrated from time zero (when the slug is removed) to a
later time, t. to give:

H, 2KL
In- = - I or
Ho r/ln(R,.IR) ,

2 KL
InH(I) : - , 1+lnHo. (7.13)
r; In( R"IR)

Equation 7.13 provides a relationship Ihat pennits estimation of

hydraulic conductivity given measurements of time, t, and H(t). The
nalural logarithm of H can be plotted versus lime. Equation 7.13 indicates
that the plot should be a straight line with slope -2 KL I [r,} In(R.,IR)] and
intercept In Ho.
Values for L. re, and R can be gleaned from records kept during well
construction. The radial extent of the slug perturbation, Re, is not known.
One assumption is to set it equal to L. Because of the logarithmic
dependence of K on Re errors associated with this assumption have only a
small impact on the estimation of K. With this assumption and an estimate
of the slope of the straight line relationship between In H versus t, the
hydraulic conductivity can be calculated.

-(sfope)xr/ In(L! R)
2L
K =
. (7.14)

For example, data collected for a slug test on well near Oyster, Virginia
(Figure 7.18) are fit with a straight line with slope -0.19 S-I . The well
casing has an inside diameter of 7.6 cm and the borehole radius is
estimated to be 12.7 cm. The length of the well screen is 5 m. From
Equation 7.14, we estimate the hydraulic conductivity for this shallow
(sand) aquifer to be

248
 1.0 --------,

0.5

0.0

_ -0.5
�
c:
-

-1.0

Slope = 0 1 9 S-I
-1.5
- .

-2 .0

-2.5 :---;:-
a 2 -;-
-

4 -;:-
6 --:
8:--
-
--::
1 :--
0 -:-;:-
12 :-;-
1 4 --:
-;:-
1 6 --
1 8-!.
-

Elapsed time (seconds)

Figure 7.18 Slug test results for a well ncar Oyster, Virginia, USA.

K � (0 . 1 9 s- ' )(0.076 m) ' ln(5 mlO. 127 m)l( I O m) � 4. l x I 0-4 ms- ' .

7.5.2 The Cooper-Jacob method

Slug tests to determine hydraulic conductivity in the field are quite useful
in many ways. They are easy to carry out, the duration of a test is short,
and only one well is involved. The measurements reflect a relatively small
volume of aquifer material around the well, however. In cases where an
average response over a larger area is the aim, tests that reflect the aquifer
properties between two wells may be preferred. The basic idea is that one
well is pumped and the hydraulic head is measured as a function oftime in
observation wells, say at distances rl and r2 from the pumping well (Figure
7.14). The heads hi and h2 will change with time as the pumping proceeds.
The equation that describes thc variation of head with respect to distance
fr om the pumping well and time is an extension of the Laplace equation
(Equation 7.4) to account for time-varying conditions. (Details can be
found in hydrogeology texts such as Fetter, 2000.)

249
 °.:;
'"
,Ph +'0-
(7. t 5)
s ail
T al axl ay!
=

The Cooper-Jacob ( 1946) method can be used to estimate S and T from a

solution to Equation 7. 1 5.

, ,1 = Q" ,nl + Q" , 2.2ST .

4lCT 4lCT r'S
s= 10- n , (7.16)

In a field test, pumping is started and elapsed time is measured. The

drawdown, s, is the difference between the original height of the
potentiometric surface in the observation well and the measured hydraulic
head as a function of elapsed time. Equation 7.16 indicates that a plot ofs
versus the In r should be a straight line with slope Qrl47fT and intercept
Q" 2.25T
I" ,
4nT r1S
Miah and Rushton ( 1 997) report data for pumping tests conducted in
the Madhupur aquifer in Kapasia, Bangladesh (Figure 7. 19). A 10-cm
diameter well was pumped at a rate of 5140 mJ day- ! for several days.
Drawdown was measured in an observation well 100 m from the pumped
well. The slope of the drawdown-In f plot is 0.36 m and the intercept is
1 .82 m (Figure 7 . 1 9). The Cooper-Jacob solutions give

QL
= -,---
"' '
I'---,- - 5 1 40 m day- ' = 1 1 30 ' d ' .
T m ay-
41r(slope) (4 x Jt x 0.36 m)

Once transmissivity has been estimated, the storativity can be obtained

using the intercept of the straight line fit to the data.

( )
2.2ST (2.2SX 1 130 m2 day- 1 )
S
=
r2exp (intercept)
41fT - ("
(100 m)�exp -
�I .8:::
2 ;;:
: " :.: ,,, ,,;-
4If",
..'" X" X" m '.::
3 0:..
I ' ::: d;:,
'Y
,- ')
-..2.
Q" 5140 m 3 day- I
= 0.0010.

Notice that, unlike the Cooper-Jacob method, the slug test provides an
estimate only of the hydraulic conductivity and not of the storativity. This
is due to the fact that the slug test is based on a water balance equation for
the well (Equation 7. 1 1 ), in which the flow from the aquifer is calculated
without accounting for the unsteady character of groundwater flow. In fact,
Equation 7.9 is integrated assuming that Q is constant with respect to r.
This assumption (i.e., 8Qlor = KfYhla?- = 0) corresponds to setting to zero
the left-hand side of Equation 7. 1 5. Because the transient character of the

250
slug test's response is expressed by the water balance of the well and not
by the unsteady groundwater flow, this test does not provide any
information on the "inertia" of groundwater flow (i.e., of storativity or
specific yield).

7.6 Contaminant Hydrogeology

Among the vel)' important environmental issues today are how to prevent
contamination of groundwater and, should contamination occur, how to
clean up aquifers. Groundwater flow is relatively slow in human terms, so
once an aquifer is contaminated, clean up can be extremely difficult or
even essentially impossible using reasonable resources. Knowledge needed
to determine effective protective measures and to explore the feasibility of
clean-up options includes insights from physical hydrologists regarding
rates and directions of groundwater flow.

1.8 ,-------:

1.6

1.4
E
-

- 1.2
•
-

"
� 1.0

l 0.8
c
.5 = 0.36In(I)+1 .82
0.6

0.4

0.2 L--
-5 -4 -3 -2 -1 a
In(/), time in days
Figure 7.19 Pumping test results for the Madhupur aquifer in Kapasia,
Bangladesh, for the Cooper-Jacob method.

Advection is the process of transport of contaminants by the movement

251
of the water in which they are dissolved or suspended. The contaminants
"go with the flow" and travel along groundwater flow paths. For example,
a contaminant injected into the groundwater at the water table will move
by advection along a path detennined by the flow net if the flow is steady
(Figure 7.20). The speed at which the contaminant moves can be estimated
using Darcy's law provided the hydraulic conductivity, K, and the
porosity, ¢, of the aquifer is known. Although the specific discharge, q,
given by Darcy's law has dimensions of velocity, it is not a true velocity.
The specific discharge is discharge per total cross-sectional area of aquifer,
but the flow occurs only through the openings in the rock. Recalling that
porosity represents the fraction of the total volume of rock that is available
to water, the average linear velocity of the groundwater flow can be
estimated as v = ql¢. If we take the schematic in Figure 7.20 to represent a
local flow system in a sandy aquifer with L = 1 0 km, ¢ = 0.25, K = 1 0-6 m
S-I, and a head gradient along the flow path A _ B of 0.033 mJm, the time
for a contaminant to travel 100 m would be

�-, o ..
0.3

0.2:::::
"

0.1

I:---01�
00
--= ..L,O'::___'� :"___--;,-
03
,-""":T:
0. '
--' 0.:'--
5
__:+:
O'
--07::-
---: � ::-'-----:T
0.'
:----4:1 .00 . 0
O'
xlL
Figure 7.20 A contaminant entering the groundwater at point A will be
carried by advection toward point B.

palh lenglh 100 m

t
1n"...1
- - = 7 5 x I 06 s S7 days.
(1 0-0 m S-I x 0.033 m/m)/0.25
. ""

Even with this relatively high head gradient and hydraulic conductivity,
we would estimate that the time for a contaminant to travel the length of
the domain would approach 25 years. For lower gradients and longer
paths, the times would be longer. Flushing a contaminated aquifer by
natural recharge of fresh water can require very long time periods.
Advection is not the only process that occurs as contaminants move in
groundwater. The flow paths through pores, cracks, and fractures in the

252
aquifer are convoluted and water moving along various paths mixes as the
paths converge, diverge, join, and splil. These mixing processes result in
spreading of the contaminant by dispersion. A pulse of contaminant will
spread out to fonn a plume as it moves along the flow path (Figure 7.2 1 ).
Groundwater contamination can originate from many sources, for
example from landfills that were constructed before we understood how
contaminants can leach from them and move to groundwater. Over time,
extensive plumes can fonn, threatening natural ecosystems and the quality
of streams into which the groundwater drains. Hydrologists need to
understand how transport occurs to address potentially serious
environmental problems. The U.S. Geological Survey has been studying
hydrological and biogeochemical process at a landfill near Nonnan,
Oklahoma (Christenson and Cozzarelli, 2003). The leachate plume has
spread several hundred meters from the landfill and has affected a nearby
wetland. Detennination of flow paths and rates of flow require estimates of
hydraulic conductivity. A t the Nonnan landfill site, slug tests were used to
characterize the heterogeneity of the sediments and gage the impact on
groundwater contaminant transport (Scholl and Christenson, 1998).

7--, 0. '
0.3

0.2:::!
"

0.1

Contaminant disperses
to fo rm a plume
Figure 7.21 Dispersion spreads a contaminant out as it is advected along a
groundwater flow path.

7.7 Land Subsidence

A common misconception concerning confined aquifers has been that

confining layers are perfectly impervious. However, no geological
materials are truly impervious and confining units such as aquitards (e.g.,

253
clays, shales) may contain large amounts of water that can move given
enough time. Figure 7.8b illustrates this possibility. In instances where
overlying units (or interlayers within the aquifer system) are clays,
pumping the confined aquifer has two effects. The first is to increase the
effective stress on the aquifer and the second is to de-water the clay units.
What happens when clays are de-watered? You may have observed the
cracks thai appear on the bottoms of mud puddles that have dried up.
These cracks reftect the shrinkage of clays upon dl)'ing. The same thing
can (and does) happen when clay units above or within aquifers have part
of the water in them removed: they shrink. The result is called land
subsidence.
As discussed, the cone of depression in confined aquifers can be quite
large (e.g., Figure 7 . 1 6), and land subsidence due to pumping can be
significant ovcr considerable areas. [n the San Joaquin Basin, California,
subsidence has affected an area of at least 4200 square miles; maximum
subsidence approaches 9 m in some areas (Helm, 1 982).
In Mexico City the lowering of the potentiometric surface in the
Mexico City Aquifer has resulted in the removal of water from the
overlying clays. The land surface has subsided by some 7.5 m in the
central part of Mexico City (Figure 7.22; NRC, I 995b). Old well casings
in the city extend several meters above the ground surface because the land
has subsided around the casings. The result of this subsidence has been
extensive damage to the city's infrastructure, including building
foundations and the sewer system. Anothcr serious problem in Mexico
City relates to flooding. The city is bordered to the east by Texcoco Lake
-the natural low point of the southern portion of the Basin of Mexico. In
1 900, the lake bottom was 3 meters below the median level of the city
center. By 1974, the lake bottom was 2 meters higher than the city! This
change is thc rcsult of greatcr land subsidence within the city and has
aggravated flooding problems. A complex drainage system (including
excavations to [ower Texcoco Lake) has evolved to control flooding.

254
 2241
2240
e � 2239
-
-

0 >
.- �
.. .. 2238
- -

> �
� �
" e
..
2237
"
u �
-l!! E 2236
��
<II > 2235
' 0
"
e .0
..
..
...J E 2234
-

2233
2232 •• •

1900 1910 1920 1930 1940 1950 1960 1970 1980 1990
Figure 7.22 Land subsidence in Mexico City.
Data from Onega-Guerrero et a1. (1993).

7.S Groundwater Recession

Groundwater levels i n unconfined aquifers tend 10 rise after rainy periods

when recharge occurs and to decline gradually as the water discharges into
streams and other bodies of surface water following recharge events. The
declining portion of a stream hydrograph following a stonn, snowmelt, or
wei season reflects the decline in groundwater levels through time because
surface-water runoff 10 the stream ceases in a relatively short time after the
precipitation stops. This decline in groundwater input to a stream is known
as groundwater recession.
Recession curves often have the shape of a negative exponential. In
regions with a very pronounced seasonal rainfall distribution this can be
seen very clearly. In Pescadero Creek, south of San Francisco, for
example, a semi-logarithmic plot of streamflow recession shows that a
straight-line segment (an exponential relationship) fits the streamflow data
reasonably well from early May through early July ( Figure 7.23). A
straight line also fits the data from July through August, but with a
different slope than the earlier data. This result suggests that, for Pescadero
Creek, two groundwater reservoirs with different time constants control

255
the base-flow recession. The fonn of the curve for the recessions is:

(a) Arithmetic axes (b) Logarithmic vertical axis

0.8 , .0
0.8
0
0.7
o
0.6
7' 0.6
" o
00
E 0.5
"
0.4
o
_

..
'" o 0.3
; 0.4 o
.r:::
u o
" a> 0.2
is 0.3

0.2
'lr"
Ufo
..
.. .. .
o. , o. ,
-

May Jun Jul Aug Sep May Jun Jul Aug Sep
Month in year 201 0
Figure 7.23 Groundwater recession for Pescadero Creek, California,
showing both arithmetic (a) and logarithmic (b) axes.

(7.17)

where Q (L3 11] is the discharge at time I after recession begins, Qo is the
discharge at the beginning of the recession period, I is time measured from
the beginning of the period, and c [I I] is a recession constant.
The appropriateness of this equation can be investigated by considering
the conservation of mass. During recession, the inflow to the entire
groundwater reservoir can be assumed to be zero and we would expect that
(he outflow would be a function of some measure of the groundwaler
elevation, Ii [L]. In other words, the hydraulic head (and hydraulic
gradient) should decrease with time, and, thereby, reduce the driving force
for groundwater discharge. We could write this as:

q, = [(II), (1.18)

where qr is the discharge per meter width of stream [L2 11]. Conservation
of mass requires the outflow to be balanced by the change in groundwater

256
storage. The time rate of change of groundwater stored should depend on
;;, the storativity S or specific yield �v (depending on whethcr the aquifcr is
confined or unconfined), and the length of the aquifer L (Figure 7.24). The
appropriate conservation equation would then be:

dh -
SL- =-q, = -f(h). (7.19)
d1
Thc Icft sidc of this cquation is thc timc rate of change of watcr stored
in the aquifer per meter stream width. Now to integrate Equation 7.19 we
need a specific fonn of/(/1). The simplest relationship that might be used
is a direct proportionality:

(7.20)

where c is a proportionality constant r.1]. Substituting Equation 7.20 into

7.19 gives:

dh
=-eh. (7.21)
d1

which can be integrated to give:

II = "oe-�', (7.22)

where 110 is the average hydraulic head at the initial time. Then, from
Equation 7.20 we have:

Figure A simple model of groundwatcr recession.

257
 -

0'

(7.23)

Equation 7.23 is directly comparable with Equation 7 . 1 7. This

comparison indicates that the use of 7.17 implies that the groundwater
reservoir behaves as a linear reservoir, that is, that groundwater outflow is
directly proportional to the amount of water stored. Analysis ofbasef1ow
recession curves can be useful for estimating flows throughout a dry
season and thus can be the basis for planning water use in areas with
seasonal precipitation regimes.

7.9 Concluding Remarks

Natural groundwater flow pattems and rates are controlled by basin

geometry, water-table configuration, and how hydraulic conductivity is
distributed in the subsurface. Flow nets, or the modern equivalent,
numerical groundwater models, are useful tools for describing these
patterns and rates. Groundwater hydrology also relics on the analysis of
well hydrographs. The changes in water levels and the consequences of
these changes can be of profound importance. An excellent example is the
case of Mexico City. Mexico City is dependent on groundwater; only 2%
of its water supply is from surface water within the basin. Several
125 km,
aqueducts bring surface water from other basins as far away as
but these supply only about 26% of the demand. Thus, about 72% of
Mexico City's water supply comes from groundwater (NRC, \ 995b).
Planning for sustainable use of the aquifer is obviously critical.
Despite the realization in the early part of this century that groundwater
resources in the Basin of Mexico were being rapidly depleted, over­
exploitation ofthc aquifer continues to cause problems ofland subsidence
and increased vulnerability to contamination. To predict a useful life for
the aquifer under current pumping rates (or to plan future uses to keep
from depleting the resource), the drawdown rates of the aquifer must be
measured accurately and the hydraulic parameters of the aquifer must be
estimated. The National Research Council ( 1 995b) recommended
implementation of a program to develop the hydrological data necessary
for supporting sustainable management of the resource. Analysis of well
hydrographs will be an important part of this program.
In many places, conserving groundwater resources is an important

258
concern. Numerical models can be used to construct flow nets describing
past or present groundwater flow patterns. Several kinds of information are
required to model groundwater flow accurately, some of which havc been
described in this chapter and Chapter 6, namely: porosity, intrinsic
penneability, and storativity or specific yield of basin materials; basin
geometry and water-table configuration; and a record of water levels thaI
can be compared with model predictions. Once the prcsent conditions have
been established, these models can be used to predict the conscqucnccs of
future or continued changes or management practices. The High Plains
Aquifer in the central United States was mentioned in the introduction as a
heavily exploited aquifer system. There is obvious concern for this
resource because of the dramatic changes in water levels that have been
observed over thc past 50 years. A numerical modeling study ofthc impact
of future (1 998-2020) management strategies for this aquifer system
predicted water-level declines of over 30 meters for some portion of the
region (Luckey et aI., 1 999). The extent of the predicted region of
declining watcr lcvels depends on the particular pumping rates assumed.
Despite the vastness of the region underlain by this aquifer system, the
groundwater resource does have a limit, and significant depletion remains
a concern for the millions of people who depend on it.

7.10 Key Points

• Groundwater flow patterns are characterized by recharge and discharge

areas; water moves downward across the water table in recharge areas,
and upward in discharge areas. {Section 7.2 }

• Thc watcr table is often a subdued replica of the land-surface topography,

especially in humid areas. { Section 7.2}

• Flow nets may be used to determine the natural basin yield-the quantity
of water, on average, that discharges from a basin. If the flow is steady,
this quantity will be balanced by the recharge rate. {Section 7.2}

• The pattern and rate of groundwater flow are controlled by the basin
aspect ratio, the water-table topography, and the distribution of
permcability. {Section 7.3 }

• The basin aspect ratio is the ratio of basin length to basin dcpth or
thickness. Shallow basins are characterized by mostly horizontal flow,
and deeper basins by significant vertical flow components and relatively

259
 higher natural basin yield. { Section 7.3. I }

• Variations in the height of the water table drive groundwater flow. The
topography of the water table may produce patterns of local,
intcnnediate, and regional flow. { Section 7.3.2}

• Hydrological basins are generally heterogeneous; the distribution of

aquifers and aquitards influences the pattern of groundwater flow.
Aquifers tend to focus flow and provide relatively rapid pathways for
water movement toward discharge areas. {Section 7.3 .3 }

• A well hydrograph is a record of water level in a well lhrough time. Well

hydrographs show variations due to daily and seasonal variations in
evapotranspiration rate, pumping of wells, and any other process
affecting the subsurface water balance. {Section 7.4 }

• Changes in groundwater storage are quantified using the specific yield

(unconfined aquifers) and storativity (confined aquifers). Both may be
defined as the quantity ofwaler produced per unil aquifer area per unit
decline in the hydraulic head. Specific yield values tend 10 be much
larger than values of the storativity, because declines in head in
unconfined aquifers are accompanied by drainage of a portion of the
aquifer; in confined aquifers, the material is nOI drained and water is only
released from storage due to cxpansion of thc watcr and compression of
the aquifer. { Sections 7.4. 1 and 7.4.2 }

• Water removed from aquifer storage through pumping produces a cone

of depression in the water table or potentiometric surface. The size of the
cone of depression is related to the pumping rate, the aquifer
transmissivity, and thc storagc coefficient (specific yield or storativity).
{Section 7.4.2}

• Withdrawing water from a well reduces the hydraulic head and fluid
pressure at that point. As a result, the effective stress (the total stress
minus the fluid pressure, or 0" = 0r - p) is increased; the effective stress
is that portion of the weight of the overlying material that is borne by the
solid material of the aquifer. Water is then "released from storage" in two
ways: through fluid expansion as the pressure decreases and by expUlsion
from the compressed aquifer material. {Section 7.4.2 }

• Well tests are used to detennine aquifer hydraulic properties in the field.
Slug tests perfonned in a single well give information about the

260
 hydraulic conductivity of the aquifer material surrounding the well.
{Section 7.5. 1 }

• Tests involving more than one well usc a pumping well to remove water
from an aquifer and observation wells to measure changes in hydraulic
head that occur as a result of the pumping. The Cooper-Jacob ( 1 946)
method can be used with the observation well measurements to estimate
the transmissivity and storativity of the aquifer in the vicinity of the area
of the wells used in the test. {Section 7.5.2}

• Contaminants in an aquifcr arc advcctcd by the watcr flow at an averagc

linear velocity of q/rjJ. Dispersion of the contaminant also occurs during
flow, causing spreading of the contaminant into a plume. {Section 7.6}

• Land subsidence is one result of groundwater withdrawal from pumping

wells. This subsidence is produced by compression of the aquifer
material (an increase in effective stress) and by dc-watering of clayey
aquitards, causing shrinkage. {Section 7.7}

• Strcamflow in times bctween prccipitation or snowmelt cvcnts is

generally produced by groundwater discharge. Because the groundwater
reservoir must decrease as a result (assuming no new water is being
supplied), the groundwater discharge decreases in time, often producing
an exponential groundwater recession curve (Q = Qae-i:I). {Section 7.S}

7.1 1 Example Problems

Problem I. Determine the natural basin yicld (m3 yr- I pcr metcr basin
width) for the following cases.

A. For the basin in Figure 7.4a with L = 5000 m and K = 30 m ye l

B. For the basin in Figure 7 4b with L = 5000 m and K "" 30 m yr-I

C. For the basin in Figure 7.5a with L = 200 m and K = 100 m yel . You
might want to begin by calculating the discharge through each local
flow system (dashed box).

Problem 2. Answer the following questions for the basin in Figure 7.25.

A. Is the lower unit an aquifer or an aquitard?

261
B. What fraction of the total basin yield passes through the lower unit?

x (m)
Figure 7.25 Flow net for Problem 2.

Problem 3. Consider the flow net shown in Figure 7.26. The sides of the
region are groundwater divides, the top boundary is the water table and the
bottom is an impermeable boundary. Streamlines are blue and
equipotentials are gray.

A. Label the equipotentials with the appropriate value of hydraulic head

(m).

B. Draw arrows on the streamlines indicating the direction of groundwater

flow.

C. Label all recharge and discharge areas.

D. Indicate at least one area within the flow net where flow is relatively
fast, and one area where flow is relatively slow.

E. Determine the water level (hydraulic head, m) in wells A and B.

262
 4Oj ------

30
E
-

20
"

10

o +-....:!,- '-,-.1.-.,-'--
�-'--,
- ___.,.J.
- _- ,_!_�-

o 10 20 30 40 50 60 70 80 90 100
x (m)
Figure 7.26 Flow net for Problem 3.

Problem 4. Well hydrographs for unconfined aquifers can be used to

dctcnnine the change in water stored within the aquifer. Consider the
change in water-table level observed in Coffee County, Tennessee,
between mid-May and August 2010 (Figure 7.9).

A. What is the change in water-table level during this time period (m)?
Docs this indicate an increase or a decrease in water stored within the
aquifer?

B. If the specific yield of the aquifer is 0.25, and the aquifer has an area of
600 km2, what is the change in water stored over the same time interval
(m) ?

Problem 5. Detennine the recession constant, c (day- I ), for the two

groundwater "reservoirs" contributing to Pescadero Creek during the
spring and summer of 20 I 0 (Figure 7.22).

7.12 Suggested Readings

Fetter, C.W. 2000. Applied hydrogeology, 4th ed. Upper Saddle River, NJ:
Prentice Hall, Chapters 5 and 7.
National Research Council (NRC). 1995. Mexico City's water supply:
Improving the olltlookfor slistainability. Washington: The National
Academies Press.

263
8 Water in the Unsaturated Zone

8 . 1 Introduction

8.2 Forces on Water in the Unsaturated Zone

8.2.1 Pressure head

8.3 Capillary-Pressure Head and the Moisture Characteristic

8.4 Darcy's Law

8.5 V er tical Water Movement

8.6 The Equilibrium Profile above a Water Table

8.7 The Profile of Capillary-Pressure Head as an Indicator of
Flow

8.8 The Infiltration Process

8.8.1 The Green-Ampt equation

8.9 Field Measurements in a Soil

8. 1 0 Evapotranspiration from the Unsaturated Zone

8. 1 1 Capillary Barriers

8 . 1 2 Concluding Remarks

8. 1 3 Key Points
8. 14 Example Problems

8 . 1 5 Suggested Reading

8.1 Introduction

In most areas, with the exception of bogs and swamps, the water table is
some distance below the ground surface. Between the ground surface and
the water table is a region in which the pore spaces of the rock or soil may
be partly filled with air and partly with water. This region is referred to as
the unsaturated zone or vadose zone and water in this zone is referred to
as soil moisture. Hydrologists want to be able to describe the flow of
water in the unsaturated zone to deal with a number of important issues.

264
For example, in the last chaptcr we discussed the concept of recharge to
subsurface aquifers. Recharge takes place most often through the
unsaturated zonc, either ovcrlying an unconfined aquifer or in the recharge
zone of a confined aquifer. Changing climate and land use practices have
strong effects on recharge through the vadose zone, and understanding
these effects is particularly important in arid and semi-arid areas (Scanlon
et aI., 2006).
One important aspect of water flow in the vadose zone is the water
balance of plants. Most terrestrial plants extract water from the vadose
zone. Plants wilt when soils become too dry because the forces holding the
water in the soil are too great to allow the plants access to the water (see
also Chapter 9). Related to the water balance of plants is the practice of
irrigation in agricuhure. Agriculture accounted for 90% of freshwater use
over the past century and water used for irrigation accounts for 90% of
consumptive use (return of water to the atmosphere by evapotranspiration)
of water that is withdrawn from streams, rivers, and aquifers (Scanlon et
aI., 2007). Given that water resources are already stressed in many regions
and that food demand is expected to increase by 50% as the global
population increases to 9 billion by 2050, there is a real need to improve
the efficiency with which we use water to grow crops (Rockstrom et aI.,
2007). Understanding the movement of soil water, and its uptake by plants
and "loss" through evapotranspiration and recharge to the groundwater
system, is essential in this regard.
Another example of an important problem in vadose-zone hydrology is
the usc of scmiarid locations with deep unsaturated zones for the disposal
of wastes. The Low-Level Radioactive Waste Policy Act was passed by
the United States Congress in 1980. This act provides for the fonnation of
regional compacts by states to supply sites for the safe disposal of low­
level radioactive waste (LLRW). LLRW includes test tubes, rags, rubber
gloves, tools, and so forth used in medical research and treatment, in other
research (for example, in research in environmental sciences to study the
biodegradation of organic wastes), and in nuclear power plants. The major
concern about finding a suitable site is groundwater contamination,
because the pathway through groundwater is the one that is most likely to
be the one that places human populations at risk of exposure to
contaminants. Given the concern for transport of contaminants to
groundwater, it has been recognized for some time that disposal of wastes
in the unsaturated zone in a desert environment should be a safe alternative
for such disposal (Winograd, 1 98\).
In response to the LLRW Policy Act, Arizona, California, North
Dakota, and South Dakota fonned the Southwest Compact and selected a

265
site in Ward Valley in the Mojave Desert for the first disposal site for the
Compact. Before any site can be used for the disposal of wastes, a license
must be obtaincd. The liccnsing procedurc rcquires earcful study ofthc sitc
and estimates of rates at which radionuclides might leach into the
groundwater. One of the key parts of the analysis is the determination of
flow rates in the unsaturated zone. Even with careful study, siting waste­
disposal facilities is a contentious issue and opposition can be expected to
any sitc. Opponents oftcn challenge the scientific assumptions regarding
hydrological processes. The relationships between the forces on and the
flow of water are ingredients of arguments presented by both proponents
and opponents of waste-disposal facilities. We will return to the case of
Ward Valley at the end of this chapter after we have developed the ideas
useful in describing the flow of soil moisture.
Finally, as we will see in our discussion of catchment dynamics in
Chapter 1 0, the storage within and release of water from the vadose zone
are quite important in detennining the stonnflow dynamics of a catchment.
The infiltration and movement of water in the unsaturated zone represents
one potential pathway for precipitation entering a stream. Variation in the
ability of water to infiltrate a soil or rock is, therefore, an important aspect
of catchment water dynamics.
The early literature recognized three divisions within the unsaturated
zone: the capillary fringe, the intennediate belt, and the belt of soil water
(e.g., see Meinzer, 1923). According to Meinzer, the capillary fringe is "a
zone in which the pressure is less than atmospheric, overlying the zone of
saturation and containing eapil1ary intcrstices some or all of which arc
filled with water that is continuous with the water in the zone of saturation
but is held above that zone by capillarity acting against gravity." That is,
the capillary fringe is a saturated zone above the water table where water is
affcctcd by capillary forccs. Thc uppcnnost belt, or bclt of soil water, is
"that part of the lithosphere immediately bclow the surface, from which
water is discharged into the atmosphere in perceptible quantities by the
action of plants or by soil evaporation." This definition recognizes that
plants, for the most part, extract water from a portion of the soil (the "root
zone") ncar thc surface. The intennediate bclt is "that part of a zone of
aeration [i.e., the ul/saluraled zone] that lies between the belt of soil water
and the capillary fringe." The intermediate belt is distinguished mainly by
the fact that something must be between the root zone and the capillary
fringe. The distribution of moisture abovc the water table is what
motivates the definition suggested above. The volumetric moisture content
(or simply moisture content) in the capillary fringe is the saturation value
(Figure 8. 1 ); that is, the pores are completely filled with water.

266
Volumetric moisture content, or more precisely volume wetness, is
defined as the volume of water per bulk volume of soil sample. We will
usc the symbol () to represent volumetric moisture content. After a rather
rapid decrease from saturation, the moisture content in the intermediate
belt may remain fairly constant. Field capacity is a term used to represent
this "constant" moisture content. The moisture content in the soil water
belt decreases rapidly from the field capacity due to the extraction of water
by plant roots and to direct evaporation at the soil surface.

z
Soil water
zone

Unsaturated Intermediate
or vadose zone
zone

- - - - - - - - - - - - - - - - - - - - - - - - - - - _ .

Capillary
fringe
v
8
Saturated
zone
Figure 8.1 The distribution of moisture in the vadose zone and the
classification of waters according to Meinzer ( 1 923). Water near the
surface of the soil is available for uptake by plant roots. After several days
affair weather, the moisture content in this belt of soil water (or rool zone)
decreases substantially due to evapotranspiration. Directly beneath the root
zone, the moisture content tends to be fairly constant over a depth of up to
a meter or more. The relatively constant value of moisture content i n this
region is referred to as the field capacity of the soil. Near the water table,
the pores of the soil act as ';capillary tubes" and remain saturated even

267
though the pressure head in the water is negative. This saturated zone
above the water table is the capillary fringe.

The divisions of the vadose zone often are useful i n describing general
observations of soil moisturc. Of coursc, there arc no sharp dividing lines
marked off in the field. The physical principles that we use to quantify
flows in the vadose zone do not change in moving, say, from the root zone
to the intennediate zone. The terminology introduced above is used
widely, however, and we will encounter these descriptive terms later. Our
introduction to the physics of soil moisture will hold for all of the zones.

8.2 Forces on Water in the Unsaturated Zone

The unsaturated zone is a threc-phase system consisting of soil, watcr, and

air. The physical description of the system and of the flow of water in the
system is thus more complex than for the two-phase system of the
saturated zone. A full treatment on a microscale of the diverse forces
acting on water in an unsaturated soil and the resultant motion of this
water is not feasiblc. Fortunately, empirical work shows that, as in
saturated soils, water flow in unsaturated soils is down a gradient of
hydraulic head of soil water. For flow of groundwater in the saturated
zone, the hydraulic head is composed of two terms, pressure head (P/pg)
and head due to gravity, or elevation head (z). Darcy's law states that the
flux of groundwater is proponional to the gradient in hydraulic head.
Gradients in elevation head and in pressure head also drive flows in the
vadose zone. All terrestrial water, including soil moisture, is within the
Eanh's gravitational field. Therefore, head due to gravity in the vadose
zone is once again the potential energy per unit weight, z, that is, the
elevation head. The main way that the physics for the vadose zone differs
from that for the saturated zone is in the pressure head contribution to
hydraulic head.

8.2.1 Pressure head

If water is withdrawn from a rock or soil matrix that does not shrink upon
drying, air enters the pore space, and air-water interfaces (menisci) are
present in the pore space. Such curved interfaces are maintained by
capillary forces. Surface tension acting in the interfaces provides a
mechanism of soi I-water retention against externally applied suction. This
phenomenon is seen in the rise of water in capillary tubes, for example,

268
and is explained by the attractive forces between the glass walls of the tube
and the water. The glass attracts the adjacent water molecules more
strongly than do other water molecules themselves. The water is therefore
"pulled" up the inside of the tube (Figure 8.2). This "pull" is a tension
which, in the terms that we are using, is a negative (gage) pressure. That is,
the gage pressure in unsaturated soils is Ilegative. Negative pressure heads
are developed in unsaturated rock and soil matrices. The height of rise in a
capillary tube (a measure of the ncgative pressure head) is inversely
related to the diameter of the tube. Water will rise higher in a tube with a
small diameter than it will in a tube with a large diameter. This observation
translates to soil physics in that smaller diameter pores retain water against
higher suctions than do larger pores (cf. the large capillary tube and the
small capillary tube sketched in Figure 8.2). Thus, when water drains from
a soil or rock, large pores empty first because it takes relatively less
applied suction to pull water out of larger pores. The negative pressure
produced by capillary forces, when divided by pg, is referred to as the
capillary-pressure head.
AI first glance, the idea of negative pressure head does not seem
intuitive to most people. Yet there are commonplace examples with which
most of us are familiar. If asked about the pressure head at the surface of a
pan of water in your kitchen, you should be comfortable with the answer
"zero." But what happens if you bring a dry sponge just barely into contact
with the water surface? Water rises into the pores of the sponge. The
sponge pulls water up from a state of zero pressure head. The "pull" is
capillary tension, a negative gage pressure. Conversely, if the sponge
subsequently is withdrawn from the water surface, all of the water does not
drain out. The sponge retains water against the downward action of gravity
by counteracting the downward gradient in elevation head with an upward
gradient in (negative) pressure head.

269
 Height of
capi lla ry Pair
rise is
inversely A
related
to the
radius
of the tube
-1 ---., PB Pc Pair 0
h, = = =

PA ; PB - pghz
Pcapillary = Pair PA-

Figure 8.2 Surface tension "pulls" water up into capillary tubes. Water
pressure within the tubes is less than atmospheric pressure, or is negative
in gage units. The height of water above the free surface in the tube is
equal 10 Ihe negative of the capillary-pressure head. The amount of
negative pressure head with which a capillary tube can " hold" water is
inversely related to the diameter of the tube. That is, small-diameter tubes
(and by analogy, soil pores) hold water at a more negative pressure head
than do large-diameter hlbes (or soil pores).

8.3 Capillary-Pressure Head and the Moisture

Characteristic

For areas where there are moderate fluxes ofwatcr through the vadose
zone, the two major driving forces on soil water are the gradients in the
negative capillary-pressure head and the gradient in elevation head. This
situation is exactly analogous 10 that for flow in the saturated zone, with
the only change being that the positive pressure heads encountered in
groundwater are replaced by negative capillary-pressure heads. Thus, the
hydraulic head for the vadose zone is defined to be the sum of the head
due to gravity and the (negative) capillary-pressure head. (In general,
several forces act 10 create the negative pressure heads in the unsaturated
zone. The treatment given here remains valid, but an "equivalent" negative
pressure hcad that incorporates all important forces, rather than just
capillarity, is used. See Childs ( 1 969) or Guymon ( 1 994) for further
explanation. The material that we present is strictly valid for relatively

270
moist soils and rocks and is valid for almost all circumstances with the
extended definition of negative pressure head.)
A suction (or negative pressure relative to atmospheric pressure) must
be applied to withdraw water from the unsaturaled zone above the water
table. The greater the applied suction, the more water is withdrawn, and
the lower is the soil-moisture content when the soil has reached
equilibrium with the applied suction. Our example of a sponge may
provide insight. In the kitchen, we don't have devices to exert suction to
remove watcr from a sponge. We can make an analogy bctwcen exerting
suction and "squeezing" the sponge, however. lf we exert small effort in
squeezing a sponge, a relatively small amount of the water will be forced
from the pores. To paraphrase the sentence referring to exerting suction on
a rock, the greater the applied " squeezing" to the sponge, the more water is
withdrawn, and the lower is the amount of water remaining in the sponge
when "squeezing" has ceased. The relationship between the external
suction applied to a rock and the amount of water per bulk volume (the
moisture content) that the rock retains against that de-watering suction is
called the moisture characleristic. The applied negative pressure is a
measure of the water-retaining forces of the soil and represents the
capillary-pressure head, W. The moisture characteristic generally is
presented as a plot of II' versus (}.
The moisture characteristic for a porous material can be detennined
using a pressure plate apparatus (Figure 8.3). The rock (soil) sample sits
on a porous plate made of a fine-grained material (e.g., a ceramic) that
remains saturated even at high negative pressure heads. The sample is
allowed to equilibrate at a given negative pressure head and the moisture
content associated with this capillary-pressure head is determined. The
experiment is repeated at different negative heads to obtain other points on
the moisture characteristic. The locus of all such points then defines the
moisture characteristic (Figure 8.4). The moisture characteristic is one of
the important curves that define the relationships among hydraulic
variables in the soil-water system. Another is the relationship between
moisture content and hydraulic conductivity.

271
 Soil or rock sample
/
I

Negalive head Ceramic plate

applied to sample (saturated at
(pressu re head) negative applied head)

I
Figure 8.3 A pressure plate for measunng capli\ary-prcssurc head. The
• •

rock sample is placed in contact with the ceramic plate which is saturated
with water under a negative pressure head that is set by the distance of the
free water surface to the right of the diagram below the ceramic plate. The
moisture content of the sample is recorded after the sample has come to
equilibrium with the selected capillary-pressure head. This measurement
gives one point on the moisture characteristic curve.

8.4 Darcy's Law

As water is drained from a saturated soil, the large pores fill with air but
the smaller spaces remain filled with water. The water in the smaller
capillaries is held by surface tension more tightly than is the water in the
large pores. In a moist but unsaturated soil, water flows through the water­
filled pores while avoiding the larger, air-filled spaces. For a given
moisture content, we might conceptualize this flow as if the air-filled pores
were filled instead with a solid (e.g., wax). We would thus have flow in an
equivalent, saturated porous medium and would expect Darcy's law to be
valid, but with the hydraulic conductivity appropriate for the equivalent
medium. This is, in fact, found to be the case; Darcy's law is valid for
unsaturated soil but each moisture contcnt corresponds to a different
equivalent saturated medium, and hence a different value of hydraulic
conductivity. We therefore write the hydraulic conductivity as K(f})
indicating that it is a function of moisture content.

272
 The hydraulic conductivity decreases very rapidly as the medium
becomes unsaturated (Figure 8.5). This can be explained by recalling that
Poiscuille's law (Equation 3.36) indicates that mean velocity is
proportional to the second power of the diameter of a cyl indrical tube.
Hence, as we have seen, the intrinsic penneability of a porous medium is
proportional to the square of the pore size (Equation 6.7) . Because the
discharge is equal to the mean velocity multiplied by the cross-sectional
area of now (7tlJ2) i n the case ofa cylindrical lUbe), the discharge will be
proportional to the fourth power of the tube diameter. The large pores of a
soil become air-filled first as a suction is applied to a soil, thereby
relegating now to the smaller pores which (according to Poiseuille's law)
can conduct water at milch lowerflolV rates than could be handled by the
emptied (larger) pores. This reasoning also can explain why the hydraulic
conductivity of an unsaturated clayey soil may be appreciably higher than
that of an unsaturated sandy soil (see Section 8.10).

0.4

", 0.35
"

"
-

" 0.3-
.

"
-

0
"
" 0.25-
"
�

'"
-

0.2
0
.-

E
"
.-
0.15
�

"
-

E
"
0.1
0
-

> 0.05

o
0.60 0.55 0.50 0.45 0.40 0.35 0.30
Negative pressure head, -\jI (m)
Figure 8.4 Moisture characteristic for a fine sand detennined by starting at
saturation and draining the sample. Note that for this sand, saturation is
maintained for capillary-pressure heads between 0 and -0 .36 m . The
capillary fringe in such a material would be 0.36 m high. As the capillary­
pressure head is reduced from about -0.40 m to -0.45 m, the moisture

273
content drops sharply from the saturation value of 0. 35 to about 0. 1 5 . This
steep drop is Iypical for sandy soils. Much of the pore space is in large
porcs that drain once a critical suction is exceeded. Thc moisture contcnt
therefore drops abruptly. On the other hand, moisture content drops by
only about 0.08 as capillary-pressure head drops from -0.45 m to -0.60 m.

Note that the hydraulic gradient driving flow in unsaturated materials is

the hydraulic head appropriate for unsaturated materials, II' + z:

11 = 11' +;, (8.1)

l D� ,-
---,
-
-

UI
,

10<
E
"
�

z:.
-

>
.-

10·
u
.-

"
..,
I::
0 10·
u
u
.-

"
-

..
.., 10-1-
�

,.,
:I:

lD·i-�---r-----'--r---�--,--1
0.05 0.1 0.15 0.2 0.25 0.3 0.35
Volumetric moisture content, e
Figure 8.5 Variation of K with () for a fine sand. Note that the scale for K
is logarithmic. The hydraulic conductivity decreases by orders of
magnitude as capillary-prcssurc hcad drops from -0. 35 m to -0.55 m
(moisture content decreases from saturation to about 0.06 as seen in Figure
8 . 4).

where t{J = capillary-pressure head and z = elevation head. Note that

capillary-pressure head is a function of moisture content in this expression.
Darcy's law for unsaturated conditions can thus be written:

q= -K(8) d (
dh
(8.2)

274
8.5 Vertical Water Movement

An appreciation of many important aspects of the movement of water in

the unsaturated zone can be gained by considering onc-dimensional flow
in the vertical direction. In this case, Equation 8.2 becomes:

dh
q.
. =_K(8) (8.3)
d,

or, making use of Equation 8 . 1 :

or,

(8.4)

Equation 8.4 alone is sufficient to describe the steady flow of soil

moisture . As with groundwater flow, description of transient or unsteady
processes requires the addition of a continuity equation to Darcy's law
(c.g., see Guymon, 1994). For the case of vertical flow of water the
appropriate continuity equation is:

(8.5)

The left side oflhis equation represents the rate of change of mass in a
small control volume and the right side is the difference between the
inflow rate and the outflow rate, each expressed on a per unit volume
basis. That is, the equation has the same conceptual basis as the continuity
equation that we used in earlier chapters. Combining Equations 8.4 and 8.5
gives an equation that, along with infonnation on the relationships among
0, '1', and K, describes the flow of water in unsaturated rocks . The resulting
equation, referred to as the Richards' equation, is:

Q8 �[K(8)(Q'I' Ill.
(){
=

(); a;
+ (8.6)

For steady flow, the time derivative of moisture content is zero and a
single integration of the right hand side of Equation 8.6 recaptures
Equation 8.4. That is, for steady flow the specific discharge, q_, is constant

275
and can be calculated from Equation 8.4. Of course, Equation 8.4 can be
used to calculate the flux in unsteady flow as well. Under unsteady
conditions, q_ changes with time so the calculation can be taken to give a
"snapshot" of the flow at a givcn timc. Solutions to the Richards' equation,
which accounts for time variation explicitly, yield a complete time history
of heads and fluxes for specified conditions.
A simple example illustrates the kind of calculation that can be done
using Equation 8.4. Suppose that, at noon on a day in September, moisture
content in a fine sand is measured to be 0.25 at an elevation of 3 m above
the local watcr table and to be 0. 1 5 at an clevation of 3.5 m above the
water table. Assuming that the hydraulic relationships in Figures 8.4 and
8.5 hold for the sand in question, we can estimate the direction of flow of
water and the magnitude of the flux. Consider first the direction of flow.
From Equation 8.3 we see that, if iJhliJz > 0, then the flow will be
downward (because the calculated q� will be negative). Conversely, if alii
Oz > 0, then the flow will be upward. We recognize that this conclusion
merely reiterates the main point of Darcy's law, that water flows down a
gradient in hydraulic head. So, for our case, we approximate the derivative
by a finite difference:

all 113 5 - 11) (3.5 m + Ii/,, ) - (3.0 m - li/, )

dZ 0.5 m 0.5 m
- -
,

where the subscripts refer to conditions at the different elevations above

the water table. Evaluation of the expression requires values for "'. From
Figure 8.4, we detem1ine that for a moisture content of 0.25, the capillary­
pressure head is about -0.42 m and for a moisture content of 0. 1 5 it is
about -0.45 m. The calculated hydraulic gradient is 0.94 (positive) so the
water flow is downward. To estimate the magnitude of the flux, we
multiply the gradient by the hydraulic conductivity. We know that
hydraulic conductivity varies over the 3-111 to 3 . 5-m interval because
moisture content varies. We might detcnninc an "average" value of
hydraulic conductivity in different ways. For example, we could find K(f)
= 0.25) and K(fJ = 0.15) and average these. Or we might just use K(f) =
0.20). The first method gives a value for K of about 0.000023 m S-I and
the second gives a value of about 0.00002 111 S-I . Thus, the specific
discharge is estimated to be approximately 0.00002 m S-I .

8.6 The Equilibrium Profile above a Water Table

276
Now consider how we would expect the moisture content profile to look in
a homogenous material above a static water table under conditions of zero
vertical flux. This would be the situation we might expect from the
capillary fringe through a good part of the intennediate zone after a
prolonged period without rain . In this case, Equation 8,4 becomes:

(8.7)

Dividing through by K (we can assume that K is not zero) and rearranging,
we have:

d'l' = -1
d=

0'

(8.8)

This equation can be integrated from the water table (z = 0, �, = 0) to some

arbitrary point in the unsaturated zone (z, �/) so that under zero-flux
conditions:

(8.0)

That is, the capillary-pressure head balances the head due to gravity so
there is no hydraulic gradient and thus (by Darcy's law) no moisture flux.
Equation 8.9 can be used to infer the moisture distribution if the
moisture characteristic (rp as a function or8) is known. Clearly, the
equilibrium moisture profile above a water table has exactly the same
shape as the moisture characteristic because of Equation 8 . 9 . Thus, for a
sandy soil with a very steep moisture characteristic at the dry end of the
curve, the field capacity represents the "nearly constant" moisture content
associated with the steep portion of the curve (Figure 8.6). The moisture
characteristic for a soil having a higher clay content, however, may not
exhibit such a steep characteristic and i n such cases the unambiguous
definition of a field capacity is problematic.

277
 1

EE 08

l> "
- -

, :;; .

"tI _
-
'"
"' �
" " 0.6 �
"' 16
!! �
" "
"' >
"' 0
" .c
5. '" 0.4
" c:
> .2
",
.- -

"' >
-

C»
,, "
Z w
-

"
0.2

o
o 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4
Volumetric moisture content, 8
Figure 8.6 The profile of moisture conlent above a water table under
conditions of zero venical flux of water. The equilibrium profile under
these conditions has the same shape as the moisture characteristic. For fine
sand, the moisture content drops rapidly above the capillary fringe and
remains relatively constant in the range of 0.04 to 0.08 over a sizable range
of elevation. This relatively constant moisture content is referred to as field
capacity.

8.7 The Profile of Capi\lary-Pressure Head as an

Indicator of Flow

Measurements of capillary·pressure head can be used to infer how water is

moving in an unsaturated sail or rock. In humid to subhumid climates, a
tensiometer can be used to measure the capillary-pressure head. A
tensiometer consists of a cylindrical tube, typically PVC, with a porous

278
cup mounted on the cnd (Figure 8.7). The cup, typically made of ceramic
or Teflon, is porous but with rine pores that remain saturated under the
water tensions (i.e., capillaJy-pressure heads) to be measured. The tube is
inserted into the soil, ensuring that a close contact is established between
the porous cup and the soil. The tube is filled with water and tightly
capped . A pressure gage is used to measure the pressure in the water.
Suppose the tensiometer is placed in a soil in which the capillary-pressure
head is -0.3 m. Immediately after the tube is filled with water and capped,
the pressure head inside the porous cup can be assumed to be zero.
Because the porous cup is in contact with the soil, water is drawn out of
the (otherwise sealed) tensiometer through the porous cup, lowering the
pressure in the water inside the tube. Only a small amount of water needs
to be withdrawn from the tensiometer to lower the pressure due to the low
compressibility of water; as the water pressure inside the tensiometer is
lowered, the water expands (slightly) because the porous cup remains
saturated and air cannot enter. Waler will eease to flow OUI oflhe
tensiometer when the (negative) pressure inside the tensiometer reachcs -
0.3 m, the tension in the surrounding soil water. Thus, by allowing a
tensiometer to equilibrate with the surrounding soil and then reading the
pressure gage, we get a measure of the capillary-pressure head in the soil.

Pressure gage

Porous cup

279
Figure 8.7 A tensiometer, consisting of a closed tube with a porous cup at
the end. Ncgative prcssurc, or tension, in thc unsaturated zone is rccordcd
by the pressure gage as watcr in the tube is drawn into the partially
saturated soil.

If a pair oftensiometers is installed in a soil, it is easy to see how flow

direction can be detennined. The example at the end of Section 8.5
illustrates the procedure . Recall that in that example, the capillary-pressure
head at an elevation of 3 m abovc a water table was -0.42 m and the
pressure head at 3.5 m above the water table was -0.45 m. These heads
might have been measured with tensiometers. The calculation then
illustrates how measured tensions can be used with Darcy's law to infer
flow direction.
In some instances, we might want to infer direction of flow using
measurements from a single tensiometer. Is this possible? The calculation
i n Section 8.6 shows that, when there is no flow of moisture in the vertical
direction, the capillary-pressure head decreases directly with depth, If = -z.
Thus, the profile of pressure head with height above the water table is a
straight line (Figure 8 . 8). In the field, if there is no vertical flow, a
measured capillary-pressure head should plot on the straight line
representing equilibrium. Suppose a measurement shows that capillary
pressure at a spot in a soil is smaller than it would be under no-flow
conditions (region A in Figure 8 . 8). In this case we can conclude that
water flow will bc upward ifflow is steady. If measured capillary prcssurc
is lower than the equilibrium value, the gradient in hydraulic head relative
to the water table will be upward and so we expect that water flow will be
upward because water is being drawn up from the water table. Similarly, if
measurcd pressures are greater than equilibrium values (region 8 in Figure
8.8), wc can reason that watcr is flowing downward away from thc point
of measurement (assuming steady flow).

8.8 The Infiltration Process

AI depth in a soil profilc (roughly in thc "intcmlcdiatc bclt" of Figure 8 . 1 ),

the analyses based on the equilibrium profile as presented above are often
very useful because the steady-flow approximation is reasonable.
Recharge 10 an aquifer through a thick unsaturated zone, for example, can
be approximatcd reasonably as a steady flow process. Many important
hydrological processes involve water fluxes right at the soil surface,
however, and in this area (the "soil-water zone" of Figure 8. 1 ) changes in

280
seil muisture ean be rapid and large. One ufthe inest impertant efthe
surtaee hydrelegieal preeesses is infiltratienj the meyement et‘ rain and
melting snew inte the sail. The preeess elitttiltratien at the sail surfaee is
inherently unsteady and a rigureus treatment en" the preblem usually is
apprrtaehed thrdttgh the use dt'Riehardsi equatimt, Equatimt 8.6. A
simplified treatment etitlte preblem. first presented by Green and Ampt
(It‘ll 1}. illustrates hew seil physieal prineiples ean be used te understand
the infiltratiun preeess. The Green—Ampt equatien will he eutlined after a
physieal deseriptinn et‘the int‘iltratidn preeess is sketehed.
The maximum infiltratiun rate ledmmdnly ealled the infiltratiun
eapaeity} dt'a sell is the rate at whieh water will mnye yertieally
duwnward when a supply uf water at aeru (gage) pressure is maintained at
the surlaee. The inliltratien rate in a dry seil is initially yery large. Eyen
theugh the hydraulie eenduetiyity is law. the eapillary pressure gradient is
estretnely large heeause the sail unly a few millimeters belew the surfaee
has a yery law {and negatiye) pressure head. Fer example, if the eapillary-
pressure head 10 mm beluw the surtaee is initially —2.0 m. the finite-
dil‘t‘erenee appresimatien ef the gradient at the surfaee [assuming that the
surt‘aee is at the saturatien meisture eentent} weuld he (0.0] m I 2.0
n't}.-"'U.Ol m = Ell]. Eyen with a hydraulie ee-nduetiyity at l [l ti m s", the
intiltratinn rate writtld he ahnut 0.0002 ttt s" 2 TED mm hr". A rainfall rate
dt‘TEfl mm hr‘i wmtld be quite intense! The intiltratinn rate deereases itt
time as infiltrating water muistens the surfaee layers and red tie-es the
gradient in eapillary-pressure head. Ultimately. the nearwsurt‘aee pertien el‘
the sail appreaehes saturatien, the eapillary—pressure head gradient
appruaehes aere and Equatinn 8.4 indieates that the infiltratidn rate
apprttaehes the saturated ednduetiyity {Figure 51.9]. The infiltratidn-
eapaeity eurye Fur any giyen seil. ufeettrse, wettld depend en the suil
preperties heeause the eapillary—pressure head gradient and unsaturated
yalues et‘ hydraulie eunduetiyity guyern the time prugress et' infiltratien.
�
.. . �: : . . ' .. ... . " .. " . . ..: ,. . . . . . . . : .. . . .. . . . . . , .... .. . .. . .. . . '' :",
' . " , "
. .
.
: '
.

: .
"

':
' : .
.
"
..
:
"

': . "

..
'

";:}.��. {.. :,/;.:. :.:;./�,.:..::..:.. /.. .'. ":';,: ,..:.-;:./. '.:::. , : :�:. ; ; ::,.:..::..:.-.- :/:". ,/.:.:. :. ; ";:.:,.->/:.:>. :.
: :
.

.
. .
"

"'
: :
,
.

:
. .

. . .
.
. .

"

� .
. �. ' " . . . .
. 8
.. . . . . . .. . . . ..
: : :
'

'
.
'

· .
' . '
. .. :
. .. .
.:
.. . ;., : : . ' .. .. : . . : .
.
. .

. . .
.

. . . ..
.
.
. . . . .
' '.. . :' . . . . ; ' . . : .. ' ... :' . . ; .'. . : : ..: . '.. . :' . . . ; .'.
.
.
. " . .

. . :. .:.: ... : . . .� . . . . . . . . . . . . ,. .
"
'

. . :: : : : ,...: ; . . ... . : ; :: : . : :. . ... . .:: ::: : ..: : - . . . .: : :: : : . . . : . .. . : : :: : :

.

.:
. .

: :. : . . : . . . : : : . : . . : : : : . : : . . . . : . :. . : . . : .
, . " , . , .

. .
. .

. .
.

.
: : , " : : : · ·· · ::· : �' : · ·: ' ' f:t· · · · : . . : : . . .: . .. : , : ; : : . : < : . � '. ' : . . : : . <
. . .
. .
. . .

. .
.
. .

. . : . .N h. . . . . . . . . . .,
: .

· .. ' ' . . .. . . . " . . .. ,. . ' .

o ow .. . .. . .. .. . .. " ' .. . . ... . . .' .
.
.

. : . : . " ': . : . ' . '. .

'

.
.. . . . . .
. . .
.

. : . : ': ' : > : . : ., : ': ' : ' ' :: ... .. : . : .: - :: - : : ': :

. .
.
. " .

.
.
.
' .
.

. :

>.. ' .:.'. · : .·w. en ·· . · ·:. · :.·: .· ." . > . . .. :.. : .. ." .. . >. . ' .:. . : .. : ..
'. " . .

. . .
"

' .. . .· ·· . ·. ·�.,
. " " , . " .

'
. -

: ; : : , : : :' :: ': ... , . : :: ' . : : ': : ', ' . : : :' : : : : ': ... . : :: :
.
.
.
.
.

. "
.

. .
. ':-: : .: :. :. :S. · : .ll/ . ". · .:....:.z· : . . . : : ': ' . .: : :. : : . . .: : .. ..: .: :. : : . .
'

"
. .

.. . ..
.

'.. . .
· '

.
-

:
' . .
.

... .
.

.
. . . . . . . � . ."f . . . . . . . . . .. . . .. . . . . . . . :
.

. . .. . . ..
.
': '. :. . :. . " '.'... . :: ' :. . : . : :'. ' . . ' ::'. :. . :. .. : '. ' . ' :: ' :. :. : ..
'
' .

. . .. .. . . . .. .
. . . . . .. . . . ' .
. .
.

. . . . . .
; :: : . : " . . .. . . . :: ': . .
. .
: . .'. : . : :: ' . ;.:: : ::: . . "" . :: :
. . .
' .
. . . . .
.
.

. . . ' . . . . . '.
.

. .
. .

. . .
.

:
'

' ' . . '

.

'
.

' . '. ' . . ' .

. ' . . .. ' . ' . . .. . � ' . ' '
.

:
.

. :': ::. : : :::.:.: . . . :': : :. >: : ::. : . ...... . �. : : . ."". :. '. :.: ... . :': ::. >: : ::.. : ::. . :': : . >: : .
. .
.
.
.

. .
.

.
.
.

.. . ' . . ""
.
.

·. . . . . " . . .. . ' . . ' .. . . .

.

. . . .
. . .. .. . . .\Jf. . . . .: . . .
.
..
.
·

>.:.: ." ::." : :' : . >.:.: ." ::: : :' : . : -' : " �, :: ' : : ' : . >.:.:." ::: : :' : . : -. : .: ." : " :
'
"

. . . . . .. . . .. . . . . . .. . .
.

.
. . .

. .
.

. . . .. : .. : .
. .

. ' .. � .: .. : ' . , . .. . : . : . -'- . . :

.

.
'.: ' .. : . . .. . .. . : : . . , ' .. . : . : . : '
.
. . : : . '.: ' . ';" . .. .: :
. .. :. . .. .. ". . 7
.

. > : .
.
"
Q . .' . . : '. . ".'
.

. . ' . ' " . ' . .. .

'

· . ,

'
. .
.

. ' "
.

,
. " .

. .
.
.
.
'

,
,
,
,
,
,
,
,
,
,
,
Figure 8.8 The vertical profile of capillary·pressure head under conditions
of no flow is a straight line (w = -z). If a measurement of capillary·
pressure head falls on this line, it is reasonable to infer that water flow in
the vertical at that point is negligible. If measured capillary pressure is
smaller (region A) than it would be at equilibrium (for a fixed value ofz),
it is rcasonable to infcr upward movemcnt of water at that point.
Conversely, if measured capillary pressure is greater (region B) than it
would be at equilibrium, downward flow is indicated. The inferences are
exactly correct under steady flow conditions.

282
 1.6

1.6
Infiltration
-

1.4 capacity curve

c
-
,

E 1.2
.-

E
E 1
-
Infiltration of rain at a steady
.. rate of 1 . 1 5 mm min-1
..
-

c
�
0.6
0
.-

.. 0.6
-

�
-
-

c
.-

-
0.4

0.2

a
o 10 20 30 40 50 60 70 80
Time (minutes)
Figure 8.9 Infiltration into a sandy loam. When water ponds at the surface,
the infiltration ratc (blue curve) starts out aI a very high value because of
the large downward gradients in capillary-pressure head and decreases
steadily toward the limiting value of the saturated hydraulic conductivity
of the soil, a rate that would be achieved when the downward gradient in
hydraulic head was due only to the elevation head. When water is
sprinkled on a soil surface at a relatively slow rate, the infiltration rate will
equal the sprinkling rate initially. Ifthe rainfall rate exceeds the hydraulic
conductivity of the soil and if the rainfall persists for long enough, the
ability of the soil to infiltrate water will drop below the rainfall rate after
hydraulic gradients near the surface become reduced by the wetting
process. The time at which the infiltration rate starts to drop below the
rainfall rate is known as the time to ponding. This occurs to the right of the
infiltration capacity curve because the initial rates are much lower for the
sprinkling case than for the ponding case; approximately the same total
volume of water must infiltrate in the sprinkling case as in the ponding
case before the infiltration rate starts to drop for the sprinkling case.

Now consider the case of a steady rainfall beginning at time t = O. [n

Ihis case, the rate of delivery of water to the surface is usually insufficient

283
to cause initial ponding of water, that is, the infiltration capacity exceeds
the rate of supply . Initially, all of the water reaching the surface infiltrates.
Providcd thc rainfall rate is largcr than the saturatcd hydraulic conductivity
and that rainfall continues for a long enough time, the infiltration capacity
will decrease continually as the soil becomes wet. The infiltration capacity
ultimately will be reduced below the rainfall rate. Subsequently, the rate of
infiltration will be controlled by the rate at which the soil can transmit
water and "excess water" will accumulate at the surface (Figure 8.9). This
excess water will flow rapidly over the surface to streams and can be an
important process in generating stormflow in streams and rivers, as is
discussed in Chapter 1 0.

8.8.1 The Green-\mpt equation

Observation of infiltration into dry soils, especially sandy soils, indicates

that the water tends to progress downward as a "slug." In other words, a
sharp wetting front separates the unsaturatcd soil below from the saturated
soil above and it is this front that progresses downward as infiltration
proceeds. Green and Ampt ( 1 9 1 1 ) presented an analysis of the infiltration
problem assuming that this wetting front was infinitely sharp, that is,
horizontal (Figure 8. 10). In this case, the flux evcrywhere in the saturated
upper portion must equal the infiltration rate and so the hydraulic gradient
is unifonn. Darcy's law then gives:

284
 +

=0
,
,
,
,
,

LJ. dep th 01 wetti n g

ront t ti me
:t-
.
... t.e _

: :r-:-:--:-:--:-:-� J l � t . . o .
. " . . . .. .
' : :-. " : .. . . ... . :.:... :: .... IjI = IjI'J :
. ". . . . . ' . . . : : :
. .:.> :.� ':.
'-. -
. '. . . ., .
:�. " " '.
' .
0 : : ' "" , : " . . " , .' .. .. . . . :..' : .: : :
.

.
. ..
,

<:: :: . . . . . :::: :. . '. .

:

. . . : . Moistu re profile · · . . : . : : :<:

,

.. .
, ,
.

>
.

.
. .

.
-. . . .
, ,

. .
,

. ..
•
. . . _

.. . . '.", at t,· m e I The so,· 1 . . ... .. . ... ..' ; .. .' : " �.. . . . . " >:
. . . . : ::. . . . ; ..
." : .
.
. z, ·
· ·; s
". : . ... . . . i i n itially d ry .
, : . ;

: . ; . .. .. : . : ..
,

. . " ' .: . .
.. ..

' ::>
:

.>:: . :
. . .. " . .
" . . . . . ..." ' .. . " :. .." " " .:".
. " .
-

:
.

:
..
.

" ' . ."

' . ,
. -
, .. : , ': , , ' , : : : ' :
' • - • • • • • • • • •
• - < , • • • ,
• .

Figure 8.10 Schematic diagram of wetting front movement in the

unsaturated zone, as conceptualized in the Green-Amp! model. The basic
concept is that the soil surface is held at saturation and thai water moves
downward through the soil as a sharp "wetting front." Darcy's law can be
applied under these idealized conditions to give an equation for calculating
the rate of infiltration with time.

i=-KS( -Vt+ O + I} (B.10)

where i = infiltration rate (equal to the specific discharge), V'f= the

capillary-pressure head at the wetting front, Lf = the depth to the wetting
front, and Ks = saturated conductivity (Figure 8 . 1 0). Equation 8. 1 0 can be
written:

(8.11)

Furthermore, for the type of flow envisioned by Green and Amp! (often
referred to as plug or pistol/flow), the total cumulative amount of water
infiltrated, /, is equal to the product of Lf and /J.O, the difference bctween
the saturated moisture content, Os, and the initial moisture content, 0i"
Then, because the infiltration rate, i, must equal the negative of the time

285
derivative of / (negative flux means downward motion),

i = - til =_60 tlLr (8.t2)

dt dt
and Equation 8. 1 1 bccomes:

60
tlL, _ (-'I1- + L, )
-- K\
. . (S.U)
dt Lf

This equation can be integrated from t = 0 to an arbitrary time, t, to give:

K I= [ L1
�
:..c
.' L + !/f ln l + 1 .
M " (-",, )
(8.14)

or equivalently

(8.15)

The above equation is an implicit relationship between the cumulative

infiltration, I, and timc, t. Thus, from known soil parameters (K., !J.() and
!/ff) an infiltration capacity curve similar to that shown in Figure 8.9 can be
constructed. For example, for Ks = 7.0 x 10--{i m s-1 , M) = 0.04 and Iflf= -
0.106m, thc Grccn4Ampt calculation produces thc infiltration curvc shown
in Figure 8.9 (Clapp, 1977). These parameters are appropriate for a sandy
loam soil (for which the assumption of a piston41ike wetting front should
be reasonable). Comparison of the Green4Ampt results with results from a
more complcte theory indicates that the calculated values of infiltration are
sound from a theorctical standpoint.
Even for non-sandy soils, the form of the Grcen-Amp! infiltration curve
can reproduce observations fairly well. Consider data from measurements
using an infiltrometer made on a loamy soil in central Virginia. A ring
infiltrometer is a simple device to measure infiltration rates. A metal ring
is insened into the ground and water is added to maintain a level pool of
watcr in the ring. Thc amount ofwatcr required to maintain thc level is the
rate at which water enters the soil surface, that is, the infiltration rate. Data
recorded are typically time (t) and cumulative depth of water infiltrated (I).
Equation 8.1 3 can be used to calculate an / - t curve for given values of Ks,
I{If, and !J.8. The computation is accomplished by selecting a suite of values
of / for which calculations are to be done and then using 8.13 directly to

286
calculate a corresponding value for t. For the infiltrometer data, the Green­
Ampt equation filS the measured data reasonably well with Ks = 0. 8 mm
min- I , 11'/ = -250 mm, and 110 = 0.1 (Figure 8.1 1).
The Green-Ampt method assumes that the soil column is homogeneous
and the water table is deep in a way that the wetting front does not reach
the water table during a rainfall event. In the presence of a shallow water
table or of a bedrock at a depth Ls' infiltration water may saturate the soil
column . Rainfall onto saturated soil consequently generates runoff, which
is not due to limited soil infiltration capacity but to limited soil storage
capacity, as discussed in Chapter 1 0. The soil column becomes completely
saturated when the cumulative infiltration I is equal to the initial soil
storage capacity 6,(} Lr Thus, the time (since the beginning of the
rainstonn or snowmelt event) when the soil becomes saturated-and
runofT is produced regardless of the soil infiltration capacity-can be
calculated using Equation 8. 1 5 with 1 = 6,(} Ls.

60 ,-------,

50
.!i
�

� 0
oE4
.r: E
- -

Q."
Gl Q) 30
,, -
� �
-
>-
.- -

� -
f: 20
.-

� .-
_

E
�
(.)
10

o ,
o 5 10 15 20 25 30 35
Time from start of infiltration (min)
Figure 8.11 Cumulative infiltration measured for a loamy soil in central
Virginia (blue circles) and calculated using the Green-Ampt equation
(black iiI/e).

287
8.9 Field Measurements in a Soil

Except in the most intense rains!onns, calculations of infihration based on

an assumption of saturation at the soil surface (e.g., the Grccn-Amp!
approach) are not considered sound. Numerical solutions to the Richards'
equation (Equation 8.6) can be used to gain insight into how moisture will
infiltrate a soil under various rainfall rates. The physical principles
developed so far in this chapler serve as a base for understanding processes
that occur in the field, but added complications do arise.
Let's look at some data that were collected in the soil ofa forested
catchment near Orono, Maine, USA. A sprinkler was set up at the soil
surface to simulate rainfall. Capillary-pressure heads were measured using
tensiometers (see Section 8.7) and moisture contents were measured using
the time domain reflcctomctry (TDR) tcchniquc (Figurc 8. 12).
Measurements were taken in an organic soil horizon near the soil surface
(approximately 50 mm below the surface) for a period of9 hours. The
sprinkler delivered water to the soil surface at a rate of 1 1 mm hr- l for the
first 4.5 hours and was subsequently shut off.

Soil surface

TDR probes

Figure 8.12 , (TDR)

288
device for measuring soil moisture.

The soil started at a capillary-pressure head of about -0.40 m (Figure

8 . 1 3). Between 1 5 minutes and 45 minutes after the initiation of
sprinkling, capillary-pressure heads rose rapidly to approximatcly -0. 1 5 m,
indicating passage of a sharp front. The capi lIary-pressure head continued
to increase slowly for the rest of the duration of sprinkling. This makes
sense given the principles that we have developed in this chapter. As water
infiltratcs the soil and moves downward, it fills pores that had been
occupied by air and capillary pressures increase.
Measurements of soil moisture confinn this general picture (Figurc
8. \3). The moisture content is initially about 26% and rose steadi ly from (
= 0.5 hr to t = 4.5 hr. The moisture content did not rise as rapidly as did the
capillary-pressure head, however, but after sprinkling ceased, the moisture
content and the capillary-pressure head appeared to change together in a
more direct fashion. This is not what we would expect from the previously
postulated straightforward relationship between moisture content and
capillary pressure, the moisture characteristic. What does the moisture
characteristic look like for the soil in the sprinkling experiment? We can
answer this question directly by plotting the measured valucs of moisture
content against the measured values of capillary-pressure head.

e 40
,
i;i -o
.- r
0 -- 35
E O:
u -
c v
" 0
30
._

0 -
E S
-
' u
/
g 25 , , , , , , ,
o 1 2 3 4 5 6 7 B 9
0 -------,
o
,
-
'.
::le-O· .1 . t-
e --
o.�-02
�,,� .
-
- -
0 -0' 3 ':
a�
il
-O.4 .f"
0
.t: -
1
-,--
-
2
.,--
3
---,--
-,--,--
--
5 6 7 8
--1
9
4
-

Time from start of sprinkling (hours)

Figure 8.13 Capillary-pressure head (bottom) and volumetric moisture

289
content (top) in a soil during a sprinkling experiment. Sprinkling at a rate
of I I mm h..-1 was started at time zero and continued for 4.5 hours.
Subsequent to 4.5 hours, there was essentially no flux at the soil surface
and water drained vertically downward.

The moisture characteristic for the Maine soil shows a hysteresis loop
(Figure 8.1 4). Initially, the soil is relatively dry (Y' is approximately -0.40
m, (} is approximately 26%). As the soil wets up from the sprinkling, Y'
rapidly increases to about -0. 1 6 m. As this change takes place, (} increases
relatively little, from 26% to 27%. The next phase of change is a relatively
slower increase in VJ to about -0. 1 5 m, but a correspondingly large
increase in (j to about 38%. After sprinkling ceases, the moisture content
and capillary-pressure head covary along a different characteristic curve, a
branch above the curve defined by the wetting of the soil. Hysteresis refers
to a phenomenon whereby different paths are followed on the VJ (j plot
-

depending on whether the soil is wetting or drying.

A conceptual appreciation of how the hysteresis phenomenon arises can
be gained by extending our capillary tube analogy. Consider a capillary
tube with a bulb-shaped expansion in it (Figure 8. 1 5). (This capillary-tube
analogy to explain hysteresis is often called the "ink-bottle effect" because
the shape of the capillary tube is reminiscent of the tube that was used to
transfer ink from a bottle to a pen in times past.) Suppose that the small­
diameter portion of the tube can hold water against an applied capillary­
pressure head of -0.40 m and that the large-diameter portion of the tube
can retain water against a pressure head of -o. I S m. Consider the
hwetting" of the tube from below depicted in Figure 8.15a. The water
cannot cross the large-diameter portion of the tube, which holds much of
the water in the tube, until the capillary-pressure head at the base of the
bulb-shaped expansion reaches -o.IS m. Note that the "moisture content"
in the tube will remain low until the bulb is filled. So with this analogy, as
water enters the tube from below, the pressure head increases from 0 40 - .

m to -0. 1 6 m with relatively little change in "moisture content." The

"moisture content" increases suddenly as capillary-pressure head reaches­
O . I S m at the bottom of the bulb, however, because water can then enter
the bulb-shaped portion of the tube. Conversely, if the tube is full (Figure
8. ISb; high "moisture content") and capillary pressure at the base of the
tube is reduced to drain the tube, the bulb-shaped portion of the tube will
not drain until the pressure head decreases to -0.40 m at the top of the
bulb. This is because water is held in the small-diameter part of the tube
above the bulb and water can be retained in this upper part of the tube.
This analogy indicates that for this particular capillary tube, the "moisture

290
content" will be higher for a given applied capillary-pressure head when
the tube is drajnjng than it will be for the same capillary-pressure head
when the tube is imbibing water. This is exactly the fonn of the hysteresis
loop observed for the Maine soil.

'"
0

'"
-

c
"

" . §'�
c �(;'
-

0 <
>'
u
"
"
�

.<II
-

0
-

E
u
.-

"
-

E ng
" '/'l etti
r

0 ---
-

>

-OAO -0. 5 -
0 10
Capillary-pressure head, \If (m)
.

Figure 8.14 The moisture characteristic of the organic horizon of the

Maine soil detennined from a sprinkler experiment. The hysteresis loop in
the curve is evident. As the soil imbibes water during the sprinkling, VI and
() covary along the bottom part of the loop. After sprinkling ceases and the
soil is draining, VI and () covary along the top part of the loop.

291
 (a) Inbibilion (b) D raining
Figure 8.15 Hysteresis in a capillary tube with a bulb-shaped expansion.
As shown in a, the "moisture content" oflhe tube remains low during
"imbibition" of water, because water cannot entcr the large-diameter
portion of the tube until the capillary-pressure head increases to the value
associated with the large diameter. On the other hand, as shown in h,
during "draining" the "moisture content" remains high until the capillary­
pressure head drops to the critical value for the small-diameter tube.

Hysteresis is not the only complication that arises in considering flows

in the field. Real soils are heterogeneous. As onc example, we can
consider vertical changes in soil properties. Different soil horizons (loosely
speaking, "layers") typically have different hydraulic properties. For the
sprinkling experiment in Maine, capillary·pressure heads were measured at
four different depths in the soil, in four different horizons. Considering the
observations, we see that the smooth progress of a wetting front downward
through the soil as envisioned in the Green-Ampt model is interrupted
(Figure 8.16). In Figures 8.13 and 8 . 1 4 above, we looked at the Oa horizon
near the soil surface. The Bhs horizon is the next deepest. The capillary­
pressure head builds up in the Bhs horizon from about -0.45 m to slightly
positive values. Positive pressure heads indicate that a saturated layer had

292
to build in this horizon before water "pushed through" into the next lower
horizon, Bs2. Clearly there was an impediment to flow into the Bs2
horizon that caused watcr to "back up" until positive pressures increased
the hydraulic gradient (and hence the flow) into the layer. In the C horizon,
the positive pressure heads indicate that a water table developed with a
saturated depth of 100 mm above the tensiometer. The changes in
hydraulic properties with depth influencc the dynamics of vertical soil­
water movement in layered soils. Interestingly, the "backing up" of the
water above the Bs2 horizon is probably because the Bs2 horizon is
coarser than the Bhs horizon. The resistance to flow in unsaturated
materials provided by coarse layers of sediments (described further in the
next section) is an important but somewhat non-imuitive phenomenon.
0.2

0.1
•

C horizon (0.8 m depth)

- • • • • -
_

E
. •
•
. • •

·
•
•
•
. . .
• •
•
-
• -

�
. .
- •
•
• - - -
. .
. .

'.
•
.
.

0.0
• . .
. . ..
�. . . . . .
. .
. . . . • .
• . • . . . . -

'C
.
� • •
• •

'"
• •
•

" Bhs horizon (0.2 m depth)

• • •

• • •

�
• •

',' .
•

.
• •

-0.1
• •
. •

CI.I
.

Oa horizon
. . •

(0.05 m depth)
• •

"
- • •

'"
•
•
• •

'" ·
•

·
-0.2
•

CI.I
•
•
• -
•

Q.
-
•

..
•
•
• .

� ·
•

.
. .

·
•
- -
-

�
•
. -

-0.3
•
•

Bs2 horizon (0.5 m depth)

•
-

Q.
•
.-

'"
•

·
•

U
•

-0.4
•

•
·

-0.5
•
. .

o 1 2 3 4 5 6 7 8 9
Time since start of sprinkling (hours)
Figure 8.16 Measured pressure heads in four horizons ofa forest soil in
Maine for a sprinkling experiment. Sprinkling began at time zero at a
constant rate of I I mm hr-1 and stopped after 4.5 hours. The lag in the
pressure response at the various depths is due to the time of travel of
moisture vertically downward in the soil. The pressure builds to slightly
positive values in the Bhs horizon, indicating that moisture flow into the
underlying Bs2 horizon is impeded. A water table builds up in the C

293
horizon because the base material is a glacial till with very low
penneability.

Another example of soil heterogeneity is the nearly universal presence

of macropores. Beven and Germann ( 1982) considered macropores to be
those openings in a soil that are large in relation to those in the
surrounding soil, such that the movement of water, once initiated, may be
much faster than within the surrounding soil matrix. The statement above
regarding the initiation of flow in macropores deserves further explanation.
These large pores may be thought of as large capillary tubes, and therefore
will tend to drain at (negative) pressures close to zero. If we imagine
rainfall encountering a dry soil, water will tend to be drawn by capillary
forces into the smallest pores first. The flow of water in macropores begins
only when saturation occurs locally at portions of the soil surface.
Furthennore, water may flow quickly through a macropore, and encounter
drier soil at some depth below the general wetting front. At that point,
capillary forces may once again cause water from the macropores to be
drawn into the surrounding small pores.
Soil macropores may be produced by both biotic (animal burrows,
earthwonn channels, plant roots) and non·biotic (soil pipes, shrinkage
cracks) processes. Their importance in the movement of water within
hillslopes continues to be an important area of study in catchment
hydrology.

8.10 Evapotranspiration from the Unsaturated Zone

As indicated above, moisture content in the unsaturated zone can be quite

variable in time. Infiltration of water during precipitation or snowmelt
events increases the soil moisture. Afterward, drainage of water downward
toward the water table and movement of water upward out of the soil due
to evaporation and transpiration decrease soil moisture. The amount of
water stored in the unsaturated zone therefore changes with time. A
detailed water balance for the land surface requires an accounting of
inflows, outflows, and rates of change of storage in the unsaturated zone.
Measurements that are used to keep track of moisture dynamics in
surface soils include several that we have discussed. Capillary-pressure
head can be measured with tensiometers (Section 8.7) and moisture
content can be measured using time domain reflectometry. It is quite
difficult to separate drainage rates from evapotranspiration rates using only
measurements made in the soil. One experimental technique for measuring

294
evapotranspiration involves the usc of weighing Iysimeters. A weighing
Iysimeter is essentially a large pot, filled with soil, and mounted flush with
the soil surface. A device to record the total weight of the Iysimeter is
placed beneath the Iysimeter. Plants arc allowed to grow in the Iysimeter.
The amount of precipitation that infiltrates through the surface is recorded
carefully as is the amount of water that drains from the base of the
Iysimeter. Changes in the water content within the Iysimeter are
determined by changes in the total weight. All of these measurements on a
Iysimeter can be done with high accuracy. Evapotranspiration then can be
computed directly from Equation 1.7 because all tenns except et are
known.
Rates of removal of water from the unsahlrated zone by
evapotranspiration are controlled by a number of factors. One of these
controlling factors is the wetness of the soil itself (Figure 2.1 1). Typically,
if a vegetated surface is supplied with plenty of water (e.g., a well-watered
lawn), evapotranspiration will be controlled by atmospheric conditions.
That is, evapotranspiration will proceed at the potential rate, which is a
function of solar radiation, wind speed, humidity, and so forth. I f the lawn
is not watered and if there is a prolonged period without rain, soil moisture
will decrease (outflows exceed inflows). Evapotranspiration will proceed
at the potential rate for some time, but ultimately the rate will drop. Why
does the rate drop when the atmosphere is capable of taking up water at the
potential rate? The explanation stems from the soil physical principles that
have been covered in this chapter. As water is pulled from the soil ncar a
plant root, the moisture content in the soil surrounding the root decreases.
Decreases in (} lead to large decreases in K. Therefore, by virtue of Darcy's
law, to maintain a steady flow of water to the plant root, the plant must
exert ever greater suctions (ever more negative capillary-pressure heads)
so that increases in the hydraulic gradient counterbalance the decreases in
K. At some point, the plant cannot sustain this battle with a drying soil and
the transpiration rate falls below the potential rate, as discussed in Chapter
9. The plants regulate the transpiration rate by adjusting the opening of the
stomata on the leaves (sec Section 2.4).

8.1 1 Capillary Barriers

Flow of fluids through unsaturated media often results in counterintuitive

behavior. One very important example is that flow through coarse material
is impeded relative to flow through fine material under unsaturated
conditions. That is, under unsaturated conditions, water flows more readily

295
through clayey soil than through gravel! The key to understanding how
gravel layers can be capillary barriers to flow of water in the unsaturated
zone is appreciating that gravel has almost all "large" pores. Under
negative capillary-pressure heads, these large pores fill with air and
essentially stop the transport of water. (Water actually does move by vapor
diffusion, but this is an exceedingly slow process.) Clays, on the other
hand, have almost all "small" pores. Thus, under anything but extreme low
(negative) capillary-pressure heads, many of the pores will be filled with
water and will conduct water, albeit at slow rates (relative to saturated
gravel, but at very fast rates relative to vapor diffusion). The situation can
be appreciated by looking at the hydraulic conductivity curves for gravel
versus a clayey material (Figure 8 . 1 7). At saturation, Kgravcl » Kclay, but at
moisture contents not too far below saturation, Kclay » Kgravel' That is,
flows are impeded in the gravel relative to the clay at intennediate to low
values of the moisture content.
The recognition that coarse materials serve as barriers to flow in
unsaturated regimes has several practical applications. In desert regions,
cobbles can be used to mulch agricultural fields. During intense
thunderstonns (the usual method of delivery of water to the surface in
semiarid regions), the gravel layer is wet and readily allows water to
infiltrate. Once the surface gravel dries out in the following fair weather,
however, the evaporative flux of water from the surface is essentially
prevented because the hydraulic conductivity of dry gravel is very low.
The infiltrated water is available for use by plants for transpiration.
Another application is the construction of barriers to flow into waste
disposal trenches. Ifwastes are placed in the vadose zone, a gravel cap is
placed over the trench, and fine material is filled in over the gravel, the
gravel will act as a capillary barrier and deflect the flow of infiltrating
water through the fine material and away from the waste trench (under
unsaturated conditions).

296
 K
(log
scale)
•

Decreasin.Q pressure head

Figure 8.17 The hydraulic conductivity of clay is orders of magnitude less
than that of gravel at saturation but is orders of magnitude larger at low
values of the moisture content.

8.12 Concluding Remarks

Understanding the mechanics of flow of soil moisture is prerequisite for

dealing with many problems in physical hydrology. The process of
redistribution of moisture following an infiltration event, percolation of
waler to a groundwater table, movement of soil moisture in response to an
evaporation demand, etc., are all important aspects of the field water cycle.
Furthermore, solutions to "applied" problems, such as determination of
optimal irrigation rates or siting of septic fields, also depend on a
knowledge of flow in the unsaturated zone. As one example, let's
reconsider the proposed LLRW disposal site in Ward Valley, California.
(Note that the Ward Valley site no longer is under consideration as the
California legislature cancelled the project in 2002.)
Ward Valley is in the arid Mojave Desert. Annual rainfall averages less

297
than 150 mm. The site is on a broad alluvial surface some 1 7 m above the
calculated level of the 100-year Ilood in nearby Homer Wash. The alluvial
and basin-fill deposits at the site arc about 600 m thick with the water table
about 225 m below ground surface.
In 1993, after contractors for the state of Califomia had spent more than
two years collecting data to characterize the proposed disposal site and
were ready to apply for licensing, several geologists questioned the safety
of the site for disposal of radioactive wastes. One of the concerns was the
potential for infiltrating waters to dissolve the wastes and recharge the
groundwater beneath the site at rates that would be too fast to allow
enough time for radioactive decay to render the wastes hannless. How can
the claims of (fairly rapid) water movement in the unsaturated zone be
evaluated? One set of measurements that would be useful is the vertical
distribution of capillary-pressure heads. Recall our discussion of the
equilibrium moisture profile above a water table. We concluded that, if
capillmy-pressure head declined with height above dalllm to exactly
balance the increase ill head due to gravity, there would be no movement
of water. That is, if the vertical gradicnt in hydraulic head is zcro, Darcy's
law indicates that Ilow is zero. Conversely, if the gradient in hydraulic
head is in the downward direction, recharge is indicated and if the gradient
is upward, discharge (due to evapotranspiration) is indicated. A committee
of the National Research Council concluded that the data for Ward Valley
were insufficient to estimate the gradient, but that data from similar arid
sites show upward gradients (NRC, 1995c). The committee concluded that
the dry state of the materials at the Ward Valley site did indicate that
vertical flow rates would be very slow. Detailed assessments of potential
waste-disposal sites in the vadose zone require careful measurements and
interpretation in light of the theory outlined in this chapter.

8.13 Key Points

• Pore spaces in the unsaturated zone may be partly fillcd with water and
partly fil led with air. The volumetric moisture content, e, is a measure of
the amount of water held in a soil or rock. {Section 8.1 }

• Moisture content in the unsaturated (or vadose) zone is quite variable

near the surface where interactions with the atmosphere are strong. In
particular, plants take up water from this zone of soil water. Where
unsaturated zones arc decp, below about I to 3 meters, the tcmporal
variability in moisture content caused by atmospheric changes are

298
 dampened. Water flows in this intennediate zone are approximately
steady. Near the water table, a saturated zone develops due to capillary
risc of water in the pores of the soil or rock. This saturated zone above
the water table is known as the capillary fringe. { Section 8. 1 }

• In unsaturated soils, negative pressure heads develop because ofcapillary

forces acting on curved air-water interfaces . {Section 8 .2}

• One of the important relationships needed to understand moisture

dynamics in unsaturated materials is that between capillary-pressure head
and moisture content. The moisture characteristic defines this
relationship. {Section 8.3 }

• The gradient in hydraulic head drives flow in the unsaturated zone, just
as it does in the saturated zone. In the unsaturated zone the hydraulic
head is defined by II = If + z, where If is the (negative) capillary-pressure
head. {Section 8A}

• Darcy's law for flow in the unsaturated zone has the same fonn as it does
for flow in the saturated zone. For unsaturated flow, hydraulic
conductivity is a function of moisture content, so q = K(8)(dhldl).
-

{ Section 8A}

• Darcy's law can be used to infer the direction of water flow and the
magnitude of the specific discharge. If the gradient in hydraulic head is
positive, flow is downward. If the gradient in hydraulic head is negative,
flow is upward. The specific discharge is the product of K and the
gradient in hydraulic head. {Section 8.5 }

• For the case of zero specific discharge in the vertical in the unsaturated
zone above a static water table, the gradient in elevation head must be
exactly balanced by the gradient in capillary-pressure head . Thus, If = -z
under these conditions. This indicates that the equilibrium profile of
pressure head above a water table is a straight line upward to the left at
45°. The equilibrium profile of moisture content has the same shape as
the moisture characteristic. { Section 8.6}

• Infiltration into a dry soil is very rapid initially because hydraulic

gradients can be very large. The rate of infiltration decreases with time i f
the surface conditions remain constant, and over time approaches a
constant value equal to the saturated hydraulic conductivity. { Section
8 . 8)

299
• The Green-Amp! equation can be used to calculate the infiltration-time
curve if several soil parameters can be estimated. These parameters are
the salUrated hydraulic conductivity, the wetting-front pressure head, and
the change in moisture content between saturation and the underlying
soil. {Section 8.8. 1 }

• Field conditions rarely confonn to the ideal. Hysteresis in the hydraulic

relationships for a soil occur, meaning that the relationship between I{I
and 0, for example, is different when the soil is imbibing water than
when it is draining. Heterogeneities such as macropores are common in
soils, which may be responsible for rapid movement of water in the
unsaturated zone under certain conditions. {Section 8.9}

• Coarse material can act as a barrier to flow of water in unsaturated soils

because of the relationships among I{I, 0, and K. {Section 8.1 O}

8.14 Example Problems

Problem I. Tensiometers are installed at OAm and 0.5m above the water
table in a unifonn sandy soil with the moisture characteristic and hydraulic
conductivity curves given in Figures 8.4 and 8.5, respectively. One set of
tensiometer readings indicates that the capillary-pressure head at the first
of these tensiometers is -0.45 m and at the second is -0.6 m.

A. What is the direction of water movement between the two

tensiometers?

B. Estimate the magnitude of the specific discharge between the two

tensiometers.

Problem 2. Heidmann et al. (\ 990) report that the moisture characteristic

for a sandy loam soil in Arizona can be represented accurately by the
equation w = _963.7rr·659 , where If! is in meters and 0 is expressed as a
percentage (e.g, 26% rather than 0.26). The saturation value of moisture
content for this soil is estimated to be 27%.

A. Plot the moisture characteristic for this soil.

B. What is the equilibrium profile of (a) capillary-pressure head and (b)

moisture content above a static water table in this sand?

C. Suppose that moisture content measurements from 1 0 m below the soil

300
 surface in this sandy loam to a depth of 50 m show a constant moisture
content of 3%. The water table is at a depth of80 m. Over the interval
of the moisture measurements, how do capillary pressure head and
hydraulic head vary?

Problem 3. Assume that a wetting front moves into the sandy loam soil
described in Problem 2. The moisture content at the surface of the soil is
held constant at 27%. The underlying moisture content is constant at 6%.
(fthe saturated hydraulic conductivity for this soil is 3 x 1 0-6 m S- I ,
estimate how long it will take the wetting front to move I m into the soil.

8.15 Suggested Reading

Fetter, C.W. 2000. Applied hydrogeology, 4th ed. Upper Saddle River, NJ:
Prentice Hall. Chapter 6.

301
9 Ecohydrology: Interactions between
Hydrological Processes and the Biota

9 . 1 Introduction

9.2 Hydrologic Controls on Plants and Soil Microorganisms

9.2 . 1 Plant-water relations

9.2.2 Waterlogging

9.2.3 Hydraulic redistribution

9 .2.4 Soil moisture control on microbial activity

9.3 Biotic Controls on Hydrological Processes

9.3.1 Impact of vegetation on evapotranspiration

9.3.2 Impact of vegetation on infiltration and runoff

9.3.3 Effect of vegetation on precipitation

9.4 Concluding Remarks

9.5 Key Points

9.6 Example Problems

9.7 Suggested Readings

9.1 Introduction

Hydrological processes play an important role in the dynamics of

terrestrial ecosystems because they detennine the avai lability of freshwater
resources that sustain life on Earth. At the same time, the biota affect
fundamental hydrological processes, such as evapotranspiration,
precipitation, runoff and infiltration, which have a major impact on the
watcr cycle at global, regional, and local scales. In reccnt years
hydrologists have paid increasing attention to the coupling between
ecological and hydrological processes and have become aware that a better
understanding of the dynamics underlying changes in water fluxes and
stocks in continental land masses can be achieved by investigating

302
interactions between water and the biota. The study of these interactions
has led to the emergence of ecohydrology, a subdiscipline at the
confluence between ecology and hydrology. Research in this field is
investigating both the controls exerted by hydrologic processes on the
dynamics of terrestrial ecosystems (e.g., carbon sequestration,
biogeochemical cycles, plant and microbial stress), and the effect of the
biota on water flows and stocks. Understanding these hydrologic-biotic
interactions is crucial to the prediction of the effccts of changes in water
availability and climate on ecosystems, as well as the evaluation of land
use change impacts on the water cycle.

9.2 Hydrologic Controls on Plants and Soil

Microorganisms

Most plants take up water through their roots. Water then moves through
the xylem, capillary tubes within the plant, to the leaves and is lost by the
plant as watcr vapor fluxcs (i.e., transpiration) through the stomata, small
cavities typically located beneath the leaves (Figure 9. 1 ). Plants also
assimilate CO2 from the atmosphere through the stomata during
photosynthesis. Plants can regulate the size of opening of the stomata to
control the rates oftranspiration and carbon assimilation. Thus, the water
and carbon cycles are strongly coupled through stomatal physiology.

303
 Transpiration (qlrlUISP) �

-
•
I
/
Figure 9.1 Schematic representation of water flow through the soil-plant­
atmosphere system. Plants take up water from the soil through their rools.
This water is transported through the xylem and released as water vapor
fluxes (transpiration) 10 the atmosphere through the stomata.

9.2.1 Phtnt-nater relations

Classic theories of water transport in soil, roots, and xylem assume that the
flow is laminar. Soil pores and plant vessels are so tiny and the flow
velocities arc so small that the Reynolds numbers arc definitely within the
laminar range (see Chapter 3). Therefore, plant physiologists investigate
water flow in plants using Darcy's law, which assumes the flow to be
laminar (see Chapter 6). According to Darcy's law, the flow rate is
proportional to gradients of hydraulic head or-in h t e jargon of plant
physiologists-gradients of water potential, h [L]. The hydraulic head of

304
soil water (or soil water potential, lis) is the sum of elevation head (zs) and
soil capillary·pressurc hcad (\1'.\.), which in tum dcpcnds on the soil watcr
content through the moisture characteristic curves (Figure 8.4). Water flow
within plants is partly driven by concentration gradients across
semipemleable membranes; known as osmosis, this process entails the
transport of water molecules through semipenneable membranes from
lower to higher solute concentration areas (see Box 9.1). To account for
this phenomenon, the xylem water potential ("x) is expressed as the sum
of elevation head (zx), xylem capillary·pressure head (Wx), and the osmotic
potential (0), which is zero in pure water and negative in the presence of
solutes (0 decreases with increasing solute conccntrations). In a steady
flow the rate of root uptake is equal to the transpiration rate, q,rall!lp [L il]
(see Figure 9. 1 ), and can be expressed as

q'''''''r= -c(II, - It,}. (9.1)

where hT = ZT + 'Px + Q; 1Is = z, + 'P.,; and c [ill is a constant representing

the conductivity of the soil·root-xylem system. The capillary-pressure
head in the xylem can bc thus expressed as

tr" = tp; -ll: - n - q"""'r Ie. (9.2)

where ru = zx-zs is the height above the rooting zone (Figure 9. 1) .

Based on Equation 9.2, the xylem capillary-pressure head, IJIx' decreases
with increasing plant hcight, transpiration rate, and osmotic potential. Low
values of 'Px cause conditions of plant water stress in two ways. First, to
maintain turgidity, the "stiffness" of cells that allows the non-woody parts
of plants to sland vertically, thc prcssure of the water insidc plant cells
needs to be positive. Because the water potential of intracellular water (he)
is in cquilibrium with that of xylem watcr (he::;;: "x), low values of 'Px may
cause loss of turgidity and wilting. Second, when water pressure decreases
below a critical value air bubbles fonn in the xylem. Known as cavitation,
this phenomenon is undesirable because air bubbles cause the hydraulic
failure of the xylem by clogging and damaging the conduits (embolisms).
As a result of cavitation, fewer vessels remain available for xylem flow,
thereby decreasing the conductivity, c, of the soil-root-xylem system and
further reducing 'Px (Equation 9.2). To maintain turgidity and avoid
cavitation, plants can either modify (decrease) the osmotic potential by
changing solute concentration in xylem water�a process known as
osmotic compensation-or reduce the transpiration rates (see Equation

305
9.2). We recall from Chapter 8 that the relationship between I.fIs and the
waler content is expressed by the moisture characteristic curve (Figure
8.4): the capillary-pressure head IfIs of soil water decreases as the soil
becomes drier. During a dry period, plants might use osmotic
compensation to prevent turgidity loss and hydraulic failure, but below a
critical soil moisture value, 0*, they eventually have to decrease
transpiration by closing their stomata. As a result of stomala closure,
photosynthetic uptake is also reduced because CO2 fluxes occur through
the same stomata. Under these conditions, if the rate of carbon assimilation
is insufficient to meet the metabolic demand, plants need to use their
carbohydrate reserves. Prolonged droughts may lead to the depletion of
these reserves, thereby, causing plant mortality by carbon starvation
(McDowell et aI., 2008). Overall, as soil moisture decreases below the
critical value, 8*, plant transpiration is reduced, eventually becoming zcro
at a soil moisture level, 8.. , termed permanent wilting point (Figure 9. 2).
.

The discussion above suggests that plants exposed to droughts need to

walk a fine line of using stomatal closure to prevent desiccation and
hydraulic failure while maintaining adequate rates of photosynthetic
uptake to avoid carbon starvation. Two different stomatal regulation
strategies may be used by plants (Figure 9.2): isohydric plants tend to
close the stomata and reduce photosynthesis as the soil becomes drier,
while anisohydric plants prefer maintaining higher photosynthetic
uptakes even when soil moisture is low. The isohydric strategy avoids
hydraulic failure at the cost of exposing the plant to carbon starvation.
Conversely, anisohydric plants are less prone to carbon starvation but
more susceptible to cavitation. An interesting example of the dife f rent
drought sensitivity of isohydric and anisohydric plants has been observed
in pinon-juniper woodlands across the southwestern United States. A
severe drought in 2000-2002 caused about 95% mortality in piTion trees
and 25% mortality in junipers. Scientists have explained these different
mortality rates as an effect of the drought response strategies of these two
species. Junipers exhibit anisohydric behavior (i.e., more susceptible to
hydraulic failure), while pinons are isohydric plants (more prone to carbon
starvation). There is some evidence that during the 2000-2002 drought,
pinon trees had a negative carbon budget and were likely affected by high
mortality rates due to carbon starvation (McDowell el aI., 2008).

306
 1 r
---

l-
w I
c..
- I
"
-

I
0 8
<P
• •

e w:::: e aniso e iso

Figure 9.2 Relationship between the ratio of evapotranspiration (et) to
potential evapotranspiration (PET) and soil moisture (0) for isohydric
(blue, solid) and anisohydric (black, dashed) plants. The value of (j is
maximum (and equal to the porosity, r/J) at saturation. For soil moisture
values greater than 0", the stomata are wide open and transpiration occurs
at the potential rate (PET). As soil moisture decreases below 0* the plant
starts closing the stomata thereby reducing et. At the permanent willing
point (0",), er is zero. Isohydric plants have a value of (j* greater than
anisohydric plants.

BOX 9.1 Osmotic Pressure

When ions (charged particles) in aqueous solution are nOl free to

diffuse throughout a medium in which concentration gradients are
maintained, an osmotic pressure develops. For example, in a device
in which a semi-permeable membrane (a membrane with pores that
allow water molecules to fit through, but which filter ions so they
cannot pass) maintains a concentration gradient, osmotic pressure
causcs watcr to flow from thc pure water into the solution until the
hydrostatic pressure in the pure water phase balances the osmotic
pressure (see figure below). The device sketched in the figure, a
simple osmometer, can be used to measure osmotic pressure of the
solution. Notice how this device resembles a plant that maintains
solute concentrations in the xylem higher than in soil water (osmotic
compensation). The existence of such a concentration gradient
across semipermeable membranes allows plants to enhance root
uptake, while reducing the risk of hydraulic failure due to low xylem
prcssure values.

307
 Interestingly, even though no semi-permeable membranes exist in
soils, in some clays there is a mechanism for segregating a relatively
ion-rich phasc from a relativcly ion-deficient phase. The surface of
clay particles tends to have a negative charge due to the structure of
the mineral. Because of the attraction of particles having opposite
electrical charges, an increased concentration of cations (positively
charged ions) is found in the space between clay particles. If there is
very dilute water (very low cation concentration) in the pore spaces
exterior to the clay plates, the increased ion concentration between
any two clay plates sustains an osmotic-pressure head gradient that
tends to force water in the direction of greater ion concentration;
that is, into the space between clay plates. In a fashion similar to the
osmometer (see figure), waler flow will continue until the osmotic
pressure is balanced by capillary pressure. This phenomenon
explains the expansion (shrinking) of some clays upon imbibition
(drying).

Salt solution

ii,
Semi-permeabl e
membrane

_3
.
r _______ 1i
Pure water
. . .
A dev1 ee for demonstratmg osmot1 c pressure.

9.2.2 \-\'atel'logging

Most plants need oxygen at their roots for metabolic processes. Because
the oxygen supply to the root zone occurs by diffusion through air-filled
soil pores, oxygen availability becomes restrietcd as soils bccome
saturated. Lack of oxygen makes saturated soils unsuitable for the growth

308
of many terrestrial plant species. Thus water stress in vegetation can be
caused by both a water deficit and by an excess of water (waterlogging).
Plants are sometimc classified on the basis of their tolerance of
waterlogging conditions. While most plants take up water from the
unsaturated zone, phreatophytes place part of their root system below the
water table and take up water from the groundwater. These plants rely on
watcr uptake from thc saturated zone but may still have only a limited
tolerance to waterlogging. Hydrophytes withstand prolonged periods in
which the ground surface is flooded.
The different sensitivity of plant species to flooding typically explains
their distribution along topographic gradients in flood prone areas.
Changes in hydrologic conditions may alter these distributions. For
instance, in the Florida Everglades, shifts in plant community composition
have been observed in response to altered hydrologic conditions. The
Everglades are an extensive wetland with a seasonal 60 km wide sheet­
flow delivering freshwater from Lake Okeechobee to Florida Bay over a
distance of about 150 km. This slow shallow flow takes place on a
limestone substrate partly vegetated by herbaceous plants. The landscape
exhibits a number of islands that emerge from the watcr for most of the
year. Because of their higher elevation the islands are more infrequently
flooded than the surrounding marshes and are, therefore, densely covered
by trees and other woody plants with limited flood tolerance. In 1 948 the
U.S. Congress approvcd a project that led to the construction of canals and
levies that partitioned the Everglades into basins used for agriculture,
water storage (Water Conservation Areas, WCAs), and biological
conservation (the Everglades National Park). The WCAs were established
to retain some of this freshwater and make it available for drinking and
agricultural uscs during the dry season. Thc consequent increase in
flooding frequency in portions of the water conservation areas caused a
die-off of trees in the lower elevation parts of the islands and a complete
die-off in low islands.

9.2.3 H�'d.-aulic redist.-ibutioll

Most plants keep their stomata closed at night and no transpiration occurs.
When vertical hydraulic head gradients exist along the soil profile, water is
redistributed through the roOI system from shallow 10 deep soil or vice
versa, depending on the direction of decreasing hydraulic head (see
Section 8.5) . Because of their higher hydraulic conductivity with respect to
unsaturated soils, plant roots provide a preferential pathway for the
nocturnal redistribution ofwaler (Caldwell and Richards, 1989). Irthe soil

309
is much drier at the surface than at depth (e.g., during a dry spell), then the
capillary-pressure head at the surface is much lower than in the deep soil;
this gradient in pressure head can overcome the opposite elevation head
gradient and bring deeper water up through the roots to the surface (Figure
9.3). Known as hydraulic lift, this process may facilitate the establishment
and survival of shallow-rooted plants and enhance the uptake of nutrients
stored in the shallow soil. The opposite process can also occur: right after a
rainfall event, if the deeper soil is sufficiently dry, the hydraulic gradient
from the shallow to the deep soil could draw water downward through the
root system (Figure 9.3). This phenomenon of hydraulic descent can be
used by deep-rooted plants to bring soil moisture out of the reach of
shallow rooted plants (Burgess et aI., 1 998). The occurrence of hydraulic
lift has been well documented in several forest ecosystems worldwide,
including the Amazon. Model simulations indicate that hydraulic lift may
enhance root uptake from deeper soil layers, and induce a substantial
increase in dry season transpiration and photosynthesis with important
impacts on the regional and global climate (Lee et aI., 2005).

tt tt
- -

(a) Daytime, (b) Nighttime, (c) Nighttime,

moist soil soil drier at surface soil drier at depth
Figure 9.3 The role of roots in hydraulic redistribution. During the day
water is taken up from the soil by the roots and transported to the leaves
where it is transpired (a). At night, if the soil is much drier at the surface
than at depth, water is taken up by deeper roots and released into the
ground by shallow roots (b). Conversely, if the deeper soil is much drier
than the shallow soil, at night water is drawn from the shallow soil and
released into the ground by the deeper roots (c).

310
 9_2A SoilmoistUl-e control 011microbhll acth"it)'

Soil moisture afe f cts microbial processes in multiple ways. Low soil
moisture inhibits microbial activity because the substrate (carbon and
nutrients) used by microbes is transported in the water phase and its
mobility is strongly reduced in dry soils. Moreover, in hyperarid
(extremely dry) conditions microbes are affected by dehydration. Overall
the water stress tolerance of soil micro-organisms is much higher than for
plants; that is, the wilting point is much lower for microbes than for most
types of vegetation. As water content increases above the wilting point,
microbial activity increases because availability of substrate increases. The
activity increases until effects of waterlogging become important after
which increases in moisture lead to decreases in aerobic microbial activity
(Figure 9.4).
Thc sensitivity of microbes to water logging depends on the microbial
metabolism. While anaerobic microorganisms thrive in waterlogged soils
and anoxic conditions, the activity of aerobic communities relies on the
supply of oxygen and requires soil aeration. For example, the
decomposition of soil organic matter is perfonned by aerobic
microorganisms that convcrt littcr and othcr organic rcsiduals into
inorganic compounds while releasing CO, into the atmosphere. Water
logging conditions hinder decomposition, and provide an ideal
environment for the accumulation of organic matter. For instance, peat
deposits are commonly found in wetlands where the rate of organic matter
production exceeds the decomposition rate. It has been estimated that a
large fraction of global carbon stocks is in peat deposits. In boreal regions,
defined by very cold winters and relatively short and cool summers,
peatlands are prone to drainage as an effect of climate wanning. This
phenomenon is expected to increase soil aeration and soil organic matter
decomposition, thereby causing peat loss, decreases in ground surface
elevation (subsidence), and increases in atmospheric CO2 emissions.

3II
 / "\
.
> Substrate Oxygen
-

control control
-
.-

o
III

III
-

.c
.-

e
o

:2
.-

Soil moisture
Figure 9.4 A concepruaJ representation of soil moisture control on aerobic
microbial activity.
Modified after Skopp ct al. (1990).
At the beginning of this section we indicated that the carbon and water
cycles are strongly coupled. At this point it is clear that the coupling acts
both through the process of atmospheric carhon assimilation by plants
(hydrologic controls on stomatal regulation of photosynthetic uptake) and
through soil moisture controls on organic matter decomposition.

9.3 Biotic Controls on Hydrological Processes

Hydrological processes are affected by the biota in various ways. At the

plot scale, plant roots and soil organisms enhance infiltration by providing
preferential pathways for water flow through the soil column, while plant
uptake strongly modifies soil moisture dynamics. At the watershed scale,
vegetation affects the water balance by enhancing evapotranspiration and
reducing runoff. At the global scale, the terrestrial biosphere sustains a
faster water cycle, that is, in the presence of vegetation, water cycles faster
between the Earth's surface and the atmosphere than it would in the
absence of vegetation. Using model simulations, Fraedrich el al. (\999)
compared the water cyete in a desert planet (no vegetation) and in a green
planet (with all the non-glaciated land covered by forest vegetation) and
found that in the vegetated scenario evapotranspiration from continental

3 12
land masses increased by 243% and precipitation by 93%, while runoff
decreased by 24 % with respect to the unvcgctated case. These two
(unrealistic) extreme cases give us an idca of the important role played by
vegetation in the global water cycle.
In Chapter I we explained how the water received at the Earth's surface
as precipitation leaves continental landmasses either as evapotranspiration
or mnoff. Because mosl of the evapotranspiration from terrestrial
ecosystems is contributed by plants, this flux is sometimcs tenncd "green
watcr flow," while surface and groundwater mnoff arc referred to as "blue
water flows" (Figure 9.5 ) (Falkenmark et aI., 2006). To investigate the
impact of the biosphere on the water balance we can consider how
vegetation affects the partitioning ofprecipitalion over land into
evapotranspiration and runoff (i.e., into green and bluc watcr flows).

9.3.1 ImpHct of Hgetation on e\"HpolranspirHtion

Evapotranspiration from a vegetatcd landscape occurs at a faster rate than

evaporation from bare soil. Thus, removal of vegetation reduccs grcen
water flows and increases runoff. Deforestation is one of the major
contributors 10 land cover change: it has been estimated that forests
currently cover about 3 1 % of the global land surface and that they are
currently lost at the rate of 13 million ha y..-' (FAO, 2010). Central and
South America, Africa, and South-East Asia are particularly affected by
dcforestation. At the same time other regions of the world (e.g., the eastern
U.S.) are undergoing reforestation as a result of abandonment of
agricultural land, changes in climate, and land management. I n most cases
deforestation does not cntail the replacement of a vegetated landscape with
bare soil but with pasture or cropland. This change in land cover is still
typically associated with a decrease in evapotranspiration (Figure 9.6)
because, when compared to crops or pastures, forests have (a) a greater
leaf area index (LAI, total foliage area per unit ground area), (b) deeper
root systems that give them access 10 deeper soil moisture, and (c) greater
land surface roughness that enhances evapotranspiration (sce Chapter 2) .
Moreover, forests typically have a lower albedo (i.e., they are darker),
which a[lows them to receive a higher net solar radiation than croplands
and pastures. In other words, more energy is available 10 sustain
evapotranspiration (see Equation 2.19) in a forest than in a pasture.

313
 Green water flows
I
, I
iHH Hiii
Transpiration
I I I I I
Evaporation
I I I I I
• •

• • • •
• •

• • •

Surface runoff
• • •

• • •

OJ
). c
• • • •

'"
• •

�
• •

"
• • • ·

• • • • •

'"
-
• • • • •
• Groundwater flow �

0
• • • • • • • •

• • • • • • • • · •

• -

•
• • <- • • • • • • •

'00
• • + • • • • • • •
• • • • • •

• • , . . . . • • • • • • • • • • • •

· , .. . . . . , • • • • • • • • . . . . , , • •
• • • • • • • • • • • • • • • • • • • • • • • •

• • · . . . , . . . . . . . . . • • • • •
• · . . . , • • • " . . . • • •

• • • • • • • • • + • • • • • • + • • •

Figure 9.5 Green and blue water out of continental landmasses.

While deforestation is expected to reduce evapotranspiration and

increase runoff, other trends in land use change may have the opposite
effect. For instance, irrigation takes water from streams and aquifers and
makes it avai lable for crop evapotranspiration. As a result of irrigation,
blue water flows are strongly reduced while green water flows increase.
Many rivers on earth (e.g., the Rio Grande or the Colorado River) do not
reach the ocean anymore as a result of water withdrawal for agriculture
and other uses. Even though only 20% of the global agricultural land is
currently irrigated, irrigation is expected to increase over the next few
decades (to meet the escalating food demand) thereby further reducing
river flows and increasing green water flows (Falkenmark et aI., 2006).

3 14
Higher LAI, higher roughness, Lower LAI, lower roughness,
deeper roots, lower albedo shallow roots, higher albedo

..

Forested land surface Cropland, pasture, bare soil

Figure 9.6 Major land surface changes associated with deforestation, with
forcsted conditions (left) contrasted with cropland (right).

9.3.2 Impact ofwgctation on infiltration and runoff

Rescarch in forest hydrology has clarified the effect of vegetation on the

hydrologic response of catchments to precipitation. Forest soils are very
permeable and allow high infiltration rates because roots provide
preferential pathways for water flow through the soil matrix, particularly
through the formation of macropores (i.e., relatively large cavities; see
Chapter 1 0) resulting from the decay of dead roots. Moreover, plant
canopies and litter shelter the soil surface against rainsplash compaction
(the compaction induced by raindrop impacts onto the soil surface),
thereby further enhancing the soil infiltration capacity. As a result, flow of
water across the land surface in forested watersheds tends to be restricted
to areas whcre soils are more easily saturated, typically in sites with
convergent topography located close to the channel network (see Chapter
10). The removal of forest vegetation disturbs the soil surface and exposes
it to compaction by rainsplash; it also reduces soil bioturbation and
macropore fonnation by plant roots and other soil organisms. As a result,
deforestalion causes a decrease in soil infiltration capacity and an increase
in surface runoff (Figure 9.7).
Forested watersheds exhibit higher evapotranspiration rates than their
deforested counterparts: the replacement of a forest with pasture or bare
soil reduces both interception and transpiration. The amount of water that
flows through the watershed as surface and groundwater runoff is
consequently greater in clearcut than in forested landscapes. These effects

315
are in general more noticeable during the growing season, when vegetation
plays a more active role in the water balance.

•
40
::
•

•
0
c
c
•
30
�
�
-
• •
•
•
•
•
� 20 .. •
�
• •
•

c • • • •

c .. '
• • •
• •
• • •
�
•
1 •
••
I
• • . •

-
'
c • •
•
•
• • •
• •
• •
.

•
�
'"
• • •
• •
0
c
� •
�
u
-
-10 •
c
�
�
u -20
� •
11. •
Deforestation
-30 • Reforestation •
•

-40!;---c;;--;;:--;;;;-
o - 10 --:,:---c; ;---;;:--;;;;---:,:--:e ;---:!.
20 30 40 50 60 70 80 90 100
Percent change in cover
Figure 9.7 Effect of deforestation and reforestation on water yields
(expressed as a percentage of mean annual precipitation). A decrease in
forest cover is generally associated with an increase in runoff, while most
of the reforestation experiments exhibit a decrease in runoff.
Data from Andrcassian (2004).

The impact of forest vegetation on runoff is typically investigated in

paired watershed experiments, which compare the hydrologic response of
IWO small watersheds located in the same region with similar topographic
and hydrogeological features but different land cover conditions. These
snLdies have shown that the increase in runoff induced by forest cutting
affects streamflow in both fair weather periods and during stonns. Overall
a denuded landscape has a more limited capacity to retain and store water
and tends to respond more quickly to rainfall events, in some cases with
abrupt flash-floods. Snowmelt floods are also faster and occur earlier in
the year. While in some regions the increase in rtlnoff resulting from
deforestation has the undesirable effect of increasing the potential for
nooding and soil erosion, in others the decrease in runoff resuit"ing from

316
afforestation by bush encroachment or tree plantations is recognized as a
serious threat to water yields and water availability, especially in semiarid
catchments.

9.3.3 Effect of vegetation on precipitation

The impact of terrestrial vegetation on precipitation is more dimcult to

assess than its more direct effect on evapotranspiration and runoff.
Because the effect on precipitation is expected to be detectable at scales
that are much larger than those of the paired watershed studies discussed in
the previous section « 10--100 km2), direct experimental evidence is not
available. Therefore scientists rely on process·based models that account
for the coupling between hydrological and atmospheric processes. These
model-based studies are in overall agreement: regional-scale deforestation
is expected to cause a decrease both in evapotranspiration and in regional
precipitation (e.g., Bonan, 2008). The decline in precipitation can be
explained as a direct effect of the decrease i n evapotranspiration (Section
9.3.1), and as a result of the altered surface energy balance resulting from
changes in land cover conditions. A decrease in evapotranspiration entails
a reduction i n the terrestrial supply of atmospheric water vapor that
contributes to precipitation recycling (Section 1 .3). Vegetation removal is
typically associated with an increase in albedo (Figure 9.6), hence a
decrease in the net solar radiation that is received at the land surface. As a
result, surface heating is reduced and convection-induced uplift--one of
the mechanisms causing precipitation (Chapler I }-is weakened. It has
also been reported that vegetation can affect precipitation by altering the
terrestrial supply of aerosols suitable as cloud condensation nuclei (e.g.,
aerosols resulting from biogenic gas emissions, smoke from forest fires, or
mineral dust emitted by deforested landscapes). A quantitative assessment
of these effects is a subject of ongoing research.
An interesting mechanism contributing 10 the supply of water to some
terrestrial ccosystems is associatcd with the interaction ofvcgelaled
canopies with moist air. In Chapter 1 we stressed how condensation
requires a surface on which water molecules can condense. Close to the
ground, this surface may be provided by plant canopies, thereby leading to
canopy condensation and dew formation. Alternatively, small droplets may
already be present in fog or low clouds; as this damp air moves through the
canopy, some of droplets are deposited onto leaf surfaces. Water
eventually drips from wet leaves to the ground surface, a phenomenon
known as occult precipitation (Chapter 2). In some coastal foggy regions
occult precipitation may contribute to a substantial fraction of the total

317
water input and these inputs may play an important ecological role,
particularly if they occur during the dry scason (e.g., the wcst coast of the
U.S.) or in deserts (c.g., the Namibian or thc Atacama Desert). [t is worth
noticing Ihat occult precipitation would not occur (or would be severely
reduced) in the absence of vegetation canopies (Runyan et al., 2012). To
enhance their water supply, some communities living in "fog deserts" have
built systems of nets and gutters that trap atmospheric moisture imitating
the way plant canopies function in these environments.

9.4 Concluding Remarks

In the previous sections we have seen both how plants depend on water
(Section 9.2) and how the rate of major hydrological processes can be
modified by vegetation (Section 9.3). We can now consider the case of
ecosystems in which, by altering hydrological processes, plants are able to
establish conditions that reduce their exposure to water stress caused either
by water scarcity or by waterlogging (Runyan et aI., 2012).
Fog deposition can be crucial to the survival of plant communities in
arid and semiarid ecosystems. This is an example of systems that rely on
mutual interactions between vegetation and hydrological processes. In fact,
the decrease in occult precipitation that would occur in response to the
removal of plant canopies could impede vegetation re-establishment.
Similarly, regional climate models have shown that vegetation removal
from Mongolia, the Sahel region of Africa, or other desert margins would
result in a reduction of precipitation, thereby preventing vegetation re­
establishment (Bonan, 2008). Similar impacts of vegetation on
precipitation are often invoked to explain major climate and vegetation
changes in the history of the Earth. For example, fossil records indicate
that up 10 the mid-Holocene (i.e., about 5000 years ago) large parts oflhe
Sahara used to be vegetated. It has been argued that the decline of the
"Green Sahara" was sustained by a decrease in precipitation induced by
the loss of vegetation cover. Although changes in orbital conditions appear
to have played a major role in the shift to the desert state, paleoclimate
models indicate that the decrease in precipitation was likely enhanced by
vegetation losses (Brovkin et aI., 1 998).
Analogous conditions may result from plant-water table interactions.
The typical effect of vegetation on the underlying shallow unconfined
aquifer is to lower the watcr table by taking up water from the overlying
unsaturated zone or through direct depletion by phreatophytic vegetation.
In landscapes with shallow water tables vegetation may play a crucial role

318
in draining the shallow soil, thereby preventing the emergence of water
logging conditions that would be detrimental to the survival of the existing
plant species. Some ccosystcms may strongly depend on these positive
interactions between vegetation and water table dynamics. In some cases,
the removal of native vegetation may result in a dramatic increase in water
table elevation. For example, the remov<ll of eucalyptus trees from the
Murray-Darling basin (Australia) has allowed the water table to fise so
close to the ground surface that direct evaporation of shallow groundwater
is causing salt accumulation at the soil surface, thereby making soils
unsuitable for most plants (Runyan et aI., 2012).
These are just some examples of ecosystems that strongly depend on
mutual interactions between hydrological processes and the biota.
Understanding these interactions is crucial the slUdy of ecosystem
sensitivity to changes in climate and hydrologic conditions.

9.S Key Points

• Water transport within plant vessels occurs as a laminar flow. To take up

water from the ground, plants need to exert a suction that is strong
enough to establish a hydraulic gradient between rool zone and leaves.
There are, however, some hydraulic constraints to the maximum suction
tolerated by plants: strong suctions (i.e., low negal"ive pressures) may
induce loss of turgidity and cavitation, thereby causing the failure of the
hydraulic function of the xylem. {Section 9.2. 1 }

• The water characteristic curves discussed in Chapter 8 (Figure 8.4)

indicate that drier soils exhibit a lower (negative) capillary-pressure head.
Thus in dry periods plants need to maintain a stronger suction to sustain
uptake. To avoid damage caused by cavitation, plants tend to reduce
transpiration by closing the leaf stomata, small cavities through which
water vapor (transpiration) and CO, (photosynthesis) are exchanged with
the atmosphcre. {Section 9.2 . 1 }

• As a result of stomata regulation, phOlosynthetic carbon uptake by plants

is also reduced. Thus, prolonged droughts may induce plant mortality by
carbon starvation. {Section 9.2. 1 }

• Water stress conditions can emerge not only when the ground is dry but
also in saturated soils (waterlogging). Oxygen is needed in the root zone
to sustain plant metabolic processes. Because oxygen is supplied by
diffusion through air-filled soil pores, its availability is restricted in

319
 saturated soils. Thus, most plant species do not tolerate waterlogging.
Similarly microbial reactions that require oxygen (e.g., decomposition)
arc restricted in saturated soils. {Sections 9.2.2 and 9.2.4}

• At night plant roots may redistribute soil water along decreasing

gradients of soil moisture. If the soil is much drier at the surface then at
depth, hydraulic redistribution entails an upward water flow through the
root system ("hydraulic lift"). Conversely, ifsoil moisture is much
greater in the shallow soil then at depth, downward redistribution through
the roots may occur. {Section 9.2.3}

• Vegetation enhances evapotranspiration and precipitation in continental

landmasses, thereby intensitying the terrestrial portion of the water cycle.
{Section 9.3}

• Forest removal (deforestation) reduces evapotranspiration (or "green

water flows") and increases runoff (or "blue water flows"). Conversely,
irrigation reduces blue water flows and increases green water flows.
{Section 9.3}

• The ability offorests to enhance evapotranspiration is due to the higher

leaf area index (hence greater transpiration and interception), deeper
roots, and greater roughness of forest vegetation with respect to
croplands or pastures. {Section 9.3. 1}

• Vegetation has the effect of increasing soil infiltration capacity because

plant canopies shelter the soil surface against rainsplash compaction,
while soil bioturbation by plant roots creates macropores, thereby
increasing the soil penneability. Overall, the impact of vegetation on the
hydrologic regime is to increase infiltration and decrease runoff. {Section
9.3.2)

• At the regional scale, forest vegetation increases precipitation. In fact,

vegetation enhances evapotranspiration, thereby intensitying
precipitation recycling. Moreover, because of their lower albedo, plants
modify the surface energy balance thereby enhancing surface heating and
convection. {Section 9.3.3}

• Occult precipitation (i.e., fog deposition and canopy condensation) can

be an important contributor to the water balance in regions affected by
fog and low clouds. {Section 9.3.3 }

• The ability of vegetation 10 modify the local or regional water balance

320
 may be crucial to the survival of some plant communities in water scarce
environments. In these ecosystems vegetation removal can dramatically
alter the water budget to thc point of preventing the re-establishmenl of
the same plant species. {Section 9.4}

9.6 Example Problems

Problem I . Irrigation is used in areas in which agricultural productivity

can be enhanced by maintaining the soil water content well above the
permanent wilting point (e.g., 0 > 20w) throughout the growing season.
When irrigation water is applied, root zone soil moisture should not exceed
field capacity 0lc' because, if 0 > O/c, water and fertilizer are lost from the
root zone by drainage. With this management approach soil moisture
remains below field capacity and the only losses of soil water from the
root zone are by evapotranspiration. Assuming that no rainfall occurs, the
ICmporal dynamics of depth-average soil moisture 0 in the roOI zone can
be cxprcssed by the soil water balancc equation

d8
Z-=-et,
dl

where Z is the depth of the root zone. I n other words, changes in the water
stored in the control volume (i.e., the root zone) are due to the difference
between water inputs and outputs (see Chapter I) . I n this specific case
there are no rainfall-induced inputs, while the only outflow is due to el.
The irrigation period (i.e., the time between two consecutive applications
of irrigation water) is calculated by integrating the soil water balance
equation between time I = 0 when 0 = O" (i.e., right after irrigation) and
time Ii when 0 = 00 (and it is time to irrigate again), assuming that the
f<

evapotranspiration rate linearly decreases from the potential rate PET at 0

= 0lc to zero at 0 = 0 . (i.e., e( = PET(O- O....)/(O lc - 61".»
..

Consider the case of a crop grown on a sandy soil with 61lc = 0.19 and a
crop-specific wilting point, 61 = 0.05. The root zone is 40 cm deep. After
....

each application of irrigation water, soil moisture is equal to ()/c- Assuming

that PET = 6 mm d-1 , calculate the irrigation period, Ii' i.e., how long it

32 1
would take for soil moisture to decrease from field capacity to the value (}o
= 2(}", in the absence of any rainrail input.

Problem 2. With the irrigation scheme described in the previous problem,

during each appl ication of irrigation water soil moisture is increased from
eo = 20.", to 0jc' The volume of water (per unit area) required to make this
change is Z((}jc - (}o). Calculate the amount ofwater per unit area that
should be applied every lime the area is irrigated.

9.7 Suggested Readings

Bonan, G. 2008. Ecological climatology. Cambridge, UK: Cambridge

University Press.
D'Odorico, P., and A. Porporato. 2006. Dlylalld ecohydrology. Dordrecht,
the Netherlands: Springer.
Rodriguez-lturbc, J., and A. Porporato. 2005. Ecohydrology ofwater­
con/rolled ecosystems: Soil moiS/llre alld pIC/III dynamics. New York:
Cambridge University Press.

322
10 Catchment Hydrology: The HiIIslope­
Stream Continuum

10.1 Introduction
10.2 Streamflow Hydrographs
10.3 Hydrograph Separation
10.4 Runoff Processes
10.4.1 Direct precipitation onto stream channels
10.4.2 Overland flow
10A.3 Shallow subsurface stonntlow
10.4.4 Groundwater flow
10.4.5 "Old" and "new" water
10.5 Contributing Area and Topographic Controls on
Saturation
10.5.1 Contributing area
10.5.2 Topographic index
10.5.3 Hillslope stability
10.6 Routing Water through a Catchment Using Catchment
Models

10.6.1 TOPMODEL calculations

10.6.2 Hydrograph simulation using TOPMODEL
10.7 Concluding Remarks
10.8 Key Points
10.9 Example Problems
10.10 Suggested Readings

10.1 Introduction

323
One of the problems that has occupied hydrologists for many decades is
the identification and quantification of the pathways by which rainfall and
snowmelt movc through catchments and ultimatcly producc runoff. In
Chapters 5, 7, and 8 we dcscribed the individual flow paths over and
beneath the Earth's surface that precipitation might take through a
catchment and provided the physical basis for understanding each. To
describc the catchmcnt hydrological cycle, we have to intcgrate all the
individual processcs of surface and subsurface flow of water. This is a
fonnidable problem to solve in a quantitative fashion. In this chapter we
introduce some of the ideas and tools used in catchment hydrology.
There are many reasons why hydrologists are interested in
understanding flow paths within catchments. One reason is that
understanding the routing of rainfall and snowmelt is required for
undcrstanding how water interacts chemically with rocks, sediments, and
biota within a catchment. For example, hydrologists are interested in
understanding the effects of acid rain on the chemical composition of
streams around the world. The pH of rainfall and snowfall in many parts of
North America and Europe in the latter part of the twentieth century was
typically below 4.5, which is well below the level at which water would be
considered to be acidic « 5.65). "Acid rain" is known to result from the
chemical interaction of water in the atmosphere with gaseous emissions
from power plants, factories, and automobiles. (Although the trend in the
acidity of rain has been positivc ovcr the past decade or so and the extent
of areas with rainfall pH below 4.5 has decreased considerably, concerns
remain about the impact of deposition of acidic compounds on catchments
(Burns et al., 2011 ).) In some catchments, the stream pH shows an
episodic depression (known as acidification) that at first glance might
appear to be caused by rain falling directly on the Sfream surface.
However, for many streams this explanation cannot be correct because the
pH of the stream remains low even after the rainfall has ceased (for
example, Bear Brook in Maine, Figure 1 0. 1 ). How, then, is the
composition of the rain altered when it comes into contact with soils and
rocks? Answers to such questions require knowledge of the flow paths
through and residence times within various portions of the catchment, and
are relevant to a variety of environmental issues. For example, depressions
of pH below about 5 may cause damage to certain types of fish.
Another reason for desiring an understanding of runoff generation in
catchments is to be able to predict the effects that humans will have on the
hydrological cyclc through their land-based activities. It is now known
from experimental studies at research sites throughout the world (for
example, the Hubbard Brook Experimental Forest in New Hampshire) that

324
various silvicultural practices (such as selective cutting, clear-cutting, and
others) can have dramatic effects on stream-water discharge and quality
for many years after the particular managcmcnt activity is performed.
Hydrological changes that have been demonstrated experimentally include
increases in annual water yields, decreases in evapotranspiration, increases
in nitrate concentrations, and increases in stream temperature and sediment
loads.
::-- 0.04 65
• 0
E 0.03 6

8, 0.02
0
"'- 5.5 a
0
"
•

0.01 o 00 0 0 5
•

•
u

C 4.5
re re re
0
� � � , ,

- - - - - - -
8. Ii 8 8 8 8 8. 8
N N N N N

.
E E § E � E § E E §
� ,; z ci
:§
• z ci
f •
z
$ $ $ $ $
:2 �
�
• � �
• � �
• �

Figure 10.1 Bear Brook, Maine, is acidified during a rainfall event in

1992. Rainfall produces a rising hydrograph (blue line) and a decrease in
the strcam pH (circles).

Hydrologists often conceptualize a catchment as functioning like a

group of "reservoirs" that store and release water in much the same way as
a simple flood-control reservoir (see Chapter 5). Precipitation and
snowmelt are the "inputs" 10 the catchment reservoir and
cvapotranspiration and runoff are the "outputs," This conccptual model of
a catchment can be useful in trying to intcgrate the various processes that
affect runoff. Much of our treatment of catchment hydrology in this
chapter will make use of the idea of identifying and quantifying flow in
various "reservoirs."

10.2 Streamflow Hydrographs

Consider the hydrograph for the stream draining Watershed 34 (W34) at

the Cowecta Hydrologic Laboratory in North Carolina (Figure 1 0.2).
Watershed 34 (area = 0.33 km2) is a mid-clevation catchment with
relatively deep soils (for the mountains of weslem North Carolina).
Consideration of the hydrograph for W34 leads 10 speculation that some of

325
the flow during precipitation events occurs immediately after the onset of
precipitation, while other portions of the now occur for some time after the
end of precipitation. Thesc hvo components of a streamnow hydrograph
are known as quickflow and baseflow.
The separation of a hydrograph into two components suggests that
water is being routed through two different storage "reservoirs" (Figure
1 0.3). That is, considering the input to a reservoir to be precipitation and
the outflow from the reservoir 10 be either the quickflow or the baseflow,
we observe the typical effects of flow routing through a reservoir: peaks
are delayed and attenuated. From our discussion of flow routing through
surface impoundments (Chapter 5.4), we can infer by analogy that the
baseflow reservoir has characteristics of storage and release that are quite
different from those of the quickflow reservoir. Remember that these
catchment "reservoirs" are conceptual, meant 10 represent portions of the
hydrological system that behave similarly to surface-water reservoirs, by
storing and releasing water.

E
-

g
-

;;.
•

�
�
�

8 � � :. •
8
,
f ! .!
- - - - - - - - - - - - -

,
"
•
• •

,
•
" < "
"

"

E (b) Lower curve suggests now from

-

E hypothetical 'slow' reservoir

�
"
-

•
<
u
,
•
•

�
•
0
0

8 , •
• � •
� 8
- - - - - - - - - - - - -

, ,
"
•
" , "
• • > • • "
• •
z c , <

Figure 10.2 Hyetograph (a) and hydrograph (b) for Coweeta W34, water
year 1983.

326
 TIme

•
.'l
E

+
�
•
-
- •
•

� 00
-

'"

L-J....---:Ti
me
� Time
lL '\-'-
'-
�
Time
Figure 10.3 Rainfall routed through two conceptual "reservoirs."
Although depicted here as surface-water reservoirs, in this chapter the term
reservoir refers to portions of the catchment hydrological system with
differing dynamic behavior. For example, the base flow rescrvoir may be
groundwater; because groundwater recharge and discharge arc relatively
slow processes, this reservoir behaves like the one at the right above.

The reservoirs shown in Figure 10.3 operate simultaneously. This

makes the job of separating the hydrograph into contributions from the two
(or more) reservoirs a difficult one. The observation that precipitation
appears \0 be routcd through "quickflow" and "baseflow" reservoirs in
many catchments has led to a great deal of research on flow paths through
catchments. What makes up the quickflow and baseflow reservoirs? What
are the characteristics of catchments that determine the relative amounts of
quickflow and baseflow? Does the water that flows through the different
reservoirs acquire different chemical signatures from their contact with
different geological materials?

327
10.3 Hydrograph Separation

Questions such as those posed in the previous section have nOl been
answered completely despite many decades of research. One of the
difficulties lies in the detennination of the components of the streamflow
hydrograph. After all, we don't measure two hydrographs (quickflow and
basef1ow); we measure only the total flow. The question then arises: how
can we "separate" the measured hydrograph? The smooth line in Figure
10.2 showing the baseflow was drawn "by eye." The separation seems
reasonable, but has no scientific basis. Hydrograph separation techniques
have been used by hydrologists for many years, but all of the methods are
empirical. These empirical methods are used widely to enable the
calculation of quickflow from catchments. Once the hydrograph has been
separated into quickflow and baseflow, a very simple computation, known
as the unit hydrograph method, can be used to route the portion of the
precipitation that goes through the quickflow "reservoir" to the stream, a
procedure that has application in engineering hydrology where the size of
pipes to carry stormwater is decided on the basis of such routing
calculations.
For many years, a standard practice was to assume that the components
derived from empirical hydrograph separation methods were physically
"real"-that the quickflow arose from the flow of water overland where
precipitation rates exceeded infiltration rates and that the basellow
originated as water that had infiltrated the soil surface, percolated to the
groundwater zone, and discharged into a stream channel over prolonged
periods of time. The "reservoirs" of Figure 10.3 under this view would be
the hillslope surfaces (the quickflow reservoir) and the unconfined aquifer
(the baseflow reservoir). Unfortunately, this simple picture is not valid in
many instances. The surface-water-groundwater-reservoir interpretation of
the empirical results turns out to be counter to observations made by field
hydrologists in forested catchments, for example.
An altemate method for separating hydrographs is based on differences
in the chemical composition of waters in the " reservoirs" thought to store
and release water (e.g., see Sklash and Farvolden, 1979). The assumpl"ion
behind this alternate method for separating hydrographs is that streamflow
during a storm event is a mixture of water that has been precipitated into
the catchment during the particular stonn (i.e., event rainfall or "new"
water) and water that was stored in the catchment prior to the onset of the
storm (i.e., pre-event or "old" water). Provided that the concentration of
some chemical component (i.e., tracer) of these two waters nalUrally
differs (and does not vary during the storm or val)' spatially across the

328
catchment), measurements of the concentration of the tracer in the two
components and in the mixture through time can be used to back-calculate
the contributions from the "old" and " new" reservoirs. For example,
suppose we make measurements of the chloride concentration in the rain
falling on the catchment, in the groundwater within the catchment, and in
streamflow (the mixture) during the course of a stoml. Let Q, be the total
measured streamflow (measured at a standard stream gage). Let Q" and Q"
be the flow contributions to the hydrograph associated with the "new"
water (rain) and the "old" water, respectively. Because we assume that
there are no other flow components, at any time t we know that

Q, = Q,, +Q
•. (10.1)

Note that the streamflow (Q,) and the flow components (Qn and Qol vary
with time.
If the concentration of chloride in precipitation (rain or snowmelt) is
relatively constant over the storm and if the concentration of chloride in
"old" water in the catchment is relatively constant in both time and space,
we can rcprcsent thcse concentrmions as C" and C" (note that we havc
measurements of these concentrations). Because the stream water is a
mixture of the old and the new waters, thc concentration of chloride in the
stream will change over the course of the stonn. We represent this
temporally changing concentration as Ct. Because we assume that the mass
of chloride observed in the stream is derived from only these two
reservoirs, we have:
Q,C, Q,,c,, + Q"C".
= (10.2)

Again note that we assume that we have measurements of C1 (i.e., stream

chloride concentrations) at several times over the course of the
hydrograph. Given measurements of: ( I ) QI' (2) C, (at several times), (3)
C" (assumed constant), and (4) Co (assumed constant), Equations I 0. 1 and
I 0.2 can be solved for Q" and Qo at every time for which measurements of
C1 are available:

Q =
[
(C'-C.)
" (C�-C")
Q.
] I
(10.3)

This calculation yields a hydrograph separation in terms of "old" and

"new" water.
Consider results for a storm that occurred in June 1 992, in the Shaver

329
Hollow catchment, Shenandoah National Park, Virginia. Bazemore (1993)
made hydrological measurements and chemical observations of a 0.082-
km2 sub-catchment at Shaver Hollow. The pre-event groundwater had a
chloride concentration of about 27 mol L-I and the event water that fell
through the forest canopy had a chloride concentration of about 4 mol L- I .
Chloride concentrations in the stream, measured over the storm
hydrograph, were used to separate the hydrograph into "old" and "new"
water (Figure lOA). Note that approximately 90% of the water under the
storm hydrograph was detennined to be "old" (pre-event) water and at the
time of peak discharge more than 80% of the total discharge was
detennined to be old water (Bazemore, 1 993). The use of chemical
composition of throughfall, soil water, and ground water to separate
hydrographs into components requircs a numbcr of assumptions about the
temporal and spatial variability in the various "components." The method
can be criticized because the assumptions may be poor for natural
catchments (e.g., see Harris et aI., 1995). Despite controversy about the
details of chemical hydrograph separations, the frequent observation from
such work that a large fraction of stormflow in forested, temperate-zone
catchments is "old water" is generally agreed to be valid. This approach of
using chemical signatures to identifY contributions to streamflow from
several sources can be useful for a variety of problems and is generally
termed end-member mixing analysis.

� .-------r �

30
�� 25
-' 00 0 0 Chloride
o
; 20 0
-

0 00
a
�•� 15 000
o
"Old" water '-'
discharge
.-
C
10
5
o +-�...,.,..�-,.....""
noon 6:00 p.m. midnight 6:00 a.m. noon 6:00 p.m. midnight
Figure l OA Chemical hydrograph separation for a sub-catchmcnt of
Shaver Hollow, Shenandoah National Park, Virginia. The data cover June
4 and 5, 1 992.

330
Data courtesy of David Bazemore.
The main conclusion from the application of thc chcmical·hydrograph·
separation method to this and other forested, upland catchments located in
humid, temperate regions is that observed streamflow is mostly old water.
Quickflow is not synonymous with "new" water and baseflow is not
synonymous with "old" water. An empirical separation of the hydrograph
into quickflow and baseflow is insufficient to describe actual now paths
within a catchment, which are necessary, for example, to account for
chemical reactions in a catchment. Describing the transport of solutes
within a catchment requires a detailed understanding of the physical
processes that govern the flow of water in catchments.

10.4 Runoff Processes

There are four paths by which water precipitated onto a catchment

ultimately can be discharged into a stream channel. The four now paths
(Figure 10.5) are: direct precipitation (or throughfal1) onto an active stream
channel, overland flow, shallow subsurface stonnflow, and groundwater
flow. Snowfall can be considered "delayed rainfall" once snowmelt takes
place, and can be assumed to follow the same flow paths to the channel.

10A.1 Direct precipitation onto stream channels

The contribution of direct precipitation (or throughfall) to an active stream

channel is usually quite small because the surface area of the perennial
channel system in most catchments is a small percentage of the catchment
area. However, in many catchments dramatic expansion of the channel
system can occur during wet seasons and during periods ofprecipitation.
In upland catchments, a significant fTaction of stonn runoff can result from
precipitation falling on this expanded channel network (Dunne and Black,
1 970). Overland flow caused by precipitation on saturated areas ncar the
stream is part of the variable source mechanism for runoff production
described below.

331
(a)

I I
I
/
-
Groundwater
discharge

Pre c i p i t a t i o n

(c) � � � �

I I
I
-
/
Groundwater
discharge

Figure 10.5 RunofTprocesses. Between precipitation or snowmelt events

(a), streamflow is maintained by groundwater discharge (baseflow).
Runoff is enhanced during and after precipitation events. Direct
precipitation 01110 stream channels or overland flow (b) contributes directly
to enhanced surface runoff. infiltrating rainwater or snowmelt that reaches
the groundwater system can enhance basctlow or provide return flow in

332
saturated areas (b). Saturation of penneable soil horizons also can produce
shallow lateral flow called subsurface stonnflow (e).

1 0.4.2 Overland flm\-'

Overland flow is water that flows aeross the ground surface and discharges
into a stream channel. If overland flow is to occur, water must accumulate
at the surface rather than infiltrate into the soil. Overland flow can occur
on a catchment for a number of reasons: (a) the catchment surface may be
nearly impermeable due to the presence of exposed bedrock or paving over
the surface; (b) the instantaneous rate of infiltration through the pervious
surface may be exceeded by the instantaneous rate of rainfall (or
snowmelt) onto the catchment surface, causing ponding of water at the
surface; and (c) the catchment soil upon which the rainfall is precipitated
may be saturated to the soil surface, causing ponding because the
precipitated water cannot infiltrate into an already saturated soil. Overland
flow in catchments is one of the most rapid paths that rainfall can follow to
the stream channel; therefore, overland flow is expected to contribute to
quickflow.
Overland flow dominates the response of many urbanized catchments
with large expanses of directly connected, impervious surfaces. Because
urbanization increases the fraction ofprccipitated watcr that follows the
rapid overland flow path through a catchment, incrcased frequency of
floods is one result of development. Leopold (1 994) reports that after
suburban development of a small catchment in Maryland, flood
frequencies on Watts Branch, the stream draining the catchment, increased
dramatically. Overbank flows in Watts Branch occurred about 1 . 4 times
per year before urbanization and about 6.5 times per year afterwards.
Overland flow of water from a catchment that occurs when the rainfall
or snowmelt rate exceeds the ability of the soil to allow water to infiltrate
is referred to as infLItration-excess overland flow, or Hortonian
overland now, thereby crediting the American hydrologist Robert Horton,
who reported occurrences of this runoff mechanism. Horton ( 1933) argued
that, over the time course of a stonn, the rainfall rate eventually would
exceed the rate at which water could infiltrate into the soils ofa catchment
(as in Figure 8.9), resulting in the accumulation of excess water at the
ground surface and overland flow in sheets. As discussed in Chapter 8,
infiltration into soils depends on hydraulic conductivity of the soil profiIe.
The hydraulic conductivity of the surface soil is conditioned strongly by
vegetation as well as by the texture of the mineral soil. For example,
hydrologists have observed that forest soils typically have a high

333
infi Itration capacity duc to thc presence of above-ground vegetation (and
decomposing vegetation on the surface), which protects the surface from
compaction caused by raindrop impact, and ofbclow-ground vegctation,
which provides roots that keep the soil porous and highly penneable
(Chapter 9.3.2). Infiltration-excess overland flow is not a significant
runoff-gcnerating process in such catchments. For example, up to 150 mm
of water per hour was sprinkled for several hours on an isolated block of
soil in a forested catchment in Maine with essentially no overland flow
(Hornberger et al., 1990). Conversely, when vegetation is absent from the
surface of a catchment, the surface soil tends to develop a "crust" with
relatively low hydraulic conductivity. Infiltration-excess overland flow can
be the dominant runoff-generating mechanism in catchments where the
land surface has been strongly disturbed (e.g., a plowed agricultural field
or construction site where vegetation was removed and the soil surface left
exposed) and in arid and semiarid regions with sparse vegetative cover.
Infiltration-excess overland flow generally does not occur at a constant
rate over entire catchments due to spatial differences in soil properties.
Some areas of the catchment may produce overland flow during virtually
all storms, while other areas may rarely, if ever, do so.
In many forested catchments in the temperate zone, the soil is capable
of allowing essentially all of the incident precipitation to infiltrate. I n such
catchments overland flow develops not because precipitation intensity
exceeds the infiltration rate, but because it falls on temporarily or
permanently saturated areas (wetlands) with no capacity for water to
infiltrate. Flow developing under these conditions is known as saturation­
excess overland flow (Figure I O.5b). The areas ofa catchment that are
prone to saturation tend to be near the stream channels, or where
groundwater discharge areas occur. These areas grow in size during a
stonn and shrink during extended dry periods (Dunne et aI., 1975; Beven,
1978). The areas on which saturation-excess overland flow develops,
therefore, change with time over a stonn. Because the expansion of
saturated areas occurs in a similar fashion as the expansion of the channel
network, this mechanism of runoff generation is often referred to as the
variable contributing area concept.

1 004.3 Shallo" subsurface stormflow

Water that has infiltrated the soil surface continues to be influenced by

gravity, causing water to percolate downward through the soil profile. In
general, the ability of the soils and rocks of a catchment to conduct water
(hydraulic conductivity) decreases with depth; water percolating

334
downward has thus been observed 10 "pile up," causing localized areas of
saturation in the soil. In these instances, water may move laterally toward a
stream by a process known as shallow subsurface stormOow (Figure
1O.5c). Some of the water in subsurface stormflow moves at a relatively
slow pace through the soil and contributes to the baseflow of streams,
particularly during wetter winter and spring periods. Subsurface stormflow
also may occur along prefcrrcd flow pathways (c.g., soil cracks, old animal
burrows, decayed root channcls, etc.) called macropores (Chapter 8) and
the flow to the stream channel along these pathways may be quite rapid.
Thus, subsurface stormflow can contribute significantly to quickflow. A
portion of the infiltmted water does not become subsurface stormflow; it
percolates downward until it reachcs the watcr table and becomes
groundwater.

10AA Groundwater flow

In Chapter 7 we presented several examples of groundwater movement

and discharge to topographic lows. [n general, groundwater flow is the
slowest of all flow paths through a catchment. Baseflow in low-flow
periods is composed almost entirely of groundwater discharge. As such,
bascflow tcnds to vary quitc slowly ovcr long time periods in response to
changing inputs ofwatcr through net recharge. In catchmcnts, the
groundwater discharge to a stream channel may lag the occurrence of
precipitation by days, weeks, or even years. When subsurface water
flowing downslope to the stream enters the saturated area near the stream,
some of thc water is forccd to reemcrge onto thc ground surface bccausc
the capacity of the soils and rocks to transmit all of the flowing water
downslope is insufficient. This reemerging subsurface water is known as
return now (Figure 1 O.5b). Like saturation-excess overland flow, return
flow can be quite rapid in some catchments with shallow water tables
where groundwater mounds may fonn which dramatically increase local
hydraulic gradients toward the stream (Freeze, 1 972a, I 972b; Winter,
1983). As was the case with subsurface stonnflow, groundwater flow can
contribute to either the quickflow or baseflow components of a stream
hydrograph, although in most instances the contribution to baseflow is
dominant.

1004.5 " Old" and "new" "ater

As we have noted (Section I 0.3), for strcams in forestcd upland

catchmcnts in tcmperate climates, thc bulk of quickflow (even at the peak

335
of a hydrograph) has the chemical composition of water that was resident
in the subsurface prior to the storm event. That is, the bulk of quickflow is
"old" water, indicating that watcr contained in soils and rocks in a
catchment must be able to reach the stream quickly under stonn
conditions. This observation may seem at first glance to be at odds with
the physical processes described above in which "old" groundw<lter is
viewed as contributing primarily to the (slow) baseflow portion of the
hydrograph. There arc a number of physical processes that can be called
upon to explain the observation that a large fraction of quickflow is "old"
water. Consideration of the physics of soil water suggests that processes
that cause "new" water delivered to the surface of a catchment to displace
"old" water and force it into thc stream are common. For example, even if
only a small amount of infiltrated water reaches the capillm)' fringe
quickly (e.g., through flow in macropores, especially in the riparian, or
near-stream, are<l), it can change the negative capillary-pressure head in
the capillary fringe to a positive pressure head. Such rapid increases in
positive pressure head can force " old" groundwater into the stream rapidly.
Another mechanism that contributes to the delivery of "old" water to a
stream under storm conditions is the entrainment of "old" soil water by
overland flow. Overland flow in riparian areas occurs in patches, with
surface water flowing into the soil, mixing with shallow soil water, and
then reemerging on the surface. The resulting overland flow contains a
portion of the "old" soil water with which the "new" water has mixed. The
conceptual explanations offered above appear to be consistent with
observations, but by themselves do not provide quantitative estimates for
the various flows. Quantification requires consideration of inflow, outflow,
and storage ofwatcr within a catchmcnt.

10.5 Contributing Area and Topographic Controls on

Saturation

As we saw in Chapter 7, the topography of the landscape excrts an

enonnous influence on the movement of water in the subsurface.
Topography likewise should control the development of areas of surface
saturation and runoff. lfwe could break a catchment up into blocks
("reservoirs"), we might be ablc to usc thc conservation of mass equation
to detennine the degree of saturation and potential for runoff generation
for each one. Each block would differ in its position along the hillslope
and in the slope of the land surface (and probably the water table) through
the block. Consideration of inflows, outflows, and runoff potential for all

336
of the blocks in a catchment could provide the starting point for routing
water through the catchment.

10.5.1 Contributing area

The degree or saturation of each catchment "block" depends on its water

balance. Irthe inflow to the upslope face of the block from higher ponions
of the catchment is greater than the outflow from the downslope face of
the block, the water table within the block will increase. The inflow rate
for a catchment block depends on the contributing area, the area of
catchment upslope from a given block that contributes inflow to that block.
Contributing area, A, depends on the distance to the divide above thc block
as well as whether it is convergent, divergent, or planar (Figure 1 0.6).
To define A, elevation contours are drawn at a specified contour
interval (for example, 1 0 m) for the catchment. Beginning at the base of
the catchment, lines are drawn perpendicular to each contour they cross,
fanning a network of curves similar to the flow nets described in Chapter
6. The lines perpendicular to the contour lines represent flow lines. In
areas of the catchment with a unifonn slope (planar sections), the flow
lines will have a constant spacing (Figure 10.7). In areas where the
hillslope is concave, the flow lines tend to converge as one follows them
downslope. The surface soil layer between two flow lines on a concave, or
convergent, slope is something like a converging channel. As the upstream
subsurface flow gets funneled into a smaller area, its depth increases and,
with sufficient supply, can saturate the soil. The opposite happens in a
divergent section. The increasing distance between flow lines allows the
subsurface flow to spread out and thin.
While inflow to a catchment block is proportional to contributing area,
local slope, tanp, controls outflow from the block. For example, if the
topographic and water table slopes of a block are relatively flat, the
hydraulic gradient is small and Darcy's law indicates that water movement
will be relatively slow in the absence of changes in hydraulic conductivity.
Therefore, we might expect a small outflow and an increase in water
storage through time within that block depending on the volume of inflow.
The increase in storage is even greater if the block is at the base ofa
convergent hillslope, such that a great deal of upslope flow into the block
occurs. If the water table reaches the surface, the block is completely
saturated and any additional water supplied to the block will run ofTas
saturation excess overland flow.

337
 10.5.2 Topographic iude\:

As described above, the important characteristics of a hillslope that

influence the likelihood of areas of saturalion and runoff developing arc
the upslope contributing area per unit contour length, (f = Ale, and the local
slope of the block, tan fi. These can be related to each other as discussed in
Section 10.5.3, or they can be combined into a single variable that
quantitatively captures the effect of topography, such as the topographic
index:

TI = In(altanlJ). (10.5)

<a>

Increasing contri buting area

•

Figure 10.6 Local slope and contributing area the water balance
for a catchment "block." The inflow rate is proportional to the contributing
area A, which depends on how long the hillslope is as well as whether it is
convergent, divergent, or planar (a). The local slope controls the outflow

338
from the blocks (b). lfinflow is smaller than outflow (lIpper left in b), the
water table declines. Conversely, if inflow is greater than outflow ({olver
right ill b), Ihe water table will rise and surface saturation may occur.

Specific contributing area, a = Ale, can be defined for each point within a
catchment if the topography ofa catchment is known. Although it would
be possible to estimate a from a high-resolution topographic map, most
studies of this sort use digital elevation data that can be used in
conjunction with gcographic infonnation system (GIS) software to
detennine the specific contributing area for each point in the catchment. It
is important to note that the results of an analysis such as this are highly
dependent on the quality and resolution of the digital elevation data.
Accurate identification of the channel network, in particular, depends on
using high-resolution elevation data. The current state of the art technique
for generating high-resolution topographic data is light detection and
ranging ([idar). In fact, the critical need for good topographic data to
define channels and channel networks has led to calls for lidar maps 10 be
produced to allow accurate mapping of flood plains (NRC, 2007).

339
 C I > c2 Convergent

< c2
CI
Divergent

C1 = c
2
Planar
Figure 10.7 Influence of topography on contributing area.

A map oflopographic indices for a catchment reveals areas where

nmoffprocesses such as saturation-excess overland flow are likely to
occur (Figure 10.8). High values of Ihe topographic index indicate areas
with large cOlllributing areas and rclalively flat slopes, typically at the base
of hillslopes and near the stream. These areas also correspond with
expected groundwater discharge areas (Chapter 7). Low TI values are
found at the tops of hills, where there is relatively little upslope
contributing area and slopes are steep. These areas correspond generally
with groundwaler recharge areas.

340
 Topographic index
In("/tan�) 5 6 7 8 9 10 11 12 13 14 15

(b) 0.14
•
•
n:I 0.12
"

•
0
0
� 0.1
�
u

o 0.08
-

t;-
o
II) 0.06
,
�

-!:: 0.04
•

•
>

III 0.02
�

•
a:

O 2:---4- 10
6 8 12 14 16
index
Topographic
Figure 10.8 Topographic indices for a catchment in Shenandoah National
Park. The spatial pattern (a) indicates a likelihood of saturation in the
central valley of the catchment. Thc distribution of values (b) is uscd in
TOPMODEL calculations as described in Section 1 0.6.

1 0.5.3 HiIIsJopc stabiJit�·

The hillslopes of ca\chments are not static feafUres of the landscape. Over
time, hillslopes change slowly due to erosion by runoff and rapidly due to

341
landslides (also known as debris flows). Landsliding on steep hillslopes
typically occurs in localized areas where the soils are saturated and the
water pressurc in the porc spaces is high. Surface runoff most often occurs
when soils become saturated so that precipitation can no longer infiltrate
(saturation-excess overland flow). If the surface runoff is deep enough
and/or the slope steep enough, the flow can dislodge and carry soil
particles from the hillslope to the channel, resulting in erosion of the
hillslope.
Dietrich ct al. ( 1 992) combincd simple cxpressions describing the
thresholds of saturation-excess overland flow, landsliding, and hillslope
erosion with detailed digital elevation data and careful field observations
to predict localions within a catchment where each of these processes
dominates. Assuming a constant transmissivity T of the surface soil layer,
the saturated subsurface soil discharge across a contour line of Icngth e is
given by Darcy's law as

f2"".� Tctanp. = (10.6)

The water-table slope is assumed to be equal to the surface slope. The total
amount of watcr reaching the length of contour (e) over a specified period
of time is AqlOloi> where qlOlal = R (the recharge rate, [L liD and A is the
upslopc contributing arca. In othcr words, q'QIaI is the volume of watcr pcr
unit surface area (or depth) that is moving through the hillslope per unit
time. The dilTerence between total runoff past a contour interval (qlOl(ll) and
saturated subsurface discharge (qs"bsllrjilce) is saturation-excess overland
flow. Thus, overland flow occurs when:

Aq,OO>/ > Tc tanP ( 10.7)

0'

A T
> 1M P (10.8)
c q.u
-

Erosion by overland flow will only occur in the parts of the catchment
where the overland flow is deep enough (large specific contributing area
Alc) or the slope is steep enough (large tan fJ) for the flow to dislodge the
soil grains. Dietrich et al. ( 1 992) proposed the following expression for the
erosion threshold:
A T a
-> lan/3+ .
Atan/3F
( 10.9)
c qM,w q.,•
•

342
where a [L2 II] characterizes the resistance of the soil to erosion.
Cohesionless material on a sloping surface becomes unstable, leading to
shallow landsliding, when the slope of the surface exceeds a critical value
dependent on the soil and water properties and the degree of saturation
described by:

tan!3> (p,�"up ) tan IP;. (I0.IO)

where p is water density (i.e., 1 000 kg m-3), Ps is soil density at saturation

(of the order of about 2000 kg m-3), 0 is the degree of saturation of the soil
mantle, and rPf is the internal angle of friction (a parameter expressing the
shear strength due to friction among soil grains). When the soil is saturated
(0 = I), this reduces to tan !3 > 0.5tanrPf for typical values of soil and water
density. When the soil is unsaturated (Q/Q/(l/ < QSllbsllilfJ<.()' 0 = Q,ollJ/ 1
Qy"b.llrflJc.' = AqroliJl I( Tc tan ft)·

These expressions for thresholds of saturation overland flow (Equation

10.8), erosion (Equation 10.9), and landsliding (Equation 1 0. 1 0) all depend
on the specific contributing area Ale and the slope tan ft. A plot of the
curves defining each threshold in tenns of these parameters shows their
relationship to each other and the topographic parameters. In Figure 1 0.9,
these threshold curves arc plotted for a total runoff qlQllJ1 = 50 mm day-I ,
assuming T = 10-4 m2 5-1, rPj= 35°, and a = 8 x 10-<i m2 S-I, which
produces good agreement between predictions and observations of
hillslope hydrologic characteristics in a small ( 1 .2 km2) northern
California catchment studied by Dietrich et al. ( 1 992).
A diagram such as Figure 1 0.9 can be used to determine areas of a
catchment that are susceptible to erosion and landsliding, which may serve
a variety of purposes including guiding land-usc decisions. The threshold
most susceptible to land-use practices is the erosion threshold. The value
of the parameter a characterizing the resistance of the soil to erosion
decreases rapidly with removal of vegetative soil covers and soil
disturbance. As the value decreases, the erosion threshold shifts to the left
in Figure 1 0.9, resulting in a larger portion of the catchment that is prone
10 erosion.

10.6 Routing Water through a Catchment Using

Catchment Models

343
The question thai we address in this section is how precipitation can be
rouled through a catchment to calculate streamflow. The most
straightforward way to usc the thcory developcd in prcvious chapters to
solve the catchment routing problem is to link together equations for
overland flow (e.g., Manning's equation), for flow in the unsaturated zone
(e.g., Richards' equation), and for flow in the water-table aquifer using an
equation for groundwatcr flow. The earlicst application of such a modcl
was presented by Freeze (1 9 7 1, 1972a, I 972b), who examined thc runoff
responses of a hypothetical hillslope to precipitation inputs. More recently
with advances in computational power, this approach has been used
successfully to simulate flows through catchments in great detail. For
examplc, the Pcnn Slate Integrated Hydrologic Model (PIHM, Qu and
Duffy, 2007; www.pihm.psu.edu) has bcen applicd at sevcral catchmcnts
to study coupled hydrological-biogeochemical processes.

, o' �

,
E
�

�
ti '
..
..
�

'"
c

::1 ,
.�

.c

'"
.�

c
-

0 -
c::
... ..,
... In
-
-

..,
.�

... c::
.�

..
Co �
(/) No surface runoff
,

1 0 -1 ,
Slope, Ian P

344
Figure 10.9 Regions of saturation overland flow, erosion, and landsliding.

Here we will consider a slightly different and simpler approach using

the concepts of catchmcnt "blocks" and the topographic index described in
Section 10.5. One framcwork that uses this approach is known as
TOPMOOEL, a catchment model that is based on the idea that
topography exerts a dominant control on flow routing through upland
catchments (Beven and Kirkby, 1 979). TOPMODEL uses the equation for
conservation of mass ("inflow rate minus outflow rate equals rate of
change of storage") for scveral "reservoirs" in a catchment-for example,
an "interception reservoir" and a "soil reservoir" (Figure 1 0. 1 0). Rainfall
provides the input to the interception reservoir, which is taken to have a
capacity of a few mm of water depending on the vegetation type (see
discussion of interception in Chapter 2). The outputs from the interception
reservoir are evaporation, calculated using an evaporation fonnula (see
Chapter 2), and throughfall, which then forms the input to the soil
reservoir. The conservation of mass equation again provides a method for
calculating the water balance for the soil reservoir. By linking together the
water balance equations for all of the hypothetical reservoirs in the
catchment, a routing computation can be completed.

345
 Evapotranspiration Precipitation

t t
I Interception reservoir I
Throughfall

Variable source area

(OVerlan\ IOW)

""\
I' '\
,
-
-
-
,
--, "

"- ,
\ " ,--� �) ,-' , ,
"

Soil
-
'
- "
' " .

reservo" , " \
-
" '

�
,
"

,,
, " ,
,
'
,
, ,
" , ,

���,
,

" '
, '
,

. .
Figure 10.10 Schemallc diagram of the TOPMODEL concept.

TOPMODEL perfonns the bookkeeping for the water balance

computations in the framework of topographically defined elements and
uses Darcy's law to calculate flow rates through the soil. Consider a
segment of a catchment defined by a cut along an elevation contour line at
the bottom, and "sides" running perpendicular to contours up to the
catchment divide (Figure 1 0. 1 1 ). Recall our assumption that flow is driven
by topography; hencc, this segment is just a portion of a flow net for the
catchment. The flow of subsurface water is conditioned strongly by the
local topography. The degree of convergence of "flow lines" (lines
perpendicular to the contours) detem1ines how much upslope area drains to
a unit length of contour at any given point. The local slope, the thickness
of the soil, and the hydraulic conductivity of the soil detcnnine the
"ability" of the soil to move water farther down the slope once it has
arrived at the given point. Source areas for surface runoff occur where
subsurface water accumulates-points to which large upslope areas drain
(such as convergent hillslopes or "hollows") and where the capacity to

346
drain the water downslope is limited (where slopes flatten at the base of
hollows). Conservation of mass can be applied to the segment depicted in
Figure 1 0. 1 1 to detennine the fluxes.

A
. ..
. .
Asat
/

Figure to. I I The water balance for a catchment hillslope segment.

Throughfall at rate p falls on the segment of area A and thickness D. A
portion, R, of this recharges the subsurface. Subsurface flow from the
segment occurs at rate Q"""-<I4rjllec' Surface flow, q",-er/ond' occurs from
saturated areas (saturation-excess overland flow). The local slope at the
outflow point, p, is considered to be equal to the slope of the water table.

10.6.1 TOI)MOlJEL calculations

Streamflow is the sum of subsurface flow and of overland flow from

saturated contributing areas:

(10.1 1 )

where q/olll/ is total streamflow. It has dimensions of[L il] (discharge [LJ
i l] divided by area [L2]); all of the flow quantities in TOPMODEL have
these units. The surface flow contribution is qllwrillll</ and q"ubs"r/<.ee is the
subsurface flow contribution (Figure 1 0. 1 1 ).
Surface flow is generated when precipitation falls on a saturated area
and from return flow, so:

(10.12)

347
where Asa/A is the fraction of the hillslope area that is saturated (Figure
1 0. 1 1 ), P [L [1] is the throughfall or snowmelt rate, and (jrellirn [L [1 ] is
the return flow.
We calculate (j$llbsur!l1u [L [ I] from total subsurface discharge Q"Ub.51"il1ce
= Tc tan f3 [L3 [11 (Equation 1 0.6), where T is the transmissivity of the
soil [L2 [lJ, c [L] is the contour width (length perpendicular to the flow
direction), and tan f3 is the slope. The transmissivity is equal to the soil
depth multiplicd by thc soil hydraulic conductivity. Note that the slope of
the water table is assumed to be the same as that of the land surface.
We assume that the saturated hydraulic conductivity of the soil
decreases with soil depth exponentially, a situation often observed:

(10.13)

where K(z) [L II] is the hydraulic conductivity at depth z (measured

positively in the downward direction), Ko is the hydraulic conductivity at
the surface, andf[L-I] is a parameter that governs the rate of decrease of K
with depth. To detennine the transmissivity of a saturated zone of a given
thickness (from a depth of water table z to a depth to bedrock D) Equation
1 0. 1 3 is integrated to obtain:

(10.14)

The term e-jD is generally much smaller than the term e-ft, so Equation
1 0. 1 4 can be simplified:

T= fe-f
K
t. (10.15)

Combining Equations 10.6 and 10. 1 5 gives the following equation for
subsurface flow:

(10.16)

TOPMODEL does water-balance accounting by keeping track of the

" saturation deficit," the amount of water that one would have to add to the
soil at a given point to bring the water table to the surface. Because one
has to track saturated areas if saturation-excess overland flow is to be
computed, this makes sense. To implement computations in temlS of s, the
saturation deficit, z is replaced by s/¢ where s [L] is the saturation deficit
and ¢ is the porosity of the soil. Substituting for z in Equation 1 0. 1 6 gives:

348
 Q,....�,
.. = ; -J�fclan/l
K
t? (10.17)

To make things "neater," we introduce some simplifying notation. We can

replace Kcl/with T",ux' because this term is the transmissivity when the soil
is completely saturated (s = 0). We can also replacej7if> with 11m, a soil
parameter inasmuch as/defines the decrease of K with depth and if> is
porosity. Equation 1 0. 1 7 can then be written:

Q...m....... '" T..-e--;;c lanfJ·

•

(10.18)

We may now proceed with a calculation of the water balance for a

hillslope slice (Figure 1 0. 1 J ). This will lead, in conjunction with Equation
1 0. 1 8, to expressions for As,,/A and qSllbsllrj;Ice. Equation 1 0. 1 8 gives the
subsurface flow being transmitted downslope at any point. The flow
coming into the slice at any time is:

(10.19)

where R [L II] is the recharge rate and A is the area of the hills10pe slice
-the section of hillslope that drains past the section of contour (c) in
question. The "great leap" of TOPMODEL is to assume steady-state
conditions. Then, Qsubswface = QR' or
•

RA 0:= r.....
,e--;;crnnfJ. (10.20)

This equation can be solved for 8:

s O:= -lIItn ( ) ( )
r.
R
....,
-mln a
Ian
fJ
. (10.21)

where (I = Ale, the specific contributing area. The second tenn on the right
of Equation 1 0.2 1 describes the way in which topography controls the
propensity for every point in the catchment to reach sahlration (i.e., the
propensity of each point to generate saturation-excess overland flow
during stonns). Ifs is less than or equal to zero, the soil is saturated. From
Equation 1 0.2 1 we see that this occurs most easily for points within the
catchment where the topographic index (TI = In(aitan {f)) is large.
Until this point in the discussion, we have been referring to an
individual catchment hills10pe, or hi11slope segment, defined by a pair of
streamlines and extending from the stream to the catchment divide (Figure
1 0. 1 1 ). However, we could also consider any point in the catchment and

349
calculate the upslope contributing area and the local slope. In this way, we
can compute the distribution of topographic indices for the entire
catchment. In practice, the computations arc often done for "blocks"
delineated on the basis of DEM (Digital Elevation Model) or surveying
data. The topographic index, and therefore the contributions of surface and
subsurface flow to streamflow, can be calculated for each block. The
saturation deficit can also be calculated for each block, using Equalion
10.21 . Furthennore, these quantities will be identical for two blocks with
the same topographic index, as long as R and Tm,u: are spatially constant.
To solve for the catchment-avcrage saturation deficit (x), we can
integrate Equation 10.21 over the catchment and divide by the area. Here
we assume that R and Tmax are constant over the catchment:

(10.22)

where /.. is the mean In(a/tan fi) for the catchment. Combining Equations
10.21 and 10.22 gives:

(10.23)

This equation states that the saturation deficit at any point in a catchment
is equal to the average saturation deficit for the catchment plus a soil
parameter, III , times the diffcrence between the average topographic indcx
and the local topographic index.
Now we have a way to calculate Am/A-compute s at any point and
check to see if it is less than or equal to zero. We can estimate 11/ from soil
characteristics, calculate /.. and In(altan P) from a topographic map, and can
keep track of .'II by water-balance accounting
(p, interception, el,
subsurface flow, and overland flow). Ifs < 0, the soil is completely
saturated and any rain on the surface will become overland flow. The rate
of flow produced by this mechanism is detem1ined lIsing the throughfall
intensity and fractional catchment area that is saturated (e.g., Equation
10. 1 2). Return flow occurs where s < 0, and the rate of return flow is equal
to IsIAs,,/A .
Next we develop an expression for the mean subsurface discharge, q
subsurface' Integrating Equation 10. 1 8 over the catchment area and dividing
the result by the catchment area yields:

( 1 0.24)

350
Thus, in TOPMODEL, subsurface flow is controlled by soil characteristics
CTma:< and m), topography CA.), and the average saturation deficit of the
catchment.
This is not all of TOPMODEL, but it is a summary of the main
conceptual points. Added to the formulations above are the other standard
components of the watcr budgct-evapotranspiration, snowmelt, channel
routing (e.g., a simple reservoir routing method). All of the water-balance
accounting parts of the model are simple applications of the conservation
of mass. A fuller description of TOPMODEL is available in Wolock
( 1993 ).

10.6.2 Hydrograph simulation using TOPI\10DEL

TOPMODEL simulates the runoff response of a catchment to precipitation

cvents. By tracking the change in storage within each "block" defined by a
value ofTI, the model not only routes water through the catchment, but
enables predictions of areas that will become saturated during a stonn. The
soils at the base of hillslopes (blocks with high Tis) tend to become
saturated as a stonn progresses and saturation-excess overland flow is
produced. Streamflow is taken as the sum of subsurface flow (net outflow
from the soil reservoir) and the overland flow. Evapotranspiration removes
water from the soil store according to a rate calculated with an
evapotranspiration formula.
To illustrate the use of TOPMODEL to simulate streamflow from a
catchment, we examine the Snake River (Hornberger et al., 1994). The
catchment of the Snake River ncar Montezuma, Colorado, is 1 2 km2 in
area and is mountainous, ranging in elevation from about 3350 m to 4120
m. The Continental Divide bounds the catchment on the south and east.
Approximately half of the catchment is above the tree line. The data
necessary to implement the model were taken from nearby stations. A
snowmelt model that takes melt to be proportional 10 temperature was used
in the water-balance calculations within TOPMODEL.

351
 Simulated
Observed

!.,
-
-

.,
."
E
E
'"
-

�
•

'"
.,
�

.r=
<.>
'"
.-

.g (ij )5. � § � a. U
O E; �
� �
I::
� < �
::J

� � • � Z 0
Figure 10.12 Observed discharge ( 1984) in the Snake River, Colorado,
compared with TOPMODEL calculations.
Data courtesy of Ken Bencala and Diane McKnight.

The general shape of the snowmelt hydrograph for 1984 simulated

using TOPMODEL is in accord with the measured hydrograph (Figure
10. 1 2). The initial timing of the hydrograph fisc is latc in the simulated
hydrograph; presumably, the temperature index model calculates the
initiation of significant snowmelt to be later than when the melt is actually
initiated. This likely is due to errors in the simple model that was used for
snowmelt. The remainder of the simulation follows the observed
hydrograph quite well although the late pan of the recession is
overpredicted. Overall this simulation is quite reasonable given the
limitations in the data (e.g., the need to extrapolate the discharge,
precipitation, and temperature data from downstream stations).

10.7 Concluding Remarks

AI the beginning of the chapter we suggested that solving the catchment

routing problem was important for a number of reasons. The material
presented above introduced some of the ideas used in routing Ilows

352
through a catchment and discussed some of the complexities in the
processes that give rise to observed streamflow dynamics. The advances
that have been made in calchment modeling have been substantial, but the
state of the science is still inadequate to allow acceptable routing for some
of the most serious problems (Hornberger and Boyer, 1995). One example
is predicting the flow paths of natural solutes and contaminants in
catchments, an example of which is the acidification of streams during
stonns observed in the northeastern United States (Figure 1 0. 1 ) and
elsewhere in the world.
Models such as TOPMODEL are necessary for calculation of
acidification and of the movement of other chemicals through a catchment.
A mass-balance model is required to do an accounting of chemicals, in
addition to an accounting of waler. Robson et al. ( 1 992) combined
TOPMODEL with a simple chemical mixing model 10 describe the
episodic acidification observed in a stream in Wales. This is an example of
the approaches that are being used to study a variety of pollution problems
in catchments. The transport of chemicals through a catchment is
controlled to a great extent by interactions between the water and the soil
and rocks in the catchment. Detennination of the flows through the soil
and groundwater reservoirs in the catchment is critical for realistic
predictions of the fate of pollutants.
TOPMODEL calculates not only the stream hydrograph but
infonnation that is useful for linking hydrological calculations to
hydrochemical models (Cosby el al., 1987). For example, TOPMODEL
calculates subsurface flow, overland flow, and saturation deficit (depth to
water table), quantities that can be used to detennine how waters from
different parts of a catchment mix to produce the chemical composition
observed in streams. As a final illustration, we look once again at the
Snakc River in Colorado.
In many lakc and strcam systems, the concentration and composition of
dissolved organic material (DOM) is a critical water quality characteristic.
One example of a process controlled by DOM is the formation of
trihalomethanes, compounds that are known carcinogens, in a drinking
water supply as a result of interactions between chlorine (used to treat
drinking watcr in public supplies) and components of the DOM during
water treatment. The DOM also can have indirect effects on water quality
by influencing internal processes of aquatic ecosystems, such as
photosynthesis and heterotrophic activity. Therefore, one question that
hydrologists and ecologists arc intercstcd in answering is how hydrological
processes transport dissolved organic carbon (DOC) to streams. An
example is the Snake River catchment, where DOC builds up in near-

353
surface soils beneath the snowpack due to microbiological activity (Boyer
et aI., 1997) and is released during snowmelt in the spring. Quantifying the
"flushing" of DOC from ncar�surface soils requires knowledge of the
extent of saturation of the soils and the rates of water movement through
them. TOPMODEL can provide a computation of these hydrological
items. For example, based on the TOPMODEL simulation shown in Figure
1 0. 1 2, the concentrations of DOC in the stream can be calculated (Figure
1 0 . 1 3). The calculated values early in the melt period do not rise to the
measured levels because the snowmelt was not calculated in TOPMODEL
correctly al early times (Figure 1 0. 1 2), but the general timing and
magnitude of calculated DOC levels are consistent with measurements.
These results suggest that such calculations could be useful to water
managers who might want to avoid taking water into the water�supply
system during times when DOC concentrations in the stream are high.
Forecasts oft1ushing of DOC would enable managers to plan ahead to
store treated water so they could avoid undue production of
trihalomethanes in the drinking water supply.

5 -_
+ + Simulated
+
�
° + 0 Observed
4-
�
-

en
E 3.5 -: +
� °
c: 3
0
-
..
0
. -

� 2.5: 0 ° + 0
c:
-

., + +
u 2 -: +
c: + +
0 0
u 1 .5 00
0 ++
6 00 qlQ
� fjl
0 1 -_ 0
C
0.5 -:
0 ,. . , • • •
. ' , " , .
>- c: '" c. > "
. ' '

::> "
'" ::> ::> w 0 w
- -

...,
::; ..., <i (J) 0 z 0

354
Figure 10.13 Measured and simulated dissolved organic carbon (DOC)
concentrations ( 1984) in the Snake River, Colorado.
Data courtesy of Ken 8encala and Diane McKnight.

10.8 Key Points

• Streamflow hydrographs appear to have two temporal components

(quickflow and baseflow). Separation of hydrographs into these
components is done by "eye" or by purely empirical methods. {Section
I D.2 }

• Streamflow hydrographs can also be separated into two components that

attempt 10 describe the source of stream water (e.g., old and new water).
This "chemical" separation is based on Ihc mixing of chemically
different waters. By solving mass balance equations for water and a
conservative solute, estimates of Q" and Qo can be computed for any time
for which a measurement of Cr is available by: QiQ/ ::=: [(CrCo)/Cn-Co)]
and QjQ, HQjQ,). {Sect;on ID. 3}
�

• The application of the chemical-hydrograph-separation method to many

forested, upland catchments located in humid, temperate regions shows
that observed streamflow is mostly old water. "Quickflow" is nOI
synonymous with "new" water and baseflow is not synonymous with
"old" water. {Section 1O.3 }

• Runoff from a catchment can follow four different flow paths: direct
precipitation onto stream channels, overland flow, shallow subsurface
stormflow, and groundwater flow. {Section lOA}

• There are two different mechanisms by which overland flow is produced

in catchments: infiltration-excess overland flow and saturation-excess
ovcrland flow. Infiltration-excess ovcrland flow is considered dominant
in systems where the soil profile or soil surface has been radically
disturbed (e.g., agricultural catchments), in arid and semiarid regions
where vegetation density is low, and in urban areas where the surface is
made essentially impemleable by paving or other construction.
Saturation-excess overland flow is most significant in humid regions
with dense vegetation and topographic conditions (concave slopes with
flat valley bottoms) that cause the water table to be located relatively
close to the surface. {Section IOA.2 }

355
• Shallow subsurface stormflow may occur when permeable surficial soils
become saturated. Water may then flow to the stream through these soils.
{Section 1 0.4.3 }

• Discharge of groundwater 10 streams is responsible for baseflow to

streams during fair weather periods. Also, groundwater can discharge to
saturated surface areas (return flow) or to a stream or river during
precipitation events, and thereby contribute to increased streamflow
during stoml events. {Section 1 0.4.4}

• Inflow of water to a segment of a hillslope is proportional to upslope

contributing area, which depends on the distance to the divide and the
shape of the hillslope (e.g., planar, convergent or divergent.) {Section
10.5.1 )

• The ratio of contributing area per unit contour length to the local slope of
the topography, tenned the topographic index, provides a useful measure
ofthc likelihood of saturation in a scction ofa hillslopc. {Section 1 0.S.2}

• Relationships between the ratio of contributing area per unit contour

length and local slope can be used to identifY regions of a hillslope prone
to surface runoff, erosion, and landsliding. {Section 1 0.5.3 }

• Routing of water through a catchment can be done using computer

simulation models. One such model, TOPMODEL, is based on use of a
steady-state now net to compute subsurface flow. The modcl also
calculates overland flow from saturated areas by linking topography with
water-balance computations. {Section 10.6}

10.9 Example Problems

Problem I . A hydrologist studying runoff generation in a catchment

measures the following chloride concentrations during the peak of a
rainstorm event: CII = 4.5 pmol L-1 , Co = 40.5 I1Inol L-I, and C1 = 36.0
I1mol L- I. What fractions of total streamflow arc contributed by new watcr
and old water?

Problem 2. Data on hydraulic conductivity versus depth in a soil on a

forested slope arc presented by Harr ( 1 977). Examine the data to detennine
whether the assumption that K decreases exponentially with depth is
reasonable, and, if it is, estimate Ko and!

356
,

Depth in soil (m) Hydraulic conductivity, K (m S-I)

0.10 9. 8 x 1 0-4
0.30 1.1 x 10-3
0.70 4.5 x 10-4
1.10 4.9 x 10-4
1.30 4,4 x 10-5
1 . 50 6 . 1 x 10-5

Problem 3. Two "strcamtubc" segments of a catchment (i.e., portions of a

flow net based on topography) have properties shown below:

Segment 1 Segment 1

Up§lope area, A (ml) 500 500

length of wntour at base of $egment, I; 1m) 3.5 2S

Stope at base of segment, tanfJ 0.02 0.08

Calculate the topographic index for each segment and indicate which
segment is more likely to produce saturation-excess overland flow.

Problem 4. The soil of a given catchment has a porosity (l/J) of 0.4 and
hydraulic conductivity as described in Problem 2. The catchment has an
average value of the topographic index of 3.5. Under conditions where the
average saturation deficit of the catchment is 100 mm, calculate the
subsurface flow to the stream, qs"bsllrP�c(" If the throughfall rate at this time
is 4 x 10-3 mm S-I (14 mm hr-1), then what fraction of the catchment
would have to be saturated if saturation-excess overland flow were to be
equal 10 one-half of the subsurface flow?

10,10 Suggested Readings

Dunne, T., and L.B. Leopold. 1978. Water in environmental planlling. San
Francisco: W. H. Freeman, Chapter 9, pp. 255-278.
Wolock, D.M. 1993. Simulatillg the variable-source-area concept of
streamflolV gelleralion lVith the Watershed Model TOPMODEL. U.S.
Geological Survey Water-Resources Investigations Report 93-4124.

357
11 Water, Climate, Energy, and Food

1 1 .1 The Role of Hydrology in a Changing Planet

1 1 .2 Water, Energy, and Food: Complex Interdependencies
1 1 .2 . 1 The water-energy nexus
1 1.2.2 The water-food nexus
11.2.3 Globalization of water
1 1 .3 Impacts of Changing Climate on the Water Cycle
11.3. 1 Effects of increasing temperature on precipitation
and runoff

1 1 .3.2 Effects of increased temperature and CO2 on

evapotranspiration

11.4 Challenges for the Future

11.5 Suggested Readings

11.1 The Role of Hydrology in a C hanging Planet

As noted in Chapler I , the study of hydrology began long ago. As human

populations have grown and redistributed, the pressing questions for which
an understanding of hydrological processes is valuable have shifted and
expanded. Of course the core question remains how to provide the quantity
and quality of water needed to support human populations and the
environment on which we depend. But with a growing popUlation and
changing climate, new questions have emerged that require hydrological
research and applications to move in new directions. In fact, although the
hydrological sciences are critical to addressing new and existing water­
related issues, the complex interactions and feedbacks among hydrology,
geomorphology, biogeochemistry, ecology, and climate mean that
understanding and solving problems related to water is an increasingly
interdisciplinary enterprise.
A recent study by the National Academy of Sciences (NRC, 20 12a)
examined the challenges and opportunities in the hydrological sciences in

358
the early 2 1 st century. The report notes:

Fundamental new drivers of hydrologic sciences in Ihe 2 1 st century rest on

thc realization that: (a) humans arc a dominant influence on water
suslainability both al the global and local scales, (b) Ihe world is becoming
exceedingly "flat" not only with respect to rapid dissemination of scientific
knowledge, but also with respect to learning from distant environments
currently undergoing rapid change (e.g., deforestation, drought, agricultural
expansion, etc.) and predicting future water scenarios in other parts oftne
world, and (c) the natural world is a highly non-linear system of interacting
parts at multiple scales prone to abrupt changes, tipping points, and
surprises much more onen than previously thought possible.

Among the key challenges identified in the report are impacts of

humans and climate on the water cycle, interactions between hydrological
processes and terrestrial ecosystems (see Chapter 9), and the ways in
which hydrological sciences can infonn solutions to the problems of
ensuring the quantity and quality of water nceded for drinking, agriculture
and energy (NRC, 2012a). Here we consider how hydrological research
and understanding can contribute to the first and last of these.

i 1 .2 Water, Energy, and Food: Complex

interdependencies

Several years ago, Richard Smalley of Rice University posed the question
to a variety of audiences, "What will be the top ten problems facing
humanity over the next fifty years?" The answers consistently showed
energy, wafer, and/ood as top items (http://cnst.ricc.edu) . Without a doubt,
these resources are of critical importance and will only become more so in
the future as global popUlation increases toward 9 billion and as the effects
of anthropogenic climate change become manifest. As we seek 10 manage
these resources to meet demands, we are confronted by the realizat"ion that
the development and use of each has impacts on and is impacted by Ihe
development and use of the others (Perrone and Hornberger, 2013).

11.2.1 The \\atcr-cnergy ne\.us

As one example, consider the generation of electricity. A large fraction of

the generation is performed in thermoelectric plants where water is
converted to steam (using heat from burning coal or natural gas or from
nuclear reactions), the steam is run through turbines 10 generate electricity,

359
and then the steam is condensed by cooling. The cooling process requires
water, most often fresh water, and lots of it. According to the U.S.
Geological Survey (Barber, 2009) water withdrawals in thc United States
for thennoelectric power generation in 2005 represented 49% of total
water withdrawals and exceeded the sum ofwilhdrawals for irrigation
(31 %), public water supplies ( 1 1%), and industry (4%). Much of the water
withdrawn for cooling is returned to streams, albeit at a higher
temperature, but some of the water is consumed by evaporation. In 2008,
U.S. power plants withdrew between about 225 and 650 million cubic
meters and consumed from I I to 23 million cubic meters of water (Averyt
et aI., 2 0 11) .
Thermoelectric plants are not the only form of power generation
contributing to the withdrawal and consumption of large volumes of water.
For instance, solar power generation requires substantial amounts of water
to elean the solar panels from the dust deposited on their surfaces.
Needless to say, solar panels are often deployed in predominantly clear sky
regions with prevailing cloudless conditions and consequently arid
climatc. Thus, the supply of the water rcquired to clean solar panels and
sustain solar energy production can be problematic. Although hydropower
often is considered to be fairly non-consumptive in terms of water, an
analysis of large hydroelectric dams worldwide indicates that the blue
water consumption from them amounts to about 10% of the blue water
consumed by crops (Mckonncn and Hoekstra, 2012). Also, in some
instances the release of water from dams to run the turbines can preclude
ils diversion into irrigation canals causing a conflict between water for
crops and water for energy. Consider the case of Sri Lanka, where about
40% of the electricity is generated through hydropower. In times of plenty,
there may be enough water to supply all uses, but under drought, the
competition for water between agriculture and electricity generation can
create problems, and decisions have to be made about how to apportion the
releases. The issues involved in making decisions are complex and there is
no "right answer" (e.g., see Molle et aI., 2008).
Conversely, water distribution, treatment, and end-use require quite a
significant amount of energy. A 2005 study for California indicated that
"water-related energy use consumes 1 9 percent of the state's electricity, 30
percent ofils natural gas, and 88 billion gallons [330 billion liters] of
diesel fuel every year-and this demand is growing" (Klein, 2005). This is
a tremendous amount of energy and one conclusion of the study was that
the state could most readily meet its short-tenn (5 years) energy
conservation goals by conserving waler.
The water-energy linkage is further complicated when we consider the

360
third of Smalley's top problems. The link between water and food is
obvious. In rain-fed agriculture, crops are part of the green water portion
of thc watcr cyclc whcrcas irrigation convcrts substantial portions of blue
water into green water (see Chapter 9) . Therc also are links of food to
energy, perhaps most notably with respect to the increasing demand for
biofuels which leads to displacement of land and water resources away
from growing food. Globally withdrawal of fresh water from strcams and
aquifcrs is dominatcd by irrigated agriculture, accounting for some 70% of
the total (Rosegrant et al., 2009). Considering population growth, the
escalating demand for biofuels, and shifts toward diets with more meat
protein, which requires more water than a diet dominated by grains (see
Section 1 1.2), the demand for water for irrigation is likely to grow
globally.

11.2.2 The water-rood ne\us

To explain the relationship betwccn watcr and food security, wc nccd to

stress three important points and stay away from some common
misconceptions. First, about 86% of the human appropriation of water
resources, when both blue and green water (i.e., both irrigated and rain-fed
agriculturc) are accounted for, is used by agriculturc in thc production of
food, fibers, and biofucls (Figure 1 .2). Most ofthc pcr capita water
consumption (or water footprint) is contributed by food production (600-
1 800 m3 per person each year) while a much smaller amount of water is
used for drinking (�l m3/personly) and household needs (25-360
m3/person/y; e.g., Falkenmark and Rockstr6m, 2004). Thus, the global
water crisis is about a hungry, rather than a thirsty, humanity. Second,
even though most of agriculturc (aboul 80% of the cultivated land) is rain
fed, the dcbate about water crisis often tends to mislcadingly conccntratc
on drinking and irrigation water (i.e., blue water) without regard for our
greater reliance on green water. Third, when we look at the use of blue
watcr rcsourccs, wc need to rccognize the diffcrcnce bctween watcr
withdrawal and consumption: while withdrawal refers to thc cxtraction
from surface watcr bodies and aquifcrs, consumption refers to losses by
evaporation, transpiration, or incorporation in crops, organisms, and
products. As noticed in the previous section, much of the water withdrawn
for hydropowcr production is not consumcd but returned to watcr bodies
with a lower gravitational potcntial energy. Conversely, part of the water
withdrawn by cooling plants for Ihenno-power generation is evaporated
(i.e., consumed), while the rest is returned to water bodies at higher
temperatures. Similarly, only part of blue water withdrawals for irrigation

361
is consumed, while the rest is eventually returned to water bodies. Because
transpiration sustains the productivity of natural and agro ecosystems, it is
often referred to as a productive consumptive usc of water, while
evaporation is defined as an unproductive fonn of water consumption.
Thus, water consumption in irrigated crops is partly contributed by uptake
and transpiration by plants (productive consumption) and partly lost in
evaporation (unproductive consumption); the relative importance of these
productive and unproductive forms of consumption determines the
efficiency of an irrigation technique.
The estimation of the blue water footprint of crops or other
commodities is based on consumption and not withdrawal rates (Table
1 1 . 1 ). The calculation of the overall water footprint of these commodities
should account for both blue and green water consumption. Such a
calculation requires an analysis of all the water consumed in the course of
the production cycle of thaI good (a life cycle analysis). The results of this
analysis are available for all the major agricultural and industrial
commodities. Some examples are summarized in Table 1 1 . 1 .
Animal products have in general a much bigger water footprint than
vegetables. The production of the same amount of food calories requires
on average 8 limes more waler in the case of meat (4 mJIIOOO kcal) than
for plant products (0.5 mJII 000 kcal; Falkenmark and Rockstrom, 2004).
Thus, the water footprint of food varies across societies (600-1800
m3/person/y), depending on the prevalent diet, particularly on the reliance
on meat. Falkenmark and Rockstr6m (2004) estimated that the water
footprint ofa balanced diet (20% of kcal intake from meat) is about 1300
m3/personly. Typical patterns of economic development exhibit an
increase in meat consumption as nations become wealthier. This shift in
diet and the escalating demographic growth are both responsible for the
ongoing increase in human demand for water resources. Thus, mankind's
appropriation of water resources poses some important ethical questions
on how far humanity should go in subtracting water (and land) from
natural ecosystems and depleting groundwater stores (i.e., aquifers), and
how water should be shared among different societies.

1 1 .2.3 Globalization of watcl"

While climate controls the supply of freshwater resources (e.g., see Figure
1 . 1 ), population size and the type of diet determine the demand (Figure
I I . 1 ). In regions where the demand exceeds the supply, societies might
either have to sustain smallcr consumption rates or import food products
from other, more water-rich regions (Figure 1 1 . 1 ). Exporting and

362
importing food can be thought of as an indirect trade in water; this water is
referred to as virtual water (Allan, 1 998).
Virtual water trade allows nations to virtually share and exchange water
resources. Often known as the globalization of water (Hoekstra and
Chapagain, 2008) this phenomenon can be visualized by mapping the
global network of virtual waler trade (Figure 1 1 .2). In 20 I 0 the total
volume of virtual water traded in food commodities (2.8 x lOll m) y-I)
accounted for about 24% of the average global freshwater resources used
for food production ( 1 . 1 8 x 1013 mJ y- I ; Carret al., 2012). Virtual water
trade is crucial to the food security of many countries. Although the trade
in virtual water can be viewed as benevolent in that it can reduce
malnourishment in countries where local food supplies become limited by
water availability, there is some concem that in the long run, the ability of
countries to cope with drought might be compromised by an over-reliance
on virtual water (D'Odorico et aI., 2010).

Table 1 1 . 1 . The water footprint (in mJ kg-I) of some major food

commodities

363
 Producing Country
Glob�1
Product
Aver�ge
Austr�lI� Br��il Chln� USA

Maize Green 0,75 1.62 0.79 0.52 0.95

Blue 0.68 0.00 0.Q7 0.06 0.08

Total 1.43 1.62 0.B6 0. 59 1.03

Rice, paddy Green 0.25 1.94 0.55 0.42 1.15

Blue 1.15 0.36 0.25 0.85 0.34

Total 1.40 2.30 0 79
. 1 . 27 1.49

Wheat Green 2,00 1.99 0,82 1.87 1.28

Blue 0.02 0.00 0.47 0.09 0.34

Total 2.01 1.99 1.29 1.96 1.62

Soy beans Green 1.89 2.18 2.55 1.56 2.04

Blue 0.00 0.00 0,25 0.09 0.07

Totol 1.89 2.18 2.80 1.65 2.U

Tomatoes Green 0.04 0.07 0.18 0.03 0.11

Blut.' 0,07 0.02 0.00 0.07 0.06

Totol 0.11 0.09 0.19 0.11 0.17

Swine meat Green 3.86 4.44 3.69 2.99 3.58

Blut.' 0.88 0.54 0,29 0.46 0.33
Totol 4.74 4.97 3.97 3.46 3.91

Bovine meat Green 10 .30 13.65 9.08 9. 18 10.23

Blue 0,4) 0.12 0,)5 0.37 0,)9

Totol 10.13 13.78 9.43 9.55 10.62

Source: Based on Mekonnen and Hoekstra (2010).

If demand :> supply If suppl :> demand

I
Water deficit Water sumlus

r '---Import
-l
Consumption < demand Virtual Export

I
of food water trade of food

Water stress: famine,

I
Consumption = demand
malnourishment
Figure 11.1 Demand and supply of water for food production.

364
 ..

Figure 11.2 The global network ofvinual water trade for the year 2010,
based on data from FAOSTAT. Only the top links, which transfer 50% of
the total virtual water flux, are shown. Net exporters are shown in dark
blue; net importers in light blue.
Courtesy of Joel A. Carr.

There is no obvious solution to the likely water shortages facing

populations living in regions of chronic water scarcity or temporary
drought. Overall, there are only a few options to meet the water needs of a
constantly growing global population while reducing malnourishment and
famine (e.g., Falkenmark and Rockstrom, 2004):

• Increase the lise o/blue waterfor irrigation. Crop yields can be greatly
enhanced by switching from rain fed to irrigated agricuhure. However,
most of the suitable freshwater resources are already overexploited and
overcommitted. Many water courses are left with only minimal flows,
and some of them do not even make it to the ocean anymore. Therefore
there are only limited opportunities 10 increase water withdrawals for
irrigation.

• Expand agricultural lalld at the expellse ofnatllral eco�ystems.

Cultivated land area is not likely to substantially increase (Fedoroff et al.,
2010). The clearing of most tropical forests results in agricultural soils
that are unsuitable to sustain good long-tenn crop yields. Considering the
environmental costs associated with CO2 emissions and biodiversity loss,
the benefits of new conversions to agriculture are not sufficient to justify
further expansion of cultivated lands (Foley et al., 2011).

365
• "More crop per drop." Food production can be enhanced by using
genetically modified crops with higher water use efficiency and "climate
sman" agriculture.

• Reduce lion-productive \vater consumption. The usc of more efficient

irrigation techniques (e.g., drip irrigation) and of new soil and water
management methods should allow for a decrease in evaporation while
sustaining productive water consumption (transpiration).

These approaches are not mutually exclusive. The combined adoption

of a variety of methods aimed at increasing agricultural yields through an
integrated use of new crop and irrigation technology while enhancing food
awareness through educational programs appears to be the most promising
strategy to meet the increasing demand for water and food (e.g., Tilman et
aI., 201 I).

11.3 Impacts of Changing Climate on the Water Cycle

It is clear from Section 1 1 .2 and elsewhere in this text that humans are
directly affecting the water cycle in dramatic ways through agriculture,
energy production, groundwatcr extraction, dcforcstation (and
reforestation), and urbanization, to mention just a few. Humans are also
affecting the water cycle more indirectly by increasing atmospheric levels
of carbon dioxide (C02), which alters air temperature and transpiration. It
has been suggested that these alterations may cause an acceleration of the
water cycle (see Box 1 1 . 1 ) and global redistribution of water, leading to
more extreme weather events that may increase the number of floods in
some regions while making other regions are more drought-prone. Because
the water cycle is tightly linked to biogeochemical cycles (e.g., carbon and
nitrogen) and vegetation distributions, changes in the water cycle due to
climate change are likely to have far-reaching effects on ecosystems and
landscapes.

1 1-3.1 Effects of incn�asin� temperature on preCipitation and

runoff

As atmospheric CO2 levels rise, so will air temperature. Climate models

help to quantitatively project future global surface warming in response to
specific scenarios of future emissions of CO2 and other greenhouse gases.
Low to high emission scenarios lead to predicted increases in global mean

366
surface air temperature of 1.8°C to 3.4°C by 2100 (Meehl et aI., 2007).
These increases will not be uniformly distributed across the globe, with
somc areas, such as ponions ofthc southwest U.S., cxpericncing a largc
increase in the number of days with a heat index > 100°F (37.8°C) by the
end of the century (NRC, 20 12b).
Changes in air temperature will affect precipitation, evaporation and
transpiration. Higher temperatures will increase evaporation rates of
surface water (from oceans, lakes, or soil). In addition, wanner air can
hold more watcr (sec Figure 2 . 1 ). An increase in cvaporation will decrease
soil moisture, which could reduce transpiration, but changes in air
temperature and vapor pressure will also affect transpiration rates. Because
some of these changes can be offsetting, sorting out how precipitation and
evapotranspiration respond to increascs in air temperature is challenging.
Hydrologists and other environmental scientists are combining theory,
models and measurements (in the lab and field) to make the best forecasts
of future change in precipitation, evapotranspiration and runoff given our
current understanding (see Box 1 1 .1 ) .

BOX 11.1 Intensification orthe Global Water Cycle

As noted in Chapter I , water is conserved in the global hydrological

cycle. Therefore, globally and over long time scales, precipitation
(P) equals evapotranspiration (el). Using equation (2.8) we can then
express precipitation as

p = el = K'e...A I - ele"",),

where K' is a factor expressing the (global) conductance of the

Earth's surface (see Equation 2.9), ewu is the saturation vapor
pressure, and e is the actual water vapor pressure of the air, which,
as notcd in Chapter 2, is proportional to thc water vapor density and
is a good indicator of atmospheric humidity. An increase in global
temperatures is expected to increase both the atmospheric humidity
(or e}-because of the more intense evapotranspiration-and the
saturated vapor density (see Figure 2. 1 ), which measures the amount
of moisture the air can hold in unsaturated conditions. Therefore, it
has been argued that the ratio ele.wl, also known as relative
humidity, will not substantially change as an effect of climate
wanning. Thus, only the tenn eWI on the right-hand side of the above
equation is expected to change with T. The dependency of eM on

367
 temperature is shown in Figure 2 . 1 : because e.al is an increasing
function of T, climate warming is expected to increase global
precipitation, an effect known as the intensification of the global
water cycle. The dependency between eSIJI and T suggests that global
precipitation should increase by 6.8% per each degree Celsius of
global wamling. Climate change models and observations, however,
indicate that the increase in precipitation associated with IOC of
wanning is more in the 2-3% range, presumably owing to other
factors affecting the global conductance parameter K' (e.g., Katul et
aI., 2012). This analysis is based on global mean values of
temperature and precipitation. Climate change research has shown
that changes in both temperature and precipitation are not expected
to be unifonn around the globe (see Figure 1 1 .3).

-,

Figure 11.3 Annual mean change in precipitation (in inches of water per
year) for 2081 -2100 relative to 1950-2000 calculated bascd on the
Geophysical Fluid Dynamics Laboratory (GFDL) CM2.1 model in which
carbon dioxide was increased from 370 to 7 1 7 ppm. Positive values
correspond to wetter conditions.
Redrawn from an image developed by NOAA GFDL (www.gfdl.noaa.gov).

Climate models have been used to predict the change in amounts and
global distribution of precipitation as the air warms. Figure 1 1 .3 shows the
distribution of predicted changes in annual mean precipitation in 208 1-
2 1 00 relative 10 1950-2000 calculated based on the Geophysical Fluid
Dynamics Laboratory (GFDL) CM2.1 model in which carbon dioxide was
increased from 370 to 7 1 7 ppm. Whereas high latitudes are likcly to

368
experience more precipitation (as rain or snow), low latitudes, except in
the immediate vicinity of the equator, tend to become drier.
Not surprisingly, these patterns are generally mirrored in predicted
changes in runoff in thc 2 1 st century (Figure 1 1 .4; Milly et aI., 2005).
Incrcases in evaporation caused by increasing temperatures coupled with
decreases in precipitation lead to decreases in runoff in many parts of the
world. In other parts, increases in precipitation are leading to increases in
runolT. For example, reconstruction of monthly discharge from the world's
largest rivers indicates that global continental runolT increased during the
last century (Labat et ai., 2004), most likely due to increases in
precipitation in the watersheds feeding these rivers. Regardless of whether
precipitation is increasing or decreasing, observations suggest an increase
in the intensity of the largest rainfall events.

1 1 .3.2 Effects of increased temperature and CO2 on

c\"apolranspi ration

Understanding the ways in which climate change will affect

evapotranspiration is more difficult than for precipitation and runoff
because of feedbacks between soil moisture and evapotranspiration:
evapotranspiration dries soils but drier soils slow evapotranspiration (sec
Figure 2. 1 1 ). In addition, increases in atmospheric CO, have been found to
decrease stomatal conductance (Leakey et ai., 2009), which could decrease
evapotranspiration rates and thereby increase soil moisture and runoff
(Gedney et aI., 2006), although the magnitude of this effect is uncertain.
The challenge of predicting how evapotranspiration will change as climate
changes is compounded by the lack of relatively long-term, globally
distributed measurements such as those available for precipitation and
runoff. Even the records that do exist are complicated by changes in land
use during the last century, such as deforestation, that could overwhelm
climate signals at many sites (e.g., see Figure 9.7).

369
 40
30
20
10
2
-2

Figure 11.4 Annual mean change in runoff for 2041-2060 as a percentage

relative to 1900-- 1 970 calculated as the ensemble mean of 35 model runs
using 1 2 different models. Positive values correspond to higher runoff.
Adapted by permission from Macmillan Publishers Ltd: Nature, from Milly ct al.
(2005).

1 1 .4 Challenges for the Future

As the Danish physicist Niels Bohr once said, "Prediction is very difficult,
especially about the future." Nevertheless, it is clear that we face serious
issues with regard to what are called social-ecological systems in the
coming decades, and that many of these issues will be framed by the topics
above--ensliring water and food security in the face of a changing climate.
Whether predictions (or forecasts or sccnarios) about thc futurc arc borne
out will depend on actions taken by individuals and collectives of
individuals, including nation states. We believe that students of
environmental science should have a broad understanding of the
fundamentals of physical hydrology so they can help shape the future in
ways that arc infonllCd by scicncc. Thc NRC (21 02a) rcport notcd at thc
beginning of this chapter concludes that: "Compelling challenges and
opportunities lie ahead in understanding, quantifying, and predicting water
cycle dynamics, the interaction of water and life, and how to build a path
to the sustained provision of clean watcr for pcople and ecosystems." The
hydrological underpinnings for mccting the ehallcnges and taking
advantage of the opportunities fonn the substance of this book.

1 1 .5 Suggested Readings

Falkenmark, M., and J. Rockstrom. 2004. Balancillg waterfor humalls al/d

nafllre. London: Earthscan.

370
National Research Council (NRC). 2012a. Challenges and opportunities in
the hydrologic sciences. Washington, DC: The National Academies
Press.
National Research Council (NRC). 2012b. Climate change: Evidence.
impacts. alld choices. Washington, DC: The National Academies Press.

371
APPENDIX I
Units, Dimensions, and Conversions

Hydrological Quantities

Units and Dimensions

Significant Figures and Precision

Unit Conversions

Hydrological Quantities

We can separate the quantities that we encounter in hydrology into two

classes. Basic measurements constitute the first class. For example, we
might measure the height of water in a manometer, piezometer, or water
well, the velocity in a stream, the mass of water in a precipitation collector,
or the temperature of water in a lake. The second class includes the
discharge in a stream (calculated from the mean velocity and stream cross­
sectional area), or the total volume of precipitation that falls on a
catchment during a storm (calculated from the catchment area and the
mean precipitation depth). These arc derived quantities, because they are
not measured directly but are calculated from measured variables using an
equation representing a relationship between variables. Regardless of their
type, hydrological quantities have an associated unit, such as centimeters
per second, and dimension, such as length per time (although some
quantities may bc unitless and dimensionless). In recording basic
measurements or manipulating these values to derive other quantities, we
need to be concerned with several things, such as precision and
appropriate units and dimensions, which are the subjects of this appendix.

Units and Dimensions

Consider a simple example. A field investigation in a small catchment

records some basic infonnatiotl, including the depth of water in a single
precipitation collector placed within the catchment. We would like to
know, as precisely and accurately as we can with the available data, the

372
total amount (volume) of water delivered to the catchment during a single
cvent:

Volume = Depth x Area.

The measure of precipitation has an associated dimensioll, in Ihis case

length or [L], and Illlit, for example, centimeters or inches. We usc as
fundamental dimensions length [L], mass [M], time [T], and temperature
[0]. Many quantities have a dimension that is some combination of these
fundamental dimensions (Table A 1 . 1 ). For example, the volume of
precipitation falling on a catchment has dimensions of length cubed [L3].
To continue with our calculation, we can first check to see if the
relationship between depth and volume is dimensionally homogeneolls,
that is, whether the dimensions on both sides of the above equation are the
same:

As we would expect, this equation is dimensionally homogeneous.

Dimensional homogeneity does not guarantee that the equation will give
an accurate or correct result; for example, what if the length of the stream
were used erroneously in place of the precipitation depth in the equation?
However, an equation that completely and accurately describes a physical
relation must be dimensionally homogeneous. It is always useful to check
that any equation you use is dimensionally homogeneous.
There are a variety of units that correspond to each dimensional
quantity. Length might be given in meIers, centimeters, inches, or even
gallons per square foot. Therefore, care must be taken to make sure any
calculation is based on a lInitarily homogeneolls fonn of an equation. For
the example of precipitation over a catchment,

Table A1 . 1 Base and derived units relevant to hydrology in SI

measurement

373
Quantity Dimension Unit 51 Symbol Formula

Sase units

Length ILl meter m

Mass IMI kilogram k,
Temperature lei kelvi n ,

Time ITI seco nd ,

Oerlved units

Area l
(l ] square meter m'

Volu me (tJ] cubic meter m'

Velocity (L T-<] meter per second m '�

Acceleration IL T-2) meter per second squared m,'

Derlsily (M L�) kilogram per cubic meter kg m-J
Force 1M L T-2] rlewtorl N kg m s-2
Pressure IM L.... T-2] pascal " N m'

Stress (M l.... T-2] pascal p, N m'

Energy (M llT-2) joule J N·m

Quantity of heat 1M L' T-2] joule J N·m
Work 1M L' T-2] joule J N·m
Power 1M LlT-3] w,tt W , ,4

Viscosity, dynamic 1M L.... T....) pascal-second Pa-s

Viscosity, kinematic It' T....] square meter per second m's....
Specific heat (Ll S....T-2] Joule per kilogram-kelvin J kg-< K....

Table A1.2 Prefixes used in the S] system

374
 Prefix 51 symbol Multiplication factor

tera T 1 000 000 000 000 = 1012

giga G 1 000 000 000 = 109

mega M 1 000 000 = 106

kilo k 1 000 = 10)

hecto· h 100 = lOZ

deka· da 10 "" 101

deci' d 0 .1= 10-1

centi· c 0.01 "" 10�

mUll m 0.001 = 10-3

micro � 0.000 0 0 1 "" 10-6

na no n 0.000 000001 = 10-'3

pico P 0.000 000 000 001 = 10-12

femto f 0.000 000 000 000 001 = 10-l�

atto a 0.000 000 000 000 000001 = 10-18

aAvoid use of this prefix where possible.

Volume (mJ) = Depth (m) x Area (mz).

Needless to say, a wrong answer will result ifthe units on one side of an
equation are different from those on the other side.
The most common system of units employed today is the SI (System
illternational d'Unites; Table A 1 . 1 ). Other widely used systems of units
include the Ellglish !.ystem (foot, pound) and the cgs !.ystem (centimeters,
grams, seconds). The SI system includes a sequence of standard prefixes to
indicate magnitude (Table AI .2). For example, a kilogram is equal to 1 000
grams, and a nanogram is equal to 10-9 grams.

Significant Figures and Precision

To complete the simple example with which we began, we will use

measured quantities to derive the volume of precipitation (in mJ) given a
precipitation depth of 1 3 mm (0.013 m) and a catchment area of2.065 x
10' m2. The volume of water received by the catchment is 0.0 13 m x 2.065

375
x 105 m2 = 2684.5 mJ. This answer is correct mathematically, but it is
expressed with greater relative precision than is justified by the measured
values. There arc too many significant figures in the answer. This requires
some explanation. First, the measured quantities have a certain absolute
precisioll. For the measured precipitation depth, the absolute precision is I
mm. It is likely that the rain gage could only measure precipitation depth
to the nearest 111111. The absolute precision for the area (206,500 m2) is 100
m2 . This is apparent because we wrote the value initially using scientific
notation. For a number written as 206,500, it is not clear whether the
absolute precision is 100 m, 1 0 m, or I m. The relative precision is best
thought of in terms of sigllificalllfigures. In a given quantity, a significant
figure is any given digit, except for zeros to the left of the first nonzero
digit, which serves only to fix the position of the decimal point. Some
examples are given in Table A I .3 .

Table A1.3 Examples of quantities and their significant figures

Quantitv Significant figures Scientific notation

650,000 2' 6.5x l0s

30 l' 3 X l01

30. 2 3.0x 101

30.0 3 3.00 X 101

30.01 4 3.001 x 101

.01 1 l x lO-.1

0.01 1 1 x lO-.1

0.010 2 1.0 X 10 -l

0.00500 3 5.00x 10�

1,000.0010 8 1.0000010 x 10)

RTrailing zeroes without a decimal point arc ambiguous.

There are a number of rules for dealing with derived quantities to

ensure that answers arc expressed with the correct relative precision or
number of significant figures. When multiplying or dividing numbers, the
answer should be expressed using the same number of significant figures
as the least relatively precise lIumber involved ill the calculation, that is,
the one with the fewest significant figures. In our example, the
precipitation depth ( 1 3 mm) has Iwo significant figures and the catchment

376
area (2.065 x 105 ml) has four. Therefore, our answer should have only
two significant figures, 2.7 x 103 m3.
When adding or subtracting numbers, the rule is that the answer should
be expressed using the same number of significant figures as the number
with the/ewest decimal places. Finally, numbers should not be rounded to
the appropriate precision in a calculation with several steps unti! the vely
end.

Unit Conversions

Measurements that describe hydrological quantities may be expressed in a

variety of different units. As a result, one often has to convert a quantity
from one unit to another. For example, suppose that a value of hydraulic
conductivity (see Chapter 6) found in a research report is given as 6.2 feet
per day. We would like to express {his value in the 51 units of meters per
second. One approach is to multiply the value by ratios of equivalent units.
The ratios are formed such that the old units cancel, leaving the new units.
The procedure is illustrated below:
ft 1 m ,
I day m
x x = ? ? x 10
3 .281 ft
6.2 _._ - .
day 86,400 s s

Note that the answer is expressed in scientific notation, using the same
number of significant figures (2) as the least relatively precise number,
which in this case is the hydraulic conductivity value. [t should also be
obvious that the multiplication has resulted in the original units (ft, day)
being " canceled."
Tables A 1.4 through A 1. 8 provide equivalent values for many
quantities used frequently in hydrology. To use the tables, look down the
left column to find the unit you are interested in using to express a
quantity. In the example above, we wanted to convert feet to meters.
Looking down the first column of Table A 1 .4 you will find a row
beginning with "meter." Scanning across this row, you will find that 1
meter is equivalent 10 3.281 feet, which is the value used to constmct the
ratio in the example above.

Table AI.4 Equivalent units for length

377
 '"1.'''''' '

- - ' . I .'" - - - I" 'e' " -

"..; , �..
, , ',0]' 11' •.DO],., 0.00'000 '.D�,a-< D.Io>,. . ,a-<

KO ... , ',OIn ..� lUx,a-< 1S,11x,a-<

- m " , .- ]OUxla-< ,!"hla-<

- ,� l'I.11 1.>0' , .., .,�" "a-<

'u.,�... '.000,000 l'I,)'III ".. ,� , 0..,,.

- 1.'09,000 �� ". ,- ,- ,

aln Tables A 1.4 through A I ,8, values are shown to four significant figures, and the
S I expression, in base or derived units, is in italics.

Table A 1.5 Equivalent units for area

. ....., .�

- - - - - �,- .- -

"0' , 0._' "'.>x ,a-< ",.. "a-< ..." " ,... "', ' x ,a"" '''1 ' , ,a""

-, ", , tn.O"lCt" lUhl(l"' ',290)_ W" 91.1(1<,(1"' 1Uhla-<

- ... ,.. , 1<01.1 ,,(I"' � .� ]86..1, ,a-<

•• 6,V] < .0' U5:b.O' -, , ...0:01 ,,_, �,�

��. '�'OK .0' u,,�x .O' W 1.'11 , �, 0""".1

,-� '.,.0.,0' '.O'h .0' W ",. ,. , .�,

- "''''10' nux ,.' a90�I(I' � m ,� ,

Table A1.6 Equivalent units for volume

'...-

•• -, - � .- ,-' ,� -� --

KO' , D.O" '" .-� '1I.1xllt'" 2l.... .a-< 10, ,,, " "" 11.H. W""

- ..� , •.� CI.01U, .,oo.lOI ,� ".,'xla-<

�.S-_ 11UI om • D.Ul1 ._, 0.00111' l.Q6.t.,(1"'
-' "" " .. ,�, , .m� ...101>. II.'1fiXla-<

,., .- �. - " , .- 61U_W"

-' .... ,� �., .." o� , 0'0.• <la-<

aa.·_ 1'-1h.0' " llh ,," l.IS h .(1' <."h.D' li" "" ,

Table AI.7 Equivalent units for discharge

..,""' .. �

- .- - - """'"<Ir' --- '-' .. - -' - �

- .......
� , ••m O.Q(l<'" 0.00>>1' " .C9 x ,a-<

--- .s-a, , �= o.onn ,�

ouo·_ ...... ,., .... , ._, 0...,.,.

-' .. C' « .,. "" ,�, , D.ol..

l

--- 15,1500 ,� - �, ,

Table A 1.8 Equivalent values for velocity

378
 1....-

•• - .." .'".. � _ ." _ ." -..... , �

_ .." , "."'x ,O"" U-'''' , or< ,...." ' - l--'IlI. _

�." �,. , 0.911) 0.62,. 0.1111

_." �- u,,' , 0..... �

- ..� 'lO.1IID ,� ,�, , D."'"

_." "'� ,- J.181 1.1'7 ,

379
APPENDIX 2
Properties of Water

General Properties
Water Density and Viscosity

General Properties

The Earth's hydrosphere contains almost 1.4 billion km3 of water. In

nearly all of ils physical properties, water is either unique or at the extreme
end of the range ofa property. Water remains a liquid within the
temperature range most suited 10 life processes.
In basic structure, the water molecule, composed of one atom of oxygen
and IWO of hydrogen, has a small dipole moment. Water will dissolve
almost anything to some degree, although the extent may be very small for
some substances. Once dissolved, substances tend to remain dissolved
because of one of water's exceptional properties. The dielectric constant
for water is greater than that for any other substance. The dielectric
constant (multiplied by the square of the distance separating ions) indicates
the magnitude of the force of attraction between positive and negative
ions. Ionic substances in solution are therefore tightly held by the
surrounding water dipoles. As a result, water found in nature contains
some amount of dissolved material.
Water also has the greatest specific heat capacity and latent heat of
vaporization known among liquids, and has a higher thermal conductivity
than any other liquid except mercury. It also has a high heat of fusion. One
consequence of these properties is that water at the Earth 's surface tends to
moderate effects of hot or cold air temperatures.

Water Density and Viscosity

Liquid water contracts as it cools, reaching a maximum density (P) of 1 000

kg m-3 at 3.98°e (Figure A2.1 ; Table A2.1 ). Between 3.98 and ooe, water
expands slightly, but once frozen, ice has a density of about 920 kg m-3.
Because of the temperature dependence of water density, wam1 waler will

380
float on top of cooler water at temperatures above 3.98°C, while below
3.98°C, cooler water will float on top of wanner water. As a result, ice will
lend 10 form and remain at the surface of a water body. When it melts, the
waler becomes denser again and sinks below the surface. This process
leads to spring mixing of surface and deeper water in many mid- to high­
latitude lakes.
The viscosity of water (p, often referred to as the dynamic viscosity)
also varies with temperature (Figure A2. 1 ; Table A2. 1). In fact, the
variation in water viscosity is much greater than thai of density over the
temperature range most commonly encountered in hydrology. For
example, from 1 0 to 30°C, waler density decreases from 999.73 kg m-3 to
997.07 kg m-3, a decrease of 0.27%. Over the same temperature range, the
viscosity of water decreases from 1 . 307 x 10-3 Pa . s (at 10°C) to 0.7975 x
1 0-3 Pa ' s (at 30°C), a decrease of39%. For this reason, hydrologists are
often coneemed with the temperature effects on water viscosity when
making their calculations. [n the Sl system (see Appendix I ), the units of

�__�""':::::::: ==��=:�
:
viscosity are Pa ' s.

1000 ____
_ - 1 .9 x l 0-a
3
- 1 .8xl0-
\ 1.7 xl0-3
999
3
- 1 .6xl0-
-
7
t!'
-

. 1 . 5 x l 0-
3
..
E 998
'" - 1 . 4 x l 0-
3
e:.
""
�
·
- 3
,..
1 .3x l 0-

- 1 .2xl0- 8
- '"
.- 3
'"
c
997
:£
_

.. 1 . 1 x l 0-3
C
_

- 1 .Ox l 0-3
996 : 3
O.9xl0-
3
- O.8x l 0-

995 O.7xl0-3
o 5 10 15 20 25 30
Temperature (oC)
Figure A2.1 The density (P) and viscosity (P) of water as a function of
temperature, from 0 to 30°C.

Table A2.1 The density (p) and viscosity (�l) ofwllter as a function of
temperature, from 0 10 50°C

381
Temperature (-e) Density (kg m-J) Viscosity (Pa·s)

0 999.87 1.787 x lo-J

3.98 1000

5 999.99 1.519 x 10...,]

10 999.73 1.307 x 10...,]

15 999.13 1.139 x 10...,]

20 998.23 1.002 x 10...,]

25 997.07 8.904 x 1�

30 995.67 7.975 x 10-4

40 992.24 6.529 x 10-4

50 988.07 5.468 x 10-4

382
APPENDIX 3
Basic Statistics in Hydrology

Often, hydrologists have only point measurements of quantities such as

precipitation, which clearly vary temporally and/or spatially. Due to the
complexity of processes such as precipitation, we must often take a
probabilistic approach to studying them, based on observation rather than
physical theory or mechanistic causes.
In a probabilistic approach, processes are viewed as being somewhat
random, or as having some element of uncertainty or unpredictability.
Common examples used to discuss probability are rolling dice or flipping
a coin. In these cases, the outcome of the random process is a number ( I
through 6 on each die) or "heads" or "tails." The numbers of doIS on a die
face and " heads" or "tails" arc examples of discrele random variables,
because only a discrete number of values are possible. Many hydrological
measurements or quantities (i.e., average annual precipitation) are
considered cOllfinuous random variables, because the quantity of interest
may have any (reasonable) value.
Consider the data in Table A3 . 1 . There seems to be a general tendency
for annual precipitation to be around one meter, with two "extreme"
values, 780 and 1 1 92 mm. How would we use this infonnation to make a
prediction about future precipitation? With continuous random variables
such as annual precipitation, the question we might ask is: What is the
probability that precipitation will be greater than 1 1 00 mm? To address
this question, we need to consider how probability is "distributed," or what
probabilities are associated with what measured or expected values.

Table A3. t Annual precipitation for a ten-year period

• •

Year Annual precipitation (mm)

1975 1 020
1 976 987
1977 894
1978 1 040

383
 1 979 995
1 980 780
1981 1004
1982 930
1983 1 192
1 984 950
• •

Continuous random variables are described using the probability

densityfllllClioll (pdl), J{x). The cumlllalive distriblltion fllllc/ioll (cdf),
F(u), is the integral of the probability density function:

F(o) = P(xSo)= J:f(X)d\', (A3.1)

In other words, the cdr for a chosen value, a (1 100 mm, in the example
above), is the probability thai the outcome ofa random process, x
(precipitation next year), will be less thall or eqllal lo the chosen value.
Notice that the limits of integration are negative infinity and Ihe value a;
the integration is adding up all of the probabilities associated with all of
the values less than and including a. Conversely, the compleme1/f(fl)'
Cl/IIII/Ialille distriblltiolljilllclioll, often written as G(a), is defined as:

G(a)= P(x >a)= I - F(a):: r f(x)dx, (A3.2)

which, in the example, provides the probability that precipitation will be

greater thall 1 100 mm. We refer to this as the exceedallce probability.
Several relationships can be inferred from the definitions above:

p(a Sx S b):: F(b)- F(a)= rf(x)dr:. (AJ.J)

p(x:a)= L" f(x)dl;.O. (Al.4)

P(__ :S;.\·:S;oo)= J:f(.t)du l. (A}.5)

Equation A3.3 describes the probability of the outcome of a random

process being within a fixed range of values. Equation A3.4 simply points
out that, with continuous random variables, the probability of a single
definite outcome is zero; that is why we always refer to the probability of
an outcome being greater than, or less than, or within a range of values.

384
Equation A3.5 indicates that the total probability is equal to I (annual
precipitation must have sOllie value).
The distribution of a sample of values of a random variable can exhibit
a varicty of forms and can often be describcd by relatively simple
probability density functions. Examples include the normal distribution,
the log-nomlal distribution, and the exponential distribution.
The normal distribution can be described with two parameters: the
mean (pJ and the standard deviation (oJ. Given a sample of data (e.g.,
several decades of annual precipitation), the mean and standard deviation
are approximately x (for /Ix) and Sx (for ox):

LX,
"

,
x=
,
(A3.6)
.

"

L(x, -",)'
"

, ,
" (A3.7J
.

/1-1
. •

where n is the number of observations. For the data in Table A3 . 1 , the

samplc mean is 979 mm and the sample standard deviation is 1 0 1 mm. The
pdf and cdf for the normal distribution are:
jJ-e,t
f(x;J.l C1� )
I '.'
vI2 JrC1..
-e - .
_

.. . =
(A3.')

F(X;J.l .. . C1.. ) = P(X � x ) = J�f(X:,u•. C1.. }. (A3.•)

The normal distribution is a "bell-shaped curve" (Figure A3. 1 ). The mean

provides a measure of the central tendency of the variable. For normally­
distributed data, the mean equals the median (the middle value, if all the
values are arranged from lowest to highest or vice versa). The probability
that a value will lie within one standard deviation of the mean is about
two-thirds (0.68) and the probability that a value wi1l lie within two
standard deviations of the mean is about 0.95 (refer to the gray and dark
blue regions in Figure A3. 1 ).
Often, data arc lIormalized by calculating z-vailies. Data tables that
provide the cumulative distribution function of the standard I/ormal
distribution are based on z-values. Nonnalization of data involves shifting
the entire set of data by the amount required to bring the mean to zero, and
then shrinking or expanding the distribution so that the standard deviation

385
is I (Figure A3 .2). Values of the cumulative distribution funclion (Table
A3.2) or complementary cumulative distribution function of the standard
normal distribution can be used in calculations. Thc z-value represeOls the
nonnalized outcome for which the probability is desired ( 1 100 mm, in the
earlier example). This value is calculated according to:

-20 -Ox J1x

x
Figure A3.1 The normal distribution.

(AJ.IO)
a-x

For the example data in Table A3 . 1 , with a = I 100 mm, x = 979 mm, s" =
I 0 I mm, the calculated z-value using equation (A3.1 0) is 1 .20. (Note that
the z-value is dimensionless.) The edf for this z-value (Table A3.2) is
0.8849 (i.e., F(z) = 0.8849), which indicates that the probability that
precipitation in any year will be less than or equal to 1 1 00 mm is about 0.9
(Figure A3.3).

386
 (a) (b)
"N'
t.;!.
, F(x)=F(z)
�
�

"
--< --<
�
" �
"

Jlx o -1 0 1
0.4

0.35

0.3

0.25
N-
o:::; 0.2

0.15

0.1

0.05

-4 -1 1 4

Figure A3.2 Transformation to the standard nonnal distribution (c)

involves altering both the mean (a) and the standard deviation (b) of the
original distribution.

387
 For l. = 1 .26

F(z)

-4 -3 -2 -1 o 1 2 3 4
z
Figure A3.3 The standard normal distribution showing the portions of the
distribution given by the cumulative distribution function, F(z), and
complementary cumulative probability, G(z).

To find the excecdance probability, we subtract this value from one

[equation (A3.2)] to get approximately a 0.1 chance that precipitation will
equal or exceed 1 1 00 mm in any year. In other words, there is a 10%
chance of annual precipitation exceeding 1 1 00 mm. That is, on average,
1 100 mm or more of precipitation occurs once every 1 0 years. The return
period for annual precipitation equal to or greater than 1 100 mm is 1 0
years:

T
1 1
= =10
",,,,,,, excecdance probability 0.1 yr�l
_
yr. (A3.1 1 )

This entire procedure can also be used, in reverse, to detennine the

value or range of values associated with a given return period or
exceedance probability. For example, using the example precipitation data,
we can calculate the annual precipitation amount that we expect to be
exceeded l out of 3 years. Problems such as this are common in water
resource planning. From the given information, we know that G(z) is equal
to approximately 0.33. Table A3.2 can be used to find the z�value for F(z)
= I- G(z) = 0.67, which is 0.44. Subsl"ituting this value into equation
(A3.10) and rearranging to solve for a (wc know the samplc mcan and
standard deviation) gives a value of 1021 mm.
This brief review only scratches the surface of applications of statistics
to hydrological problems, but does give some idea of how the methods

388
might be used to predict possible behavior. Some of the concepts
introduced, such as probability, probability distribution, exceedance
probability, and return period, arc used throughout the book.

Table A3.2 The cumulative distribution function of the standard normal

distribution, F(z), for z > 0

389
 • 0.00 0.01 0.02 0.0] 0.04 0.05 0.06 0.07 0.08 •."

,., ' ''''' 0.5040 0.5080 a.SIlO 0.51(;0 0.5199 0.5239 O.S179 0.5319 0.5359

,.. 0.5398 0.5438 0.5478 0.5517 0.5557 0.5596 0.5636 0.5675 0.5714 0.5753

,.. 0.5793 0.5832 0.5871 0.5910 0.5948 0.59117 0.6026 0.6064 0.610] 0.6141

,., 0.6179 0.6217 0.6255 0.6291 0.6331 0.6]68 0.&406 0.6443 0.6480 0.6517

, .• 0.6554 Q,WU 0.6628 0.6664 0.67(1) 0.6736 0.6772 0.6808 0.6844 0.6879

OS 0.6915 0.6950 0.6985 0.7019 0.7054 0.7088 0.7123 0.7157 0.7190 0.1224

,.• 0.7257 0.7291 0.7324 0.7357 0.7389 0.7422 0.7454 0.7486 0.7517 0.7549

,., 0.7580 0.76U 0.7642 0.7673 0.7704 0.7734 0.7764 0.7794 0.7823 0.7852

,.• 0.7881 0.7910 0.7939 0.7967 0.7995 0.8023 0.805\ 0.8078 0.8106 0.8133

,.• 0.8159 0.8186 0.8212 0.8238 0.8264 0.8289 0.83\5 0,8340 0.8365 0.8389

.., 0.8.413 0.8438 0.8461 0.8485 0.8508 0.8531 0.8554 0.8577 0.8599 0.8621

U 0.8641 0.8665 0.8686 0.8708 0.8729 0.8749 0.8770 0.8790 0.8810 0.8830

.., 0.8849 0.8869 0.8888 0.8907 0.8925 0.8!144 0.8962 0.8980 0.8997 0.9015

U 0.9032 0.9049 ' '''' 0.9082 0.9099 0.9115 0.9131 0.9147 0.9162 0.9177

L. 0.9192 0.9207 0.9222 0.9236 0.9251 0.9265 0.9279 0.9292 0.9306 0.9319

L' 0.93l2 0.9345 0.9357 0.9370 0.9382 0.9394 0.9406 0.9418 0.9429 0.9441

••• 0.9452 0.!I463 0.9474 0.9484 0.9495 0.9505 0.9515 0.9525 0.9535 0.9545

U 0.9554 0.9*, 0.9573 0.9582 0.9591 0.9599 0.9608 0.9616 0.9625 0.9633

L' 0.9641 0.9649 0.9656 0.9664 0.9671 0.9678 0.9686 0.9693 0.9699 0.9706

L' 0.9713 0.9719 0.9726 0.9732 0.9738 0.9744 0.97SO 0.9756 0.9761 0.9767

,., 0.9772 0.9778 0.9783 0.9788 0.9793 0.9798 0.9803 0.9808 0.9812 0.9817

U 0.9821 0.9826 0.9830 0.9834 0.9838 0.9842 0.9846 0.9850 0.9854 0.9857

>.> 0.9861 0.98&4 0.98&8 0.9871 0.9875 0.9878 0.9881 0.9884 0.9a87 0.9890

U 0.9893 0.9896 0.9898 0.9901 ,."" 0.9906 •."'" 0!�911 0.99H 0.9916

, .• 0.9918 0.9920 0.9922 0.9925 0.9921 0.9929 0.9911 0.9932 0.9934 0.9936

,.. 0.9938 0.9940 0.9941 0.9943 0.9945 0.9946 0.9948 0.9949 0.9951 0.9952

U 0.9953 0.9955 0.9956 0.9957 0.9959 0.99&0 0.9961 0.9962 0.9963 0.9964

,., 0.99&5 0.9966 0.9967 0.99&8 0.9969 0.9970 0.9971 0.9972 0.9973 0.9974

U 0.9974 0.9975 0.9976 0.9977 0.9977 0.9978 0.9979 0.9979 0.9980 0.9981

U 0.9981 0.9982 0.'�982 0.9983 0.9984 0.9984 0.9985 0.9985 0.9986 0.9986

Note: To use the table, scan the left column to locate the z-value to the first
decimal place, and then scan across the row to find F(z) for the second decimal
place. Note that for z < 0, use F(z) I F(IzI).= -

390
Answers to Example Problems

Chapter 1
Chapter 2
Chapter 3

Chapter 4

Chapter 5
Chapter 6

Chapter 7
Chapter 8

Chapter 9

Chapter 10

Chapter I

Problem 1.1. Precipitation is typically measured as a volume [Ll] per unit

area [Ll], which has dimensions of length [L]. In the United States, the
average annual precipitation varies from a minimum at Death Valley,
California ( 1 . 6 inches), to a maximum on MI. Waialeale on the island of
Kauai in Hawaii (460 inches). What is the average annual precipitation (in
millimeters, mm) at each of these locations?

From Table AlA, 1.0 inch is equivalent to 25.4 mm. The conversion
is as follows:

. _).
,- 4 mm
Death Valley: 1 .6 m x = 4 l mm
1 in
25.4 mm
nun
.
= 1 . 1 7 x l O,
1
Mt. Waialeale: 460 m x
in

Problem 1.2. In the United States, stream discharge is often measured in

units of cubic feet per second (W S- I , or "cfs"). In most other countries,
discharge is measured in cubic meters per second (m3 S-I ). What is the

391
equivalent flow (in m3 S- I) of 18.2 ft3 S-l ? (You might want to review
Appendix I on units, dimensions, and conversions.)

From Table A 1 .7, 1.0 rtl S-I is equivalent to 0.02832 m3 S-I. The
conversion is as follow:

0.02832 m) S-I
? ft3 s-I x
1 8._ J = 5 . 1 5 x 10-1 m3 s- 1
I ft S-I

(Note that three significant figures appear in the solution, following

tbe rules described in Appendix I.)

Problem 1.3. In an average year, 1 .0 meter of precipitation falls on a

catchment with an area of 1 000 (or 1 03) km2.

A. What is the volume of water received during an average year in cubic

meters?

106 m1
Volume = Depth x A rea = 1.0 III X \03 km1 x = 109 m).
I km1
B. In gallons?

264.2 gal
Volume = I 09 m) x = 3 x 10" gal.
I mJ
Problem 1.4. The polar ice caps (area = 1.6 x 107 km2) are estimated to
contain a total equivalent volume of2.4 x 107 km) of liquid water. The
average annual precipitation over the ice caps is estimated to be 5 inches
per year. Estimate the residence time of water in the polar ice caps,
assuming their volume remains constant in time.

The residence time for a reservoir at steady state is defined as the

reservoir volume divided by the volumetric inflow or outflow rate:

, 1m , ' 106 m2 .... " J ,

J = 5 m yr- x x l .6 x l O km- x = 2.0..,2 x I 0 - m yr- .
.

39.37 in I kml

V 2 4 x l 016 m3
Residence time = Tr = - = . 3 = 1 .O x I 04 yrs.
J 2.032 x 1 0 12 In yr- I

392
 =
Problem 1.5. In an average year, a small (area 3.0 km2) agricultural
catchment receives 950 mm of precipitation. The catchment is drained by a
stream, and a continuous record of stream discharge is available. The total
amount of surface-water runoff for the year, detennined from the stream
discharge record, is 1 . 1 x 106 mJ.

A. What is the volume of water (in mJ) evapotranspired for the year
(assume no change in water stored in the catchment)?

r, = 1 . 1 x I 06 mJ•

p = 950 mlll X
1m
x 3 .0 km2 x
1 06 m2 ) = 2.85 x I 06 m3.
1 000 mm I km2

B. What is the depth of water (in mm) evapotranspired for the year (again,
assuming no change in water stored in the catchment)?

To calculate the depth, divide the volume by the area:

- -'-' '- -';c ::-;-

1 . 75 x 1 06 m J
' O k
--,- = 0_58 m X
1000 mm
= 5 80 mm.
2 1 06 m 2 1m
J. ' 111 X 2
I km

(Note that the volume calculated in A was reported with two

significant figures, hence 1.75 x 106 mJ became 1.8 x 106 mJ. For B,
the original result was used in the calculation, and the final answer
was rounded to two significant figures.)

C. What is the runoITratio CF/Ji) for the catchment?

�= .,.
.:.
1 -:.:x I ::.
0..
1 .:..:. ' m
:..
.:.: ' --:-
=- � = 1 . 1 x 106 m 3 . .
= 0 39
P
_ __
____

1 m 106 m2 2.85x 1 06 m3
950 rmn x x3.0 km2 x
I km-
7
1 000 mm

Chapter 2

Problem 2.1. Two tipping bucket rain gages arc used to collect the
following rainfall data:

393
 Cumulative precipitation (mm) Cumulative precipitation (mm)
Station III Station 112
Time

4:00a.m. 0.0 0.0

6:00a.m. 0.0 0.0
8:00 a.m. 1.0 1.0
10:00 a.m. 4.0 3.0
12:00 noon 13 11
2:00 p.m. 17 15
4:00 p.m. 19 16

6:00 p.m. 19 17

8:00 p.m. 19 17

10:00 p.m. 19 17
12:00 midnight 19 17

2:00 a.m. 19 17

4:00a.m. 19 17

A. Calculate the mean daily rainfall intensity for each station (mm hr-I).

Station # 1 : intensity = 1 9 mm/24 hr = 0.79 mm hrl.

Station #2: intensity = 1 7 mm/24 hr = 0.71 mm hr-I .

B. Calculate the maximum 2-hour rainfall intensity for each station (mm
hr-1 ),
By inspection, tbe greatest 2-hr rainfall occurred between 10:00 a.m.
and noon:

Station # I : intensity = ( 1 3 - 4) mm/2.0 hr = 4.5 mm hr- I .

Station #2: intensity = ( 1 1 - 3) mm/2.0 hr = 4.0 mm hr-I .

C. Calculate the maximum 6-hour rainfall intensity for each station (mm
h,.-I).

The greatest 6-hr rainfall occurred between 8:00 a.m. and 2:00 p.m.:

Station # I : intensity = ( 1 7 - 1.0) mm/6.0 hr = 2.7 mm hrl.

Station #2: intensity = ( 1 5 - 1.0) mm/6.0 hr = 2.3 mm hr-l.

D. Using the arithmetic average method and knowing that the drainage
basin area is 176 mi2, calculate the total volume of rainfall (m3)

394
 delivered to the basin during the event.

p = ( l 9 + 17)12.0 = 1 8 mill.

( . , 2.S90 X 10' m'

V = A x p = 176rm-x
)( 18 mmx
1m ) = 8.2 x l 0" m· .
I mi2 1000 mm

Problem 2.2. Measurement of changes in volume of water in an

evaporation pan is a standard technique for estimating potential
evapotranspiration. United States Class A evaporation pans are cylindrical
with the following dimensions: dcpth = 10.0 inches and diameter = 47.5
inches. An evaporation pan can be considered a hydrological system with
an inflow, outflow, and storage volume. Evaporation from pans is not the
same as evaporation from natural surfaces for a variety of reasons. For
example, water temperatures in shallow pans will be much more variable
than temperatures in a nearby lake. Evaporarion measured in pans is
adjusted by a factor called a pan coefficient to convert to an estimate of
potential evapotranspiration (see Brutsaert, 1 982).

A. Calculate the cross-sectional area (m2) ofa United States Class A

evaporation pan through which inflows and outflows of water can pass.
Also, calculate the total storage volume of the pan (m3).

47.5 in 2.54 em 1m ' )

(
.
Cylrnder area= rrr2 = rr(dI2)2 = rr x . x = 1 . 1 4 m2
2 1 m 100 em

, (
Cylindervolume=areaxdepth= I. 1 4 m- x IO.O rnx
.
.
2.54 em
I In
x
1 m
100 em
) = 0.290 m,

B . Initially, the pan contains 10.0 U.S. gallons of water. Calculate the
depth of water in the pan (mm).

0.003785 on'
J
VoI lime = 1 0 .0 gaI x = 3.79x 1 0-2 m .
I gal

Depth = volume/area = 3.79 x 1 0-2 m3/ 1 . 1 4 m2 = 0.0332 m = 33.2 mm.

e. Assuming a water density of997.07 kg m-3 (25°C), calculate the mass

(kg) of water in the pan

Mass = p x volume = (997.07 kg m-3)(3.79 x 1 0-2 m3) = 37.8kg.

395
D. After 24 hours in an open field (no precipitation), the pan is checked
and the volume of water left in the pan is detennined to be 9.25 gallons.
Calculate the average evaporation rate (mm hr-1 ) from the pan.

dV1d1 = / - a = (pan area) x (p - et) or el = - (dVIdi}/(pan area),

where p and et are the average rate of precipitation and

evaporation expressed as a depth per time,

dV 0.75 gal O.003785 m3

-
x = - 1 . 1 8 x l O"""""' m3 hr-l.
dl 24 hr I gal

el = - (-1 . 1 8 x l O-' m' hr-I )!( 1 . 1 4 m')= 1 .04 x 1 0-' m hr-I .

el=O.lOmm hr-1
E. The pan is emptied and refilled with 10.0 gallons of water and left in an
open field for another 24 hours. During this period, rain fell for a 3-hour
period at a constant intensity of 2.5 mm hr- 1 ; after 24 hours, the volume
of water in the pan was 1 1 .50 gallons. Calculate the average
evaporation rate (mm hr-1) from the pan during this period.

et = p - (dV/dt)/(pan area).

The total precipitation for the 24-hour period is (3 hr)(2.5 rum hr-I)
= 7.5 rum. Over the 24-hour period, this is equal to an average
precipitation rate of:

p= (7.5 mm)/(24hr) =0.3125mm hr-l=3.125x 10"""""' m hr-l.

dV 1 .50 gal 0.003785 m3

x = 2.37 x 10"" m3 hr-I.
_

dl 24 hr I gal

ef = (3.125 x 10"" m hr- I ) - (2.37 x 10"" mJ hr- I )/( 1 . 1 4 1112 ).

et = 1 .05 X 10"" 111 hr-I = 0.1 1 111 111 hr I.-

F. If the evaporation rate calculated in E remains constant and no

additional precipitation occurs, estimate the time (days) for the pan to

396
 empty as a result of evaporation.

el =p- (dV/dl)/(pan area).

dl = --;-:
.,-'-
----:­
dV
(el X pan area )

0.003785 1n3
dV = - I I .SO gal x = -4 .3Sx 10-2 In ) .
I gal

-4.35 x I O-2 1n3

d1 = .> h r= 1 5 days.
= "63
( 1 .05 x I 0-4 m hr- ' X I . 1 4 m 2 )
Problem 2.3. For Problem 2.2 part D, calculate the nux of latent heat from
the water in the pan to the atmosphere (W m-2). Use a watcr dcnsity, Ph' =
1000.0 kg m-3.

et P.. �,.
E
I ,
=

PM)'"
el or E, =

I hr
ct = 1 .04x I O""" m hr-1 x = 2.89 x I O-s ml s-l.
3600 s

E, = (2.89x 10-' m ,-')(1 000 kg m-')(2.45x 1 0' J kg- ' )= 70.8 W m-'.

Problem 2.4. A small (area = 300 ha) catchment in Iowa absorbs a mean
RIO = 330 W m-2 during the month of June. In this problem, you will apply
the energy balance approach to estimate evapotranspiration from the
catchment during the month of June.

A. Write a complete energy balance equation (i.e., including all terms) for
the catchment for the month of June.

dQ
= = R. - G - H - E,.
dl
B. Neglecting conduction to thc ground (G) and thc changc in energy
stored (dQ/dl = 0), simplify your energy balance equation for the
catchment so that it can be solved for the latent heat flux, E/. Also,
replace the term H (the sensible heat flux) with B x E/ (where B is the

397
 Bowen ratio).

R,, - H - E, =O.

E -
R"
,
_

B+ I
C. Using a mean Bowen ratio of 0.20 for the catchment, calculate the mean
daily nux of latent heat to the atmosphere (W m-2) and the mean
evapotranspiration rate (mm day-I) from the catchment. Use a water
density, Pw = 1000.0 kg m-3.

, :::: 330 W m -1 275 W m-1 = 280 W m -2

E =

1 .20

EI ' 75 W m -,
-
el :::; = 1 . 1 2 x lO-7 m s-l•
P.)" ( \ 000 kg m lX2.4Sx 106 J kg I )

1 000 mm 86400 s
e/ ::: 1 . 1 2 x 1 0-7 m S-I x x = 9.68 mill day- I =9.7 mill day-I .
1 m I day

D. Calculate the total evapotranspiration from the catchment during the

month of June (mm).

Total evapotranspiration = 9.68 mm day-l x 30 days = 290 mm.

Chapter 3

Problem 3.1. The following questions make use of the hydrostatic

equation.

A. What is the gage pressure (Pa) at a depth of 10.0 m in a lake with a

water temperature of 1 5°C?

AI lSoC P = 999.1 kg m-3, so

p = pgd = (999.1 kg m-3)(9.8 1 m 5-2)(1 0.0 m) = 9.80 x 104 kg m-I s-2 =

98.0 kPa.

398
B. Would the pressure change significantly if the water temperature was
22°C instead?

At 25°C P = 997.1 kg m-J, so

p = pgd = (997.1 kg m-3)(9. 8 1 m 5-2)(1 O.Om) = 9.78 x 104 kg m-I S-2 =

97.8 kPa.

There is a small difference in the third significant figure.

C. At what depth (m) is the gage pressure 300 kPa?

d = p/pg = 300 kPa/9.80 kN m-J = (3.00 x 105 kg m- I s-2)/(9. 80 x 10J

kg m-2 5-2) .

d = 30.6m.

D. What depth (m) of mercury, with a unit wcight of 133 kN m-J, would
be required to produce a pressure ofJOO kPa?

d = p/(Pg)�lg = 300kPa/ 133 kN m-3 = 2.26 m.

Problem 3.2. A plate is pulled over a horizontal layer of water that is 10.0
mm deep (Figure 3. 1 ). The temperature of the water is 20°C. If the plate
exerts a shcar stress of 0.0 I N m-2 on the upper surface of the water, what
is thc spccd (m S- I) ofthc platc?

From Equation 3.1 lIplale = (FIA)(d4/). FIA is the shear stress, equal
,

in this case to 0.01 N m-2 or 0.01 kg m-1 s-2. The viscosity at 20°C is
1.00 x 1 0-3 Pa . s (Table 3.1) or 1.00 x 10-3 kg m-I S-I . Therefore,

up/ale = (0.01 kg m-I s-2)(0.0IOm/1.00 x 10-3 kg m-1 S-I ) = 0.10 m S- I .

Problem 3.3. Observations show that flow in a circular pipe of diameter D

remains laminar up to a Reynolds number Rpipe = pUDlp = 2000 (Figure
3 .10). What about flows in other flow in channels or pipes with different
geometries? Consider the example of flow between two flat plates shown
in Figure 3. 1 . In this case, the appropriate length scale for the Reynolds
number is L = d. It also makes sense to use plate speed rather than mean
flow velocity speed as thc characteristic velocity in R, giving Rp1dle =
pUp/ate dip. Changes in length and velocity scales can alter the upper limit
for laminar flow, but for the case of flow between two parallel plates, flow

399
is again laminar up to RI'/ille =
2000. [A parameter called the hydraulic
radius (Chapler 4.5) can be used 10 find values of R corresponding 10 Ihe
laminar -Iurbulent transition for different flow cross-sections. This is
explored further in an example problem in Chapter 4.]

A. For 20°C water between 2 plates separated by a distance of 4.0 mm

(i.e., d 4.0 mm; Figure 3. 1 ), what is the maximum speed that the
=

upper plate can move and still maintain laminar flow?

Rearranging Rl'lal pllplate dip to solve for IIplate gives lip/lite Rl'lal"
= =
"

p/(Ptf). Tbe maximum value of IIplllM to maintain laminar flow is

=
found when Rplate 2000:

B. Is the flow in Problem 2 laminar? [Note: if not, equation 3. 1 is no

longer correct.]

plI",""d (1 000 kg m-'XO.I 0 In s-' X I O.O x 1 0-' m)

Rplilk
= = =
1000.
J1 ( 1 .00 x l O-3 kg m- l s-l)

Since Rl'llIle < 2000, the flow is laminar.

Problem 3.4. Surface temperature in a river is measured by a thennometer

drifting with the water at a rate of I km hr -I. The water in the river as a
whole is warming at a rale of O.2°C hr -I, and the temperature along the
stream increases by 0.1 °C every kilometer in the downstream direction.
What change in temperature eC) docs the thermometer record in 6 hours?

dT aT aT
- -:;- + II- (0.2°C hr- I ) + (I OOO In hr- 1 XI x 1 0....oC m- I ) O.3°C hr- I.
df 01 a.\"
= = =

The change over 6 hours is (6 hr)(0.3°C hr-1) = 1.8°C.

Problem 3.5. A tank like the one pictured in Figure 3.6 is filled to a
constant level ofO.70m. The center of the outflow opening near the bottom
is 0.1 Om above the bottom of the tank. What is the velocity (m S-I) of flow
exiting from the outflow opening?

d=
The depth between the water surface in the tank and the outflow
opening is: =
Zl - Z2 0.70 m - 0.10 m 0.60 m. From Equation =

400
 This is the velocity of flow exiting the tank.

Problem 3.6. The pressure drop through a well-designed constriction can

be used to measure the velocity of flow through a pipe. If the pressure drop
from a O.I-m diameter cross section 10 a O.OS-m diameter cross section is
7.5 kPa, what is the velocity (m S�I) in the O.I-m diameter section of the
pipe? Hint: Usc the conservation of mass equation to relate the velocity at
the smaller cross section to that at the larger cross section.

From the Bernoulli equation

- Ul-Vj •

2g

From conservation of mass,

Substituting V2 = 4VI into the Bernoulli equation gives:

PI- P2 U,'-(4U,)' 15Ul

•

pg 2g 2g

Solving for VI:

"'
_
2 (p,-p,) 2(7.5x 10 ' I'a)
U- -
1 15p 1.5x 104 kg m-J

Problem 3.7. A steady discharge of2.0 1 0-4 m ) S-l is flowing through a

x

20-mm diameter hose. The viscosity of the water is 1.0 x 10- 3 Pa . s, and
the density of the water is 1000.0 kg m -3•

A. Calculate the Reynolds number. Is the flow laminar or turbulent?

U=QIA =(2.0 x 1 0-4 m 3 s- l )1[n( 1.0x 1 0-2 m 2)] =0.64 m S-l

R=pUDIIi =( 1 OOO.Okg m-3)(0.64m s-I)(2.0x 1 0-2 m)/(I.Ox 10-3 Pa .

s)=l.3xI0 4

401
 Because R > 4000, the flow is turbulent.

B. What is the friction factor and the head loss per unit length for this flow
(both are dimensionless)?

From Figure 3.10,/= 0.03. We could also calculate/using Equation

3.39, because the flow is turbulent and 4000 < R < 100,000. The
head loss per unit length may be calculated as:

!!b.: f2.. V' : 0.03

I (0.64 m , - ')'
: 0.03.
L D 211 (2x 1 0 -' m) 2(9.81 m , - ' )

C. What is the change in pressure (Pa) over a 10-m length of the hose?

p,-p, : pg(h/lL)L = (9.81 kN m-')( O.03)(1 0 m) = 3kPa.

Problem 3.8. Lava, with a density of2700 kg m-3 and viscosity of 1 .0 x

10 3 Pa. s flows through a conduit that is circular in cross-section. The
diameter of the conduit is 1.0m. Flow of lava through the conduit is driven
by a pressure gradient of -2.0 k.Pa m-I. Assuming the flow is laminar,
what is the discharge of lava (m3 S-I)? Is the assumption of laminar flow
valid?

(1 .0111) 1
u=_ p 02 = -(-2 0x I Ol Pa m -1)
d = O.063 m s- l .
dx3211 32(LOxIO'Pa·,)

Q: VA: V(",'): (0.063 on ,-')[,,(0.50 on)']: 0.049 m' ,-'.

: pVD : (2700 kg on-'XO.063 on '-'XLO on): . .

R 0 17
II (LOxIO' Pa·,)

The Reynolds number is much less than 2000, so the assumption

that the flow is laminar is valjd.

Chapter 4

Problem 4.1. Water flows in a 3.00-m wide rectangular channel. The

water depth is 1.50 m and Ihe discharge is 1 .50 m3 S-I. The channel bottom
drops smoothly by O.IOOm over a short distance (a step down in the

402
bottom) with no head loss or change in the width of the channel.

A. Calculate the specific discharge (m2 S-I) and specific energy (m) at the
upstream station.
1
- 0 .)-00m' S-1 .
Q 1.50m} S
q.., - -
_ _ - _

w 3.00m

U2 q;. (0.500 m1 5- 1)2

E= +h= +h= � � +1.50m= I .) l m.
_

2iZ 2gh' 2(9.81m s - X1.50m)-

B. Calculate the specific discharge (m! S-I) and specific energy (m) at the
downstream station.

The specific discharge, q....'. is constant because there is no change in

channel width.

£2 = EI + 0.100 m = 1.61 m.

Note that the channel drops, so the specific energy illcreWies.

C. Calculate the water depth (m) at the downstream station

Determine the upstream flow criticality by calculating the Froude

number:

l =0.500m2 s-I/I.50 m = 0.333 III S-I .

U=q...h

0.333 m S-l
r U 0.086.
hh J(9.81m s-2X1.50 m)
= = =

The downstream flow will be subcritical also, because we are

moving along the subcritical limb of the specific energy diagram in
the direction of increasing E. To find the water depth at the
downstream location, we usc an iterative procedure:

where "n" indicates the iteration number. Choose an initial "guess"

for the downstream depth, ,,�O),to be 2.0 m and substitute this into
the right side of the iteration equation to get ,,�I):

403
 (0.5 m' s·' ) '
- = 1.606 m -
" ll) 2(9.81 m s·' X2.0 111)'
= 1 . 6028 m.

Continue the iteration by substituting the newly calculated value of

,,�I) into the right side of the iteration equation:
(0.5 ml 5-1 )2
h(12 )= 1.606 m _ = 1.601 m.
2(9.81 m s· 'XI.6028 m)'

Continuing the iteration, we find that a good estimate for"2 is 1.6

m.

Problem 4.2. A discharge of2.0 m3 S-l is carried in a canal with the cross
section shown in Figure 4. 1 5. The canal is 1400 m long and drops 0.50 m
in elevation over that distance. Manning's II for the channel is estimated to
be 0.020. What is the value of IV (m) for this canal?

Discharge is mean velocity times cross-sectional area. From

Manning's equation:

k
Q=llwh= - RJPSV211'h= - RNJS1l2�
k \\,2

fI II 2.5
where the hydraulic radius is given by:

11111 w2 \112 II'

RH -- -
211+w 2.5 r" + 2( 11'/2.5)1 4.5\11 4 .5

Therefore,
213
1 ml13s-1
'"
m' -
'"
-
0.5 m ",'
,
- ,
s 0.020 4.5 1400 t11 2 .5

from which ,vlw2l3 = 14.42, so w = 2.7 IU.

404
 w

1V/2.S

.
Figure 4.15 Canal cross sectIOn for Problem 4.2.

Problem 4.3. As nmed in Problem 3.3, channels and pipes of varying

geometl)' will be characterized by different formulations of the Reynolds
number. For example. Rp;pe = pUDljI, where D is pipe diameter, whereas a
morc appropriate fonn for channels might be R pUR'//-l, where RH is
=

hydraulic radius. Use of a different length scale in R will generally alter

the upper limit of R for laminar flow relative to that found for pipes. In
Ihis problem, we refonnulale R using hydraulic radius as the length scale
to obtain a morc general fonn of R that we can use to relate the laminar­
turbulent transition in pipes to olher flow geometries.

A. Find an expression for the hydraulic radius RH in terms of pipe diameter

D for a pipe with a circular cross-section.

A ,,(012)' D
R II
�

P ,,0 4
� �

B. Rearrange the expression found in 3A to get a relationship for D in

terms of Rlf. Substitute this into Rpipe pUDlp to get an expression for
=

R in terms of R".

Rearranging the answer from 4.3A gives D = 4RII• Substituting this

into Rl'ipe gives:

405
C. Set the expression found in 3 8 equal to 2000, the critical Reynolds
number for pipes. Rearrange this to find the equivalent critical value for
RRH = pURfi,l.

pURH pURH 2000

-4
2000 - or R RJI-crll - - 500
_ _

JI JI
- -

4
•

D. Use the equation you developed in 3C to find the critical Reynolds

number for the transition from laminar to turbulent flow in a relatively
wide, rectangular channel (such that RH;::; II).

If RII;::;: II, then the critical Reynolds number for the channel flow
will be 500 as in 4.3C.

Problem 4.4. Use Equation 4.32 to estimate the depth and mean velocity
of a flow in a channel with a slope S = 0.003, width IV = 1 5 m, discharge Q
1
= 1.0 m3 s - and Manning's /I = 0.075. Does the assumption that RH;::; II

seem reasonable for this flow?

Specific discharge q.. = Qlw = (1.0 m3 s-I)1(15 m) = 0.067 m2 S-I.

.

Rearranging Equation (4.32) to solve for" gives:

.,
11= q,,11 (0.067 rn
'
s-'XO.075) = 0.24
kS1l2
m.
(I 011/3 s-IXO.003)"2

U = q•.lh = (0.067 Ill' s-, )/(0.24 Ill) 0.28

=
'
rn s- .

For w = 15m and h = O.24m, RH = (15m)(0.24m)/I(15m) + 2{0.24m)l

= 0.23m. This is within 5% of the value of h, so the assumption that
RII;::;" is reasonable for this flow.

Chapter 5

Problem 5.1. Irthe flood used in the reservoir example delivered the same
volume of water in a shorter amount of time (shorter duration with higher
peak dischargc), as given by the inflow hydrograph in the table below,
how would the outflow hydrograph change? Complete the table below,
assuming the initial conditions and other reservoir parameters remain the
same.

406
 Time t. 2Vp 2Vp:1
I. I.+I.�,
(days)
Step n -0. +° + 0••, t••,
'" At .1
A B C 0 ,

1 0.00 0.7 4.7 342,6 347.3 0.66 0.25

2 0 25
. 4.0 22.0 346.0 368.0 5.31 0.50

3 0.5{) 18.0 28.0 357.3 385.3 10.0 0 75

.
4 0 75
. 10.0 15.6 365.3 380.9 8.75 1.00

5 1.00 5.6 B.6 363.4 372.0 6.35 1.25

6 1 25
. 3D 4.5 359.3 363.8 4.30 1.50

7 1.50 1.5 2.2 355.2 357.4 2.81 1.75

B 1.75 0.7 1.4 351.8 353.2 1.88 2.0

The completed table for this larger inflow shows that the peak
outflow is a smaller fraction of the peak inflow (O!'cak/l!'cak = 0.56)
compared to the example in Table 5.3 and Figure 5. 1 0 (O!'eak/1peak =
0.63).

Problem 5.2. The Muskingum routing coefficients for a stream reach are
detem1ined to be: Co = 0.26, C1 = 0.55, C2 = 0.19. For the inflow
hydrograph given in the table below, complete the calculation of the
predicted outflow hydrograph.

Time IhT) Inflow 1m3 s-', Outflow 1m3 s...,

0000 10 10

0600 50 20

1200 130 6S

1800 110 112

2400 70 100

Problem 5.3. In Figure 5.14, a 40 +year annual series ( 1 950-1989) of

floods on the Eel River, California, is plotted against the fraction [riCI/ +
I)] of floods with discharges greater than or equal to each value

A. Fit a line through the data and determine the return period of an 8000
m 3 S-1 flood.

The value of the exceedance probability indicated by a best-fit line

for Q = 8000 mJ S-1 is approximately 0.20 (Figure PS.I). The return

407
 period T
....'urn
= 110.20 = 5 yr.

B. Estimate the magnitude of the IOO-year flood.

408
 o
0.05
� 0.10
.g 0.20
.g 0.30
0.40
�
c.
..
u
c:
0.70
..
"C

a: 0.80
l:l
w

o
o

1
'
Discharge, Q (m s-')

0.01
0.02

0.05
0.10
,.,
!::
.
-

.
.c
-

..

.g 0.30
0.40
�
c.
..
u
c:
..
�

�
"
w 0.90
0.95

0.99
10' lcr 103 104 10'
Discharge, Q (ml 5-1)

409
Figure PS.I Probability plot for Eel River, California, peak annual
discharge data, 1950--1 989 (a), and 1917-1 996 (b).

A 100-year flood has an exceedance probability of 1/100 = 0.01.

From the best-fit Hne (Figure PS.I), the corresponding discharge is
approximately 1.7 x 104 mJ S- I . Note that this is smaller than the
largest discharge on record (over 2.1 x 104 mJ S- I, in 1965). If the
annual series data for the Eel River were truly log-normally
distributed, this would suggest that the 1965 flood has a return
period longer than the 40-year record length. In fact, the full record
for the Eel River begins in 1917. The 1965 flood is the largest of the
82-year record from 1917 to 1989. The best-fit line through the 82-
year record still puts the 100-year flood at approximately 1.7 x 104
m3 S-I.

Chapter 6

Problem 6.1. You arc charged with designing a very simple filtration
system for a community water supply, using cylindrical sand columns (K =
5.0m day- I). The filter needs to be 3.0m long to adequately trap
particulates in the water, and since the system will be driven by gravity,
the pressure heads at the top and bottom of the (vertically oriented) filter
wi11 be zero.

A. What diameter filter is required to treat 4.0 x 103 gallons of water per
day? Is this value feasible (anything larger than about I m is not
feasible)?

Because the pressure heads at the top and bottom of the filter are
zero, the hydraulic gradient must be equal to I. Therefore:

I rn3
Q=4.0x I03gai day-Ix =15.1 mJday-'.
264.2 gal

,
dh
=-KA-=
o
Q 5.0mday-lxJr -
dz 2
,

or

410
 112
4 15.1 mJday-l
D= 2 .0 m.
!C 5.0
=

m day-l

This value is too large to be feasible.

B. Consider each of the alternatives and how you might modify your
design:

i. Lengthen the sand filter (how long?)

Lengthening the sand filter will not increase the discharge, since
the hydraulic gradient would still be equal to I.
ii. Raise the hydraulic head at the inflow (how high?)
Increasing the hydraulic gradient from 1 to 4 would meet the
requirement of 4.0 x 103 gallons per day of treated water. This
would require a hydraulic head at the top orlhe 3.0-m long
column of 12.0m (almost 40 feet), which may be impractical.
iii. Usc several filters (how many? what size?)
You could use four I-m diameter filters in parallel to treat 4.0 x

103 gallons of water each day.

Problem 6.2. A penneameter is used to measure the hydraulic

conductivity of a porous medium (see Figure 6.3). The penneameter is
perfectly round in cross section, with a diameter of 50.0 mm. The
following parameters are measured: zl = 220.0 mm, z2 = 1 50.0mm, Pl/pg =
2 3 0 0mm, p/pg = 280.0 mm, with L = 200.0 mm, and a discharge at the
.

lower end of the column Q = 500.0 mm3 min- I.

A. Which way is water flowing in the column? [s water flowing from high
to low hydraulic head? From high to low pressure?

liz = p/pg + z2 = 430.0 mm.

Therefore, the water is flowing from high to low hydraulic head, as

we would expect. In this case, the water is flowing from low to high
pressure; hydraulic head is decreasing along the now because the
elevation head is decreasing faster than the pressure head is
increasIng.
• •

B. Calculate the specific discharge, q (mm min- I).

41 1
 q = Q/A = (500.0 mm3 min-I)/[lI'(25.0 mm)2] = 0.25 mm min-I

C. Calculate the hydraulic conductivity, K, of the material (m S-I)

K q dl -0.25 mm min-I x [200.0 mm/(430.0 m m -450.0 mm)].

dh
= - =

K=2.50mm min-I.

K 2.50 mm Illin-I X I min/60s X I mil 000 mm =4.2 X 10-s III

=
l
S- .

D. Calculate the intrinsic permeability, k (m2).

Assume that the water bas a density p = 1000.0 kg m-3 and a

viscosity 11= 1.0 x 10-3 Pa . s. Then,

k= KJ!.... =4.2 x 10-' rn 5-1 x [1.0 x 10-3/(1000.0 kg rn-J x 9.81 m 5-2)] =4.3 x 10-12 m l.
pg

Problem 6.3. Consider two piezometers placed side-by-side but open in

different aquifers at depth (i.e., the two piezometers on the right side of
Figure 6.6 ). The following measurements are made:
PIezometer 111 Piezometer 112

Elevanon of piezometer (m above mean sea level) 200 200

Oepth of pIezometer (m) 60 20
Oepth to water in piezometer (m) 20 18

A. Calculate the elevation and hydraulic heads (rclativc to mean sea level),
pressure head, and fluid pressure in the two piezometers, and fill in the
table below.

Piezometer n Piezometer #2

Elevation head, Z 1m) 140 180

Pressure head, plpg 1m) 40 2

Hydraulic head, h=p/pg+z (m) 180 182

Pressure, p=pgd (Pa) 3.92xl0s 1.96x 104

B. Calculate the vertical hydraulic gradient between the two piezometers.

412
 Is the flow of water upward or downward?

dhidF (h, - h,)I(z, - z,) � (180 - I 82 )1( 140 - 1 8 0 ) � -21-40 � 0.05.

Therefore, since the gradient is positive upward, the flow is

downward (from piezometer #2 to piezometer #1)

Problem 6.4. Consider the now net for a drainage problem shown in
Figure 6. 1 5. Drains such as pipes and culverts placed in a wet field may be
uscd to removc groundwater by ercating a "sink" or area of low hydraulic
head. In the figure, a cross section through such a field is shown. The
hydraulic conductivity, K, of the surficial material is 1 .0 x 10- 5 m S-I. The
thick black lines represent impenneable boundaries; a constant head is
assigned to the top and lower left side. The cross seclion is 20 m long by
1 0 m deep. The gray lines are equipotentials and the blue lines are
streamlines.

. . ..
. .
. . .
. .

. . : . : . .. ... . . . : . . ' . ' .

. . - . .. . . - ...
.

..
. .
.. . : . . . .

.. . . ..... .. . ... .. .. . .. . . . .. ..... . .

.
, �,
.
.
. .

. '. . ' . .
. . .
'. '

.
.

,
. . '
' . . .
·
. . .
' . . .

.
.. ' ..

. . .- .
'
: ,
:.
.

...
. . ' . . '
' '
,
. '. ' "

.. . .... . . .. . ... . . . . . . . ... . - . ... -

"
' . . _ . ' ''
. .
. : . . . . . :
:

.. . . . .
.
.
. ' ' .
,
. . '. ' " ' ' . '. .

. . . ..
·
.
'
,". " ' '
' ' .

, :-<.; : . ..-. . . .:... : ..:..; ., . ...:

. .
, ,
. . . .

.'
. : .
. . : . . . :
' . . . .. . . . '
,
. . . . . .
. . . " . _ . .

,, . . . . . , . .. . . . .. . . . .
.
.
: -
' ' "
"
. . "
' .
. .

. .. . . : ... . .
' . .
.
'. ' .
, . .
. . " ' .' . .'

; -. :-:- :,-, : : :
: . . . :
.
. ' . . . ' '

.
. .

. .
-- . . . - . .
'
. " ' .

.. : . - .. .:.
. . " : , ,

.. . . .. . : ..
' . ' "
. .

. ..
-
.

:. "
: - . :
:

,. . : .. . .. ... : .: : .. . :
. . . . -.
. . ' ' '.
.

. ..
. . ' . ' .
. . . . .

.. .
. . . .
. . . . . . .

. .. . ... .. . ... .
. . . .
.

.. . .
.
·

. . . .

.",
. .
•
. ' .
. . .
.
'
'. . .

A. Place labels on the equipotentials, indicating the value of the hydraulic

head along the line.

Beginning with the uppermost equipotential, they should be labeled

as 8 m� 6 m, 4 m, and 2 m.

B. Calculate the discharge through each streamtube, and the total discharge
or ratc at which thc field is draincd (m 3 day-I pcr m width of material).

413
 Qs = Kb dh = 10- 5 m S-I x I m x 2 m = 2 x 10- 5 m 3 S-I x 86400 S day-I
= 1.7 m3 day-I.

There are 5 streamtubes, so Q=5Q,. = 8.6 mJ day-I.

Chapter 7

Problem 7.1. Oetennine the natural basin yield (m3 yr-I per meter basin
width) for the following cases.

A. For the basin in Figure 7.4a with L = 5000 m and K = 30 m yr-I.

The basin is 5000 m long, so the equipotential spacing, dh, is 15 m.

Therefore, the discharge through each streamtube, Qj, is (30 m yr­
I) x (IS m) or 450 ml y..-I per meter width. There are slightly more
than 2 streamtubes, so the natural basin yield is slightly more than
900 mJ yr-I per meter basin width.

B. For the basin in Figure 7.4b with L = 5000 m and K = 30 m yr-I.

Because Land K are the same as in part A, the discharge through

each streamtube is the same, Qj = 450 mJ yr-I per meter width.
There are slightly less than 8 complete streamtubes, so the natural
basin yield is slightly less than 3600 mJ yr-I per meter basin width.

C. For the basin in Figure 7.5a with L = 200 m and K= 100 m yel. You
might want to begin by calculating the discharge through each local
flow system (dashed box).

The basin is 200 m long, so the equipotential spacing, dh, is 0.6 m.

Therefore, the discharge through each strcamtube, Qs' is (tOO m yr­
I) x (0.6 m) or 60 ml yr-I per meter width. There are slightly more
tban 4 streamtubes witbin each local flow system, so the natural
basin yield from each is slightly more than 240 m3 yr-I per meter
basin width. There are 5 local flow systems in the basin, so the total
natural basin yield is approximately 1200 m3 y..-I per meter basin
width.

Problem 7.2. Answer the following questions for the basin in Figure 7.25.

414
 e
-

x (m)

Figure 7.25 Flow net for Problem 7.2.

A. Is the lower unit an aquifer or an aquitard?

Groundwater flow is being focused into the lower unit, which

indicates that it is an aquifer. There are other clues thai indicate
this as welJ. Downward flow across the upper unit indicates that it
is less permeable than the lower unit; the flow is mostly horizontal
in the aquifer (lower unit). The streamlines and equipotential lines
form curvilinear squares in the upper unit, but not in the lower
unit. This is a good indication that the hydraulic conductivity of the
two units differs. In the lower unit, the streamlines and
equipotentials form rectangles with their long axes along the
streamlines. Therefore, the lower unit must be more permeable.

B. What fraction of the total basin yield passes through the lower unit?

The total number of streamtubes in the flow net is II.

Approximately 9.5 streamtubes (approximately 85% of the flow)
pass through the lower unit.

Problem 7.3. Consider the flow net shown in Figure 7.26. The sides of the
region are groundwater divides, the top boundary is the water table, and
the bottom is an impermeable boundary. Streamlines are blue and
equipotentials are gray.

The answers to parts A through D are shown in Figure PS.2.

A. Label the equipotentials with the appropriate value of hydraulic head

(m).

B. Draw arrows on the streamlines indicating the direction of groundwater

flow.

C. Label all recharge and discharge areas.

415
D. Indicate at least one area within the flow net where flow is relatively
fast, and one area where flow is relatively slow.

E. Determine the water level (hydraulic head, m) in wells A and B.

The open portions of the wells intersect the 12.5 m (well A) and 30
m (\\'ell B) equipotentials.

50 DISCHARGE' R E CHA R G E 'D I S C H A R G E 'RECHARGE

, , ,
, , , 45
40 , , ,
, 42.

"
30
E
-

N ".
20
A

'0

s
0 +- ��
'� 5 �
__ .5 2��
71 �� o �
n . 5�25"-
� 27��
5. ��
� ____ *32�
.5"--.
____,;35"--,__ ___

o 10 20 30 40 50 60 70 80 90 100
x (m)
Figure PS.2 Answers to Problem 7.3, parts A through D.

Problem 7.4. Well hydrographs for unconfined aquifers can be used to

determine the change in water stored within the aquifcr. Consider the
change in water-table level observed in Coffee County, Tennessee,
between mid-May and August 2010 (Figure 7.9).

A. What is the change in water-table level during this time period (m)?
Does this indicate an increase or a decrease in water stored within the
aquifer?

The water-table level decreases by approximately 15 m over this

time interval. This indicates that the amount of water in storage has
decreased over this time interval, assuming that the observed
change in water-table level is representative of that in the aquifer.

B. If the specific yield of the aquifer is 0.25, and the aquifer has an area of
600 km2, what is the change in water stored over the same time interval
(m))?

416
 Again, assuming that the change observed in the one well is
representative of the aquifer as a whole, the change in water stored
is found by multiplying the specific yield by the area and the
change in water level:

t:J.V= Sy x dll x A = (0.25)(-15 m)(6.0 x 108 rn2) = 2.25 x 109 m J.

Problem 7.5 Determine the recession constant, c (day- I), for the two
groundwater "reservoirs" contributing to Pescadero Creek during the
spring and summer of 20 I 0 (Figure 7.23).

The equation for the recession curve is (7.17). In logarithmic form

this would be:

InQ - InQo = -C(

or, rearranging to solve for c:

c = "I"
:.",
Q Til Q",.
- -'.",
.! o_
I

Choosing points on the two straight lines in Figure PS.3, we can

solve this expression for c. For the early "reservoir," choosing June
and July:

= InQo-lnQ =In(0.40)-ln(0.19)= -o.92-( -J.66)= 5X O_'

c 2. I days- '.
f 30 days 30 days

For the later "reservoir":

c= InQo -lnQ=In(0.14)-ln(0.1 15)=-1.97 -(-2.1 )=

6 6.3x 10-3 days-I.
I 30 days 30 days

417
 1.0

0.8

0.6
�
-

in
0.42.
E 0.4

0.14- 1-
- _ I
0.115
- L _ I-
0.1
May Jun Jul Aug Sep
Month in year 2010
Figure PS.3 Groundwater recession analysis for Pescadero Creek,
California, showing points used in analysis of recession constants
(Problem 7.5).

Chapter 8

Problem 8.1. Tensiometers are installed at OAm and O.Sm above the water
table in a uniform sandy soil with the moisture characteristic and hydraulic
conductivity curves given in Figures 8.4 and 8.5, respectively. One set of
tensiometer readings indicates that the capillary-pressure head al lhc !irs!
of these tcnsiometers is 0 45 In and at the second is -0.6 m.
-
.

418
A. What is the direction of water movement between the two
tensiometers?

If the gradient in hydraulic head is positive, water flow is

downward (specific discharge is negative). Conversely, if the
gradient in hydraulic head is negative, water flow is upward
(specific discharge is positive).

Z!!. = 11o.j-/Io� = (0.5 m+ljfos)-(O.4 m+IjfO.4) = [CO.5 -0.6)-(0.4-0.45) = - 5

0. .
d= O.lm O.lm O.lm

Therefore, the flow is upward.

B. Estimate the magnitude of the specific discharge between the two

tensiometers.

From Darcy's law, q = -K x (--0.5) = O.SK. The estimate requires a

value for K. From Figure 8.4, we determine that the moisture
content is about 0.14 at the O.4-m tensiometer and about 0.05 at the
0.5-m tensiometer. From Figure 8.5 we determine K values for the
two tensiometer locations to be about 5 x 10-{i m S- I and 6 x 10-8 m
S-I. We conclude that the upward specific discharge is between 2.5
x 10--6 and 3 x 10--8 m S- .
I

Problem 8.2. Heidmann et al. ( 1 990) report that the moisture

characteristic for a sandy loam soil in Arizona can be represented
accurately by the equalion I{J = -963. 7() -4.659, where I{J is in meters and () is
expressed as a percentage (e.g., 26% rather than 0.26). The saturation
value of moisture content for this soil is estimated to be 27%.

A. Plot the moisture characteristic for this soil.

Choose values of moisture content from 2% to 27% and calculate

capillary pressure head from the equation given. The moisture
characteristic is the plot of capillary-pressure head versus moisture
content (Figure PS.4).

419
 30

'"
-

-25
'"
- "

r::
"
-
r:: 20
0
"
"
�
:J 15
-
...
0
-

E
." 10
-
-

"
E
:J 5
0
-

>

o
102 10 ' -1 10 1 10 2 104 10 ....
Capillary-pressure head, 'V (m)
_ _ - - _ - _ -

Figure PS.4 Moisture characteristic for Problem 8.2, part A.

B. What is the equilibrium profile of (a) capillary-pressure head and (b)

moisture content above a static water table in this sand?

(a) The term equilibrium in this case is interpreted as zero now.

For zero flow, the hydraulic head is constant (no gradient) so '" =
-z. (b) The equilibrium profile of moisture content has the same
shape as the moisture characteristic that is given by the equation
presented by Heidmann ct al. (1990) but with 'I' replaced by-z
(Figure PS.S).

420
 20 20 ,------r--,

0-1------
� 20 � 10 0 o 1 2 3 4 5 6
Capillary-pressure Volumetric moisture
head, III (m) content, e (%)
Figure PS.S Distribution ofcapillary-pressure head (let
f ) and volumetric
moisture content (right) for Problem 8.2, part B.

C. Suppose that moisture content measurements from 1 0 Tn below the soil

surface i n this sandy loam to a depth of 50 III show a constant moisture
content of 3%. The water table is at a depth or80 m. Over the interval
orthe moisture measurements, how do capillary-pressure head and
hydraulic head vary?

Moisture content is constant at 3% over the interval so capillary­

pressure head is also constant: �I =-963.7 x 3--4·659 = -5.77 m.
Hydraulic head is the sum of pressure and elevation heads.
Elevation head varies from 30 m to 70 m to which is added the
constant -5.77 m to get hydraulic head. Note thai 'he gradient in
hydraulic head is positive in this case (slope of h versus z curve is
upward to the right; Figure PS.6), indicating downward water flow.

421
 70 .....,. --, 70 , ,-
...,
-, -
__ _ _ _ _ _ _
____

30-,
-L __ ...
, _ ---,
, __ �
-7 -<l -5 20 30 40 50 60 70
Capillary-pressure Hydraulic head (m)
head, \11 1m)
Figure PS.6 Distribution of capillary-pressure head (left) and hydraulic
head (right) for Problem 8.2, part C.

Problem 8.3. Assume that a wetting front moves into the sandy loam soil
described in Problem 8.2. The moisture content at the surface of the soil is
held constant at 27%. The underlying moisture content is constant al 6%.
If the saturaled hydraulic conductivity for this soil is 3 x 10--6 m 5-1 ,
estimate how long it will take the wetting front to move 1 m into the soil.

Assume that the Green-Ampt equation is valid for this case.

Estimate the wetting front capillary-pressure head as IIf = -963.7 x
6--4·659 = -0.23 m. Then, noting that I = Lffj,O,

and substituting appropriate numerical values, we get t';::; 12 hr.

Chapter 9

Problem 9.1. Irrigation is used in areas in which agricultural productivity

can be enhanced by maintaining the soil water content well above the
pennanent wilting point (e.g., 0 > 20w) throughout the growing season.
When irrigation water is applied, root zone soil moisture should not exceed
field capacity 0fe' because, if 0 > Of,"' water and fertilizer are lost from the
root zone by drainage. With this management approach soil moisture
remains below field capacity and the only losses of soil water from the

422
root zone are by evapotranspiration. Assuming that no rainfall occurs, the
temporal dynamics of depth-average soil moisture 0 in the root zone can
be expressed by the soil wmer balancc equation.

dO
Z-=-el'
dl
where Z is the depth of the root zone. In other words, changes in the water
stored in the control volume (i.e., the root zone) are due to the difference
betwcen water inputs and outputs (sec Chapter I). In this specific case
there are no rainfall-induced inputs, while the only outflow is due to el.
The irrigation period (i.e., the time between two consecutive applications
of irrigation water) is calculated by integrating the soil water balance
equation between time' = 0 when 0 = Ofc (i.c., right after irrigation) and
time 'i when 0 = 00 (and it is time to irrigate again), assuming that the
evapotranspiration rate linearly decreases from the potential rate PET at 0
= O/<: to zero at 0 = Ow (i.e., et = PET(O-O)..)J(Oc
j -O ».
...

Consider the ease of a crop grown on a sandy soil with Ofc = 0.19 and a
crop-specific wilting point, Ow = 0.05. The root zone is 40 cm deep. After
cach application of irrigation water, soil moisture is equal to Ofe. Assuming
that PET = 6 mm d- 1, calculate the irrigation pcriod, Ii' i.e., how long it
would take for soil moisture to decrease from field capacity to the value 00
= 20", in the absence of any rainfall input.

Using the above equation with PET = 6 rum d-1, Of<. = 0.19, OK' =
0.05,00 = 20 , and Z = 0.40
... 01, we find: I;::::; 10 days.

Problem 9.2. With the irrigation scheme dcscribed in the previous

problcm during each application of irrigation watcr soil moisture is
increased from 00 = 20", to Ofe' The volume of water (per unit area)
required to makc this change is Z(Ofe- (0), Calculate the amount of water
per unit area that should be applied every time the area is irrigatcd.
Assume that during thc irrigation process about 20% ofthc irrigation watcr
is lost in evaporation bcfore il infiitrates.

During each application of irrigation water soU moisture is

423
 increased from 00 = 20", to
0fc' The volume of water (per unit area)
required to make this cbange is Z{(Jjc - 00) = 36 mm. Because of
evaporation losses the irrigation method has an 80% efficiency (i.e.,
only 80% of the irrigation water infiltrates). Thus the amount of
water that needs to be spent in irrigation is 36/0.80 = 45 mm every
10 days.

Chapter 10

Problem 10. t. A hydrologist studying runoff generation in a catchment

measures the following chloride concentrations during the peak of a
rainstonn event: Cn = 4.5 /lmol L-1, Co = 40.5 limoI L- 1, and Ct= 36.0
pmol L-1. What fractions of total streamflow are contributed by new water
and old water?

From Equation 10.3,

(36.0 - 40.5) -4.5

Q = Q = 0.13Q
(4.5-40.5) t -36.0 ' t

or Q,/Q, = 0.13 = 13%. Therefore, Q/Q, = J-Q,IQ,=O.87 = 87%.

Thus, 13% of the flow is contributed by new water and 87% by old
water.

Problem 10.2 . Data on hydraulic conductivity versus depth in a soil on a

forested slope are presented by Harr ( 1 977). Examine the data to determine
whether the assumption that K decreases exponentially with depth is
reasonable, and, jf it is, estimate Ko and!
•

Depth in soil (m) Hydraulic conductivity, K (m S-I)

0.10 9.8 x 10-4
0.30 1.1 x 10-3
0.70 4.5 x 1 0-4
1.10 4.9 x 10-4
1.30 4.4 x 10-5
1.50 6.1 x 10-5
•

424
 If K decreases exponentially with depth, the data should define a
straight line on a plot of K on a logarithmic scale and z 011 a linear
scale. The relationship can be inspected visually and the
parameters of Equation 10.13 (Ko and}) can be found as the slope
and intercept of a line drawn through the data (Figure PS.7).

10-'-.-----

Intercept (Ko) = , ,85x1Q 3 mS

'

,.,
-

.� 10-3-_ o
>

o
.-

tJ
::I
't:1
r::
o Slope of line if) -2.2 m-1
o =

.!1 10-4.

10-5-��� �������� �

o 0.5 1 1.5
Soil depth (m)
Figure PS.7 Data and fitted line for Problem 10,2.

Problem 10.3. Two "streamtube" segments of a catchment (i.e" portions

of a flow net based on topography) have properties shown below:

Segment 1 Segment 2

Upslope area, A (m') 500 500

length of contour at base of segment, c (m) 3.S 25

Slope at base of segment, tanfJ 0.02 O.Og

Calculate the topographic index for each segment and indicate which
segment is more likely to produce satumtion-excess overland flow.

Equation 10.5 is used:

425
 TI = In(a/tan fl).

For segment 1, TI=lnl(SOO m2/3.Sm)IO.021 = 8.9. (Note that the units

of the topographic index i n this case are In(m).) For segment 2, TI =
Inl(SOO/2S)/0.081 = 5.5. Segment J has a larger TI and is more likely
to produce saturation-excess overland flow.

Problem 10.4. The soil ofa given catchment has a porosity (t/J) of 0.4 and
hydraulic conductivity as described in Problem 10.2. The catchment has an
average value of the topographic index of 3.5. Under conditions where the
average saturation deficit of the catchment is 100 mm, calculate the
subsurface flow to the stream, qSllhs!lrJace.lfthe throughfall rate at this lime
is 4 x 10-3 mm S-I ( 1 4 mm hr-1), what fraction of the catchment would
have to be saturated if saturation-excess overland flow were to be equal to
one-half of the subsurface flow?

Equation 10.24 can be used:

TIHu.� = K.lf= 1.85 x 10-3 m s-1/2.2 m-1 = 8.4 x 10-4 m2 S-l. Thc soil
parameter, m, is ,plf= 0.4/2.2 m-I = 0.2m. Thus, CJsubsurJac£= 1.5 x lO­
I l
sm S- = I.S x 10-2 mOl S- .
[\'cn if the entire catchment were
saturated (in which case -; would be zero, not 100 mOl) the
contribution of overland flow cannot be half of the subsurface flow.

426
Glossary

actual evapotranspiration the real rate of evapolranspirmiol1 from a land

surface. {Section 2.4}

advection the transport of a contaminant in groundwater due to the

physical flow ofwalcr {Section 7.6}

alternate depths the two physically allowable water depths for flow in a
channel for a given specific energy and a given specific discharge.
{Section 4.2}

anisohydric plants plants that lend to maintain the stomata open and
sustain higher photosynthetic uptakes even when soil moisture is low.
{Section 9.2. 1 }

anisotropic a material whose properties (such as intrinsic permeability)

depend on the direction of measurement. {Section 6.5.2}

aquiclude a safUralcd geological fannalian that may contain waler but

does not transmit significant quantities. {Section 6.4}

aquifer a saturated geological fonnation that contains and transmits

significant quantities of water under normal field conditions. {Section
6.4}

aquifer test an experiment designed to measure the in situ properties of an

aquifer, based on the response of the water level in a well to pumping
or injecting watcr. {Section 6.3. 1 }

aquifuge a saturated geological formation that neither contains nor

transmits significant quantities of water. {Section 6.4}

aquitard a saturated geological formation that is of relatively low

permeability; see aquiclude, aquiu
f ge. {Section 6.4}

artesian aquifer see confilled aquie

f r. {Section 6.4}

assimilative capacity the ability of a river, lake, aquifer, or other water

427
 body to process waste discharges such that they are harmless. {Section
l.2}

average linear velocity v=q/¢ [L II], the average velocity of fluid within
the pores of a porous medium, equal to the specific discharge divided
by the porosity. {Section 6.3.1 and Section 7.6}

baseflow background low-flow conditions in a stream { Section 5.2}; see

also { Sections 7.6 and IO.2}

basin aspect ratio the ratio of basin length (the direction parallel to flow)
to basin depth or thickness (above a low-pemleability unit). { Section
7.3.1}

Bernoulli equation (1I2/2g) + z + (P/pg) = H, stating that the sum of the

pressure, elevation, and velocity heads in a frictionless fluid is constant
along a streamline. {Section 3.5.2}

body forces forces that act unifonnly on each fluid element; examples
inelude gravitational forces and electromagnetic forces. { Section 3.3}

Bowen ratio B [dimensionless], the ratio of the sensible heal flux (H) to
the lafell1 Ilea/jlux (E,). { Section 2.4.2}

capillary barrier coarse layers of sediments that impede the movement of

water under unsaturated conditions. {Section 8.11}

capillary forces forces that are exerted on soil water due to the strong
attraction of water by soil minerals; see capillmy-pressllre head.
{Section 8.2.1}

capillary fringe the zone immediately above the water table in which the
pores are filled with water but the water is under pressure less than
atmospheric. {Section 8.11

capillary-pressure head VI [L], the (negative) pressure head in the

unsaturated zone that develops across curved air-water menisci
because water is attracted more strongly to soil minerals than it is to
olhcr water molecules. { Section 8.2.1}

catchment an area of land, bounded by a divide, in which watcr flowing

across the surface will drain into a stream or river and flow out of the
area through a specified point on that stream or river. { Section 1.4.2}

428
cavitation fonnation of vapor bubbles in (liquid) water under low
(negative) pressure (i.e., strong suction). { Section 9.2.1}

Chczy equation U = CJSRH' an equation relating channel mean velocity

to channel slope and hydraulic radius via a roughness coefficient C
called the Chi!zy /llImber. { Section 4.5}

Chezy number C [LI f2 JI], an empirical roughness coefficient used in the

Chhy eqllation describing open channel flow. { Section 4.5}

coalescence a physical process occurring in clouds by which falling drops

of water repeatedly collide with other drops or droplets, thus fonning
larger drops. {Section 2.2}

combination method evapotranspiration methods that combine the energy

balance and masNransjer me/hods. {Section 2.4}

cone of depression the spatial pattern of drawdown in an aquifer as the

result of pumping. { Section 7.4.1}

confined aquifer a permeable fonnation whose upper boundary is an

aqllitard; water in a well within a confined aquifer will rise above the
top of the aquifer. { Section 6.4}

conservation of mass dMldt= I' - 0', the law that states that for any
particular compartment (usually referred to as a control volume), the
time rate of change of mass stored within the compartment is equal to
the difference between the inflow ratc and the outflow rate. { Section
1.4}

continentality effcct the decrease in mean annual precipitation with

increasing distance from the oceans. {Section 2.2}

continuity equation an expression of conservation ojmass in a flow,

stating that the inflow rate minus the outflow rate equals the change in
storage. {Section 3.5.3}

contributing area the area of catchment upslope from a given block that
contributes inflow to that block. { Section 10.5.1}

convective acceleration the spatial component of acceleration at a fixed

time, e.g., u(olilox) for flow in the x-direction [L 12]. { Section 3.5.1}

429
critical flow flow that occurs at the minimum value of specific energy
allowed for a given specific discharge. The Froude number equals I
for critical flow. { Section 4.2.2}

Darcy's law q = -K(dhldl), states that the specific discharge through a

porous medium is proportional to the hydraulic gradient. {Section 6.3}

density p [M L - 3], the mass per unit volume of a substance, defined at a

point. { Section 3.2}

diffusion a transport process driven by concentration gradients. { Section

2.4.2}

discharge Q = VA [L3 ,1], the volume flux of water. {Section 1.4.2}

discharge area a region in which ground\Vater is moving upward across

the \Vater tabie, thereby discharging into the unsaturated zone above,
or to the land surface or a surface-water body such as a lake or stream.
{Section 7.2}

dispersion the transport of a grO/lI/dh'ater contaminant by processes that

mix waters that flow along paths with various linear flow velocities.
{ Section 7.6}

displacemcnt hcight z(( [L] a parameter expressing the effect of canopy

height on the near surface air flow. { Section 2.4.2}

divide the boundary of a catchment. Typically topographic highs or ridges,

a divide separates an area of land that should drain toward a particular
point on a stream or river from surrounding land areas that do not.
{Section 1.4.2}

drawdown the change in water level in a pumping well or nearby

observation well or piezometer. { Section 7.4. 1}

effective stress [M L- 1 ,2], an upward stress (force per unit area)

0..

exerted by aquifer solids in the subsurface. { Section 7.4.2}

elevation head z [L], a component of the total head that may bc thought of
as the potential energy per unit fluid weight. {Section 3.S.2}

equipotential a line of constant hydralilic head; equipotentials and

streamlines together constitute ajlo\V net. {Section 6.S}

430
cvaporation the physical process involving a phase change from liquid to
vapor by which water is returned to the atmosphere. { Sections 2.1 and
2.4)

cvapotranspiration el [L JI], the sum of all processes by which water

changes phase (from solid or liquid) to vapor and is returned to the
atmosphere. { Section 2.4}

cxceedance probability [dimensionless], the relative frequency associated

with a random variable attaining a value greater than some specified
value. {Section 2.2.3}

field capacity the relatively constant moisture content that a sandy soil
tends to attain following drainage. { Section 8.1 }

flood a peak in stream discharge due to inflow of water following a rain or

snowmelt event. { Section 5.2}

flood routing detennination of river discharge at a point based on

knowledge of the discharge at some upstream location (inflow) and the
characteristics of the intervening river channel or reservoir. { Section
5.1)

flow nct a two-dimensional map of equipolelllials (lines of constant

hydraulic head) and streamlilles in a region of groundwaler flow.
{Section 6.5}

fluid a substance thaI has no resistance to defonnation when subjected to a

shearing force. {Section 3.2}

free surface the upper boundary of an open channel flow, between the
watcr and thc atmosphcre. { Scction 4.2}

frequency analysis a statistical tcchniquc used by hydrologists for

estimating the average rate at which floods, droughts, stonns, stores,
rainfall events, etc., of a specified magnitude recur. { Section 2.2.3 }

friction factor I[dimensionless], an empirically detennined,

dimensionless quantity relating head loss to flow properties such as
velocity and diameter or depth [U1L/(2gD) for pipe flow]. A friction
factor diagram gives values of friction factorIas a function of
ReYllolds !lumber. { Section 3.6.1 }

431
friction slope the slope of the imaginary line that is a distance U2/2g
above the water surface in an open channel. {Section 4.4}

Fronde number F = UIJgh, a dimensionless number that is used to

define critical, subcritical, and supercritical flows in channels. {Section
4.2.2)

gage pressure the fluid pressure relative to atmospheric pressure. The

absolute pressure is the sum of the gage pressure and atmospheric
pressure. {Section 3.4}

gaging station a facility installed at a selected river site to collect long­

term records of flow depth as a function of time. {Section 5.2}

groundwater water found in the saturated zolle of the subsurface.

{Section 6.1 }

groundwater divide an impermeable vertical or near-vertical boundary or

a ridge in the potentiometric surface separating groundwater flow
systems. {Section 7.2}

groundwater mining prolonged pumping of groundwater at rates that

exceed the rate of replenishment by recharge. {Section 7.1 }

groundwater recession a decline in groundwater input to a stream

through time. {Seclion 7.S}

groundwater runoff groundwater that discharges to streams or to the

ocean. {Section 1 .3}

head loss ilL [L}, the loss in head due to friction within the fluid and
between the fluid and the side walls of a pipe or channel. {Section 3.6}

heterogeneity the condition of spatial variation in physical properties;

often applied to the intrinsic permeabiliry of rocks and soil; adjectival
form is heterogeneous. {Section 6.5.2}

hollow-stem auger auger for drilling wells that has a hollow core that
allows placement ofa casing and construction of the well inside the
augers, which are then removed, letting the casing remain in place.
{Section 6.4. 1 }

homogeneity the spatial constancy of physical properties, such as density

432
 or ill1rinsic permeability; adjectival form is homogeneous. {Sections
3.2 and 6.5.1}

Hortonian overland flow see infiltration-excess overlandjlolV. { Section

I0.4.2 }

hydraulic conductivity K = k(pgIJl) [L .1], the ability of a porous

medium to transmit fluid, dependent on both fluid and porous medium
properties. { Section 6.3}

hydraulic descent nocturnal transfer of water from the shallow (and

wetter) to the deep (and drier) soil through the root system. { Section
9.2.3 }

hydraulic gradient dhldl [dimensionless], the change in hydraulic head

per unit distance; the driving force in flow through porous media.
{ Section 6.3}

hydraulic head h =(p/pg)+z [L), the mechanical energy per unit fluid
weight, used in the study of flow through porous media. { Section 6.2}

hydraulic lift nocturnal transfer of water from the deep (and wetter) to the
shallow (and drier) soil through the root system. {Section 9.2.3}

hydraulic radius RH [L], the ralio of the cross-sectional area of flow in a

channel to the wetted perimeter. { Section 4.S}

hydrograph a continuous record of streamflow (stage or discharge) as a

function of time. { Section S .2}

hydrological cycle the global-scale, endless recirculatory process linking

water in the atmosphere, on the continents, and in the oceans. { Section
l.3}

hydrology the study of the occurrence and movement of water on and

beneath the surface of the Earth, the properties of water, and its
relationship with the living and material components of the
environment. { Section 1.1}

hydrophytes plants that grow partly or completely submerged by water.

{ Section 9.2.2}

hydrostatic equation an equation describing the relationship between

433
 pressure and depth in a fluid at rest. For a homogeneous fluid, the
hydrostatic equation is p = pgd. { Section 3.4}

hyetograph a graph of precipitation versus time. { Section 2.2.1}

hysteresis the loop-like curve that relates pairs of hydraulic properties of

an unsaturated porous medium because volumetric moisture comenl,
capillary-presslITe head, and hydraulic condllctivity covary along
different curves depending on whether the soil is imbibing water or
draining. {Section 8.9 }

incompressible fluid a fluid for which density is not a function of

pressure. {Section 3.2}

infiltration the movement of rain or melting snow into the soil at the
Earth's surface. { Section 8.8}

infiltration capacity the maximum rate at which water can infiltrate into a
soil. { Section 8.8}

infiltration-excess overland flow also known as Hortonian overland flow

(after its proponent, Robert Horton), a mechanism of runoff generation
in which the infiltratioll capacity of a catchment or portion of a
catchment is exceeded by the rainfall intensity, which results in
ponding of precipitation at the soil surface and flow across the surface
either in sheets or in small rivulets. { Section I 0.4.2}

infiltrometer a device to measure infi/tratioll rates with water ponded

inside a ring driven into the ground. { Section 8.8.1 }

interception storage the process by which precipitation (either liquid or

solid or both) is temporarily stored either on vegetation surfaces
(canopy interception) or on litter surfaces (litter interception);
intercepted water either can retum to the atmosphere as evaporation or
can become stemjlow or through/all. { Section 2.3 }

intermediate flow system a groundwater flow system that is smaller than

a regiolla/flow system and that is characterized by flow from a water­
table high to a non-adjacent water-table low. {Section 7.3.2}

internal energy E" [M L2 "12], the component of the total energy of a

substance that is due to the kinetic and potential energy of the
individual molecules. { Section 2.4.2}

434
intrinsic permeability k = K(jI/pg) [L2], the ability of a porous medium to
transmit fluid, independent of the fluid properties. { Section 6. 3. 1 }

isohydric plants plants that tend to close the stomata and reduce
photosynthesis as the soil becomes drier. { Section 9.2 . 1}

isohyet line of equal precipitation, in the isohyctal method of estimaring

areal precipitation. {Section 2.2.1 }

isohyetal method a technique for estimating areal precipitation to a

catchment based on representing precipitation structure using lines of
equal precipitation known as isohyets. { Section 2.2.1 }

isotropic a material whose properties (such as intrinsic permeability) do

not depend on the direction ofmcasuremenl. {Section 6.5.2}

laminar flow a smooth, regular flow in which disturbances arc damped

oul by viscous forces. Laminar flows in pipes and channels occur al
Reynolds numbers less than 2000 and 500, respectively. { Sections 3. 1
and 3.7}

land subsidence the decline of the land surface produced by pumping

groundwater wells. { Section 7.7}

Laplace equation an expression of conservation of mass combined with

Darcy 's law that describes steady two-dimensional groundwater flow
in a homogeneolls region. Flow nets are graphical solutions to the
Laplace equation. {Section 7.2}

latent heat [M L2 ,2], the portion of the inrernal ellergy ofa substance
that cannot be "sensed" (i.e., is not proportional to absolute
temperature); latent heat is the inlemal energy that is released or
absorbed during a phase change at constant temperature. { Section
2.4.2}

latent heat flux £, [M ,3], the rate per unit area at which latent heal is
transferred to the atmosphere. { Section 2.4.2}

latent heat of vaporization ).� [L2 ,2], the amount of energy per unit
mass absorbed during a phase change from liquid to vapor at constant
temperature. For evaporation of water at O°C, ...1.v = 2.5 x 106 J kg-I .
{Section 2.4.2}

4 35
leaky aquifer an aquifer that is not perfectly confined but that has leakage
across the surrounding confining layers. { Section 7.4.2}

local acceleration the change of velocity with time at a fixed location, aliI
at [L " l. (Section 3.5.1 }

local flow system a groundwater flow system characterized by flow from

a water-table high to an adjacent low. { Section 7.3.2}

macropore a relatively large pore, such as a soil pipe, animal burrow or

shrinkage crack, in an otherwise fine-grained soil. { Section 8.9}

Manning coefficient n [dimensionless], an empirical channel roughness

used in Manning's eqllation describing open channel flow. { Section
4.S}

Manning's equation U = kRhst/11' an equation commonly used to

calculate channel mean velocity, based on channel geometry and
roughness; II is the Mannillg coejjiciellf. { Section 4.5}

manometer a device used to measure fluid pressure, consisting of a tube

filled with fluid and open at one end; the hydrostatic equation is used
to relate the pressure at the open end to an unknown pressure at the
measurement point. { Section 3.4}

mass-transfer methods methods providing a discrete representation of

diffusion processes such as evaporation or transpiration. { Section 2.4}

mean velocity U= QIA [L i l], the cross-sectionally averaged fluid

velocity. { Section 3.5.3}

moisture characteristic the relationship between moisture content and

capillmy-press/fre head for a porous medium. {Section 8.3}

Muskingum method a numerical method for routing a flood through a

river channel. Given an inflow hydrograph, the method predicts the
outflow hydrograph for a given stream reach. { Section 5.4.2}

natural basin yield the average rate of discharge From a basin under
natural or undisturbed conditions (i.e., in the absence of anthropogenic
groundwater withdrawals or changes in climate or vegetation).
{Section 7.2}

436
normal stress a force per unit area oriented perpendicular to a surface of a
fluid or solid object. Pressure is a nonnal stress. {Section 3.3}

numerical method a method for solving equations by transfonning them

inlo one or more algebraic equations that can be solved more easily
than the original equation, usually on a computer. { Section 5.4}

occult precipitation the condensation of dew and/or deposition of waler

droplets from fog and low clouds onto leaf surfaces and the subsequent
dripping of this water down to the forest floor. {Section 2.2}

orographic effect the increase in mean annual precipitation with elevation

typically obscrvcd in many mountain ranges around the world.
{Section 2.2}

osmometer a device used to measure the osmotic pressure. {Box 9. 1 }

osmosis transport of water molecules through semipenneable membranes

from lower to higher solute concentration areas. {Box 9.1}

osmotic compensation increase in xylem water potential attained by

increasing solute concentration in xylem water, thereby decreasing the
xylem osmotic potential. {Section, 9.2.1}

osmotic pressure n [L], pressure that develops in the presence of

concentration gradients across a scmipcmleablc membrane {Box 9.1}

paired watersheds a set of two watersheds of similar size, in the same

region, and with similar topographic and hydrogeological features but
different land cover conditions; typically used to investigate the impact
of land cover on streamflow. {Section 9.3.2}

Penman's method see cOlllbillatioll llletllod. { Section 2.4.4}

permanent wilting point the driest soil moisture conditions that a plant
can withstand without wilting. { Section 9.2. 1 }

permeameter a device used to measure the flow rate through and the
hydraulic conductivity of a porous medium. {Section 6.3.1 }

phreatophytes deep rooted plants that draw part of their water from
beneath the water table. { Section 9.2.2}

piezometer a single tube manometer used to measure pressure head (and

437
 thereby hydraulic head) at a point in the subsurface. {Section 6.4}

piezometric surface see potentiometric slIIface. {Section 6.4}

Poiseuille's law U = -{dpld,)(D2/32ji), an equation for the velocity of a

laminar pipe now. { Section 3.7. 1 }

porosity If = V/V/ [dimensionless], the fraction of the total volume ofa

porous medium occupied by void space. { Section 6.3.1 }

porous medium a rock, sediment, or soil that contains pores or void

spaces. { Section 6.2}

potential evapotranspiration PET, the maximum rate of

evapotranspiration from a vegetated catchment under conditions of
unlimited moisture supply. {Section 2.4.2}

potentiometric surface a surface that depicts the distribution of hydralilic

heads in a confined aquifer; the water in a well or piezomeler
penetrating a confined aquifer defines the surface. { Section 6.4}

precipitation the dominant process by which water vapor in the

atmosphere is returned to the Earth's surface either as liquid drops
(e.g., rain) or solid particles (e.g., snow) under the influence of gravity.
{ Section 2.2}

precipitation intensity [L y- l], a measure of the rate of precipitation,

commonly computed for a specified duration. { Section 2.2.3}

precipitation recycling water that enters the atmosphere from

evapotranspiration in a region and later contributes to precipitation in
the same region. { Sections 1 .3 and 2.2}

pressure p [M L-] [2], the force per unit area acting perpendicular to a
surface, or normal stress. { Section 3.3}

pressure head p/pg [L], a component of the total head that may be thought
of as the "flow work" or the work due to pressure per unit fluid weight.
{Section 3.5.2}

pressure transducer an instrument that converts pressure into a

proportional electrical signal that can be recorded. { Section 6.4.2}

438
pumping test a technique to estimate aquifer transmissivity and storativity
by pumping water from onc well and obscrving induced changcs in
watcr level ovcr time in anothcr wcll. {Scction 7.S.2}

quiekflow onc of two components into which a flood hydrograph can be

separated (the other is known as baseflow). { Section 10.2 }

rain shadow low rainfall area on the leeward side ofa mountain range.
{ Section 2.2}

rating curve a relationship between srage and discharge used to convert

continuous measurements of stream depth (stagc hydrograph) to a
discharge hydrograph. {Section S.2}

reach a segment of a stream or river channel. { Section 4.2}

recharge area a region in which water is crossing the water table

downward, hence recharging the groundwater system. { Section 7.2 }

recurrence interval the interval between two events associated with a

random variable attaining a value greater than some specified value.
{ Section 2.2A}

regional flow system a groundwater flow system characterized by flow

from a regional water-table high to a regional water-table low.
{Section 7.3.2}

residence time Tr = VII [T], a measure of the average time a molecule of

water spends in a rcservoir. The residence time defined for steady-state
systems is equal to the reservoir volume divided by the inflow or
outflow rate. { Section IA.1}

return flow the process by which groundwater reemerges from the soil at
a saturated area and flows downslope as overland flow. { Section
IOAA}

return period Trel,,,,, [T], a measurc of how often (on averagc) an cvent
(precipitation, flood, etc.) will occur that is greater than some chosen
valuc; the invcrsc of the exceedancc probability. { Section 2.2.3}

Reynolds number R = pUDlp [dimensionlcss], a dimcnsionlcss number

representing the ratio of inertial to viscous forces in a flow. Flows with
low Reynolds numbers « 2000 for pipe flow) are laminar; flows with

439
 high Reynolds numbers (>4000 for pipe flow) are turbulent. Flows of
different fluids with the same Reynolds number will be similar.
{ Section 3.7}

Richards' equation an expression of mass conservation in the unsaturated

zone, incorporating Darcy 's law. Solutions to the Richards' equation
provide a history of the pressure distribution in a vertical soil column.
{ Section 8. 5}

roughness irregularities or protmsions along the boundary ofa flow (e.g.,

inner pipe wall or stream channel bed) that increase frictional flow
resistance. { Sections 3.7.2, 4.4 and 4.5 }

roughness height Zo ell, a parameter that accounts for the roughness of

the land surface. {Section 2.4.2}

runoff ratio the ratio of average annual surface runoff to average annual
precipitation for a given land area V/"j5) { Section 1.4. 1 }

saturated zone a region of the subsurface where pores are completely

filled with waler; the saturated zone is bounded at the top by the waleI'
lable. {Section 6.4}

saluration value of moisture content the volumetric moisture content

when all pores ofa soil or rock are filled with water. { Section 8 . 1 }

saturation vapor pressure es,,' [M L- I ]2], in a system in which both

liquid water and waler vapor are present, the partial pressure exerted
by the water vapor during an equilibrium condition in which the rates
of vaporization and condensation are equal. { Section 2.2 }

saturation-excess overland flow a mechanism of runoff generation that is

particularly important in vegetated catchments in humid regions in
which a shallow water table intersects the ground surface, causing
ponding of water at the soil surface and flow across the surface either
in sheets or in small rivulets. { Section I DA.2 }

sensible heat [M L2 ]2], that portion of the illlernal energy of a substance

that can be sensed (i.e., is proportional to absolute temperature).
{Section 2.4.2}

shallow subsurface stormflow a mechanism of runoff gencration

whereby water flows through a shallow, penneable soil horizon, such

440
 as when a perched water table forms above a layer of tile soil with low
permeability; some of the flow may occur along preferred pathways
known as macropores. { Section IOA.3}

shear stress t [M L-I 'I2J, a langential force per unit area acting on the
surface ofa solid or fluid. { Sections 3.2 and 3.3}

slug test a technique to estimate hydraulic conductivity of a fonnation by

removing a "slug" of water from a well and observing the rate of
recovery of the water level in the well. { Section 7.5.1 }

snowpack accumulated snow that melts seasonally in many mountain

regions but which, in some instances, is pennanent and fonns glaciers.
{Section 1 .3}

soil horizon a soil layer defined on the basis of physical and chemical
properties and on the history of its fonnation. { Section 8.9}

soil moisture water that is held in soils and rocks under pressures less than
atmospheric; water in the unsaturated zone. { Section 8. 1 }

soil water potential lis [L], the hydraulic head of soil water. {Section
9.2.1 }

solar energy energy deriving from radiation from the sun. It is the driver
of the hydrological cycle. { Section 1 .3 }

specific contributing area upslope contributing area per unit contour

length, a = Alc. { Section 1O.5.2}

specific discharge (groundwater now) the discharge per unit cross­

sectional area of flow through porous media, q = QIA [L J l ]. { Section
6.3}

specific discharge (open channel flow) the discharge per unit width of
channel in a rectangular open channel, q", = Qlw = Uil [L2 ' ].
11
{ Section 4.2}

specific energ.y [L], the energy per unit weight of water in a channel with
respect to the channel bottom as datum. The total energy H [L] is the
sum of tile bottom elevation and the specific energy. { Section 4.2 }

specific energy diagram a graph of specific energy versus water depth for

441
 a given specific discharge. The diagram shows the physically
allowable water depths for a given specific discharge and a given
specific cncrgy. { Scction 4.2}

specific heat capacity cp [L 2 0 -1 [2}, a proportionality constant that

relates the change in illlernal energy of a substance to a change in
absolute temperature. {Section 2.4.2}

specific yield Sy [dimensionless}, the volume ofwatcr produced from an

IInconfined aquifer per unit aquifer area per unit decline in the water
lable. {Section 7.4.1 }

stage the dcpth of flow in a stream. { Section 5.2 }

steady now a flow that is constant in time at each location in the flow. [n
steady flow, the local acceleration is zero. { Section 3.5.1 }

stemnow a physical process by which water is transferred from

inrerceptioll storage to the soil surface by flowing along the stem or
trunk of a tree. { Section 2.3 }

stomata tiny pores in the leaves of vascular plants by which gases

(including carbon dioxide, oxygen, and water) are exchanged with the
atmosphere. { Section 2.4}

storativity S [dimensionless}, the volume of watcr produced from a

f
confined aquier per unit aquifer area per unit decline in the
potell1iometric slilface. { Section 7.4.2}

streamline a path defined by the motion of fluid elements in a flow; at any

point along a streamline, the flow direction is tangent to the streamline.
Conceptually, water may not cross streamlines. In describing
groundwater flow, streamlines and equipotentials together constitute a
flow lIet. {Sections 3.5.2 and 6.5 }

streamtube a region within aflm v lief between two streamlines. { Section

6.S. I }

stress a force per unit area (S[ units: N m-2 or Pal. { Section 3 .3 }

subcritical flow the (relatively) deep and slow flow corresponding to a

given specijic energy and a given specific discharge. The Frollde
!lllmber is less than unity for subcritical flow. { Section 4.2.2}

442
sublimation the physical process by which water in the solid phase
changes to water vapor and is directly returned to the atmosphere.
{ Section 2.4.2}

supercritical flow the (relatively) shallow and rapid now corresponding to

a given specific energy and a given specific discharge. The Frollde
/Illmber exceeds unity for supercritical flow. { Section 4.2.2}

surface forces forces that act through direct contact on specific surfaces of
a fluid or solid body. { Section 3.3}

surface runoff waler from rainfall or snowmelt thaI runs over the surface
of the Earth in sheets, rivulets, streams, and rivers. { Section 1 .3 }

tensiometer a device for measuring negative capil!m},-press/lre heads in

soils. { Section 8.7 }

throughfall a physical process by which water is transferred from

interception storage to the soil surface by dripping off the leaves of the
canopy. { Section 2.3 }

time domain renectometry (TDR) a method for measuring soil moisture

by timing the movement of a high-frequency electromagnel'ic wave
reflected from the opcn end of two steel rods inserted into the soil.
{Section 2.4.2}

TQPMQDEL a hydrological catchment model based on land-surface

topography. { Section I O.6}

topographic index TI = In(altan P), where a is the upslope contributing

area per contour length and tan P is the local slope; used by the
catchment model TOPMODEL to calculate the water balance for
individual blocks within a catchment. { Section I O.5.2 }

total head H [L], the sum of the pressure, elevation, and velocity heads in
a frictionless fluid. { Section 3.5.2}

total stress (Ir [M L-I [2], the weight (a force) of material overlying a
plane of unit cross-sectional area in the subsurface. {Section 7.4.2 }

transmissivity T = Kb [L2 '11], a measure of the ability of an aqlli{er of

thickness b to transmit water. {Section 6.5 . 1 }

443
transpiration the physical process by which water changes phase from
liquid to vapor, is released through the stomata ofa plant, and returns
to thc atmosphcre. { Section 2.1 }

turbulent flow flow with rapid, irregular fluctuations of velocity in space

and time. Turbulent flows in pipes and channels occur at large
Reynolds numbers. { Sections 3. 1 and 3.7}

unconfined aquifer a permeable fonnation whose upper boundary is the

�'v'ater table. { Section 6.4}

uniform flow a flow that does not change from place to place along the
flow path; the convective acceleration is zero. { Section 3.5.1}

unit weight y[M L-2 T- 2], the gravitational force per unit volume, pg,
acting on a fluid or solid (SI units: N m-3). { Section 3.2}

unsaturated zone the zone in a soil or rock between the Earth's surface
and the waleI' table; pores in the unsaturated zone are partly filled with
water and partly filled with air. { Section 8.1 }

vadose zone see III/satl/rated zone. { Section 8. 1 }

vapor pressure e [M L-l j2] the actual partial pressure exerted by a

,

vapor within an air mass; related to the concentration of water vapor in

air. { Section 2.2}

variable contributing area concept the idea that catchment areas where
saturation-excess overlandjlolV develops expand and contract with
time over a stonn. {Section 10.4.2}

velocity head [L], a component of the total head that may be thought of as
the kinetic energy per unit fluid weight. {Section 3.5.2}

velocity profile the variation of velocity with distance away from a

boundary. { Section 3.2}

viscosity If [M L-lr-l], a measure ofa fluid's ability to resist defonnation.

Fluids of high viscosity flow more slowly than low viscosity fluids,
everything else being equal (SI units: Pa . s). { Section 3.2}

volumetric moisture content e [dimensionless], the volume of water held

in a soil or rock per bulk volume of the sample. {Section 8.1 }

444
watcr budgct a calculation of the inflows, outflows, and change in storage
for a particular control volume (such as a lake or a catchment) over a
particular timc pcriod. { Section 1 .4}

waterlogging saturation ofthc shallow soil rcsulting from the rising of the
water lable to the ground surface. {Section 9.2.2 }

water table a surface separating the salurated and /ll/saturaled zones of

the subsurface, defined as a surface at which the fluid pressure is
atmospheric (or zero gage pressure). { Section 6.4 }

water-table aquirer see Ilnconfined aquifer. { Section 6.4}

water vapor density [M L-3], mass of water vapor per unit volume.
{ Scction 2.4.2}

weir an artificial obstruction such as a stcp or dam ovcr which all thc
water in a channel must flow and that can be used 10 measure stream
discharge. { Section 4.3 }

well casing a pipe, typically steel or PYC, that serves to line a well.
{ Section 6.4. 1 }

well hydrograph a record of the variation in water level in a well through

time. { Section 7.4. 1 }

well screen a section at the end of a well casing that allows water to enter
the wcll; construction may be holes drilled or slots cut in a pipe or may
be a wire mesh material. { Section 6.4. 1 }

xylem capillary tubes within the plant. {Section 9.2}

xylem water potential k, [L], the sum of elevation head, capillary

pressure head, and osmotic potential. {Section 9.2. 1 }

445
References

Allan, J.A. 1998. Virtual water: A strategic resource global solutions to

regional deficits. Ground Water 36:545-546.
Andreassian, V. 2004. Waters and forests: From historical controversy to
scientific debate. Journal ofHydrology 29 1 : 1-27.
Averyl, K., J. Fisher, A. Huber-Lee, A. Lewis, J. Macknick, N. Madden, J.
Rogers, and S. Tcl1inghuiscn. 20 I I . Freshwater use by u.s. pOlVer
planls: Electricil), 's IhirSIJor a precious resource. A report of the
Energy and Water in a Wanning World initiative. Cambridge, MA:
Union of Concerned Scientists.
Barber, N.L. 2009. Summary of estimated water use in the United States in
2005. U.S. Geological Survey Fact Sheet 2009-3098.
Barnes, H. H", Jr. 1967. Roughness characteristics ofnatural channels.
U.S. Geological Survey Water-Supply Paper 1 849.
Bazemore, D.E. 1993. The role of soil water in stormflow generation in a
forested headwater eatehmenl. M.S. Thesis, Departmenl of
Environmental Sciences, University of Virginia.
Beven, K., and P. Gennann. 1982. Macropores and water flow in soils.
Water Resources Research 1 8 : 1 3 1 1-1325.
Beven, K.J. 1978. The hydrological response of headwater and sideslope
areas. Hydrological Sciences Bulletin 23:419-437.
Beven, K.J., J. Gilman, and M. Newson. 1979. Flow and flow routing in
upland channel networks. Hydrological Science Bulletin 24:303-325.
Beven, K.J., and M.K. Kirkby 1979. A physically based, variable
contributing area model of basin hydrology. Hydrological Sciences
Bulle/ill 24:43-69.
Biswas, A.K. 1972. His/olY of hydrology. Amsterdam: North-Holland.
Bonan, G. 2008. Ecological climatology. Cambridge, UK: Cambridge
University Press.
Boyer, E.W., G.M. Hornberger, K.E. Bencala, and D.M. McKnight. 1997.
Response characteristics of DOC flushing in an alpine catchment.
Hydrological Processes 1 1 : 1635-1647.
Brovkin, V., M. Claussen, V. Petoukhov, and A. Ganopolski. 1998. On the
stability of the atmosphere-vegetation system in the Sahara-Sahel
Region. JOllmal ofGeopllY�"ical Research. 103:316 t 3-3 1 624.

446
Brown A.E., L. Zhang, T.A. McMahon, A.W. Western, and R.A. Vertessy.
2005. A review of paired catchment studies for determining changes in
water yield resulting from alterations in vegetation. Journal oj
Hydrology 3 10:28-6 1 .
Brutsaert, W. 1982. Evaporation into the atmosphere. Dordrecht, the
Netherlands: D. Reidel.
Brutsaert, W. 2005. Hydrology: An introduction. New York: Cambridge
University Press.
Burgess, S.S.O., M.A. Adams, N.C. Turner, and e.K. Ong. 1998. The
redistribution of soil water by tree root systems. Oec% gia 1 15:306--
311.
Bums, D.A., J.A. Lynch, B.J. Cosby, M.E. Fenn, and J.S. Baron, and US
EPA Clcan Air Markets Div. 20 1 1 . National Acid Precipitation
Assesslllelll Program repor/ IO Congress 201 I: An il/legrated
assessment. Washington, DC. National Science and Technology
Council.
Caldwell, M.M., and J.H. Richards. 1989. Hydraulic lift water effiux from
upper roots improves cffectiveness of water uptake by deep roots.
Oecolagia 79: 1-5.
Campbell, G.S., and J.M. Norman. 1998. All illlroeil/ction to ellvironmelllal
biophysics. New York: Springer-Verlag.
Carr, J.A., P. D'Odorico, F. Laio, and L. Ridolfi. 2012. On the temporal
variability of the virtual water network. Geophysical Research Lerrers.
doi: 39, L06404.
Chanson, H. 2004. Environmental hydralllics ofopen channelflolVs.
Oxford: Elsevier-Butterworth-Heinemann.
Childs, E.C. 1969. The physical basis ofsoil waleI' phenomella. New York:
John Wiley & Sons.
Christenson, S.c., and 10M. Cozzarelli. 2003. The Norman Landfill
Environmental Research Site: What happens to the waste in landfills?
U.S. Geological Survey Fact Sheet 040-03. http://pubs.usgs.gov/fs/fs-
040-03/.
Clapp, R.B. 1977. Infiltration in relation to rainfall intensity: An
investigation of an empirical equation using simulated data. M.S.
Thesis, Department of Environmental Sciences, University of Virginia.
Cooper, H.H., and C.E. Jacob. 1946. A generalized graphical method for
evaluating fonnation constants and summarizing well field history.
Transactiolls oJthe American Geophysical Ullioll 27:526-534.
Cosby, 8.J., G.M. Hornberger, D.M. Wolock, and P.F. Ryan. 1 987.
Calibration and coupling of conceptual rainfall-runoff/chemical flux

447
 models for long-term simulation of catchment response to acidic
deposition. In Systems analysis in water quality management. edited by
M.B. Back, pp. 1 5 1-160. Oxford: Pcrgamon Prcss.
Court, A. 1960. Reliability of hourly precipitation data. JOllmal of
Geophysical Research 65:4017-4024.
Darcy, H. 1 856. LesJon/ailles pllbliqlles de la ville de Dijoll. Paris: Victor
Dalmont.
DePaul, V.T., D.E. Rice, and O.S. Zapecza. 2008. Water-level changes in
aquifers of the Atlantic Coastal Plain, predevelopment to 2000: U.S.
Gcological Survey Scientific Investigations Report 2007-5247.
DeWalle, D., and A. Rango. 2008. Principles ojsnow hydrology. New
York: Cambridge University Press.
Dietrich, W.E., c.J. Wilson, D.R. Montgomery, J. McKean, and R. Bauer.
1992. Erosion thresholds and land surface morphology. Geology
20,675-679.
D'Odorico, P., F. Laio, and L Ridolfi. 2010. Does globalization of water
reduce societal resilience 10 drought? Geophysical Research Letters
37,L i3403.
Domenico, P. A., and F.W. Schwartz. 1990. Physical and chemical
hydrogeology. New York: John Wilcy & Sons.
Dunnc, T, and R.D. Black. 1970. Partial area contributions to stonn runoff
in a small New England watershed. Waler Resources Research 6: 1 296-
131 I .
Dunne, T, and LB. Leopold. 1978. Water ill ellvironmental planning. San
Francisco: W. H. Freeman.
Dunne, T, T.R. Moore, and C.H. Taylor. 1975. Recognition and prediction
of nmofT-producing zones in humid regions. Hydrological Sciences
Bulletin 20:305-327.
Eaglcson, P.S. 1970. DYllamic hydrology. Ncw York: McGraw-Hill.
Eltahir A.B.B., and R.L. Bras. 1996. Precipitation recycling. Reviews oj
Geophysics 34:367-378.
Falkcnmark, M., and J. Rockstrom. 2006. The new blue and grcen water
paradigm: Breaking new ground for water resources planning and
management. Journal oj Water Resources Planning and Managemellt
I EI29-132.
Falkenmark, M., A. SemIen, A. Jagerskog, J. Lundqvisl, M. Matz, and H.
Tropp. 2007. 011 the verge ofa lIew water scarcity: A callJor good
governallce alld human ingenllity. SIWI Policy Brief.
Falkenmark, M., and J. Rockstrom. 2004. Balancillg walerJor humalls alld
nature. London: Earthscan.

448
Food and Agriculture Organization of the United Nations (FAO). 2010.
Global forest resource assessment 20 I 0: Key findings.
hllp:llwww.fao.orglforcstry/fra/fra20 I Olen!.
Fedoroff, N.V., D.S. Battisti, R.N. Beachy, P.J.M. Cooper, D.A. Fischhoff,
C.N. Hodges, V.c. Knauf, D. Lobell, B.J. Mazur, D. Molden, M.P.
Reynolds, P.c. Ronald, M.W. Rosegrant, P.A. Sanchez, A. Vonshak, J.­
K. Zhu. 20 I O. Radically rethinking agriculture for the 21st century.
Sciellce 327:833-834.
Feller C.W. 2000. Applied hydrogeology, 4th cd. Upper Saddle River, NJ:
Prentice Hall.
Foley J.A., N. Ramankutty, K.A. Brauman, E.S. Cassidy, J.S. Gerber, M.
Johnston, N.D. Mueller, C. O'Connell, O.K. Ray, P.c. West, C. Balzer,
E.M. Bennett, S.R. Carpenter, J. Hill, C. Monfreda, S. Polasky, J.
RockstTom, J. Sheehan, S. Siebert, O. Tilman, and D.P.M. Zaks. 201 1 .
Solutions for a cultivated planet. Nature 478:337-342.
Fraedrich, K, A. Kleidon, and F. Lunkeit. 1999. A green planet versus a
desert world: Estimating the effect of vegetalion extremes on the
atmosphere. Journal of Climate 12: 3 1 56-3 163.
Freeze, R.A. 1971. Three-dimensional, transient, saturated-unsaturated
flow in a groundwater basin. Water Resources Research 7:929-941 .
Freeze, R.A. 1972a. Role of subsurface flow in generating surface mnoff:
1 . Baseflow contributions to channel flow. Wafer Resources Research
8:609-623.
Freeze, R.A. 1972b. Role of subsurface flow in generating surface runoff:
2. Upstream source areas. Water Resources Research 8: 1 272-1283.
Freeze, R.A., and J.A. Cherry. 1979. Groundwater. New York: Prentice­
Hall.
Freeze, R. A., and P. A. Witherspoon. 1966. Theoretical analysis of
regional groundwater flow: 1 . Analytical and numerical solutions to the
mathematical model. Water Resources Research 2:641--656.
Freeze, R. A., and P. A. Witherspoon. 1967. Theoretical analysis of
regional groundwater flow: 2. Effect of water-table configuration and
subsurface penneability variation. Wafer Resources Research 3:623-
634.
Freeze, R. A., and P. A. Witherspoon. 1968. Theoretical analysis of
regional groundwater flow: 3. Quantitative interpretations. Water
Resources Research 4:581-590.
Gedney, N., P.M. Cox, R.A. Betts, O. Boucher, C. Huntinglford, and P.A.
Stott. 2006. Detection of a direct carbon dioxide effect in continental
river runoff records. Nature 439:835-838.

449
Giordano, M. 2009. Global groundwater? Issues and solutions. Annual
Review ofEnvironment and Resources 34: 153-178.
Gleiek, P.H. (Editor). 1993. Water in crisis: A guide to the world 'sfresh
water resources. New York: Oxford University Press.
Green, W.H., and G.A. Ampt. 191 1. Studies on soil physics: I . The flow
of air and water through soiIs. Journal ofAgric/lllllral Science 4: 1-24.
Guymon, G.L. 1994. Unsaturated zone hydrology. Englewood Cliffs, NJ:
PTR Prentice Hall.
Haan, c.T. 2002. Statistical methods ill hydrology, New York: J. Wiley &
Sons.
Hamilton, P.A., and R.J. Shedlock. 1992. Arefertilizers andpesticides in
the groundwater? A case study ofthe Delmarva Peninsula: Delm'v'are.
Marylalld. and Virginia. U.S. Gco[ogica[ Survey Circular [080.
Harr, R.D., 1977. Water flux in a soil and subsoil on a steep forested slope.
Journal of Hydrology 33:37-58.
Harris, D.M., J.J. McDonnell, and A. Rodhe. 1995. Hydrograph separation
using continuous open system isotope mixing. Water ResOllrces
Research 3 1 : 1 57-1 7 1 .
Heidmann, L.J., M.G. Harrington, and R.M. King. 1990. Comparisoll of
moisture retentioll cun'es in representative basaltic and sedimentmy
soils ill Arizona prepared by two methods. Research Note RM·500,
USDA Forest Service, Rocky Mountain Forest and Range Experiment
Station.
Helm, D.C. 1982. Conceptual aspects of subsidence due to fluid
withdrawal. In Recent trends in hydrogeology, edited by T. N.
Narasimhan. Geological Society of America Special Paper 189, pp.
103-139.
Helton, J.c., D. R. Anderson, M.G Marietta, and R. Rechard. 1997.
Perfonnance assessment for the waste isolation pilot plant: From
regulation to calculation for 40 CFR 1 9 1 . 1 3 . Operations Research
45,1 57-177.
Hoekstra, A.Y. and A.K. Chapagain. 2008. Globalization ofn;ater:
Sharing the planet'sfreshwater resources. Malden, MA: Wiley·
Blackwell.
Hornberger, G.M., K.E. Bencala, and D.M. McKnight. 1994. Hydrological
controls on dissolved organic carbon during snowmelt in the Snake
River near Montezuma, Colorado. Biogeochemistly 25: 147-165.
Hornberger, G.M., KJ. Beven, and P.F. Gennann. 1990. Inferences about
solute transport in macroporous forest soils from time series models.
Geoderma 46:249-262.

450
Hamberger, G.M., and E.W. Boyer. 1995. Recent advances in watershed
modeling. In u.s. Natiollal Report to IUGG. 1991-1995. Contributions
/11 Hydrology. pp 949-957. Washington: American Geophysical Union.
Horton, R.E. 1933. The role of infiltration in the hydrologic cyclc.
Transactiolls ojthe Amer/can Geophysical Vllioll 14:446-460.
Hsieh, P.A. 200 1 . Topodrive and Particleflow: Two computer models for
simulation and visualization of groundwater flow and transport of fluid
particles in two dimensions. U.S. Geological Survey Open File Report
0 1 -286.
Hubbert, M.K. 1940. The theory of ground-water motion. Journal oj
Geology 48:785-944.
Jonkman, S.N. 2005. Global perspectives on loss of human life caused by
floods. Nat/lral Hazards 34: 1 5 1-175.
Jury, W.A., and H. Vaux. 2005. The role of science in solving the world's
emerging water problems. Proceedings oJthe National Academy oj
Sciences (PNAS) \02: 1571 5-15720.
Jury, W.A., and H.J. Vaux, Jr. 2007. The emerging global water crisis:
Managing scarcity and conflict between water users. Advances in
Agronomy 95: 1-76.
Katul, G.G., R. Oren, S. Manzoni, C. Higgins, and M.B. Parlangc. 2012.
Evapotranspiration: A process driving mass transport and energy
exchange in the soil-plant-atmosphere-climate system. Reviews 0/
Geophysics 50:RG3002. doi: I 0.1 029/20 1 1 RG000366.
Klein, G. 2005. Califomia's Water-Energy Relationship. California
Energy Commission, Final Staff Report, CEC-700-2005-0 I I-SF.
Kohn, M.S. 2012. Bathymetry of Totten Reservoir, Montezuma County,
Colorado, 20 I I : U.S. Gcological Survcy Scientific Invcstigations Map
3203. hup://pubs.usgs.gov/sim/3203/.
Konikow, L. F. 201 1 . Contribution of global groundwater depletion since
1900 to sea-level rise. Geophysical Research Lef/ers 38:L 17401.
Konikow, L.F., T.E. Reilly, P.M. Barlow, and c.1. Voss. 2006.
Groundwater modeling. In The handbook o/groundwater engineering,
2nd ed., edited by J. W. Del1eur, Chapter 23. Boca Raton: CRC Press.
L'vovich, M.I. 1979. World water resources and tileirjilfllre. translated by
R.L. Naec. Washington: American Geophysical Union.
Labat, D., Y. Godderis, J.L. Probst, and J.L. Guyo!. 2004. Evidence for
global runoff increase related to climate warming. Adval/ces ill Waler
Resources 27:63 1-642.
Leakey, A.D.B, E.A. Ainsworth, C.J. Bamacchi, A. Rogers, S.P. Long,

451
 and D.R. Ort. 2009. Elevated CO2 effects on plant carbon, nitrogen, and
water relations: Six important lessons from FACE. JOIIl"l1al of
Experimental Botany 60:2859-2876.
Lee, J.-E., R.S. Oliveira, T.E. Dawson, and I. Fung. 2005. Root
functioning modifies seasonal climate. Proceedings ofthe National
Academy ofSciences 102: 1 7576-1 7581.
Leopold, L.B. 1994. A view ofthe river. Cambridge, MA: Harvard
University Press.
Leopold, L.B., and M.G. Wolman. 1957. River channel patterns: Braided.
meandering, and straight. U.S, Geological Survey Professional Paper
282-B, pp. 39-85.
Leopold, L.B., M.G. Wolman, and J.P. Miller. 1964. Fluvial processes in
geomOlp!lOlogy. San Francisco: W. H. Freeman.
Luckey, R.R., and Becker, M.F. 1999. Hydrogeology, water use, and
simulation of flow in the High Plains aquifer in northwestern
Oklahoma, southeastern Colorado, southwestern Kansas, northeastern
New Mexico, and northwestern Texas: U.S. Geological Survey Waler­
Resources Investigations Report 99-4104.
Maidmenl, D.R. 1993. Hydrology. In Handbook ofHydrology, edited by
D. R. Maidment, pp. 1 . 1-1.15. New York: McGraw-Hill.
Margat, J., S. Foster, and A Droubi. 2006. Concept and importance of
non-renewable resources. In NOI/-renewable Groundwater Resources: A
Guidebook 01/ Socially-SuslailJable Managemclllfor Water-Policy
Makers. edited by S. Foster, and D. P. Loucks, IHP-VI Ser.
Groundwater, vol. 10, pp. 1 3-24. Paris: UNESCO.
McDowell, N., W.T. Pockman, C.D. Allen, D.O. Breshears, N. Cobb, T.
Kolb, J. Klaut, J. Sperry, A. West, D.G. Williams, and E.A. Yepez.
2008. Mechanisms of plant survival and mortality during drought: Why
do some plants survive while others succumb to drought? New
Phytofogist 1 78:719-739.
McGuire, V.L. 201 1 . Water-level changes in the High Plains aquifer,
predevelopment to 2009, 2007-{)8, and 2008-{)9, and change in water
in storage, predevelopment to 2009: U.S. Geological Survey Scientific
Investigations Report 20 I \-5089, hup://pubs.usgs.gov/sir/20 \ 1/5089/.
Meehl, G.A., T.F, Stocker, W.O. Collins, P. Friedlingstein, AT Gaye,
J.M. Gregory, A Kitoh, R, Knutti, J.M. Murphy, A Noda, S,C.B.
Raper, I.G, Watterson, AJ, Weaver, and Z.-c. Zhao, 2007, Global
climate projections. In Climate Cha/Jge 2007: The Physical Science
Basis. Contribution of Working Group I fo the Fourth Assessment
Report ofthe Illfergo�'erl/lnen1a1 Panel 01/ Climate Challge. edited by S.

452
 Solomon, D. Qin, M. Manning, Z. Chen, M. Marquis, K.B. Averyt, M.
Tignor and H.L. Miller. Cambridge, UK: Cambridge University Press.
Meinzer, O.E. 1923. The occurrence 0/ground lVater in the United States.
U.S. Geological Survey Water-Supply Paper 489.
Mekonnen, M.M., and AV. Hoekstra. 2010. The green, blue and grey
water footprint of crops and derived crop products. Value of Water
Research Report Series No. 47, Delft, the Netherlands: UNESCO-lHE.
hup://www.waterfootprint.orglReportslReport47-WaterFootprintCrops­
Voll .pdf.
Mekonnen, M.M., and AY. Hoekstra. 2012. The Blue Water footprint of
electricity from Hydropower. Hydrology and Earth System Sciel/ces
1 6 ; 1 79-1 87.
Miah, M.M., and K. R. Rushton. 1 997. Exploitation of alluvial aquifers
having an overlying zone of low permeability: Examples from
Bangladesh. Hydrological Sciel/ces JOllrnaI 42:67-79.
Milly, P.C.D., K.A Dunne, and A.V. Vecchia. 2005. Global pattern of
trends in streamflow and water availability in a changing climate.
Natllre 43S:347-350.
Moile, F, P. Jayakody, R. Ariyaratne, and H.S. Somatilake. 200S.
Irrigation versus hydropower: Sectoral conflicts in southern Sri Lanka.
Water Policy 1 0 (Supplement I ) : 37-50. doi: 1 O.2166/wp.200S.051 .
Moran, M.S., and R.D. Jackson. 1991. Assessing the spalial distribution of
evapotranspiration using remotely sensed inputs. Journal of
Environmental Quality 20:725-737.
National Research COllncil (NRC). 1 995a. Flood risk management and the
American River Basill. Washington: The National Academies Press.
National Research Council (NRC). 1 995b. Mexico City's lVater supply:
Improving the olltlookfor sllstainability. Washington: The National
Academies Press.
National Research Council (NRC). 1 995c. Ward Valley: An e.xaminatioll
o/seven issues in earth sciences and ecology. Washington: The
National Academies Press.
National Research Council (NRC). 2000. Seeing illto the Earth:
Noninvasive characterization a/the shallolV slIbsllljacefor
envirOl/mental alld el/gineering applications. Washington: The National
Academies Press.
National Research Council (NRC). 2007. Elevation daraforfloodplain
mapping. Washington: The National Academies Press.
National Research Council (NRC). 2012a. Challenges and opportunities in
the hydrologic sciences. Washington: The National Academies Press.

453
National Research Council (NRC). 2012b. Climate change: Evidence.
impacts. alld choices. Washington: The National Academics Press.
National Research Council (NRC). 2012c. Water reuse: Potential/or
expanding the natioll 's water supply through reuse a/municipal
wastewater. Washington: The National Academies Press.
Ortega-Guerrero, A., J.A. Cherry, and D.L. Rudolph. 1993. Large-scale
aquitard consolidation near Mexico City. Ground Waler 3 I :708-71 8 .
Perrone, D., and G.M. Hornberger. 2014. Water, food, and energy
security: Scrambling for resources or solutions? WIREs Water 1 :49--68.
doi: 10. I 002/wat2. I 004.
Qu, Y., and C.J. Duffy. 2007. A semidiscretc finite volume formulation for
mUltiprocess watershed simulation. Water Resources Research
43:W0841 9. doi: l 0.1 029/2006WR005752.
Robson, A., K.J. Beven, and C. Neal. 1992. Towards identifying sources
of subsurface flow: A comparison of components identified by a
physically based runoff model and those determined by chemical
mixing techniques. Hydrological Processes 6: 199-2 14.
Rockstrom, J., M. Lannerstad, and M. Falkcnmark. 2007. Assessing the
water challenge of a new green revolution in developing countries.
Proceedings a/the Nationul Academy a/Sciences (PNAS) 104:6253-
6260.
Rosegrant, M.W., C. Ringler, and T. Zhu. 2009. Water for agriculture:
Maintaining food security under growing scarcity. Anllllal Review 0/
Ellvirollmell1 alld Resources 34:205-222.
Runyan, C.W., P. D'Odorico, and D. Lawrence. 2012. Physical and
biological feedbacks on deforestation. Reviews a/Geophysics
50:RG4006.
Scanlon, B. R., I. Jolly, M. Sophoc1eous, and L. Zhang. 2007. Global
impacts of conversions from natural to agricultural ecosystems on water
resources: Quantity versus quality. Waler Resources Research
43:W03437.
Scanlon, B.R., K.E. Keese, A.L. Flint, L.E. Flint, c.B. Gaye, W.M.
Edmunds, and I. Simmers. 2006. Global synthesis of groundwater
recharge in semiarid and arid regions. Hydrological Processes
20:3335-3370.
Scholl, M.A .. and S.c. Christenson. 1998. Spatial variation in hydraulic
conductivity determined by slug tests in the Canadian River alluvium
near the N0n11an Landfil1, Norman, Oklahoma. Water Resources
Investigations Report 97-4292.
Shah, T., D. Moldcn, R. Sakthivadivel, and D. Seckler. 2000. The global

454
 groundwater simatiol/: Overview ofopportunities and challenges.
Colombo, Sri Lanka: International Water Management Institute.
Shiklomanov, I.A. 1999. World water resources and their use: A joint
SHlIUNESCO product.
http://webworld.unesco.orglwalerJihpJdb/shiklomanov/index.shlml.
Siebert, S., J. Burke, J.M. Faures, K. Frenken, J. Hoogeveen, P. 0611, and
F.T. Portmann. 2010. Groundwater use for irrigation: A global
inventory. Hydrology alld Earth System Sciences 14: 1 863-1880.
Sklash, M.G., and R.N. Farvolden. 1979. The role of groundwater in storm
runoff. Journal ofHydrology 43:45-65.
Skopp, J., M.D. Jawson, and J.W. Doran. 1990. Steady-state aerobic
microbial activity as a function of soil water content. Soil Sciellce
Society ofAmerica J01/rnal 54: 1 6 1 9-1625.
Smith, J.A., M.L. Baeck, M. Steiner, and A.J. Miller. 1 996. Catastrophic
rainfall from an upslope thunderstorm in the central Appalachians: The
Rapidan storm of June 27, 1995. Water Resources Research 32:3099-
3 1 13.
Smith, J.A., OJ. Seo, M.L. Baeck, and M.D. Hudlow. 1996. An
intercomparison study ofNEXRAD precipitation estimates. Water
Resources Research 32:2035-2045.
Smith, T.R. 1 974. A derivation of the hydraulic geometry of steady-state
channels from conservation principles and sediment transport laws.
Journal of Geology 82:98-104.
Soeder, D.J., J.P. Raffensperger, and M.R. Nardi. 2007. Effects of
withdrawals on ground-water levels in southern Maryland and the
adjacent Eastern Shore, 1980-2005. U.S. Geological Survey Scientific
Investigations Report 2007-5249.
Stednick, J.D. 1996. Monitoring the effects of timber harvest on annual
water yield. Journal of Hydrology 176:79-95.
Tilman D., C. Balzer, J. Hill, and B.L. Befort. 201 1 . Global food demand
and the sustainable intensification of agriculture. Proceedings ofthe
National Academy ofSciences (PNAS) 108:20260-20264.
T6th, J. 1962. A theory of groundwater motion in small drainage basins in
central Alberta, Canada. Journal of Geophysical Research 67:4375-
4387.
T6th, J. 1963. A theoretical analysis of groundwater flow in small drainage
basins. Jo/(rnal of Geophysical Research 68:4795-4812.
Wada, Y., L.P.H. van Beek, C.M. van Kempen, J.W.T.M. Reckman, S.
Vasak, and M.F.P. Bierkens. 2010. Global depletion of groundwater
resources. Geophysical Research Letters 37:L20402.

455
Winograd, 1.1. 1981. Radioactive waste disposal in thick unsaturated
zones. Science 2 1 2 : 1457-1464.
Winter, T.e. 1983. The interaction of lakes with variably saturated porous
media. WaleI' Resuurces Research 19: 1203-1 218.
Wolman, M.G. 1955. The natural channel oj Brandywine Creek
Pennsylvania. U.S. Geological Survey Professional Paper 27 1 .
Wolock, D.M. 1993. Simulating the variable-source-area concept oj
streamflow generation with the watershed model TOPMODEL. U.S.
Geological Survey Water-Resources Investigations Report 93-4124.

456
Index

Page numbers in italics refer to figures and tables.

Absolute pressure. See Pressure: absolute

Acceleration, 67-68
convective, 67
local, 67
Acidification, 5, 258, 258, 2 8 1
Acid rain, 5, 258, 258
Actual evapotranspiration. See Evapotranspiration: actual
Advanced Very High Resolution Radiometer (AVHRR), 28
Advection, 199, 200
Albedo, 249, 250, 252
Alternate depths, 94, 95, 98. See also Specific energy
American River, California, 1 1 8, 1 32
Anisotropy, 167
Aquiclude, ISS , J 56
Aquifer, 1 5 5-1 56, 156, 175, 183-1 85, 185
aquifer test, 152, 1 93-198
confined, 1 56, 159-16 1, 161, 1 89-193, 190, 191, 192
High Plains aquifer, 174, 206
leaky, 1 9 1
Madhupur Aquifer, Bangladesh, 1 98, 199
Mexico City aquifer, 174, 205
slOrativity, 1 89-1 9 1 , 190
unconfined, 1 55-1 56, 186-189, 186, 188, 1 92
Aquifuge, 1 5 5, 156
Aquitard, 1 5 5, 167, 184, 185, 190, 1 9 1
Artesian aquifer. See Aquifer: confined
Artificial recharge. See Recharge: artificial
Atmospheric CO2. 241, 248, 290, 291-294, 293
Atmospheric water, 7, 13
Auger, hollow stem, 1 57, 158

457
Average linear velocity, 1 54, 199

Baseflow, 1 2 1, 145-146, 187, 259--260, 260, 261-263, 264

Baseflow recession. See Groundwater: recession
Basin aspect ratio, 180, 181, 182, 183
Basin yield, natural, 179-180, 1 84
Bernoulli equation, 68-77
applications of, 73-76, 74, 1 00-- 1 02
assumptions in deriving, 72-73
derivation of, 69, 70--73
channel flow and, 92-93, 95-99, 100--102
frictionless flow and, 69, 91 , 92-93, 96, 100
head loss and, 77, 78, 79-80
specific energy and, 92-93, 95-99
Biogeochemical cycle, 291
Bluc walcr, 4, 248-250, 249, 287-288, 289
Body force. See Force: body
Boundary conditions: constant-head, 177- 178
impermeable or no-flow, 177
Bowen ratio, 50
Brandywine Creek, Pennsylvania, I I I , 1 12, 1 1 3
at Comog, I l l , 1 12

Canopy interception, 39, 40

Capillary barrier, 235-236, 236
Capillary force, 2 1 3-2 1 4
capil1ary tension, 2 1 4--2 1 5, 214
Capil1ary fringe, 2 1 1-21 3, 212, 217, 220, 267
Capil1ary pressure gradient, 223-226
Capil1ary-pressure head, 2 1 4-221 , 214, 216, 217
capil1ary barriers and, 235, 236
Darcy's law and, 2 1 7, 2 1 8, 220--222
hysteresis loop and, 230--233, 231
infiltration and, 223, 224
macropores and, 234, 267
moisture characteristic curves and, 2 1 5-216, 217, 221
soil properties and, 233-235, 233
soil water potential and, 242

458
 tensiometer based measurement of, 222, 222
vertical flow and, 220-2 2 1 , 222-223
Capillary tube, 147-148, 148
Carbon assimilation, 241 , 244
Carbon dioxide, atmospheric, 241 , 248, 290-291, 291-294, 293
Catchment, 1 3-14, 15
Coweeta, North Carolina, 259
Hubbard Brook, New Hampshire, 258
Catchment divide. See Divide
Catchment hydrology, 257-282
cgs system, 299
Channel bottom roughness. See Roughness: channel bed
Channel flow, 90, 92, 103, 104, 108 , 11 3
Chezy equation, 106
Kannan-Prandtl equation, 108, 109
Manning equation, 106, 1 1 0, 1 1 2
Chczy equation, 106
Climate change, 291-294, 293, 294
Colorado River, 2 1, 140
Combination method, 50
Condensation, 22-23, 252
Cone of depression, 189, 1 91-1 93, 192, 194, 201
Confined aquifer. See Aquifer: confined
Conservation of mass, 9-12, 75-76, 76
flow nets and, 167, 176. See also Continuity equation
Consumptive water use, 2 1 0, 286, 288, 291
Contamination, of water supplies, 5, 146, 163, 198-200, 2 1 1, 281
Continuity equation: for steady flow, 75-76, 76, 93-94
for unsteady channel flow, 125
for vertical groundwater flow, 2 1 9
Continuum assumption, 6 1
Contributing area, 268-270, 269, 270
variable contributing area, 266. See also Specific contributing area
Control volume, 9, 1 1 , 13, 42, 45, 47, 48, 125, 176, 2 1 9
Convective acceleration. See Acceleration: convective
Cooper-Jacob method, 197-198, 199
Coweeta, North Carolina, 259
Critical flow, 94, 97, 98, 99-100, 101

459
Cumulative distribution function, 308

Dam, groundwater flow beneath, 166, 166

Dams, 1 26-127, 127, 140-141
Folsom Dam, Califomia, 1 1 8-1 1 9, 132-133
Glen Canyon Dam, Arizona, 140
Darcy, Henry, 149
Darcy's law, 149-153
flow nets and, 176
hillslope flow and, 268, 269
hillslope stability and, 272
natural basin yield and, 180
plams, 242
restrictions on, 1 54-1 55
soil anisotropy and, 167
steady flow and, 159, 1 6 1
streamtube, 1 64, 179
TOPMODEL and, 275
unsaturated soils, 2 1 6-2 1 8, 235
Decomposition, 248
Deforestation, 53, 249, 250-251, 250, 251, 285
Delmarva Peninsula, 163
Density, 1 1, 6 1, 62, 305, 306
Diffusion, 42, 49, 245
turbulent, 8 1
Digital Elevation Model (DEM), 269-270, 272, 278
Dimensional homogeneity, 298
Dimensions, 297-299, 298
Discharge, in channels, 75, 90, 91, 1 07, 1 10, 1 12, 1/2
bankfull discharge, 1/3, 1 14
discharge hydrograph, 1 19-122, 120, 122, 123
measurement, 1 00-102, I 10, 110
stage-discharge curves, 1 20, 122
Discharge, of groundwater, 149-1 50, 164-165
discharge, total, 1 64-167, 179. See also Specific discharge, of
groundwater
Discharge area, 1 77, 178, 1 80, 183- 1 84
Dispersion, 200, 201

460
Dissolved organic material (DOM), 281 , 282
Divide, 14-1 5, 15
Drainage basin. See Catchment
Drawdown, 189, 1 9 1 , 198, 199
Drought, 243-245, 285, 287, 290, 291

Ecohydrology, 240-256
Effective stress, 1 89- 1 93, 191, 20 I
Elevation head. See Head: elevation
Energy: balance, 41 , 45-49, 47
production, 286--287
water-energy nexus, 286--287
English system, 299
Equipotential, 1 6 1- 1 62, 162, 164, 165, 166, 1 67, 177
Erosion, 272-273, 274
Eutrophication, S
Evaporation, 7-8, 8, 9, 21
increasing air temperature and, 291
water consumption and, 286, 287-288. See also Evapotranspiration
Evapotranspiration, 7-9, 20-2 1 , 40
actual, 41 , 52, 52
in a catchment, 1 5-16, 1 5
climate change, 293-294
energy balance and, 45-49
in global water budget, 9, 10, 1 1-1 3 , 293
green and blue water flows and, 249, 249
latent head flux and, 48
plant water relations and, 242
potential, 41 , 52-53, 52
soil moisture and, 245, 245
unsaturated soil, 234, 235
vegetation and, 248-250
water balance and, 41 , 42
water cycle and, 7-9, 8, 9
water table fluctuation and, 187, 187
water vapor diffusion and, 42-43
Evapotranspiration, methods of estimation: Bowen, 50
combination, 50

461
 energy balance, 45--49, 47
mass transfer, 42--45, 43
Penman, 50
water balance, 4 1 --42
Event water. See Hydrograph separation: "old" water
Everglades, Florida, 245-246
Exceedance probability, 30-3 3, 35-36, 1 3 7, 139, 308, 3 1 1 , 311
Extreme events, 34-36, 35, 1 3 6, 140, 291

Field capacity, 212, 2 1 3. See also Soil moisture; Volumetric moisture

content
Floods, 6, 6, 1 1 8-1 1 9, 1 2 1 , 122
control of, 1 1 8-1 1 9, 1 32- 1 3 3
frequency of, 1 36- 140, 139
routing of, 1 1 9, 125-1 36
Flood wave: attenuation, 123- 1 24, 124, 1 3 1 -1 32, 132, 1 34- 1 36, 135
propagation, 122- 1 24, /24, 1 34-1 36, 135
Flow criticality. See Critical now
Flow nct, 1 6 1- 164, 162, 163, 1 64-167, 165, 166
anisotropy and, 167- 168
and natural basin yield, 175- 1 80
Flow path, 257-258, 263-267, 281
Fluid, definition, 59
Fluid dynamics, 58-59, 66-73
Fluid statics, 62-66, 65, 66
Folsom Dam, California, 1 1 8-1 19, 132- 1 3 3
Food, 286-291, 289
Force: balance of, 63, 63, 68, 70-7 1 , 1 55, 189-190
body, 62, 6B, 70
capil1ary, 2 1 3-214
inertial, 83
surface, 62, 70
surface tension, 2 1 4, 214
viscous, 83
Free surface, 92-93, 1 5 6
Frequency analysis, 30-34, 31, 136-140, 138, 139
Freshwater resources, 2-5, 4, 13, 288-29 1 . See also Water resources
Friction, 69, 77-78, 78

462
 friction factor, 79-80, 80, 83, 84, 105
friction slope, 103, 103, 1 04, 106
Frictionless flow, 69, 72, 76
specific energy, 92-94
Froude number, 100
critical depth, 94, 101 , 101
critical flow, 94, 99, 100, 101, subcritical flow, 94, 97, 98, 99, 100
supercritical flow, 94, 97, 99, 100

Gage pressure. See Pressure: gage

Gaging station, 1 20, 121
Gaussian distribution. See Normal distribution
Geophysical techniques, 1 58, 160
Glaciers. See Snow
Green�Ampt equation, 226--229, 226, 228
Grecn water, 4, 248-250, 249, 287-288, 289
Ground�penetrating radar, 160
Groundwater, 7, 13, 1 45-1 72, 173-206
contamination, 198-200, 200, 201
depiction, 174, 175
divide, 1 77, 1 78
mining, 1 74-175
recession, 202-205, 203, 204. See also Groundwater runorf
Groundwater flow, 1 49-155, 159-168
models of, 168, 1 76, 1 77, 184, 205
quantifying using flow nets, 164-168, 1 75-1 80
in unsaturated zone, 2 1 8-220. See also Darcy's law
Groundwater mnoff, 7, 8, 9, 1 5-16, 264, 266--267
Gumbel distribution, 36

Head: capillary�pressure, 2 1 4-223, 216, 217

elevation, 72-73, 93, 149, 1 56
hydraulic, 84, 148, 149, 159-1 6 1 , 177- 1 79, 179
pressure, 72-73, 149, 156
total, 72-73, 78, 93, 103
velocity, 72, 73. See also Capillary-pressure head
Hydraulic head
Head loss, 78, 79, 103-1 05, 103, 146

463
Heterogeneity, 1 67, 168
effect on groundwater flow systems, 183-1 85, 185
Heterogeneous. See Heterogeneity
High Plains aquifer, 1 74, 206
Hillslope: contributing area, 268, 269
convergent, 268, 270
divergent, 268, 270
runoff, 263-267, 264, 276
stability, 272-273, 274
subsurface flow, 1 63, /63, 177-\78, 272, 276-277
topographic index, 268-271 , 27/
water balance, 278-279, 276
Hollow-stem auger, 1 57, /58
Homogeneity: homogeneous fluid, 6 1 , 73
homogeneous porous media, 1 67, 177, 1 80, 228
Homogeneous. See Homogeneity
Hortonian overland flow. See Overland flow: infiltration-excess
Humidity. See Water vapor: density
Hydraulic conductivity, 149, 151, /53
basin yield and, I 76, 1 79- 180
capillary barriers and, 235, 236, 236
Darcy's law and, 149, 1 5 1- 153, 151
flow nets and, 164-1 67
infiltration and, 226-227
soil moisture and, 2 1 8, 218
soil properties and, 152, 153
transmissivity and, 1 64-165
unsaturated flow and, 2 1 6-2 1 8, 218
Hydraulic geometry, 1 1 3-1 14, JJ3
Hydraulic gradient: Darcy's law and, 150-1 52, 150, 15I
depression cone and, 189, 195
regional groundwater flow and, 1 8 1-1 82
Hydraulic head, 84
Cooper Jacob method and, 1 97-198, 199
Darcy's law and, 150
flow nets and, 1 6 1
groundwater flow and, 148, 149
laminar flow and, 84

464
 Poiseuille's law and, 84
water table and, 1 56, 156, 188
Hydraulic radius, 105-106, 105
Hydraulic redistribution, 246-247, 246
hydraulic descent, 247
hydraulic lift, 246
Hydrograph, 1 19-122, 120, 122, 123, 259-260
discharge, 120, 120, 258, 259
rating curve, 120, 122
stage, 120, 120
TOPMODEL simulation of, 279-280, 280. See also Floods
Hydrograph separation, 261-263, 260
baseflow, 12 1 , 259-260, 259, 260, 261-263
chemical, 261-263, 263
"new" water, 261-263, 267
"old" watcr, 261-263, 267, 263
quickflow, 259-260, 260, 261-263
Hydrological cycle, 7-9, 8, 9, 10, 291-294
global hydrological cycle, J 0, 1 2, 13, 292. See also Water cyete
Hydrology, 1
Hydrostatic equation, 63-66, 74, 93
Hyetograph, 25, 25, 26, 122, 259
Hystercsis, 23 1-232, 231, 232

Impenneable boundary, 177

Incompressible fluid, 62, 72
inertial force. See Force: inertial
Infiltration, 7, 8, 187, 223-226, 225
Green-Ampt equation, 226-229, 226, 228
infiltration capacity, 223-226, 225, 250
infiltration-excess overland flow, 261 , 265-266
vegetation effect on, 250-251
Intiltration-excess overland flow. See Overland flow: infiltration-excess
Infiltrometer, 228
Intcrception, 39, 40, 274
canopy, 39
litter, 39
Intennediate flow system, 182-183, 183

465
Internal energy, 45--46
Intrinsic permeability, 152-1 53, 153, 155, 1 67-168
Irrigation, 1 74, 2 1 0, 249-250, 290-291
period, 255
withdrawals for, 286-288
Isohyet, 27, 28, 29
isohyetal method, 27, 28
Isotropic. See Isotropy
Isotropy, 167

Karman-Prandtl equation, 108, 109

Laminar flow, 59, 8 1 , 81, 82-83, 109

Poiseuille's law, 84, 148-149, 1 5 1 , 155, 2 1 6
Landsliding, 272-273, 274
Land subsidence, 201-202, 202
Laplace equation, 177
Latent heat, 46-48
flux, 47, 48
of melting, 47
of sublimation, 47
of vaporization, 47
Leaf area index, 249, 250
Leaky aquifer, 191
Linear reservoir, 205
Litter interception, 39
Local acceleration, 67
Local flow system, 1 82-1 83, 182, /83
Low-Level Radioactive Waste Policy Act, 2 1 1

Macroporc, 234, 250

Manning equation, 106-107, 1 10, 1 1 2
Manning coefficient, 106, /07, 1 10- 1 1 1, I I /
Manometer, 149, 156
Mass transfer method, 42--45, 43
Mean, 31 , 33, 308-309, 309
Mean velocity, 75
in a channel, 1 06-107, 108-1 10

466
 in a pipe, 83-84
Mexico City, 174, 1 84, 201-202, 202, 205
Moisture characteristic, 2 1 5, 216, 217, 221 , 221
field measurements, 230-231 , 231
Moisture content. See Soil moisture; Volumetric moisture content
Moisture profile, equilibrium, 221 , 221
Murray·Dariing basin, 253
Muskingum method, 1 34-136, 135

Natural basin yield, 180

Newton's law of shear, 60, 60
Newton's second law, 70, 189
No·flow boundary, 1 77
Normal distribution, 308-312, 309, 310, 311, 312
log·normal distribution and flood frequency, 137
and precipitation, 30-33, 33
Normal stress, 62
Numerical method, 125-126

Open channel flow. See Channel flow

Orange County, California. 173
Orographic effect, 24
Osmotic compensation, 242
Osmotic pressure, 244, 244-245
Overdraft, of groundwater, 174, 184
Overland flow, 264, 265-266, 267
and contributing area, 268
and hillslope stability, 272-273, 274
Hortonian, 265-266
infiltration-excess, 265-266
saturation-excess, 265-266. 278
in TOPMODEL, 275, 276, 276, 278-279
and topographic index, 270

Penman's method, 50
Pcnneability. See Hydraulic conductivity; Intrinsic penneability
Pcnncameter, 152

467
Pescadero Creek, California, 202-203
Piezometer, 1 56, J 56. See also Well
Piezometric surface. See Potentiometric surface
Plants and soil water, 241, 242-246, 243, 246
anisohydric plants, 243, 243
hydrophytes, 245
isohydric plants, 243, 243
phreatophytes, 245
Poiseuille's law, 84, 1 5 1 , 1 55
implications regarding permeability, 1 5 1-1 52, 2 1 6
simi larity with Darcy's law, 148-149, 1 5 1 , 155
Pollution. See Contamination, of water supplies; Radioactive waste
Porosity, 154
Porous medium, 146-149, 147, 148, 1 5 1-1 54
Potenlial evapotranspiration, 41, 52-53, 52, 245
Potentiometric surface, 156, 157, 158, 161-162, 189-193, 190, 192
Power generation, 286-287
Precipitation, 7-9
catchment water budget and, 14-16, 14
climate change, 291-293, 293
direct, 263-265, 264
extreme events, 34-36, 35
hyelograph, 25, 25, 26
intensity, 30
isohyet, 28, 27, 29
occult, 23, 252-254
physics of, 22-24
recycling, 8-9, 23, 252
satellite data, 28
statistics of, 30-39, 31, 33, 34, 37
vegetation effect, 252, 253
water budget and, I 1-1 3
water cycle and, 7-9, 9, 10
weighted average, 27-28
Precipitation gage, 24
Pressure, 62, 63-66, 65, 69, 70
absolute, 64
atmospheric, 193

468
 gage, 64, 156
hydrostatic, 63--65, 65, 66
transducer, 157
Pressure gage, 222, 222
Pressure gradient, 70, 79
Pressure head. See Head: pressure
Pressure plate, 2 1 5, 2 I 6
Probability, 307-3 1 2
cumulative distribution function, 308, 3 I I, 3 I 2
density function, 34, 34 308
exceedance, 30-33, 137, 308, 3 1 1, 3 I I
Gaussian distribution, 30
Gumbel distribution, 36
normal distribution, 30, 137, 308-3 12, 309, 310, 31 1, 312
random variable, 30, 307-308
Probability and floods, 136--140, 139
Probability and precipitation, 29-39, 33, 34, 35
Productive consumption usc. See Consumptive water use
Psychrometric constant, 44, 50
Pump test, 197-1 98, 199

Quickflow, 259-260, 260, 261-263, 265, 266, 267

Radar, estimates of precipitation, 28, 29

NEXRAD, 28
Radioactive waste, 146, 168
Rain. See Precipitation
Random variable, 30, 307-308
continuous, 30, 307
discrete, 307
Rating curve, 120, 122
Reach, 92, 135
Recession, groundwater, 202-205, 203, 204
recession constant, 203
Recharge area, 156, 1 77, 1 78, 180, 183
Recharge of groundwater, 173, 186-- 1 87
artificial, 1 87
Recurrence interval. See Return period

469
Regional groundwater flow, 1 80-185, 182, 183
regional flow system, 182-183, 183
Reservoir: concept for catchment water routing, 259, 260
concept in TOPMODEL, 274-275, 275
flood control, 1 19, 126, 127, 1 3 2-133, 132
Folsom Reservoir, 1 1 9, 1 3 3
routing, 126-1 3 3
Totten Reservoir, 127-128. See also Dams
Residence time, 1 2, 13, 258
Return flow, 264, 267
Return period, 3 I , 35-36, 35, 137
Reynolds experiment, 80, 81, 82
Reynolds number, 80, 82, 106, 154
Richards' equation, 2 1 9
Rivers: American River, California, 1 1 8-1 1 9, 1 3 2-133
Brandywine Creek, Pennsylvania, I I I , 1 1 2-1 1 3
Colorado River, 2 1
Pescadero Creek, California, 202-203
Powell River, Virginia, 137
Snake River, Colorado, 279-280, 281-282
Root: hydraulic redistribution, 246-247, 246
water uptake, 241 , 241, 242-244
zone, 2 1 2, 212, 241, 245, 249, 250, 250
Roughness: channel bed, 84-85, 105-107, 107, 108, I I I, 1 1 1
element, 84, 85
height, 45, 46
pipe, 80, 84, 85
surface, 43, 249, 250
Routing, catchment, 273-280, 275
Routing, flood, 1 19, 125-136
Muskingum method, 134-136, 135
reservoir, 126-1 3 3
river, 134-136, 135
Runoff, 7, 8, 1 0
catchment water budget, 1 5-16, 15
global water budget, 1 1- 1 3, 14
vegetation effect, 250-2 5 1 , 251. See also Surface mnoff; Groundwater
mnofT

470
Runoff processes, 263-267, 264
Runoffratio, 1 3, 14

Sahara, 253
San Joaquin Basin, 20 I
Saturated zone, 145, 1 55, 21 2
Saturation deficit, 277-279
Saturation-excess overland flow. See Overland flow: saturation-excess
Saturation value, of moisture conlent, 213
Saturation vapor pressure. See Vapor pressure: saturation
Sensible heat, 46
flux, 47, 48
Shallow subsurface slonnflow, 264, 266
Shear stress, 59, 62
Shenandoah National Park, 262-263, 271
Significant figures, 299-300, 300
Slug test, 194-1 97, 195, 197
Snake River, Colorado. See Rivers: Snake River
Snow, 8, 26
snowfall rates, 26, 28
snowmelt runoff, 8
snowpack, 7, 8
Soil: erosion, 272-273, 274
horizon, 233-234, 233
water potential, 242. See also Hydraulic conductivity; Lntrinsic
penneability; Porosity
Soil moisture, 210, 212, 2 1 9, 223
climate change and, 292, 293-294
equilibrium profile, 221, 221
measurement, 229-230, 229
saturation value of, 2 1 3
water redistribution by plants, 245-248, 246. See also Volumetric
moisture content
Solar energy, 7
Solar radiation, 4 1
Specific contributing area, 269-270, 272-273, 274, 278
Specific discharge, of groundwater, 149-150, 1 5 1-153, 151, 159, 1 64-- 1 65
Specific discharge, in open channels, 94, 96, 107

471
Specific energy, 92-100, 103
alternate depths, 94, 95, 98
diagram, 94, 95, 96-97, 97
Specific heat capacity, 46
Specific yield, 1 88-189, 188
Spreading basing, 173
Sri Lanka, 287
Stage, 119-120, 120, 12/
Standard deviation, 31, 33, 308-309, 309
Steady flow, 68
flow nets, 175-178, 176
of groundwater, 159, 166
in pipes or channels, 71, 73, 75
soil moisture, 219, 223
Steady state, 1 1
Stemflow, 39, 40
Stomata, 40, 45, 241, 243, 243, 294
Storage of water in a confined aquifer, 189-193
Storativity, 189-193, 190
Streamflow 90, 91, 106, 110, 1 /0, 112-114
baseflow, 12 1 , 145, 1 87, 202-205, 259-260, 260
rating curve, 120, 122
roughness coefficient, 105-107, 107, 108, I I I , I I I
stage, 119-120, 120, 12J See also Channel flow
.

Rivers
Streamline, 73, 161, 162, 164-165, 165, 1 76-178
Streamtube, 164-167, 176-180
Stress, 59, 62
effective, 1 89-193, 191, 201
normal, 62
shear, 59, 62
total, 189, 191
Subcritical flow, 94, 97, 99, 100
Sublimation, 47
Subsurface flow, 266, 272-273, 276-279, 276. See also Groundwater flow
Suction, 215
Supercritical flow, 94, 97, 99, 100
Surface discharge. See Surface runoff

472
Surface force. See Force, surface
Surface mnoff, 7, 8, 1 1-1 3 , 14, 1 5-16, 15, 264, 265-266
climate change, 291-293, 294
Hortonian overland flow, 265-266
infiltration-excess overland flow, 265-266
saturation-excess overland flow, 264, 265-266, 268, 270, 272-273, 274,
275
in TOPMODEL, 276, 276, 278-279
vegetation effect, 250-251 , 251
Surface tension, 2 1 3-2 14, 214
System International d'Unites (SI), 298, 299, 299

Tensiometer, 222-223, 222

Tension. See Surface tension
Throughfall, 39, 40, 262, 274, 275, 276, 279
Time domain reflectometry (TOR), 229, 229
TOPMODEL, 274-282, 275, 280
Topographic index, 268-270, 27/, 274, 278-279
TOIaI head. See Head: total
Total stress. See Siress, lotal
Transmissivity, 164-1 65, 189, 272-273, 276-279
Transpiration, 7, 8, 2 [ , 1 8 7, 241-243, 241 See also Evapotranspiration
.

Turbulent flow, 59, 8 1-83, 81, 84, 106, /09

Kannan-Prandtl equation, 108
Turin, italy, 36-39

Unconfined aquifer. See Aquifer: unconfined

Unifonn flow, 68, 93, 103-104, 103, 106, 176
Units, 297-299, 298, 299
conversions, 300-301, 302-304
Unit weight, 6 1, 64
Unsaturated zone, 155, 177, 2 1 0-237, 212, 226
capillary barriers, 235-236
capillary fringe, 2 1 1-2 1 3, 212
capillary-pressure head, 2 1 4, 2 1 5, 217, 222, 229-230, 230
equilibrium profile of moisture content, 220-22 1 , 221
hydraulic conductivity, 2 1 6, 218
no-flow conditions, 220, 223, 224

473
 Richards' equation, 2 1 9. See also Soil moisture
Unsteady flow, 1 25, 21 9. See also Steady flow

Vadose zone. See Unsaturated zone

Vapor diffusion, 235-236
Vapor pressure, 22, 23, 292
saturation, 22, 23, 292
Velocity gradient, 60, 78-79, 78
Velocity head. See Head: velocity
Velocity profile, 60, 60, 78, 1 08, 109
Virtual water, 288-290, 289, 290
Viscosity, 59, 62, 152, 305, 306. See also Newton's law of shear
Viscous force, 83
Volumctric moisture contcnt, 2 1 3, 221 , 221
field measurements, 229-231, 230, 231
and hydraulic conductivity, 2 16-2 1 9, 2 1 7
hysteresis and 231-234, 231. 232. See also Soil moisture

Ward Valley, California, 2 1 1 , 236--2 37

Waste Isolation Pilot Plant (WI.PP), 146, 153, 168
Water, properties of, 305-306
density, 1 1 , 61, 62, 305, 306
dielectric constant, 305
viscosity, 59, 62, 305, 306
Water balance. See Water budget
Water budget, 9-1 6, 14, 2 1 , 41
catchment, 1 3-16, 2 1
global, 9, 1 1-1 3 , 13
Water consumption, 2-5, 4, 5, 286-288, 289. 291 . See also Water footprint
Water cycle, 8-9, 8, 9, 10
biotic controls, 248
climate change effects on, 291-294
intensification, 292. See also Hydrological cycle
Water-energy nexus, 286-287
Water-food nexus, 287-288
Water footprint, 287-288 , 289
Water levels in wells. See Well: hydrograph
Waterlogging, 245-246, 247

474
Water management, 2-6
Water quality, 4-6
Water resources, 1 , 2--6, 3, 4, 173-174, 205, 286-291
Watershed. See Catchment
Water stress: and human consumption, 3, 289
and microbial activity, 247-248, 247
and plants, 242, 245, 253
Water table, 156, 163
depth to, 1 86-189, /86, /87
groundwater flow and, 1 77-1 78, / 78, 1 80-183
hydraulic head and, /79
Mexico City, 174
and specific yield, 1 88, /88
Water-table aquifer. See Aquifer: unconfined
Water-table topography, 1 81-1 83, 182, /83, /84
Water vapor, 1 , 7-9, 1 1 , 22
change of phase, 47-48
density, 22, 42, 292
fluxes, 42-43, 43
Water year, 25
Weighing Iysimeter, 234
Weighted average, 27
Weir, 1 00-102, 101, 126-127
Weir equation, 102, 125, 126-127
Well: casing, 157, 158
construction, I 57, 158
hydrograph, 185-193, 186, 187, 193
tests for aquifer properties, 193-198, 195
water-level measurements, 1 57-1 58
Wetlands, 245, 266
Wetting front, 226-228, 226
Wilting point, 242, 243, 243, 247

Xylem, 241-244, 241

capillary-pressure head, 242
cavitation, 242-244
water potential, 242

475