Hydrology For Engineers

HYDROLOGY FOR ENGINEERS
RA Y K. LIN LEY, JR.
..._ ._
Profe !lor of Hydraulic Engiue ring

Stanford University
.;":
."
MAX A. KOHLER
Chi f Res arch Hydrologist
United States Weather Burea
. . ~. ~O~rlt
, '__,_ __
..../
~
",.,
~f:~ ' ' ' /
/~
I'
/
j
\"""
1"
JO EPH L. H. PAULH
.' .0
,0.
.';
"
\.
,
,\ _
Staff Hydrologist
United States Weather Bureau
MeGRA W-HIJ...L BOOK
New York
MPANY
Toronto London
1958
NBSS&LUP
Re&ional Centre LIbrary
BftogaJore 56\) 024
Aeces.ion No
~>
.....,
HYDROLOGY FOR ENGINEERS
MeGRA W-HILL CIVIL ENGINEER INO SERIES

HARMER
E. DAVIS, Consulting Editor
BABDITT, DOLAND, AND CLEASBY . Wnter upply Engineering

BENJAMIN' Statically Indeterminate tructures
CHOW' pen-channel Hydraulics
DAVIS, rrROXELL, AND WISKO IL . The Te tillg and In pection of
Engine ring Material
DUNHAM ' Advanced Reinforced Concrete
D NHAM . Foundation of tructures
DUNHAM . The Theory and Practice of Reinforced Concrete
DUNHAM AND YOUNG' Contract , Specifications, and Law for Engineers
HALLER'r . Photogrammetl'Y
HENNES AND 'KSE' Fundamentals of Transportation Engin ering
KRYNINE AND J UDD ' Principles of ' ngineering Geology and Geotechnic8
LEONARDS . Foundation Engineering
LIN LEY, I OHLER, AND PAULHUS' Applied Hydrology
LIN LEY, KOHLER, AND PAULHUS' Hydrology for Engineers
LUEDER' Aerial Photograhic Interpretation
MAT ON, S UTH, AND HURD' Traffic Engineering
MEAD, MEAD, AND AKERMAN . Contract , Specifications, and
Engineering Relations
NOlUUS, HANSEN, HOLLEY, BIGGS, NAMYET, AND MINAMI' Structural
Design for Dynamic Lads
PEURIFOY' on truction Planning, Equipment, and Methods
PEURU'OY . E timating Construction Costs
TROXELL AND D AVIS' Comp,o ition and Properties of Concr te
TSCHEBO'l'ARIOFF' oil l\1echanic l I ounqa.tions, and Earth Structures
WANG AD' CKEL . Elementn.ry Theory of Structure
WINTER, URQ HART, O'ROURKE;'AND NILSON' Design of Concrete
Structures
'
HYDROLOGY F R ENGINEERS. Copyright @ 1958 by the McGrawHill Book Company, Inc. Printed in the United States of America. All rights reserved. This book, or parts thereof, may not be reproduced in any form without
permi
n oPthe publish rs. Libr4rg of Gongress Catalog Card Number 58-9999
19 20- MAMB - 7 5 4 3 2
87968
To the memory of
MERRILL BERNARD
friend and colleague
whose enthusiasm was a source
of inspiration to the authors
~'
.'" "'-J
~.\:~.
"r
-~ C'
" - ____ ---.
:.
,"
~ I ~'r'" '" rt '

-.;_
The publication in 1949 of "Applied Hydrol gy" wa well re eiv d,

and that book has found ext nsive u e us a text in school of engin ring_
No single book can meet all ne ds, and "Hydrology for Engin 'rs" has
b en written specifically to erve as an elementary text. Th emphasis
throughout is on quantitative method of arriving at an w rs to hydrologic problems. The handbook approach, a exemplified by the many
empirical formulas widely used in the early days of hydrology, has been
avoided.
"Hydrology for Engineers" is not a mere ond nation of "Applied
Hydrology." While there i much similarity in the organization, 1.11
text has been completely rewritten. Where appropriate, new methods
and con epts developed since 1949 have been included. The experience
of the authors in teaching hydrology over s veral years ha be n utilized
as a basis for. electing topics to be included and methods of 1 r sentatioll.
The student should find hydrology an interesting subject but one much
different from mo t of his engineering cour es. The natural phenomenu.
with which hydrology is concerned do not lend them elves to rigorous
analyses such a are pos 'ible in engineering mechanic. There is, th r fore, a greater variety of methods, more latitude for judgment, and a
seeming lack of accuracy in problem solution. Actually, the accuracy
of sound hydrologic solutions compare favorably with other type of
engine ring computations.
ncertainties in engineering are fr quently
hidden by use of factors of safety, rigidly standardized working procedures, and conservative as umptions regarding properties of materials.
The authors gratefully acknowledge the splendid cooperation of their
many friends and colleague whose helpful Bugge. tions have added much
to this text.
pecial appreciation goes to Walter T . Wilson and avid
Rer hfield of the U.S. Weather Bureau for review and comment on
frequency analysis, to Profes or Stanley N. Davis of tanford niversity
for his helpful review of the chapter on groundwater, and to T. J. Nordenson of the Weather Bureau for many uggestions. Profes or J. B.
Franzini of tanford reviewed the entire manuscript, and many of his
worthwhile suggestions are incorporated in the final text. Miss Dianne
Linsley prepared the final manuscript.
Ray K. Linsley, Jr.
Max A. Kohler
Joseph L. H. Pattlhus
vn
CONTENTS
Preface
vii
Symbols and Abbreviations

Chapter I.
Xl
Introduction
The hydrologic cycle-History-Hydrology

Subject matter of hydrology
Chapter 2.
engineel'ing--
Weather and Hydrology
General circulation- Temperatur

Chapter 3.
1U
Humidity- Wind
23
Precipitation
Formation-Forms-Measurement- Interpretation of precipitation data- Variation in precipi tation- now cover and <:n owfall
Chapter 4.
Water stage-Discharg
Chapter 5.
52
Streamflow
Interpretation of streamflow data
Evaporation and Transpiration
90
Factors controlling evaporation- Estimates of evaporationEvaporation control- Transpiration- Evapotran spirati on- Potential evapotranspiration
Chapter 6.
122
Groundwater
Occurrence of groundwater-Moisture in the zone of aerationAquifers- Movement of groundwate Determination of permeability-Sources and discharge of groundwate Hydraulics of
wells- Groundwater yield
Chapter 7.
Characteristics of the Hydrograph
149
Components of runoff-Recessions- Hydrograph separation

Chapter 8.
162
Runoff Relation
Phenomena of runoff-Estimating the volume of storm runofiht
CONTENTS
Eatimating snowmelt runoff- easonal- and annual-runoff relations

Chapter 9.
Hydrograph of Runoff
193
Unit bydrographs- Overland Bow-Flood formulas

Chapter 10.
Streamflow Routing
216
Wave movement- hannel storage-Reservoir routing- Channel routing- Deriving basin outflow by routing-Gage relations
Chapter 11.
Frequency and Duration Stu die
245
Frequency analysis- Generalization of frequency data- Related

studi s
hapter 12.
Sedimentation
278
Erosion- Suspended sediment- Bed load- Sediment measurement- Re ervoir sedimentation

hapter 13.
Application of Hydrologic Technique
292
Storage-reservoir design- pillway design- Flood-control-reservoir design- Storm-drain design- River forecasting
Appendix A.
Graphical Correlation
311
Appendix B. Physical Con tants, Conversion Tables, and

Equivalents
323
Name Index
329
Subjed Index
333
SYMBOLS AND ABBREVIATIONS

SYMBOLS
A
area
a coefficien t
B width
b coefficient
C Chazy coefficient
C runoff co fficien t
C el ctrical capacitanc
Cp synthetic unit-hydrograph coefficient of peak
Ct synthetic unit-hydrograph coefficient of lag
c coefficien t
c. sediment concentration
D depth
D degree days
d diameter
d coefficien t
E evaporation
E erosion
E voltage
ET evapotranspiration
e vapor pressure
e base of Napierian logarithms
e. atmospheric vapor pressure
e. saturation vapor pressure
F fall
F force
F total infiltration
f relative humidity
f() function of
f. final infiltration capacity
fi infiltration rate
fo initial infiltration capacity
fp infiltration capacity
G safe yield of a groundwater basin
Gi bed-load transport
g gage height
g acceleration of gravi ty
H . heat of vaporization
h height, head
I inflow
1 antecedent-precipitation index
i rainfall intensity
electric current
i. supply rate (rainfall less 'r etention)

J probability of occurrence
j portion of streambed subject to shear
K Muakingum storage constant
K frequency factor
K oompaction coefficient
Kb base-How recession constant
Ka direct-runoff recession constant
K" coeffioient of permeability
Kr recession coefficient
K .. conductivity
k coeffici 'nt
L length
L. distance from outlet to center of a basin
L. length of overland flow
M snowmelt
M; suspend'd 8 diment transport
m coofficient or exponent
m rank
N normal precipitation
N number
n Manning roughness coeffici nt
n coefficient or exponent
n number
o outflow
V. subsurface seepage
P probability of nonoccurrence
P precipitation
p pressure
p porosity
pF logarithm of capillary potential in centimeters of wat.ar
Q volume of discharge or runoff
Qb net long-wave radiation
Q. energy used for evaporation
Q. groundwater-fiow volume
Q. sensible-heat transfer
\1.. )let radiant energy
Qr reBected short-wave radiation
Qs volume of surface streamflow
Q. short-wave radiation
Q. advected energy
Q9 change in energy storage
q discharge rate
q6 base-flow discharge
qd direct-runoff discharg
q. equilibrium flow rate
qA specific humidity
q" p a:k discharge
q. sediment discharge rate
R hydraulic radius
R Bowen's ratio
R
R.
T
S
S
s.
s.
S,
8
Bb
8.
8.
T
T
T
TL
Td
T.
'I'VI
t.
t,
tp
tR
t.
U
u
u
V.
Vo
v
v.
v,
W
Wp
W(u)
w.
w",
X
g
x
x
Y
Y
Y
y
y
1/-
resistance
gas constant
radius
storage
volume of surface retention
storage constant of au aquifer
groundwater storage
surface storage
slope
slope of channel bottom
slope of overland flow plane
dg /u dt
temperature
transmissibili ty
time base of unit hydrograpb
lag time
dewpoint temperat.ure
return period or re urrence interval
wet-bulb temperature
time
time of concentration
time to equilibrium
basin lag
duration of rain
unit duration of rain for synth ti' unit hydrograph
unit.-hydrograph ordinate
wave celerity
a factor in well hydraulics
volume of surface d tention at quilibrium
volume of surface detention when i = 0
velocity
horizontal velocity of sediment particles
settling velocity
infiltration index
precipitable water
well function of u
sp cific weight
specific weight of sediment
ultimate specifio weight of sediment deposits
a variable
the mean of X
the mode of X
distance
a constant or exponent
a variable
a vertical distance
t he mean of Y
a vertical distance
a reduced variate in frequency analysis
a statistical factor in frequency analysis
drawdown in a well
xiii
Itlv
z a. vertical distance
c:
evaporation portion of advected energy
fJ constant
A slope of vapor pre8sure va. temperature curve
A an inorement
e mixing coefficient
II an angle
A total potential
p a.bsolute viscosity
. J! . kinematic visoosity
. t {atio of average sediment load to bottom concentration
"I ll~wen'l:r r~lb coefficient
'1r
3.1416 .. . "
p density
2: summntion
, .', Ilt.Andard deviation'
" sh
1'~ shear at ch'artnel bed

"
.1'"
critical shear
. ' .. ~.' du .Bo;,ts: coc.ffl i nt
4> infiltration index
'ar
x v./V(jl58
,p capillary potential
ABBREVIATIONS
aore-ft acre-foot
atm atmosphere
Bt,u British thermal unit
cal calories
Co centigrade d grees
cu cubic
efs cubio feet p r B cond
csm oubic f et per second per square mile
em oentimeters
FO Fahrenheit degr es
Fig. Figure
ft feet
ips feet per second
g gram
gpd gallons per day
gpm gallons per minute
hr hour
Hg mercury (chemical symbol)
in. incbes
log, logarithm to bas i)
loglo logarithm to base 10
ly Langley
mb millibar
III met r
SYM80LS AND A88REVIATIONS
mi
min
mgd
mo
mph
mst
oz
ppm
pt
sec
sfd
sfm
sq
yr
mile
minute
million gallons per day
month
miles per hour
mean sea level
ounce
parts per million
pint
s conds
second-foot-day
sccond-foot--minute
square
year
Xf
.,r-
INTRODUCTION
Hydrology is that branch of physical geography which is concern d

with the origin, distribution, and properties of the waters of the eo.rth.
Engineering hydrology includes those segments of the very broad field
of hydrology pertinent to the design and operation of engineering projects
for the control and use of water. The boundaries between hydrology
and other earth sciences such as meteorology, oceanography, and geology
are indistinct, and no good purpose is served by attempting to define
them rigidly. Likewise, the distinctions between engineering hydrology
and other branches of applied hydrology are vague. Indeed, the engineer
owes much of his present knowledge of hydrology to agriculturists,
foresters, meteorologists, geologi 'ts, and others in a variety of fields.
I- I. The hydrologic cycle. The concept of the hydrologic cycle is a
useful, if somewhat academic, point from which to begin the study of
hydrology. This cycle (Fig. 1-1) is visualized as beginning with the
evaporation of water from the oceans. The resulting vapor is transported
by moving air masses. Under the proper conditions, the vapor is
condensed to form clouds, which in turn may result in precipitation.
The precipitation which falls upon land is dispersed in several ways.
The greater part is temporarily retained in the soil near where it falls
and is ultimately returned to the atmosphere by evaporation and transpiration by plants. A portion of the water finds its way over and througb
the surface soil to stream channels, while other water penetrates farther
into the ground to become part of the earth's groundwater supply.
Under the influence of gravity, both surface streamflow and groundwater
move toward lower elevations and may eventually discharge into the
ocean. However, substantial quantities of surface and underground
water are returned to the atmosphere by evaporation and transpiration
before reaching the oceans.
This description of the hydrologic ycle and the schematic diagram of
Fig. 1-1 are enormously oversimplified. For example, some water which
enters surface streams may percolate to the groundwater, while in other
INTRODUCTION
cases groundwater is a source of surface streamflow.

ome precipitation
may remain on tIle ground as SllOW for many months before melting
releases the water to the streams or groundwater. The hydrologic cycle
is a convenient means for a rough delineation of the scope of hydrology
as that portion between pre 'ipitation on the land and the return of this
water to the atmo phere or the cean. The hydrologic cycle serves also
to eropha ize the four ba ic phll,'o of int r st to the hydrologi t: precipitation, evaporation and tran piration, surface treamflow, and groundwater. These topics are thc subje t of much more detailed di cus ion
later in this tcxt.
FIG. 1 1. The hydrologic cycle.
(Courtesy of U.S. Geological Survey.)
If the discussion of the hydrologic cycle gives any impre sion of a

continuous me hanism throu rh wru h water moves eadily at a constant
rate, this impr ssion should be di pelled at once. The movement of
water through the variou phase of the cy Ie is most errati , both in
time and ar 11. On oc asion, natur seems to work overtime to provide
torrential rains which tax surfac - hannel capaciti s to the utmost. At
other times it eems that the rna runery of the cycle has stopped completely and, with it, precipitation and str amflow. In adjacent areas
the variation in the y Ie roay be quite different. It is pI' i ely these
extremes of flood and drought that are often of most interest to the
engineering hydrologi t, for many of our hydraulic engineering projects
are d ign d to prote t against th ill efT t of extr meso The reasons
for the climatic extremes are found in the science of meteorology and
HISTORY
should be understood, in broad d tilll at least, by the hydrologist. This

aspect of hydrology is discus' d in the following chapter.
The hydrologist i interested in more than obtaining a qualitative
understanding of the hydrologic cycl and m u uring the quantiti of
water in transit in this cycle. He mu t be able to deal quantitatively
with the interrelations between the various factors so that he can accurately prediot the influenoe of man-made works on these relationships.
He must also concern himself with the frequency with whioh the various
extremes of the cycle may occur, for this is the basis of economic analysis
which is, or should be, the final d terminant for all hydraulic projects.
The final chapters of this text deal with the e quantitative probl ms.
1-2. History. The first hydraulic project has Leen 10 t in the mists
of prehistury. Perhaps some prehistoric man found that a pile of rooks
across a stream would raise the water level sufficiently to overflow the
land which wus the source of his wild food plants and thus water them
during a drought. Whatever the early history of hydraulics, abundant
evidence exists to show that the builders understood little hydrology.
Abandoned irrigation projects the world over, including Indian works in
the southwest United States dating from about A.D. 1100, nrc believed
to be evidence of developments inadequate to sustain a permanent
civilization.
Early Greek and Roman writings indicate that these people CQt.lld
accept the oceans as the ultimate source of all water but they could not
visualize precipitation equaling or exceeding strea.mflow. Typica.l of the
ideas of the time was a view that sea water moved underground to the
base of the mountains. There a natural still desalted the water and the
vapor rose through conduits to the mountain tops where it condensed
and escaped at the source springs of the streams. Leonardo da Vinci
(ca. A.D. 1500) seems to have been one of the first to recognize the hydrologic cycle as we accept it today, but Perreault of France offered the
first recorded proof about A.D. 1650. Using crude instruments, he
measured the flow of the Seine River and found it to be only one-sixth
of the precipitation. About A.D . 1700 the English astronomer Halley
confirmed thll,t oceanic evaporation was an adequate source of moisture
for precipitation.
Precipitation was measured in India as early as the fourth century
B.C., but satisfactory methods for measuring streamflow were a much
later development. Frontinus, water commissioner of Rome in A.D. 97,
based estimates of flow on cross-sectional area alone without regard to
velocity. In the United tates, organized measurement of precipitation
started under the Surgeon General of the Army in 1819, was transferred
to the Signal Corps in 1870, and finally to a n wly organized U.S. Weather
Bureau in 1891. Scattered streamflow measurements were made on the
INTRODUCTION
Mississippi River as early as 1848, but a systematic program was not

started until 1888 when the U.S. Geological urvey undertook this work.
It is not surprising, therefore, that little quantitative work in hydrology
was done before the early years of the twentieth century, when men such
as Horton, Mead, and herman b gan to explore the field. The great
expansion of activity in flood control, irrigation, soil conservation, and
related fields which began about 1930 gave the first real impetus to
organized research in hydrology, as the need for more preci e design
data became evident. Most of our present-day concepts of hydrology
date since 1930. Hydrology is, therefore, a young science with many
important problems only JIllQerfectly understood and much research
still ahead.
1-3. Hydrology in engineering. Hydrology is used in engineering
laainly in connection with the design and operation of hydraulic structures. What flood flows may be expected at a spillway or highway culvert
or in a city drainage system? What reservoir capacity is required to
assure adequate water for irrigation or municipal wn.ter supply during
droughts? What effect will reservoirs, levees, and other control works
exert on flood flows in a stream? These are typical of the questions
which the hydrologist is exp cted to answer.
Large organizations such as Federal and state water agencies can
maintain staffs of hydrologic specialists to analyze their problems, but
smaller offices often have insufficient hydrologic work for full-time
specialists. Hence, many civil engineers are called upon for occasional
hydrologic studies. It is probable that these civil engineers deal with a
larger number of project (without regard to size) than do the specialists,
although in respect to annual dollar volume the situation may be reversed.
In any event, it seems that knowledge of the fundamentals of hydrology
is an essential part of the civil engineer's training.
1-4. The subj ect mat ter of h ydrology. Hydrology deals with
many topics. Ihe subject matter as presented in this book can be
broadly classified into two phases: data collection and methods of
analysis and applict1tion. Chapters 2 to 6 deal with the basic data of
hydrology. Adequate basic data are essential to any science, and
hydrology is no exception. In fact, the complex features of the natural
proces es involved in hydrologic phenomena make it difficult to treat
many hydrologic pro e ses by rigorous deductive reasoning. One cannot always start with a basic physical law and from this determine the
hydrologic result to be expected. Rather, it is necessary to start with
a mass of observed facts, analyze these facts statistically, and from this
analysis establish the systematic pattern that governs these vents.
Thus, without adequate historical data for the particult1r problem area,
the hydrologist is in t1 difficult position. The collection of hydrologic
THE 3U8JECT MATTER OF HYDROLOGY
data has been the life work of many hydrologist and i a primary function
of the U .. G ological urvey and the U.. Weather Bureau. It is
importlmt, therefore, that the student learn how these data. are collected
and published, th limitations on their accuracy, and the proper methods
of interpretation and adjustment.
Generally, each hydrologic problem is unique in that it deals with a
distinct set of physical conditions within a specific river basin. Hence,
the quantitative conclusions of one analysis are often not directly transferable to another problem. However, the general solution for most
problems can be developed from the application of a few relatively
standard proc dures.
hapters 6 to 12 describe these procedures and
explain how they are utilized to solve specific phases of a hydrologic
problem. Chapter 13 summarizes the preceding material by describing
how the various steps are combined in the solu tion of typical engineering
problems.
BIBLIOGRAPHY
Hydrology Handbook, ASCE Manual 28, 1949.

Johnstone, Don, and W. P. Cross: "Elements of Applied Hydrology," Ronald, New
York, 1949.
Linsley, R. K., M. A. Kohler, and J. L. H. Paulhus: "Applied Hydrology," McGrawHill, New York, 1949.
Mead, D. W.: "Hydrology," 1st ed., McGraw-Hill, New York, 1919 (revised by H.
W. Mead, 2d ed., 19qO).
Meinzer, O. E. (ed.): "Hydrology," Vol. IX, Physics of the Earth Series, McGrawHill, New York, 1942 (reprinted Dover Publicatioll$, New York, 1949).
Meyer, A. F.: "Elements of Hydrology," 2d ed ., Wiley, New York, 1928.
Wisler. C. 0 ., and E. F. Bratcr : " Hydrology, " Wiley, N w York, 1949.
PROBI ,EMS
1-1. List the agencies in your state which have responsibilities of a hydrologic
nature. What is the special problem of each agenc,Y?
1-2. Repeat Prob. 1-1 for Federal agencies.
1-3. List the major hydraulic projects in your arcll. What spacific hydrologic
problems did each project involve?
2
WEATHER AND HYDROLOGY
The hydrologic characteristics of a region are det rmined largely by

its climate and its g ological structure. Among the climatic factors that
establish the hydrologic feature of a region are the amount and distribution of precipitation; the occurrence of snow and ice; and the effects
of wind, temp rature, and humidity on evaporation and snowmelt.
Consequently, the design and operation of hydraulic projects involve
meteorological considerations. Hydrologic problems in which meteorology plays an important role include determination of probable maximum precipitation and optimum snowmelt condition for spillway design,
forecasts of pr cipitation and nowmelt for re ervoir operation, and determination of probable maximum wind over water surfaces for evaluating
resulting waves in connection with the design of dam and levees. Obviously, the hydrologist should have some under tanding of the meteorological processes determining a regional climate. The general features of
climatology are discussed in till chapter. Because of it special importance in hydrology, precipitation is treated separately and in more detail
in Chap. 3.
THE GENERAL
IR ULATJON
If th earth were a nonrotating phere,

a purely thermal circulation (Fig.
2-1) would result. The equator receives more solar radiation than the
higher latitudes. Equatorial air,
being warmer, is lighter and tends
to rise. As it rise , it is replaced
by cooler air from higher latitudes.
FIG. 21. Simple thermal circulation on
The only way the air from the
non rotating earth (Northern Hemisphere).
higher latitudes can be replaced is
from above-by the poleward flow of air rising from the equator. The
true circulation differ from that of Fig. 2-1 becau e of the earth's rotation and the effects of land and sea distribution and land forms.
2-1. Thermal cir ulation.
THE GENERAL CIRCULATION
2-2. Factor m odifying the thermal circulation. The earth

rotates fTom west to ea.st, and a point at the equator move at about
1500 fps while one at 60 lat. moves at one-half this speed. From th
principle of conservation of angular momentum, it follows that a parcel
of air at rest relativ to the earth's surface at the equator would attain
a theoretical eastward velocity of 2250 ips (relative to the earth' surface)
if moved northward to 60 lat. Conversely, if a parcel of air at the
North Pole were moved southward to 60 lat., it would reach a theoreti 'al
westward velocity of 750 fps . flowever, wind p :leds of this magnitud
are never observed in nature because of friction . The force that would
OQ L---------------------------~
FIG. 2-2. General circulation of Northern Hemisphere.
be required to produce these changes in velocity is known as the Coriolis

This apparent force always acts to the right in the N ortbern
Hemisphere and to the left in the outhem Hemispher .
The observed pattern of the general circulation in the Northern
Hemisphere is shown in Fig. 2-2. Unfortunately, the physical reasons
for this circulation are only partly known. The ri sing equatorial air
acquires an eastward component as it moves northward. At about 30
lat., it tends to subside because of cooling. Th subsiding air splits
into two currents, one moving southward as the northeast trade winds
and the second continuing northward and eastward.
In the polar cell, loss of heat in the lower layers results in subsidence,
the subsiding air spreading southward and westward. As it moves
southward the air is warmed, and at about 60 lat. it rises and returns
poleward as a southwesterly current aloft.
In the middle cell the southwe terly current in the lower layers me ts
the southern edge of the polar cell and is forc d upward over the colder
westward-moving air. This circulation would result in an accumulation of air in the polar cell were it not for outbreaks of excess polar air
southward.
The idealized circulation of Fig. 2-2 implies belts of low pressure
{surface) at the equator and at about 60 lat. where warmer air is rising.
force.
Similarly, high pressure would be expected at about 300 lat. and at the
poles. This pressure pattern (Fig. 2-3) is greatly distorted by the
effects of water and land masses. These effects are the results of differences in the specific heats, reflectivity, and mixing properties of water
and land and of the existence of barriers to air flow. Heat gains and
'20
FIG. 2-3. M ean-pressure pattern of the Northern Hemisphere.
lossos are distributed through rclativ ly great depths in large bodies of

water by mixing, while land is aft' cted only near the surface. Consequently, land-surface temperatures are far less equabl than those of the
surface f large bodies f water. This condition is further emphasized
by the lower specific heat of the soil and its higher albedo, especially
in winter when snow cover reflects most of the incident radiation back
to space. In winter there is a tendency for the accumulation of cold
dense air over land masses and warm air over the oceans. In summer,
the situation is reversed.
TEMPERATURE
2- 3. Migratory y t m. The emipermanent featur of the general, or mean, circulation (Fig. 2-3) are statisti 'al and at any time may
be distorted or displaced by trn.n itory, or migratory, systems. Both
semipermanent and tran itory feature ar cla ified as cyclones or
anticyclones. A cyclone is a mol' or Ie s 'ircular area of low atmospheric
pressure in which the winds blow counterclockwise in the North rn
Hemisphere. Tropical cyclones form at low latitudes and may develop
into violent hurricanes or typhoons with winds exceeding 75 mph over
areas as large as 200 mi in diameter. Extratropical cycloncs form along
fronts, the boundaries between warm and cold air mass s.
uch cyclones
are usually larger than tropical cyclone and may produce precipitation
over thousands of square miles. An anticyclone is an area of relatively
high pressure in which the winds tend to blow spirally outward in a
clockwise direction in the North rn H emisph reo Details on the g neral
circulation and on the structure of 'yclones an d an ticyclones can be
found in meteorological textbooks.
TF.MI'ERATU HE
2- 4. Measurement of temperature. In order to measme ail' temperature properly, the thermometers must be placed where air circulation
is relatively unobstructed and yet they must be proLected from the
direct rays of the sun and from precipitation. In the United tates
thermometers are placed in white, louvered, wooden boxes, called instru,ment shelters (Fig. 2-4), through which the air can move readily. The
shelter location must be typical of the area for which the measured
temperatures are to be representative. Because of marked vertical
temperature gradients ju t above the soil surfa 'e, the shelters should be
about the same height above the ground for the recorded temperatures
to be comparable. In the United States shelters are set about 4Yz ft
above the ground.
Th re are about 6000 stations in the United States for which th
Weather Bill'eau compiles temperature records. Except for a few hundred
stations equipped or staffed to obtain continuous or hourly temperatures,
most make a daily observation consisting of the current, maximum,
and minimum temperatures. The minimum thermometer, of the alcobolin-glass type, has an index which remains at the lowest temperature
occurring since its last setting. The maximum thermometer has a constriction near the bulb which prevents the mercury from returning to the
bulb as the temperature falls and thus regi sters the highest temperature
since its last setting. The thermograph, with either a bimetallic strip
or a metal tube filled with 0.1 ohol or mercury for its thermometric element, makes an autographic record on a ruled chart wrapped o.round
a clock-driven cylinder. Electrical-resistance thermometers, thermo-
10
couples, gas-bulb thermometers, and other types of instruments are used

for special purposes.
2-5. Terminology. A knowledge of terminology and methods of
computation is requir d in order to avoid misuse of published temperature
fiG . 2-4. Instrument shelter with maximum and minimum thermometers and psychrometer.
(U.S. Weather Bureau.)
data. The terms average, mean, and normal are all arithmetic means.
The first two are used interchangeably, but the normal, generally used
as a standard of compari on, is the average value for a particular date,
month, season, or year over 0. specific 30-yr period (1921 to 1950 as of
1958). Plans call for recomputing the 30-yr normals every decade.
dropping ofr" the first 10 yr and adding the most recent 10 yr.
The mean daily temperature is the av rage of the daUy maximum and
minimum temperatures. In the United tates, this yields a value usually
TEMPERATURE
11
less than a degree above the true daily average.

nee-daily temp ratur
observation are u ually made about 7A.M. r 5 P.M. Tempero.ture a
publi hed as of tbe da,te of the r ading even though the maximum or
minimum may hav occurred on the PI' ding day. Mean temperatures
computed from evening readings tend to b lightly higher tho.n thos
from midnight readings. Morning readings yield mean temperatur s
with a negative bias, but the difference is less than that for avelling
readings. 1
The normal daily temperature is the average daily mean temperature
for a given date computed for a specific 30-yr period. The daily range
in temperature is the difference between the higbest and lowest temperatures recorded on a particular day. Th mean monthly temp rature is the
average of the mean monthly maximum and minimum temperatures.
The mean annual temperature is the avern.ge of the monthly means for the
year.
The degree day is a departure of one degree for one day in the mean
daily temperature from a specified base temperature. For snowmelt
computations, the number of degree days for a day is qual to the mean
daily temperature minus the base temperature, all negative differen 'es
being taken as zero. The number of degree days in a month or other
time interval is the total of the daily values. Published degree-day
values are for heating purposes and are based on departures below 65F.
2-6. Lapse rates. The lapse rate, or verti 'al temperature gradient,
is the rate of change of temperatme with height in the free atmosphere.
The mean lapse rate is a decrease of about 3.6 FO per 1000 ft increase in
height. The greatest variations in lapse rate are found in the layer of
air just above the land surface. The earth radiates heat energy to space
at 11 relatively constant rate which is a function of its absolute temperature. Incoming ra.diation at night is less than the outgoing, and the
temperature of the earth's surface and of the air immediately above it
decreases. This sUl'face cooling sometimes leads to an increase of
temperature with altitude, 01' temperature inversion, in the surfac layer.
This condition usually occms on still, clear nights because there is little
turbulent mixing of air and because outgoing radiation is unhampered by
clouds. Temperature inversions are also observed at higher levels when
warm air currents overrun colder air.
In the daytime there is a tendency for steep lapse rates because of the
relatively high temperatures of the air near the ground. This daytime
heating usually destroys a surlace radiation inversion by early forenoon.
As the heating continues, the lapse rate in the lower layers of the air
steepens until it may reach the dry-adiabatic lapse rate (5.4 FO per 1000
1 W. F, Rumbaugh, The Effect of Time of Observation on the Mean Temperature
Monthly Weather Rev., Vol. 62, pp. 375-376, Octob~r, 1934.
12
WEATHER AND HYDROLOG(
it) which is the rate of temperature change of unsaturated air resulting

from expansion or compre sion as the air rises (lowering pressure) or
descends (in l'easing pres ure) without heat being add d or removed.
Air having a dry-a.diabatic lap e rate mixes readily, whereas a temperature inversion indicates a stable ondition in whi'h warm lighter air
overlies cold den er air. Und r optimum sW'face heating conditions the
air near the ground may be heated Buffici ntly so that the lapse rate in
the lowest 1l1yers becomes super-adiabatic, i.e., exceeding 5.4 FO per 1000 ft.
This is an unstable condition since any parcel of airlifted dry-adiabatically remains warmer and lighter than the surrounding air and thus has a
tendency to continue rising.
If a parcel of saturated air is lifted adiabatically, its temperatW'e will
decrease and its wat r vapor will condense, releasing latent heat of
vaporization. This heat reduces the cooling rate of the ascending air.
Hence, the saturated-adiabatic lapse rate is less than the dry-adiabatic,
being about 3.0 FO per 1000 ft in the lower layers. At very low temperatures or at high altitudes there is little difference between the two lapse
rates because of the very small amounts of water vapor available.
If the moisture in the rising air is precipitated as it is condensed, the
temp rature of the air will decrease at the pseudo-adiabatic lapse rate
which differs very little from the saturated-adiabatic. Actually, the
process is not strictly adiabati as heat i carried away by the falling
precipitation. A layer of saturated air having a 'aturated- or pseudoadiabatic lapse rate is said to be in neutral equilibrium. If its lapse rate
is less than the satumted- or pseudo-adiabatic, the ail' is stable; if greater,
unstable.
2- 7. Geograph ic di tribu tio n of temperatur. In general, temperature tends to be highest at low latitudes and to decrease poleward.
How ver, this trend is greatly distorted by the influence of land and
water mas es, topography, and vegetation. In the interior of large
islands and ontinents, temperatures are higher in summer and lower in
winter than on oasts at C01'1' sponding latitudes. Temperatures at
high elevations are lower than at low level, and southern slopes have
warmer temperatures than northern slopes. The average rate of decrease
of surface air temp ratw'e with height is usually betw en 3 and 5 FO pel'
1000 ft. Forested:11' as hav higher daily minimum and lower daily
maximum temp ratures than do barren ar as. The mean temperature
in a forested area may be 2 to 4 FO lower than that in comparable open
country, the difference being gr ater in the summer.
The heat from a large city, which may roughly equal one-third of the
tlolar radiation reaching it, produces local distortions l in the temperature
1 F. S. Duckworth and J.
. Snndb rg, The Effect of Cii iea upon Hcrizonta.l nnd
Vertical Temperature Gre.dj nta, Bull. Am. MeteoTol. Soc., Vol. 35, pp. 19 207, May,
1954.
HUMIDITY
13
pattern so that temperatures recorded in cities may not repr ent the
surrounding region. The mean annual temperature of cities av rages
about 2 FO higher than that of the surrounding region, mo t of the difference resulting from higher daily minima in the cities. Any comparison of city and country temperatures must allow for differen es in
exposure of thermometers. In cities the instrument shelt rs are oft n
located on roofs. On still, clear nights, when radiational co ling is
particularly effective, the temperature on the ground may be as much as
15 FO lower than that at an elevation of 100 ft. A slight. gmdient in the
opposite direction is ob el'ved on windy or cloudy nights. Daytime
maxima tend to be lower at rooftop level than at the ground. In general,
the average t.emperature from roof expo ures is slightly lower than that
on the ground.
2-8. Time variati.on of temperature. In continental r gion. the
warmest and coldest points of th annual temperature cycle lag behind
the solstices by about one month. In the United tates, January is
usually the coldest month and July the warmest. At oceanic stations
the lag is nearer two months, and the temperature difference between
warme t and coldest months is much less.
The daily variation of temperature lags slightly behind the daily
vu,riation of solar radiation. The temperature begins to rise shortly
after sunrise, reaches a peak 1 to 3 hI' (about 72 hr at oceanic stations)
after the sun has reached its highest altitude, and falls through th night
to a minimum about sunrise. Th daily range of temperature is affected
by the state of the sky. On cloudy days the maximum temp ratur is
lower because of reduceo insolation, and the minimum is high r beeuuse
of reduced outgoing radiation. The daily range is also smaller over
oceans.
HUMIDITY
2-9. Properties of water vapor. The.process by which liquid water

is converted into vapor is called vaporization or evaporation. Molecules
of water having sufficient kinetic energy to overcome the attractive
forces tending to hold them within the body of liquid water are projected
through the water surface.
ince kinetic energy increases and surface
tension decreases as temperature rises, the rate of evaporation increases
with temperature. Most atmospheric vapor is the product of evaporation
from water surfaces. Molecules may leave a snow or ice surface in the
same manner as they leave n. liquid. The process whereby n. solid is
transformed directly to the vn.por state, and vice versa, is called sublimation.
In any mixture of gases, each gas exerts a partial pressure independent
of the other gases. The partial pres ure exerted by water vapor is
called vapor pressure. If all the water vapor in a closed container of
moist air with an initial total pressure p were removed, the final pressure
14
p' of the dry air alone would be less than p. The vapor pressure e would
be the difference between the pressure of the moist air and that of the dry
aIr,
or p - p.I
Practically speaking, the maximum amount of water vapor that can.
exist in any given space is a function of temperature and is independent
of the coexistence of other ga es. When the maximum amount of water
vapor for a given temperature is contained in a given space, the space
is said to be 8aturated. The more common expression "the air is saturated" is not strictly correct. The pressure exerted by the vapor in a
saturated spnee is called the 8aturation vapor pre8sure, which, for all
practical purposes, is the maximum vapor pressure po'sible at a given
tempemture (Appendix B).
The process by which vapor chf),nges to th liquid or solid state is
called condensation. In a spac in contact with a water surface, condensation and vaporization always go on sin ultaneously. If the space
is not saturated, the rate of vaporization will exce d the rate of condensation, resulting in a net evaporation. l If the space is saturated, the rates
of vaporization and condensation balance, provided that the water and
air temperatw'es are the same.
Since the saturation vapor pre sure over ice is less thnn that over
water at the same temperature, the introduction of ice into a space
saturated with resp ct to liquid water at the sam or higher temperature
will result in condensation of th vapor on the ice. This is an important
factor in the production of heavy rain.
Vaporization removes heat from the li qui d being vaporized, while
condensation adds heat. The latent heat of vaporization is the amount
of heat absorb d by a unit mass of a substance, without chf:t.llge in
temperature, while passing from the liquid to the vapor state. The
change from vapor to the liquid state releases an quivalent amount of
heat known as the latent heat of conden8ation.
The heat of vaporization of water H. (cal/g) varies with temperature
but may be determined accurately up to 40C by
H.
= 597.3 -
0.56T
(2-1)
where T is the temperature in degrees centigrade.

The latent heat of f1tBion for water is the amount of heat required to
convert one gram of ice to liquid water at the same temperature. When
1 g of liquid water at 0 freezes into ice at the same temperature, the
latent heat of fusion (79.7 cal/g) i liberated.
The latent heat of sublimation for water is the amount of heat required
to convert one gram of ice into vapor at the same temperature without
passing through the intermediate liquid state. It is equal to the sum of
1
In hydrology, net evaporation is termed simply "evaporation."
HUMIDITY
15
the latent heat of vaporization and the latent heat of fu ion. At OC,
it is about 677 caljg. Direct condensation of vapor into ice at the same
temperature liberates an equivalent amount of heat.
The specific gravity of water vapor is 0.622 that of dry air at th same
temperature and plessure. The density of water vapor p. in grams per
cubic centimeter is
P.
O 622 Ra'l'
=.
1"2-2)
\
where Tithe ab olute temperature (CO) and R u, the gas constant,

equals 2.87 X 10 3 when the vapor pressure e is in millibars.l
The density of dry air Pd in grams per cubic centimeter is
Pd
Pd
= Ro'l'
(2-3
where Ptl is the pressure in millibars.

The density of moist air is equal to the mass of water vapor plus the
mass of dry air in a unit volume of the mixture. If PrJ is the total pressure of the moist air, PrJ - e will be th partial pressure of the dry air
alone. Adding Eqs. (2-2) and (2-3) and substituting PrJ - e for Pd,
Pa =
~a'1' (1 - 0.378 ; )
(2-4)
This equation shows that moist air is lighter than dry air.
2-10. Terminology. There are many expressions U ' d for indicating
the moisture content of the atmosphere. Each serves special purposes,
and only th se expressions common to hydrologic uses are discus ed
here. Vapor pressure e in millibars can be computed from the empirical
psychrometric equation
e = e, - O.000367pu('l' - Tw)
(1+ T~5~132)
(2-5 )
where T and T ware the dry- and wet-bulb temperatures (OF), respectively,
and e, is the saturation vapor pressure in millibars conesponding to 1'....
The relative humidity f is the percentage ratio of the actual to the
saturation vapor pressure and is therefore a ratio of the amount of
moisture in a given space to the amount the space could contain if
saturo.ted.
f = 100!!.
e,
(2-6)
I The millibar is the standard unit of pressure in meteorology.

It is quivalent to a
forco of 1000 dynes/sq em, 0.0143 psi, or 0.0295 in. of mercury. Mean-sea-level aiT
pressure is 1013 millibarB.
16
T he dewpoint Td is the temperature at wruch the air becomes saturated

when cooled under constant pressure and with constant water-vapor
content. It is thus the temperature having a saturation vapor pressure
e. equal to the existing vapor pressure e.
The specific humidity q", usually expres ed in grams per kilogram, is
the mass of water vapor per unit mass of moist air.
e
q" = 622 pa - .
0 37 e ::::: 622 .!.
pa
(2-7)
where Pa is the total pressure (millibars) of the air.

The total amount of water vapor in a layer of air is often expressed
as the depth of precipitable water W", in inches, even though there is no
natural process capable of precipitating the entire moisture content of
the layer. The amount of precipitable water in any air column of
considerable height is computed l by increments of pressure from
(2- )
where the pr ssure po is in millibars and qh, in grams per kilogram, is th

average of the specific humidities a.t the top an d bottom of ea.ch layer.
Figure 2-5 gives the depth of pI' cipitable water in 0. column of saturated
air with its base at the lOOO-millibar lev I and its top anywhere up to
200 millibar. Tabl s for oomputing 2 pr cipitable water and charts 3
of m an precipitable water in the atmosphere over the United States
are available.
2-11. Measurement of humidity. In general, official measurem nts of humidity in the urface lay rs of the atmosphere are made witb
a psychrometer, whioh consists of two thermometers, one with it bulb
covered by a jacket of clean muslin saturated with water. The tbermomet 1'8 are ventilated by whirling or by use of a fan. Because of the
cooling effect of evaporation, the moistened, or wet-bulb, thermometer
reads lower than th dry, the differ nc in degrees being known as the
wet-b1tlb depression. The air temperature and wet-bulb d pression are
used to obtain various expressions of humidity by reference to special
tables. 4 Condensed psychrometric tables are included in Appendix B.
The hair hygrometer makes use of the fact that length of hair varies
with relative humidity. The changes are transmitted to a pointer indiIS. Bolot, Computation of Depth of Precipitable Water in a Column of Air, Monthly
W eather Rev., Vol. 67, pp. 100- 103, April, ]939.
I Tabl s of Pr cipitabl Water, U.S. W eather Bur. T ech. Paper 14, 1951.
S Mean Pre 'ipitable Water in the United States, U.S. Weather Bur. Tech. J:>aper 10,
1949.
'''Psychrom trie Tables," W.B. 235, U.S. Weather Bureau, 1941.
Hygrometrio and Psychrometrio Tables, Sec. VII in " mithsooian McteorolClgical
Tables," rev., 6th ed., pp. 347- 372, Smithsonian Institution, Washington, D.C., 1951.
17
HUMIDITY
SIIrfOC6 d6wpo/nf
'''4.0 J/.OJ9.14&.4 $1.1 $7./60.' &4.4
200'C -(()-4014 & , 10 It 14 16 "
66.0
71.6
11.4 'F
'1'1.1
10
II
"
lf-
't:
1/~.~9
U-Jd
1I.5e-f-l;8
I-
fO.9r 36
'1(.).34
.14
9.57
.Jp
17
II>
0
.c 400
II
:
~
:::>
e! 500
.'
a.
.' . '
II
600
".,
u
'/S. t
U
700
@ 800
:
'"::;
~
a.
I I ,I I
'H -I-... J...l

I I I
I .{
I
......
II>
II>
I ;/
900
....
I .
./
./
,/
,/
V
,/
,/"
./
;0.6( .
:22
i
{llt \
'"
;'er~'\ -7.et
~kl~t~~:~t
H~
.AS
,.~~t;u.
. Dotted cllrve shows ' I!res~lI:e af which 0 '0,:"_,
ISlatffln~d~Y ' j
I
I
1.0
'2.44
,83_-+.6
/'
/'
;;-
f. o
./
H'c
3. 05
./
7'
0.5
//
/I 'f
J.d/, :.1 / f
./
1/ 'I,
/
11//1/. 'I, '/ 1', /:...-1 ./:
:0 /':: ;...-:: ....-::l......-j
t::;::::'
I {
/J I '.1 /
" .f .,
1.5
2.0
,.'B
2.5
Precipitable water (Inches)
FIG. 2-5. Depths of precipitable water in a column of air of any height aboye 1000 mlllibars as a fundion of dewpoint. assuming saturation and pseudo-adiabatic lapse rate.
Weather Bureau.)
iu.s.
18
lJating the relative humidity on a graduated scale. The hair hygrograph

is a hair hygrometer operating a pen marking a trace on a chart wrapped
around a revolving drum. The hygrothermograph, combining th features
of both the hair hygrograph and the thermograph, records botb relative
humidity and temperature on ODe chart. A dewpoint hygrometer, which
measures dew point directly and is used mostly for laboratory purpo es,
consists of a highly polished metal vessel containing a suitable liquid
which is cooled by any of several methods. The temperature of the
liquid at the time condensation begins to occur on the exterior of the
metal vessel is the dewpoint. The spectro copic hygrometer! measur s
the selective absorption of light in certain bands of the spectrum by water
vapor. With the sun as a light source, it has been used to measure total
atmospberic moisture. 2
Measur ment of humidity is one of the least accurate instrumental
procedur s in meteorology. The standard psychrometer invites many
bs rvational crrors. The two thermometers double the chance of
mi reading. At low temperatures, misreading by a few tenths of a
degr e can lead to absurd results. Also, there is always the chance
that the readings are not made when the wet-bulb i at its lowest tom
perature. In addition, there are errors with a po itive bias re ulting from
insufficient ventilation, dirty or too thick muslin, and impure water.
Any in trument using a hair element is subject to appreciabl error.
The hair expands with increasing temperature, and its response to
changes in humidity is very slow, the lag increasing with decreasing
t mp rature until it becomes almost infinite at about -40F. This
can lead to significant error in upper-air soundings where large ranges
in t mperature and sharp variations of humidity with altitude are
observed.
onscquently) soundiug instruments are now equipped with
electrical hygrometers. 3
2-1 2. Geographic di tribution of h umidity. Atmospheric moisture tend to decrease with increa ing latitude, but relative humidity,
being an inverse function of temperature, tends to increase. Atmopherie moi ture is greatest over oceans and d creases with di tance
inland. It also decreases with elevation and is greater over vegetation
than over barren soil.
2-13. Ti m e vari ation in humidity. Like temperature, atmopheria moisture is at a maximum in summer and at a minimum in winter.
1 L. W. Foskett, N. B. Foster, W. R. Thiekstun, and R. C. Wood, Infrared Absorption Hygromoter, Monthly Weather Rev., Vol. 81, pp. 267- 277, May, 1953.
J N. B. Foster and L. W. Foskett, A
pectrophotoroct r fo r the D termination of
the Water Vapor in a. Vertical Column of the Atmosphere, J . Opt. Soc. Am., Vol. 35,
pp. 601-610, September, 1945.
a V. B. Morris, Jr.) and F. Sobel, Some Experiments on the Speed of Response of
the Electric Hygrometer, Bull. Am. Metcorol. Soc., Vol. 35, pp. 226-229, May, ] 954.
WIND
19
Relative humidity, however, varies in reverse. Th diurnal variation

of atmospheric moi ture is nOl"mally small, exc pt wher land and s a
breezes bring air of diffel"ing characteristics.
rear the ground surface,
condensation of dew at night and reevaporation during the day may
result in a miniIDum moisture content near sunrise o.nd a maximum by
noon. Relative humidity, of course, behave in a manner opposite to
that of temperature, being at a maximum in the carly morning and at a
minimum in the afternoon.
WIND
Wind, which i' air in motion, i a very influential factor in veral

hydrometeorological processes. Moisture and heat are r'adily transferred
to and from air, which tends to adopt the thermal and moisture conditions
of the surfaces with which it is in contact. Thus, stagnant air in conto.ot
with a water surface eventually assumes the vapor PI' sure of the surfaoe
so that no evaporation takes place.
imilarly, stagnant air over a now
or ice surface eventually assumes the temperature and vapor pressure
of the surface so that melting by convection I1nd condensation ea es.
Consequently, wind exerts considerable influence in evaporativ and
snowmelt processes. It is also important in the production of precipitation, since it is only through sustained inflow of moist air into a storm
that precipitation Can be maintained.
2-14. Measurement. Wind has both speed and direction. The
wind direction is the direction from which it is blowing. Direction is usually expressed in terms of 16 compass points (N, NNE, NE, ENE, etc.)
for surface winds, and for winds aloft in degr es from north, measured
clockwise. Wind speed is usually given in miles per hour, meters per econd, or knots (1 m/ sec = 2.2 mph and 1 knot = 1H mph, approximately).
Wind speed is measured by instruments called anemometers, of which
there are several types. The three- or four-cup anemometer with a
vertical axis of rotation is most commonly used for official observations.
It tends to regist l' too high a mean speed in a variable wind because the
cups accelerate faster than they lose speed. Vertical currents (turbulence) tend to rotate the cups and cau e overregistration of horizontal
speeds. Most cup anemometers will not record speeds below 1 or 2
mph because of starting friction. Less common is the propeller anemometer with a horizontal axis of rotation. Pressure-tube anemometers, of
which the Dines is the best known, operate on the Pi tot-tube prin iple.
While wind spe d varies greatly with height above the ground, no
standard anemometer level has been adopted. Differences in wind
speed resulting from diJIerences in anemometer height, which may
range anywhere from thirty to several hundred feet above the ground,
usually exceed the errors from instrumental deficiencies. However,
approximate adjustment can be made for differences in height [Eq. (2-9) ].
20
2-15. Geographic variation of wind. Because of its location in the

general circulation, the United tat 8 has prevailing we terly wind .
However, the winds are generally variable since most of the country is
affected by migratory pressure systems. The upper-air wind tend to
increase with latitude, but this tendency is not noticeable near the surface.
In winter there is a tendency for surface winds to blow from the colder
interior of land masses toward the warmer oceaus ( ec. 2-2). Couver ely,
in summer the winds t nd to blow from the cool r bodies of water toward
the warmer land.
imilarly, diurnal land and sea breezes may result
from temperature contrasts between land and water.
On mountain ridges and summits wind speeds (30 ft or more above the
groUl1d) are higher than in the fre air at corresponding elevations because
of th convergence of the air forced by the orographic barriers. On lee
slopes and in sheltered valleys wind speeds are light. Wind direction is
grco.tly influenced by orientation of orogfl1phic barriers. With a weak
pres ure system, diurnal variation of wind direction may occur in mountain regions, the winds blowing upslope in the daytime and down lope
at night.
Wind speeds are reduced and directions deflected in the lower layers
of the atmosphere becau e of fri tion produced by trees, buildings, and
other obstacles. These effects becom negligible above about 2000 ft,
and this lower layer is referred to as the friction layer. Over land the
urface wind sp ed averages about 40 per cent of that just above the
friction layer, and at sea about 70 per cent. The I' lationship between
wind speed I1t anomometer height and that at some higher level in the
friction 1l1yor may be expressed by the empirical formula
(2-9)
where v is the wind speed at height z above the ground, Vo is the wind speed
at anemometer level Zo, and k is often taken as "7'7.
2-16. Time variation of wind. Wind speeds are highest and most
vl1riable in winter, wh ren. middle I1nd late summer is the calmest period
of t.he year. In winter westerly winds prevail over the United States
up to at leu t 20,000 ft, except near the Gulf of Mexico where there is a
tend n 'Y for outheo.sterly winds up to about 5000 ft. In summer, while
there is still a tend ncy for prevailing we t rly winds, there is generally
more variation of dir ction with altitude. In the plains west of the
Mississippi River there is a tendency for southerly winds up to about
5000 ft, and on the Pacific Coast the lower winds are frequently from the
northwest.
The diurnal variation of wind is significant only neal' the ground and
is most pronouncEld during the summer, Surface wind speed is usually
21
WIND
at a. minimum about sunrise and increas t a maximum in early after

At about 2000 ft above the ground, the maximum occurs at
night and the minimum in the daytime.!
110011.
BlBLIOGRAPIlY
Benton, G. S., R. T. Blackburn, and V. . Snead: The Role of the Atmosphere in the
Hydrologic Cycle, Trans. Am. GeophY8. Union, Vol. 31, pp. 61- 73, :Februa.ry,
1950.
Bernard, M.: The Primary Role of Meteorology in Flood Flow Estimating, Tran8.
AS E, Vol. 109, pp. 311- 382, 1944.
Bernard, M.: The Role of Hydrometeorology in Planning th Water E onomy of the
We t, Trans. Am. Goophys. Union, Vol. 30, pp. 263-271, April, ]949.
ByeTs, H. R.: "General Meteorology," pp. 215- 234, McGraw-Hill, ew York, 1944.
Fletcher, R. D.: Hydromet orology in the United tate, in T. F. Malone (ed.),
"Compf\ndium of Meteorology," pp.1033- 1047, American Meteorological ci ty,
Boston, 1951.
Landsberg, H.: Climatology, Sec. XII, in F. A. Berry, Jr., E. BoUay, and N. R. Beers
(eds.), "Handbook of Meteorology," pp. 937- 973, McGraw-Hili, New York, 1945.
Landsberg, H.: "Physical Climatology," Pennsylvania State University, Univ raity
l)ark, Pa., 1941.
Linsley, It. 1<.: The Hydrologic Cycle and Its Relation to Meteorology, in T. F .
Malone (ed.), "Compendium of Meteorology," pp. 104-8- 1054, American
Meteorological Society, Boston, ]951.
Me lendon, E. W.: The Role of Meteorology in Projects of the Corps of Engin era
in the Missouri River Basin, Bull. Am. Meteorol. Soc., Vol. 31, pp. 238-243,
September, 1950.
Middleton, W. E. K., and A. F. Spilhaus: "Meteorological Instruments," 3d ed. rev.,
pp. 57-117, ]35-196, University of Toronto Press, Toronto, 1953.
Rossby, C. G.: The Scientific Basis of Modern Meteorology, in "Climate aud Man,"
U.S. Dept. Agr. Yearbook, pp. 599 55, 1941.
Thiessen, A. II.: Weather Glossary, U.S. Weather Bur. Publ. 1445, 1045.
Willett, H. C.: "Descriptive Meteorology," Academic Press, 1nc., New York, 1944.
DATA
OURCES
The main sources of data on temperature, humidity, and wind are thc monthly
bulletins entitled Climatological Data published by the U.S. Weather Bureau. Th
Climatic Summ.ary of the United State8 summarizes monthly and annual data from tho
beginning of record to 1950. Monthly Weather Rev'iew contains maps summarizing
the weather of the previous month for the country. A summary of normals and
extremes may be found in" ormal Weather for the United Statcs" (by J. B. Kincer,
U.S. Weather Bureau, 1943) and "Climate and Man" (U.S. Dept. Agr. Yearbook,
1941). Special summaries arc available as follows:
Humidity
Maximum Persisting Dewpoints in Western United tates, U.S. Weather Bur. Tech.
Paper 5, 1948.
Mean Precipitable Waterin the United States, U.S . Weather Bur. Tech. Paper 10,1949.
1 C. Hellman, 'OblJr de D 'W gung der Luft jn den unt rston Schichten der Atmosphii.re, Meteorol. Z ., Vol. 34, pp. 272-285, 1917.
22
WEATHER AND H'fDROLOGY
Wind
" irway Meteorological Atlas for the United States," U.S. Weather Bur au, 194.1.
"Normal Surface Wind Dat.a for the United States," U.. Weather Bureau, 1942.
PROBLEM
2-1. Show why the theoretical enstward velocity (constant angulnr momentum) of
air at rest relative to the earth's surface at th equator would be 22S0 fps if the parcel
were displaced to 60 N. lat.

2-2. How many degree daya above 32F arc th re in a day with a minimum temperature of 2soF and a maximum of 4S0F?
2-3. A pnrcel of moist air at SOF initially at 1000 ft mean aea level ia forced to pass
ov r a mountain ridge at SOOO it mean sea level and then descends to its original
el vation. Assuming that a lift of 2000 ft produces saturation and precipitation,
what is the final temperature of the pare I?
2-4. What is the heat of vaporization, in calories por gram, for wat.er at 20C?
2-5. What is the d naity, in grams per cubic centimeter, of (a) dry air at 3SoC and
a pressure of 1000 millibnrs nnd (b) moist air at the same temperature and pressure
with 70 per cent relative humidity?
2-6. Assuming dry- and wet-bulb temperatures of 75 and 63F, respectively, and
lIsing tll psycllrom tric tables of Appendix B, determine (a) dewpoint temperature,
(b) r lative humidity, (c) saturation vapor pressure, and (d) actual vapor pressure.
2-7. A radio Bounding in saturated air ahows temperatures of lS.I, 13.S, and S.6C
at the 900, SOO, and 700-millibar levels, respectively. Compute the pre ipitable
water in tbe layer between 900 and 700 millibars and compare these results with
those obta.ined from Fig. 2-S. (The temperature of 18.1 at the OOO-millibar level
r duces pseudo-adiabatically to 22.0C at 1000 millibars.)
2-8. A snowmelt formula requires wind speed 50 ft above the ground. What is
the estimated spe d at tbis lev 1 if an anemometer 30 ft above the ground indicates
IS mph?
2-9. A pilot-balloon ob ervation shows wind speed of 60 mph at 1000 ft above the
ground. What is th probable wind speed at 20 ft above the ground?
2-10. Compute the weight (Ih/cu ft) of dry air under standard atmosphere conditions at sea level (Appendix B). Wllat would the wight. b if the temp ratur
were iner asod 20 FO at constant pressure?
2-11. How many calories of heat ar r quired to evaporate 50 gal of water at 100F?
How many pounds of ice at 20F would the same amount of heat melt?
2-12. The wind speed at 10 it above the ground is 20 mph. Plot the probabJe
velocity proillo to a height of 500 ft.
2-13. How many ca10ri s per square loot are required (a) to melt a I-ft layer of
ice of 90 p r cent density at 25F Md (b) to evaporate the resulting melt wator without raising its temperature above 32F?
3
PRECIPITATION
To the fiydrologist, precipitation is the gen ral term for all forms of
moisture emanating from the clouds and falling to the ground. From
the time of its formation in the atmosph l'e until it reaches th gr und,
precipitation is of more interest to the met ol'ologist than to the hydrologist. However, once it reaches the ground, it becomes a basic element
of hydrology.
3-1. Formation and types of pr eipitation. While water vapor
in the atmosphere is a nee' sary factor in the formation of precipitation,
it is by no means the only l' quirem nt. Moisture is always pI' S nt in
the atmosphere, even on cloudless days. For precipitation to occur,
some mechanism is required to cool the air sufficiently to cause condensation and droplet growth. Condensation nuclei are also necessary, but
they are usually present in the atmosphere in adequat,e quantities.
The large-scale cooling needed for significant amounts of pI' cipitation
is achieved by lifting the air. PI' cipitation is often typed according
to the factor responsible for the lifting. Thus, cyclonic precipitation
results from the lifting of air converging into a low-pressure area, or
cyclone. Most general storms in plains region s are of this type. Convective precipitation is caus d by the natural rising of warmer, lighter
air in colder, denser surroundings. The difference in temperature may
result from unequal heating at the surface, unequal cooling at the top
of the air layer, or mechanical lifting when the air is forced to pass over
a denser, colder air mass or over a mountain barrier. Convective
precipitation is spotty, and its intensity may range from light showers
to cloudbur ts. Orographic precipitation results from mechanical lifting
over mountain barriers. In rugged terrain the orographic influence is
so marked that storm pI' cipitation patterns tend to resemble that of
mean annual precipitation. In nature, the effects of these various types
of cooling are often interrelated, and the resulting pI' cipitation cannot
be identified as being of anyone type.
23
PRfCIPITA nON
3-2. Forms of precipitation. In the middle latitudes preci.pitation

occurs in many forms, depending on the existing meteorological conditions. Drizzle consists of waterdrops under 0.02-in. diameter, and
its intensity is usually less than 0.04 in./br. Rain consists of drop
usually greater than 0.02-in. diameter. Drops greater than 0.25-in.
diameter tend to break up as they fall through the air, so that 0.25 in.
may be accepted as a practical upper limit of raindrop size. Glaze is the
ice coating formed when drizzle or rain freezes as it comes in contact
with cold objects at the ground.
leet is frozen raindrops cool d to the
ice stage while falling through air at subfreezing temperatures. Snow
is precipitation in the form of icc crystals resulting from subli mation,
i.e., from water vapor dir ,tly to ice. A snowflake is made up of a
number of ice crystals fused together. Hail is precipitation in the form
of balls or lumps of ice over 0.2-in. diameter formed by alternate freezing
and melting as they are carried up and down in highly turbulent air
currents. Single hailstones weighing over a pound have been observed.
3-3. Artificially induced precipitation. In spite of th great
interest in artifically induced precipitation since Wodd War II, this
controversial subje t is not di scussed h rein because it more properly
belongs in the field of meteorology. J efercnces lue listed in the bibliography for those who would like to 1 arn m re about it.
MEA UREMENT OF PRECU'JTATlON
A variety of instruments and techniques have been developed for

gathering information on various phases of precipitation. Instruments
for measuring amount and intensity of precipitation are the most important. Oth l' instruments include devices for measuring raindrop-size
di tribution and for determining the time of beginning and ending of
precipitation. This text des ribes briefly the gages and observational
technique for making officiul measurements of precipitation amount and
intensity in the United tates. Details on these gages and those in
common use in oth r countries and information on the use of radar for
measuring precipitation can be obtained from the references listed in the
bibli graphy.
All form of precipitation arc measured on the basis of the vertical
depth of water thai would accumulate on a level surface if the precipitation r mained where it fell. In the United tates precipitation is
measured in inches and hundredths. An amount less than 0.005 in. is
recorded as a trace.
3-4. Precipitation gages. Any open receptacle with vertical sides
is a convenient rn,in gage, but becaus of varying wind and plash elTe ts
the measur ments would n t be compf1rable unless the re eptf1cles were
PRECIPITATION GAGES
25
of the same size l and shape and similarly expo ed. The standard gage!
(Fig. 3-1) of the U.S. Weather Bureau has a coil ctor (receiver) of 8-in.
diameter. Rain passes from the coIl ctor into a cylindrical m a uring
tube inside the overflow can. The measuring tub has a eros -s ctiono1
area one-tenth that of the collector so that O.I-in. rainfall will fill the
Overflow
con
FIG. 3-1. Standard 8-in. precipitation gage.
(U .S. Weather Bureau.)
tube to a depth of 1 in. With a mcasul"ing stick ffi::trked in tenths of an

inch, rainfall can be measured to the nearest 0.01 in. The collector and
tube are removed when snow is expected . The snow caught in the outer
container, or overflow can, is melted, poured into the measuring tube,
and measured.
The two main types of recording gages in common use in the United
1 F. A. Huff, Comparison between Standard and Small Orifice Ra.ingages, Tram.
Am. Geophys. Union, Vol. 36, pp. 68 694, August, 1955.
2 I nstructions fo r ClimAtological Observers, U.S. Weather Bur. Cire. B, 10th ed.,
pp. 19-32, rev. October, 1955.
26
PRECIPITA nON
to.tes are the U .. Weather Bureau's tipping-bucket gage (Fig. 3-2) and
the weighing-type gage (Fi y. 3-3). The former is used at some Weather
Bureau first-order stations and is equipped with a remote r corder locat d
inside the office. Th 12-in. collector funnels the rain into a two-compartment tipping bucket.
ne-hundredth inch of rain will fill one compartment and overbalance it so
that it tips, emptying into a resrvoir and moving th s cond compartm nt of the bu ket into place
beneath th funnel. A the bucket
is tipped by nch 0.01 in. of rain,
it actuates an ele trical circuit
cau iug a pen to mark 011 a revolving drum. Unfortunately, this
typ of gage is not suitable for
mea 'uring snow without heating
the collector.
The weighing-typ gage weighs
the rain or snow which fal1 into
n bu ket seL on a platform of a
spring or lever balance. The increasing weight of th bucket nnd
its contents is rec rd d on the
chart. held by a clock-driv 11 drum.
Th record thus shows the accumulation of pr cipitation.
torage gages are used in mountainous regions where much of the
FIG. 3-2. Tipping-bucket rain gage.
precipitation falls as snow and
(u.S. Wea ther Bureau.)
wher fr quent servicing is impracticable. Weighing-type storage
gages operate from 1 to 2 months without servicing, and some nonre 'ording storage gag s are d signed to op rate for an entire season
without att ntlon.
inc storage gages are located in heavy-snowfall
areas, the colle tors tire usually in the form of an inverted frustum
of a cone to prevent wet sn w from clin ring to the inside walls and
clogging th OliG e.
f course, the orifie should b above the maximum snow depth expected. The tandpipe-type gage (Fig. 3-4) u ed
by the U .. Weath r Buren.u is mn.de from 12-in. thin-walled pipe in
5-ft sections, so that any height in multiples of 5 ft is possible.
torage
gag s are u tomarily charged with a calcium chI rid solution (antifreeze) to liquefy the snow and prevent damage to the gage. Interim
"(I.easurements of the gage catch are made by stick or tape, while the
PR.ECIPITATION GAGES
initial charge and final men urement of th

a onal catch are made by
weighing the contents. Lo s due to vaporation of th gage contents
are practically eliminated by n. thin film of oil. I
FIG. 3-3. Weighing -type precipitation gage-12-in. dual traverse.
(Friez Instrument Co.)
Precipitation m asurements are subject to various erro1's,2 most being

individually small but with a gen ral tendency to yield measurements
1
E. L. Hamilton and L. A. Andrews, Cont.rol of Evaporation from Rain Gages by
Oil, Bull. Am. Metcorol. Soc., Vol. 34, pp. 202- 204, May, 1953.
2 R. E. Horton, Measurement of Rainfall and Snow, Monthly Weather Rev., Vol
47, pp. 294--295, May, 1919.
28
PRECIPITATION
that are too low. Except for mistakes in reading the scale of the gage,
observational errors are usually small but cumulative; e.g., light amounts
may be neglected, and extended immersion of the measuring sti'k may
result in water creeping up the stick. Errors in scale reading, although
large, are usually random and compensating. Instrumental errors may
be quite large and are cumulative. The water displaced by the measuring
stick increases the reading about 1
per cent. Dents in the collector
rim may change its receiving area.
It i
stimated that 0.01 in. of
each rain measured with a gage
initially dry is required to moisten
the funnel and inside surfaces.
This loss could easily amount to
1 in./ yr in some areas. Another
loss results from raindrop splash
from the collector, but no estimate
of its magnitude is available.
In rainfall of 5 to 6 in./ hr the
bu ket of a tipping-bucket gage
tips every 6 to 7 sec. About 0.3
sec is required to complete the tip,
during which Borne water is still
pouring into the already filled
bucket. The reoorded rate may
be 5 per cent too low.1 However,
the water, which is all caught in
the gage reservoir, is measured
independently of the recorder
FIG. 3-4. Standpipe-type precipitation
count, and the difference is prostorage gage equipped with Alter shield.
rated
through the period of ex(U.S. Weather Bureau.)
cessive rainfall.
Of all the errors the most serious is the deficiency of measurements
due to Willd. The vertical acceleration of air forced upward over a
gage imparts an upward acceleration to precipitatiull aDOut to enter nnd
results in deficient catch. The deficiency is greater for snow 3J;lQ appar
entlY'" may 'be 'In tlle order 01 GO per cent 01 tne 'true snoWiaTI for winds of
30 mph at orin e level (Fig. 3-5). Equipping gages with windshields
increases the catch by about 20 per cent in open areas. Various type
1 D. A. Parson, Calibration of a Weather Bur au Tipping-buoket Rain Gage,
Monthly Weather Rev., Vol. 69, p. 205, July, 1941.
I W. T. Wilson, Discussion of paper Precipitation at Barrow, Alaska, Greater Than
Recorded, Trans. Am. GeophY8. Union. Vol. 35, pp. 206-207, April, 1954.
PRECIPITATION GAGES
29
of shields) hav b en used , but th Alter shi ld (Fig. 3-4) ha b n adopted

as standard by the U .. Weather Bur au. It open 'on tru tion provide
less opportunity than solid shield for the buillin up of snow on it, and
the flexible design allows wind movement to aid in keeping the shi ld
free from ac -'umulated snow. Artificial windshi Ids cannot overcome
the effect of inherently poor gage exposure. Roof in tallations and
o .-,:2
'{'3"~
\9~
8. 20
.S
' 2\\
{i
"8.,
&
Predominantly rain ---x Predominantly snow - - -
10
i .6',
2"'
-J \
-J .....
40 1----
,0 ,
LEGENO
~"
.7
I 8
-- ,
<:::
- - --7 ---1.
oj
"-
II
1.
. f!
--
I--
._?--.-- r--___
-2
10
20
30
40
50
60
Wind speed In miles per hour
1
2
3
4
5
6
Bornstein, 1884
Stade, 1901
BiUwiller, ] 910
Rostad, 1925
Koschmieder, ]934
J{orhonen, 1926
7
8
9
10
11
Curtis, 1884
Dalgo, 1950
Schoklitsch, 1937
Wilson, unpublished
Black, 1953
FIG. 3-5. Ap parent eff ect of wind speed on the catch of precipitation gages.
W ilson.)
(Afte,
windswept slopes should be avoided . The best site is on lev I ground

with bushes or trees serving as a windbreak, provided that these are not
so close as to reduce the gage catch.
When rain is falling vertically, a gage inclined 1O() from the vertical
will catch 1.5 per cent less than it should. If a gage on level ground is
inclined slightly toward the wind, it will catch more than the true amount.
Some investigators 2 feel that gages should be p rpendicular to land slopes.
However, the area of a basin is its projection on a horizontal plane, and
measurements from tilted gages must be reduced by multiplying by the
cosine of the n.ngle of inclination. Considering the variability of land
I
I
ee Bibliography at end of chapter.

E. L. Hamilton, Rainfall Sampling on Rugged Terrain, U.S. De1Jt. AQr, T"", 1ull.
l OO()" 10M.
. .".........,........ / ..........1
L ..~..'.'\.... .r'''............
.J
~
,.........
'.!"
\ ..
r
CL
CL
..,o
;,
:v
......../ ...?
.......... ~
..~
......_..
/ "
.j
", .f
...
::r
j"\.....:
04\
_'
/'
..
:E
.!!
..............\~........ .../ .
c'
.~
~~
oa
E
=>
...J_
C(
C>
.5
w
<'>0
UI -
=>
on
;"
.5
z~
Q)
a.
Q:Q)
01
ne
vC!>
V-=:;.
30
THE PRECIPITATIONGAGE NETWORK
31
slope, aspect, and wind direction, it is virtually impo sible to install 0.

network of tilted gages for general purpose. Pm tically sp akil1g,
no gage ha been designed which will give reliable measurements on
steep slopes experiencing high wind , and such sites should be avoided.
Ail U.S. Weather Bureau precipitation records are obtained from vertical
gages.
3-5. The precipitation-gage network. The U.S. W ather Bureau
networks con ist of about 3500 recording and 11,000 nonrecording gages
40
I--"
V
(I1i\~ v- V ~
~(~v v- V I--" ~
30
~
e
~
'"
<; 20
c:
U)
'"0'
0
'"u
cf
10
9
8
7
6
~
7
~v
z /
./
./
,./
~V'
Vv
./
V
400~V v V ~
/V 69-0p v
I--"
!.---i--"
,./
./
./
./
/V'
V
,
2:3
4
5 6 7 8 910
Area per gage, 100 square miles
FIG. 37. Standard error of precipitation overages as a function of network density and
area far the Muskingum Basin.
of various types at about 13,000 stations, or one station per 230 sq mi, on
an average. The catchment area of the standard 8-in. gage is about
1/ 80,000,000 sq mi. The degree of extrapolation from gage catch to
computed average depth over a large area is obvious.
The uses for which precipitation data are intended should determine
network density. A relatively sparse network of stations would suffice
for studies of large general storms or for determining annual averages
over large areas of level terrain. A very dense network is required to
determine the rainfall pattern in thunderstorms. The probability that
a storm center will be recorded by a gage varies with network density
(Fig. 3-6). A network should be so planned as to yield a representative
picture of the areal distribution of precipitation. There should be no
concentration of gages in heavy rainfall areas at the expense of dry areas,
or vice versa. Unfortunately, cost of installing and maintaining a network and acce sibility of the gage site to an observer arc always important
considerations.
The error of rainfall averages computed from n tworks of various
32
PRECIPITATION
densities has been investigated. Figure 3-7, based on an analysis l of

rainfall in the Muskingum Basin, Ohio, shows the standard error of
rainfall averages as a function of network density and area. Figure 3-8
is based on a similar study2 on a 220-sq rei area near Wilmington, Ohio,
and shows the average error of average rainfall as a function of network
density and amount of storm precipitation. These figures are examples
0.35
0.30
0.25
~u
.:
.!: 0.20
OJ
.8- 0.15
0.05
/_/
0.10
-IJ
0o
18
.2
!.J.-
---=
~::::::::: ::.-:-;,0
~
18~
o 20
of gOge;j.--
-- -
_!.-- r-
~-- --r
3 ~FJ
r-
~--f..---
10
18
1O
18
18
1O
0.40
N IflbB f
r-
~,
0.60
0.80
1.00
1.20
1.40
1.60
1.80
2.00
Slorm precipitation In inches,55-station mean

FIG. 3-8. Average error of precipitation averages as a function of network density and
true precip itation for a 220-sq mi area near Wilmington, Ohio.
and would not necessarily apply under other topographic or climatic

conditions.
INT~IlPnETATrOl'l
OF PRECIPITATION DATA
In order to avoid erroneous conclusions it is important to give the

proper interpretation to precipitation data, which often cannot be
accepted at face value. For example, a mean nnual pre 'ipitation value
for a station may have little significance if the gage site has been changed
significantly during the period for which the aver!tge is computed. Also,
there are several ways of computing averag precipitation over an area,
each of which may give a different answer.
1 Thunderstorm Rainfall, Hydromcteorol. R ept. 5, pp. 234-259, U.S. Weather
Bureau in cooperation with Corps of Engineers, 1947.
t R. K. Liusl y and M. A. Kohler, Variations in torm Rainfall over mall Areas,
Trans. Am. Geophys. Union, Vol. 32, pp. 245- 250, April, 1951.
33
DOUBLE-MASS ANAL YSIS
3-6. E timating mis jng precipitation data. Many preeipita.tion stations have short breaks in their records because of ab enc s of the
observer or beca.use of instrumental failures. It is ften ne sary to
estimate this missing record. In the proc dur 1 u ed by the U ..
Weather Bureau, pre 'ipitation is stimated from that ob erv d at three
stations as close to and as evenly paced around the tation with tho
missing record as po sibl . If the normal annual precipitation at each
of the index stations is within 10 per cent of that for the station with
the missing record, a imple arithmetic av mge of the precipitation at
the index stations provides the estimated amount.
If the normal annual precipitation at any of the index station diITers
from that at the station in question by more than 10 per cent, the normalratio method is used. In this method, the amounts at the index stt\tions
are weighted by the ratios of the norm al-annual-pre 'ipitation values.
That is, precipitation at Station X, P x , i
(Nx P + No
Nx
- PB + Nx
- P )
Nc
Px = -J -
3 NA
(3-1)
in which N is the normal aDnual precipitation.

3-7. Double-ma s analysi. Changes in gage location, exposure,
instrumentation, or observational procedure may cause a relative change
in the precipitation catch. Frequently these chang s are not dis 'losed
urrent U.S. Weather Bureau pmctice calls
in the published records.
for a new station identification whenever the gage location is changed
by as much as 5 mi and/ or 100 ft in elevation.
Double-mass analysis 2 tests the consistency of the record at a station
by comparing its accumulated annual or seas nal precipitation with the
concurrent accumulated values of mean precipitation for a group of
surrounding stations. In Fig. 3-9, for example, a change in slope about
1923 indicates a change in the precipitation regime at Hermit. A chang
due to meteorological causes would not cause a change in slope as all
base stations would be similarly affected. The station history for Hermit
discloses a change in gage location in 1923. To make the record prior to
I J. L. H. Paulhus and M. A. Kohlor, Interpolation of Missing Pr cipitaLion Records,
Monthly Weather Rev., Vol. 80, pp. 129- 133, August, 1952.
J. E . McDonald, A Note on the Precision of Estimtttion of Missing Pre ipitation
Data, TraWl. Am. GeophY8. Union, Vol. 38, pp. 657- 661, October, 1957.
t J. B. Kincer, Determination of D pendability of Rainfall Records by Comparison
with Nllarby Records, TraWl. Am. GeophY8. Union, Vol. 19, Part 1, pp. 533-538, 1938.
C. F. Merriam, Progress Report on the Analysis of RainfaU Data, 7'ran8. Am.
Geophys. Union, Vol. 19, Part 1, pp. 529-532, 1938.
M. A. Kohlor, Double-mass Analysis for Testing the Consistency of Records and
for Making Required Adjustments, Bull. Am. Mcteorol. Soc., Vol. 30, pp. 188- 189,
May, 1949.
PRECIPITA nON
1923 comparable with that for the more recent location, it should be
adjusted by the ratio of the slopes of the two segments of the double-mass
curve (0.95/1.12). The consistency of the record for each of the base
stations should be tested, and those showing inconsistent records should
be dropped before other stations are tested or adjusted.
Considerable caution should be exercised in applying the double-mass
technique. The plotted points always deviate about a mean line, and
.g
800
0
0 700
(.)
s/ope=yc,~~ /
).0
.,E 600
:I:
.~
c
.2
=a
500
~y/
~
400
~
~)
'13
f)
Q,
"0
.!2
::>
E
::>
<.>
<.>
<l
/'
'%//
300
100
200
SO
1/1
00
100
200
300
400
500
600
700
Accumulated annual precipitation (in.I,18-statian meon
FIG. 3-9. Adjustment of precipitation data for Hermit, Colorado, by double-mass curve.
changes in slope should be accepted only when marked or substantiated

by other eviden e. A procedure has been devised l to evaluate the significance of any slope chang in terms of the probability of getting such a
change purely by chance. However, the procedure itself is subject to
some error, its results are not altogether oonclusive, and its application
is somewhat cumbersome.
3-8. Average precipitation over area. The average depth of
precipitation over a specific area, either on a storm, s asonal, or annual
basis, is r quired in many types of hydrologic problems. The simplest
method of obtaining the averag depth is to average arithmetically the
gag d amounts in the area. This method yields good stimates in flat
1 L. L. Weiss and W. T. Wilson, Evaluation of Significance of Slope Changes in
Double-mass Curves, Trans. Am. Geophys. Union, Vol. 34, pp. 893- 896, December,
1953.
35
AVERAGE PRECIPITATION OVER AREA
0.65
Arithmetic mean :
2. 82
1.46 + 1.92 + 2.69 + 4.50 + 2.98 + 5.00
6
- 3.09 in.
1. 95
1. 75
(0)
Thiessen method :
Observed
preclp.
( in. l
Area
Per cent
lotal
area
(sq mi 1
0.65
1.46
1.92
2 .69
1. 54
2.98
5.00
4.50 .
\
7
120
'1 09
120
20
92
82
76
626
Welohted
precipitation (in.l
(coLI x col.31
0 .01
0.28
0.35
0.51
0.05
0.45
0.65
0.5 4
2.84
19
18
19
3
15
13
12
ToO
Averaoe ' 2.B4 in.
\/
h75
(b)
* Area of correspondino
po.lyoon within baaln boundary
Isohyetal method.:
Isohyel
( in.l
1"
5
4
3
2
1
<1
4"
Area
enclosed
(sq mil
Net
area
(sq mil
13
90
206
402
595
626
13
77
116
196
193
31
Pre cipitation'
AvO
volume
preclp.
( In.l (col.3 x col. 41
5'. 3
4 .6
3.5
2.5
1.5
O.B
69
354
406
490
290
25
.i634
1. 95
Average ' 1634 + 626 ' 2.61 in.
* Within basin boundary
( C)
FIG. 310. Areal averaging of precipitation by (a) a rithmetica l method, (b) Thiessen method ,
and (e) isohyetal method.
36
PRECIPITATION
country if the gages are uniformly distributed and the individual gage
catches do not vary widely from the mean. These limitations can be
partially overcome if topographic influence and areal representativity
arc considered in the selection of gage sites. I
The Thiessen method 2 attempts to allow for nonuniform distribution
of gages by providing a weighting factor for each gage. The stations
are plotted on a map, and connecting lines are drawn (Fig. 3-10b).
Perpendicular bisectors of these connecting lines form polygons around
each station. The sides of each polygon are the boundaries of the effective area assumed for the station. The area of each polygon is determined by planimetry and is expressed as a percentage of the total area.
Weighted average rainfall for the total area is computed by multiplying
the precipitation at each station by its assigned percentage of area and
totaling. The results are usually more accurate than those obtained by
simple arithmetical averaging. 3 The greatest limitation of the Thiessen
method is its inflexibility, a new Thiessen diagram being required every
time there is a change in the gage network. Also, the method makes no
attempt to allow for orographic influences. Actually, the method simply
assumes linear variation of precipitation between stations and assigns
ach segment of area to the nearest station.
The most accurate method of averaging precipitation over an area
is the isohyetal method. Station locations and amounts are plotted on a
suitable map, and contours of equal precipitation 4 (isohyets) are then
drawn (Fig. 3-10c). The average precipitation for an area is computed
by weighting the average precipitation between successive isohyets
(u ualJy taken as the average of the two isohyetal values) by the area
betwe n isohyets, totaling th e products, and dividing by the total
area.
The isohyetal method permits the use and interpretation of all availabl data and is well adapted to display and discussion. In constructing
an i ohyetal map the analyst can make full use of his knowledge of
orographic effects and storm morphology, and in this case the final
map should represent a more realistic precipitation pattern than could
be obtained from the gaged amounts alone. The accuracy of the
isohyetal method is highly dep ndent upon the skill of the analyst. If
1 H. G. Wilm, A. Z. Nelson, and H. C. Storoy, An Analysis of Preoipitation Measurements on Mountain Watersheds, Monthly Weather Rev., Vol. 67, pp. 163-172,
May, 1939.
I A. H. Thiessen, Precipitation for Large Areas, Monthly Weather Rev., Vol. 39,
pp. 10 2- 1084, July, 1911.
a R. E. Horton, Acouraoy of Areal Rainfall Estimates, Monthly Weather Rev., VC'1.
51, pp. 348- 353, July, 1923.
W. G. Reed and J. B. Kincer, The Preparation of Precipitation Charts, Monthlll
Weather Rev., Vol. 45, pp. 233- 235, May, 1917.
37
DEPTH.AREADURATION ANALYSIS
linear interpolation between stations is used, the results will be e entio.lly

the same as those obtained with the Thies en method. Moreover, an
improper analysis may lead to serious error.
3-9. Depth-area-duration analy i. Variou hydrologic problems
require an analysis of time as w 11 as areal distribution of storm pr cipitation. Basically, depth-area-duration analysis of a torm is performed to
determine the maximum amounts of precipitation within various dura-
100,000
80,000
60,000
36.......
6 If 182"
'"
48
.......
"Zones A 8 CO E. F, 3 Isohyef
11?
.......
\. 1" - ,
40,000
'\. ~
24
I~- ~~ j\: - ~ ~ ~
" ,,-
1 1
_ . / ~n!TA/~f,q,~~o~r --
"""'\ """
I~-\
I 2i I A ,)
1
_ _ L _of/t! . 6 ,Iso.!_lye:__ .
~ ~~
~\ _'t I\ -/~ t\..!.4\..__ ~
~
~"
10,000 -i8~ 1-24'\:.. 8,000 I--116~ "4s-::s,
~ Zones A~ 8, 8'rl;ohyel~
i=-=}6':
6,000 ::..=..::. t:~ 1-- jtP4
~8
~
~
~ "'\ 'Zone A, 8"Isohyet
4,000
-- "6-'I?f\-- iI\ - ~'" 1\-- --3~ K- r\ l~n_gAf:iit2~
1\
--
20,000
""
~-II?
.!
::.=,~
18
1?4
'\
1,000
800
600
400
18
II?\
\6 II? /8
-6 \ --- 2;;~~
48
'r
i\)
\~c
"-
\~
20
10
6\
2
TI?
1\
1\
- y"-
-10\
a;
~ -I-- ~ _
- I--- 1:, -
40
'---- ~
48
12\
200
100
80
60
36\
\"
--
\
1\ F
"' _
\ IZoneE. \ 16" /soh;et-..

-'I? 'Zone A,
--- \
.f6\
l''\'zone
A,
// " /sohyef
,., ------
48'
f3"/soh~
O"I _
\
\
~_
~_
& I \~~ -f'g

~=
;),
v.
f-
I---
12
18
24
4
7
10
II
9
5
6
8
Maximum a~eraQe depth at ralnlall in inches
J6
48
12
12
13
14
15
Explanation
"Zones A, B, 8" lsohyet" refers to maximum depth-duration values computed for the area within the 8" isohyet in Zones A and B. Plotted points
represent maximum average depth of rainfall, over the area designated, within
the number of hours indicated by figures beside points. Depths for 10-sq mi
a.rea are ma.ximum station rainfall.
FIG. 311. Moximum depth-oreodurotion curves for the storm of Jon. 18 to 21, 1935,
centered neor 80livor, Tennessee, ond Hernondo, Mississippi. (After U.S. Corps of
Engineers.)
38
PRECIPITATION
tions over areas of various sizes. The method 1 discussed here is somewhn.t arbitrary but has been standardized by the Federal agencies so
that results will be comparable. For a storm with a single major center,
the isohyets are taken as boundaries of individual areas. The average
storm pr cipitation within each isohyet is computed as described in
ec. 3-8. The storm total is distributed through sUC' s ive incr ments
of time (usually 6 hI') in accordance with the distribution recorded at
nearby stations. 2 When this has been done for each isohyet, data are
available showing the time distribution of average rainfall over area
of various sizes. From these data, the maximum rainfall for various
durations (6, 12, 18 hI', etc.) an be selected for each size of area. These
maxima are plotted (Fig. 3-11), and an enveloping depth-area curve is
drawn for each duration.
torms with mUltiple centers are divided into
zones for analysis.
VAlUATIONS IN PRECIPITATION
3-10. Geographic variation. In general, precipitation is heaviest

at the equator and decreases with increasing latitude. However, the
irregularity and orientation of the isohy ts on the mean annual precipitation map of the United tates (Fig. 3-12) indicate that the geographic
distribution of precipitation depend' on more effective factors than distance from the equator. A study 8 of maximum min storms in the United
Stat s indicated heaviest intensities at about 30o N.
The main source of moisture for precipitation is evaporation from the
surface of large bodies of water. Therefore, precipitation tends to be
heavier n ar coastlines. This characteri tic is shown by the i ohyets
of Fig. 3-12, espe ially northward of the Gulf of Mexico. Although
distorted by orography, a similar variation is observed in the Far Western
states.
in' lifting of air masses account for almost all precipitation, amounts
and fr quency are generally greater on the windward side of mountain
barriers. Conversely, since downslope motion of air results in decreased
r lative humidity, the lee sides of baniel's usually experience relatively
light precipitation. However, the continued rise of air immediately
Mauual for Depth-AT a-Duration Analy is of torm Precipitation. U.S. Weather
tudies Tech. Paper I, 19-16.
R. K. Linslcy, M. A. Kohler, and J. L. H. Paulhus, "Appli d Hydrology," pp. 79- 90,
Me raw-Hill, ew York, 1949.
I A. L.
hands and . N. Brancato, Applied M t 'orology: Ma88 Curves of Rainfall,
U.S. Weather Bur. Hydromet orol. Tech. Paper 4, 1947.
R. A. McCormick, Latitudinal Variation of Maximum Observed United States
Rainfall East of th Rocky Mountains, Trans. Am. Geophys. Union, Vol. 30, pp. 215220, April, 1949.
1
Bur. Coop.
39
PRECIPITATION
40
downwind from the ridge and the slanting fall of the precipitation produce
heavy amounts on the lee slopes near the crest.
The variation of precipitation with elevation and other topographic
factors has been investigated 1 with somewhat varying conclusions.
~r--'---r--.---r--.--.---r--'---r--.---r--.--.---,40
Q;
J!! 11
"E
S!
.~
8
~0>
"0
0>
7
6
10
(0)
'5
0-
(b)
(c)
FIG. 3-13. Relation between average October- April precipitation and topographic
parameters for western Colorado. (After Spreen.)
P rhaps the most detailed study of orographic influences on precipitation

that by preen, who correlated mean seasonal precipitation with
elevation, slope, orientation, and exposure for western Colorado (Fig.
3-13). The parameters of the relation are (1) elevation of the precipitaj
1 J. 1. Burns, Small-scale Topographic Effects on Precipitation Distribution in San

Dimas Experimental Forest, Tran8. Am. Geophy/l. Union, Vol. 34:, pp. 761- 768,
Ootober, 1953.
W. C. Spre n, Determination of th Eft' ct of Topography upon Precipitation,
Tram. Am. Geopk1l8. Union, Vol. 28, pp. 285-290, April, 1947.
GEOGRAPHIC VARIATIONS
41
t ion statiOl1 (in 1000 ft) above mean sea level ; (2) rise, differenoe in
elevation (in 1000 ft) between station and highest point wit hin a 5-mi
radius; (3) exposure, t he sum (in degrees) of those seotors of a 20-miradius circle about the station not containing a barrier 1000 ft or more
TABLE 3-1. World's Greatest Observed Point Rainfalls
(After Jennings - I
Duration
Depth, in.
1 mln ... . .. ......... .

8 min .... . . . ....... . ...
i5 min ....... .. .......... .
20 min . .. . ... . .. . . .... .
42 mln ........... . ..... ..
2 hr, 10 min ..............
2 hr, 45 min ............. ,
4 hr, 30 min . . ......... . . .
15 hr ...... . . . .. ... . .. .
18 hr . .... .. .. . . . . . ....
24 hr ................. .
39 hr .. . . ..... . ...... . .
2 day. . . .. . .. . .... . . .
2 days, 15 hr . ........ .
3 day ........... .. .......
-4 day .. .. . ..... . ... .
5 day . ........ . . ...... . ..
6 day .... . ..............
7 days .. . .. . .... .. ... .
8 days ........ . .......
15 days ........ . ..........
31 days ....... . .. . ....... .
2 mo . . .. . . . .............
3 mo . ..............
4 mo . . . . . . . ... . ..
5 mo . . . .. ... . .. .. .. .
6 mo .. . .. . ........ . . . .
1.23
4 . 96
7.80
8 . 10
12.00
19 . 00
22.00
30 . 8 +
34.50
36.40
45 . 99
62 . 39
65.79
79 .1 2
81.54
101 . 84
114.50
122 . 50
131.15
135 . 05
188 . 88
366 . 14
502 . 63
644 .44
737 . 70
803 . 62
884.03
905 . 12
1041 . 78
1605.05
11 mo
1 yr ... ...... .. .... ......

2 yr . ...... . ......... . ..
Station
Date
Unionville, Md.
July 4, 1956
May 25, 1920
May 12, 1916
Plumb Paint, Jamaica
july 7, 1889
Curtea de Arg e., Rumania
June 22, 1947
Halt, Mo.
July 18, 1889
Rockport, W. Va.
May 31, 1935
D'Hanl., Tella s (17 mi NWI
July 18, 1942
Smethport, Po.
July 17- 18, 1942
Smeth part, Po .
Sept. 9, 1921
Thrall, Tex a s
July 14- 15,1911
Ba llUio, Philippine I.,
july 14- 16, 1911
8aguio, Philippine I.,
July 19- 20, 1913
Funkiko, Formosa
July 14- 17, 1911
8agulo, Philippine, I.,
July 18- 20, 1913
Funkiko, Formo.a
June 12- 15, 1876
Cherrapunji, India
Silver Hill Plantation , Jamaica Nov. 5-9, 1909
Silver Hill Plantation, Jamaica Nov. 5- 10, 1909
June 24- 30, 193 I
Cherrapunji, India
June 2-4- July I, 1931
Cherrapunji, India
June 2-4- July 8, 1931
Cherrapunji, India
July, 1861
Cherrapunji, India
June- July, 1861
Cherrapunji, India
May- July, 1861
Cherrapunji , India
April- July, 1861
Cherrapunji, India
April- Aug., 1861
Cherrapunji, India
April- Se pt., 1861
Cherrapunji, India
Jan.-Nov., 1861
Cherrapunji, India
Aug., 1860- July, 1861
Cherrapunjl, India
1860- 1861
Cherrapunji, India
FUslen, Bavaria
A. H. Jennings, World'. Greate. t Ob'erved Paint Rainfall., Monthly Weather Rev., Vol. 78, pp.
" - 5, January, 1950.
above the station elevation; and (4) orientation, the direction t o eight
points of the compass of the greatest exposure defined above. While
elevation alone accounted for only 30 per cent of the variation in precipitation, the fi ve parameters together accounted for 85 per cent. Relations
of this type are very u eful for constructing isohyet al maps in rugged
areas having sparse dat a.
.;
.201
!!
.!!
.E
::>
.g
o
>
.5
g
Co
'u
!!
0..
L------
I
f
I
10"
ou
L
=.
I
!I.
-L: J!;i~olJ
I
___
~ I'
'"
..
~
"
.., /
..
EN -
-_L_..
""
z L
g __
.c
"
II
'
,>;'-"\'-"','./
RECORD RAINFALLS
3-H. Tim.e variation. While portions of a. precipitation reoord

may suggest an increasing or decren ing trend, there i usually a tendency
to return to the mean; abnormally wet periods tend to b balanced by
dry periods. The regularity of these fluctuations has been repeatedly
investigated. More than 100 apparent 'ycles, ranging in period from
1 to 744 yr, have been reported . l The bibliography lists several reports
on attempts to detect these vn.riations from the analysis of tree rings and
geological formations or to associate them with fluctuations in sunspots
TABLE 3-2. Maximum Recorded Rainfalls at Five Major U.S. Cities
(After Shands and Ammerman)
Duration
Station
Minutes
0.75
8/ 12/ 26
51. louis, Mo ....... .. .....
0.60
7/ 9/ 42
New Orleans, la ... . . .. . ...
1.00
2/ 5/ 55
Denver, Colo ......... . ....
0.91
7/ 14/ 12
San Francisco, Calif ....
0.33
1/ 25/ 26
New York, N.Y ......
Hours
15
30
1.63
7/ 10/ 05
1.39
8/ 8/ 23
1.90
4/ 25/ 53
1.54
7/ 14/ 12
0.65
11 / 4/ 18
2.34
8/ 12/ 26
2.56
8/ 8/ 23
3.18
4/ 25/ 53
1.72
7/ 14/ 12
0.83
3/ 4/ 12
2~
2.97
4.09
10/ 9/ 03
3.72
7/ 23/ 33
7.95
4/ 15/ 27
2.91
8/ 23 / 21
1.70
1/ 12/ 14
9.55
10/8/03
8.78
8/ 15/ 46
14.01
4/ 15/ 27
6.53
5/ 21 / 76
4.67
1/ 29/ 81
8/ 26/~7
3.47
7/ 23 / 33
4.71
4/25/53
2.20
8/ 23 / 21
1.07
3/ 4/ 12
and solar radiation. However, with the exception of diurn al l1nd seasonal
variations, no persistent regular cycles of any appreciable magnitude
have been conclusively demonstrated. 2
The sensonal distribution of precipitation varies widely within the
United States. Figure 3-14 shows typical seasonal distribution graphs
for stations in the precipitation regions defined by Ken drew .8
3-12. Record rainfall. Table 3-llists the world's greatest observed
point rainfalls. The predominance of U.S. stations for durations under
24 hr should not be interpreted as a tendency for heavier short-duration
rainfall in the United States. The large number of recording rain gages
1 Sir Napier Shaw, "Manual of Meteorology," 2d ed ., Vol. 2, pp. 320-325, Cambridge, London, 1942.
2 H. Landsberg, Climatology, Be. XII, in F . A. Berry, Jr., E. Bollay, and N. R.
Deers (cds.), "Handbook of M et. orology, " p. 964, McGraw-Hili, New York, 1945.
W. G. Kendrew, "The Climate of t,he Continents," 4th ed., p. 434, Oxford, New
York,1953.
PR.ECIPITATION
enhances the probability of measuring localized rainfalls such as thunderstorms and cloudbursts. If the values of Table 3-1 were plotted on
logarithmic paper, they would define an enveloping curve closely approximating a straight line.
The maximum raino,lls of record 1 for duration up to 24 hr at five
major U. . cities are given in Table 3-2. Table 3-3 lists maximum
TABLE 3-3. Maximum Depth-Area-Duration Data for the United States
(Average precipitation in inches'
Duration, hr
Areo, sq ml
6
10
100
200
SOO
1,000
2,000
5,000
10,000
20,000
SO,OOO
100,000
18
24
36
48
72
3S.0b
30.7b
28.7b
36.Sb
31.9b
29.7b
37 . 6b
32 . 9b
37.6b
32.9b
37.6b
3S.2e
25.61>
22.91>
26 . 61>
24.01>
20.61>
15.01>
12.1e
9.6e
6 . 3e
4.3e
12
29.8b
26.2b
24.70
19.6b
17.91>
15.41>
13.41>
11.21>
8.11>/
5.7/
4.0j
24 .31>
21.41>
18.81>
2.5.h
1.7h
4.2g
IS.71>
19.5I>
11. 11>
14.11>
10 . le
7.ge
S.3e
3.5e
7.9k
6.0k
2.Sih
Storm
Dote
a
b
July 17- 18, 1942

Sept. 8- 10, 1921
Aug. 6-9, 1940
June 27- July I, 1899
March 13-1 S, 1929
July 5- 10, 1916
April IS-18, 1900
May 22-26, 1908
Nav. 19-22, 1934
June 27-Jvly 4, 1936
April 12-16, 1927
e
d
e
f
g
i
k
30 .71>
27.61>
2S.6d
23.1d
18.7d
15.1d
11.6d
7.9.
S.6.
31 . 9c
34 . Sc
30.3e
28.8e
26.3e
33.6e
32.2e
29.Se
20.7d
17.4d
13.8d
8 . ge
6 . 6f
24.4d
21.3d
17.6d
II.Sf
8.9f
Location of center
Smethport, Po.
Thrall, Texas
Miller Island, La.
Hearne, Texal
Elba, Ala.
Bonifay, Flo.
Eutaw, Ala.
Chattanooga, Oklo.
Millry, Ala.
Bebe, Texas
Jefferson Parish, La.
depth-area-duration data for the United States and the storms producing
them. They represent the enveloping v lues for over 400 of the country's
major storms analyzed by the Corps of Engineers in cooperation with the
U.S. Weather Bureau.
1 A. L. Shands and D. Ammerman, Maximum R corded United States Point Rainfall for 5 Minutes to 24 Hours at 207 First Order Stations, U.S. Weather BUT . Tech.
Paper 2, 1947.
SNOW
NOW PACK A D
NOWFALL
3-13. Mea uremen t . Mea 'urement of the depth of ac umulated

snow on the ground is a regular function of all Weather Bureau observer ,
Where the accumulation i not large, the measurements are made with
a yardstick or rain-gage measuring sti k. In region where large a cumulations are the rule, permanent snow stakes, graduated in in 'he , are
normally used.
uch stakes should be installed where they will be
least affected by blowing or drifting now.
Scole
1001
OrlVlng
hondle
creoning
~-.....
Cutter
FIG. 3 15. MI. Rose snow sampl er.
(Leupold and Stevens Instrument Co.)
The hydrologist is usually more interested in the water equivalent

of the snow pack than in its depth. The water equivalent of the snow
pack, i.e., the depth of water that would result from melting, depends
on snow density as well as on depth. Snow densi~y, the ratio between
the volume of melt water from a sample of snow and the initial volume of
the sample, has been observed to vary from 0.004 for freshly fallen snow
at high latitudes to 0.91 for compacted snow in glaciers. An average
density of 0.10 for fr sbly fallen snow is often as umed. In regions of
heavy snow accumulation, dcnsiti s of 0.4 to 0.6 are common by the time
the spring thaws begin.
Measurements of water equivalent are usually made by sampling
with a snow tube of the type shown in Fig. 3-15. The tube is driven
vertically in 0 the snow pack, sections of tubing bing added as required.
PRECIPITATION
The cutting edge on the leading section is designed to penetrate ice

lo.yers when the tube is rotated. When the bottom of the snow pack
is reached, the snow depth is determined from the graduations on the
tube. The tube and its contents are then withdrawn and weighed to
determine the water equivalent. For COllV nience mo t tubes have a
diameter such that 1 oz is equivalent to 1 in. of water.
Because of the highly variable snow pack resulting from drifting and
nonuniform melting, 3 to 10 measurements are made at intervals of 10
to 50 ft along an esto.blish'ed line, or snow course. Courses are selected
at sites free from extensive wind effects and lateral drainage of melt
water. Even then the measurement for each course is considered to be
only an index to areal snow pack. Monthly or semimonthly snow surveys are made in late winter and early spring at over 1000 courses in the
Western mountain region by the U .. Soil Conservation Service and
cooperating agencies. The U.S. Geological Survey and cooperators
make similar surveys at over 500 courses in the Northeastern states.
Twice-weekly observations of water equivalent are made at all firstorder U.S. Weather Bureau stations, but these data are not published
on a curr nt basis.
Snow is usually composed of ice crystals and liquid water; the amount
of liquid water is referred to as the water content of the snow. The
quality of snow, i.e., the percentage by weight which is ice, can be determined by a calorimetric process. 1 Snow is inserted into a thermos bottle
containing hot water. The quality can be computed from initial and
final temperatures and weights (or volumes) of the water and melted
snow. The water content in per cent of the water equivalent is 100
minus the quality. Most evaluations of quality made to date indicate
values of 90 per cent or more, but values as low as 50 per cent have been
obtained at times of rapid melting.
3-14. Variations. The mean annual snowfall for the United State
is shown in the map (Fig. 3-16). This map may be considerably in
error in mountainous regions because of the paucity of measurements at
high elevations. As should be expected, there is a gradual increase of
snowftl,ll with latitude and elevation. In the Sierra Nevada and Cascade
Range annual snowfalls of 400 in. are not uncommon. In general,
maximum annual snowfall occurs at slightly higher elevations than
maximum annual precipitation.
Maps of mean depth of snow on the ground or mean water equivalent
for specific dates are not available. Snow depths would, of course, be
much lower than annual snowfall amounts because of compaction of old
1 M. Bernard and W. T. Wilson, A New Technique ior the Determination of the
Heat Necessary to Melt Snow, 7'ram. Am. Geophys, Union, Vol. 22, Part 1, pp. 178-
181, 1941.
0)
0)
CI)
a::
Q.
PREeIPITATlO N
snow, evaporation, and melting. Snow depth usually builds up rapidly

early in the season and then remains relatively constant as compaction
of old snow compensates for new falls. Maximum depths on the ground
are usually less than one-half the annual snowfall at high elevations and
are still less at lower elevations where intermittent melting occurs.
imilarly, the maximum. water equivalent approaches 10 per cent of the
annual snowfall at high elevations and is much less at lower elevations.
The mean water equivalent at the beginning of the melting season is a
go d index to the mean annual pre ipitation. 1
Because of drifting, consid rable variation in snow depth and water
equivalent may be observed within short distances. This variation
can be intensified by differ nces in melting rates, which are generally
greater on south slop s and in areas without fore t cover. Nevertheless,
there is a consistency in the distribution pattern from year to year as
many important factors affecting melting-slope, aspect, elevation. forest
cover, wind currents, etc.-normally undergo little change.
BIBLIOGRAJ'IIY
PrecipUation Forrn8 and Formation
Byers, H. R.: "General Meteorology," pp. 489- 507, McGraw-Hili, New York, 1944.
Petterssen, S.: "Weather Analysis and Forecasting," pp. 37- 47, MoGraw-HilI, New
York, 1940.
ArtiftCiat Induction oj Precipitation
ufottr, L., F. Hall, F. H. Ludlam, and E. J. Smith: Artificial Control of Clouds and
Hydrom teors, World Meteorol. Organization 7'ech. Note 13, 1955.
Final Report of the Advisory ommittee on Weather Control, Vols. 1 and 2, 1957.
Houghton, H. G.: An Appraisal of Cloud eeding as a Means of Increasing Precipitation, Bull. Am. Meteorol. Soc., Vol. 32, pp. 39- 46, February, 1951.
Measurement oj Precipitation
Hiatt, W. E., and R. W. Schloemer: How We Measure the Variations in Precipitation, from "Water," U.S. Dept. Agr. Yearbook, pp. 129- 133, 1955.
Kadel, B. C.: Measurem nt of PrecipitaLion U.S. Weather B~tr. Cire. E, 4th ed.,
r V., 1936.
Kurtyka, J. C.: "Precipitation Masur menta Study," State Water Survey Division,
Urbana, TIl., 1053.
Marshn.U, J. ., n. . Langill , and W. More Palmer: Mea.suT ment of Rn.infall by
Radar, Am. Afet orol. Soc. J ., Vol. 4, pp. 186- 192, December, 1947.
Middleton, W. E. K., n.nd A. F. pilhaus: "Meteorological Instruments " 3d ed., rev.,
pp. 118-131, University of Toronto Press, Toronto, 1953.
Stout G. E., and J . C. Neill: Utility of Radar in Mea uring Ar 0.1 Rainfall, Bull. Am.
MeteoTol. Soc., Vol. 34, pp. 21-27, January, 1953.
I J. L. H. Pa.ulhus, C. E. Erickson, and J. T. Ri d I, Estimation of Mean Annua.l
Precipitation from Snow-surv y Data, Trans. Am. GeophY8. Union, Vol. 33, pp. 763767, October, 1952.
SOURCES OF DATA
Depth-Area Relations
Fleteher, R. D.: A Relation b tween faxirnuro Observ d Point and Ar 0.1 Rainfall
Values, 7'rans. Am. Geophys. nion, Vol. 31, pp. 34 34, June, 1950.
Huff, F. A., and G. E. Stout: Ar'a-D pth tudies for Thund rstorm Rainfall in
Illinois, Trans. Am. Geophys. Uni:()n, Vol. 33, pp. 495-40 , August, ] 952.
anderson, E. E., and Don Johnstone: Ac 'uracy of DEo'termioalion of nnun.! Pre'ipitation over It Given Area, T!"ans. Am. GeophY8. nion, Vol. 34, pp. 49- 57, 'ebruary,1953.
Wind Shield8
Alter, J. C.: hielded torage Pr cipitation Gages, Monthly Weather Rev., Vol. 65,
pp. 26 265, July, 1937.
Brooks, C. F.: Further }!;xperien e with Shielded Prc 'ipitat,ion Gages on Blue Hill
and Mt. Washington, T,ans. Am. Geoph1l8. Union, Vol. 21, Pnrt 2, pp. 4: 2 85,
1940.
Warnick, C. C.: Experiments with Wind hields for Pr ipit.atioo Gages, 'l'ran8. Am.
Geophys. Union, Vol. 34, pp. 379-:388, June, 1953.
Wilson, W. T.: Analysils of Winter Precipitntion in the Cooperative Snow Inv stigations, Monthly Weather Rev., Vol. 82, pp. 183- 19g, July, 1954.
Precipita#on Cycles
Glock, W. S.: "Tree Growth and Rainfall; A Study of orrelation and M th ds,"
Smithsonian Institution, Washingt.on, D.C., ]950.
Haunvit z, B.: Relation bctw cn SoJaI' Activity and the Low r Atmosphere, Tran8.
Am. Geophys. Un1'on, Vol. 27, pp. 161 - 103, April, 1946.
Schulman, E.: Centuries-long Tr e Indices of Precipitt1tion in the Southw st, Bull.
Am. Meteorol. Soc., Vol. 23, pp. 148- 161,204- 217, May, 1912.
Tannehill 1. R.: "Drought," Princeton Univ rsity Press, Princeton, N.J., ]947.
Tannehill, I. R.: Is Weal her Subject to ycles?, in ""Vater," U.S. Dept. Agr. Yearbook, pp. 84-90, 1955.
sounCES OF DATA
Th main source of precipitation data is Climatological Data. published monthly by
th U .. Weather Bur au and rontaining th daily records of rainfall. Hourly rainfall intensities arc found in H yd'l'oloUir Bulletin (H}40- 194 ), Climatolo(Jical Data
(1948- 1951), and Hourly Precipitat'ion Data (1951). W ekly and monthly
maps of precipitation over the United States a.re found in Weekly Weather and Crop
B ulletin and Monthly Weather Review. The Climatic Summa:-1/ for the United Statca
summarizes month ly and annunJ data from beginning of reeoru to ] 950.
The Water Bulletin of the Internationn.! Boundary and Water Commission contains
data for th Rio Grande Basin, and Precipitation in the Tennc~8ee River BaBin published by the Tenness e Valley Authority summarizes data in thcir 8 rvice area.
Many states and local groups publish data summaries which contain informa(,ion on
precipitation.
Results of snow surveys in California are publish d by the State Department of
Water Resources in Water Conditions in California. Elsewhere in the West, snowsurvey data are published by the U.S. Soil Conservation Service in bulletins called
Federal-State Cooperative Snow Surveys and Water-supply Forecaste. In the East soow'survey da.ta. are available as follows;
so
PRECIPITATION
"New York Cooperative Snow Survey," U.S. Geological Survey.

"Snow Cover Surveys by Eastern Snow Conference," U.. Weather Bureau.
"Snow and Ie Measurements, Susquehanna Riv r 'Watershed," Fed ral-State
Forecasting ervice, W ather Bureau flic, Harrisburg, Pa.
Special summaries and data on normals and extremes of precipitation and snow may
b found in:
Atlas of Climatic Types in the United States 1900-39, U.S. Dept. AgT. Misc. Publ.
421, 1941.
"Climate and MIlJl," U.S. Dept. AgT. Yearbook, 1941.
Maps of Seasonal Precipitation and Tabl s of ormals, U.S. Weather Bureau Misc .
Publ. 1353, Hl42.
Kincer, J. B.: "Normal Weather for the United States," U.S. Weath r Bureau, 1943.
McDonald, W. F.: "Average Precipitation in the United States," U.S. Weather
Bureau, 1944.
Sha.nds, A. L., and D. Ammerman: Maximum Recorded U.S. Point Raillfall, U.S.
Weather Bur. Tech. Paper 2, 1947.
Maximum station pr cipitation for 1, 2, 3, 0, 12, and 24 Hours, U.S. Weather Bur.
Tech. Paper 15, Parts 1 to 15, 1951- 1955.
Maximum 24-hour Precipitation in the United States, U.S. Weather Bur. Tech.
Paper 16, 1952.
Rainfall Intensiti s for Loeltl Drainage Design in the United States, U.S . Weather
B1tr. Tech. Paper 24, Part], 1953, Part 2, 1954.
Rainfall Intensity-Duration-Fr 'quen y Curves, U.S. Weather B'ur. Paper 25, 1955.
Rainfall Intensities for Local Drainage Design in Western United Stat s, U.S .
Weather Bur. Tech. Paper 28, 1950.
Storm Rainfall in the United States, U.S . Corps oj Engineer8.
PROBLEMS
3-1. Assuming rain falling vertically, express the cat.ch of a gag inclined 20 from
the vertical in terms of apr en tage of the catch for the same gage installed vertically.
3-2. Precipitation tat ion X was inoperative for part of a month during which II
storm occurred. Th storm totals at three surrounding stations, A, B, and C, were,
respectively, 4.20, 3.50, and 4.80 in. Th normal annual precipitation amounts at
to.tions X, A, B, and C are, respectively, 38.50, 44.10, 30.80, and 47.20 in. Estimate
the storm precipitation for tation X.
3-3. The annual pr cipitation at Station X IlJld the average annllal precipitation
at 25 surrounding stations ure as shown in th tabJe on page 51.
(a) Determine the consistency of til r cord at tation X.
(b) In what yeM is a hange in regirn indicated?
(c) Compute th mcan annual pre ipitation for Station X at its present site for
the entire 30-yr p riod, first, without adjustment and, s condly, with the data adjusted
for the change in r gime.
3-4. The average annual precipitation for th four subbasins constituting a large
river basin is 39.70,44.20,33.40, and 28.00 in. The o.r as lire 3GO, 275, 420, and 650
sq mi, respcctively. What is the average annual precipitation for the area as a whole?
3- 5. Construct maximum depth-ar a-duration curves for the United States from
the data of Table 3-3. Tabulate maximum values for areas of 50, 7500, and 30,000
sq mi for durations of 6, 12, 18, 24, 36, 4 , and 72 hr.
3-6. P lot the 25-station av rage precipitation of Prob. 3-3 as a time series. Also
plot 5-yr moving averages and accumulated annual departur from the 36-yr mean.
Are there any apparent r.vnles or time trends? Discuss.
51
PROBLEMS
Annual precipita tion
Year
1956
1955
1954
1953
1952
1951
1950
]949
1948
1947
1946
1945
1944
1943
1942
1941
1940
1939
Annual pr cipitation
Year
ta. X
25-station averag
7.4
7 .3
12.2
11.6
8 .2
11 . 3
7.2
12.0
9 .0
8 .5
8 .8
8 ,0
11 ,2
lO A
11 ,6
8 .1
10 ,6
9 .5
11 , 2
0 .0
15 . 2
11.7
11.2
13 .
9 .3
14 . 6
9.2
11 .4
tao X
193
1037
1936
1935
1934
1933
1032
193 1
1930
192fJ
ILl
In2
9 ,7
10.4
13.1
9 .1
9 ,2
9.1
1927
1926
1925
1924
1923
1922
1921
12 ,3
6.
) 1. I
.6
9 .7
11.2
19 .0
12 . 6
10 .8
12 . 7
17 .2
15 .3
12 ,0
12 . 6
12 .9
12 . 1
11 ,9
16 . 3
25-station
IlV
rage
14 . 2
9 .2
13 . 1
9 .3
9 .9
11 .2
14.2
ILl
10 . 7
10 .8
11 .9
13 . 8
g ,O
12 ,3
11.1
12.4
11 ,0
13 , 5
3-7. Repeat Prob. 3-6 for some station in your immediate vicinity, using data from
the Climalic Summary of lhe United Slates or other source.
3-8. Compute the mean annual precipitation for some river basin selected by your
instructor. Use th arithmetic average, Thiessen network, and isohyetal map and
compare tho three values. Which do you feel is most accura te ? Row cOJlsistent
are the answers determined by each method among members of your class?
3-9. Repeat Proh. 3-8 using rainfall from some storm select ed by your instructor.
Obtain the precipitation data from Climalological Data.
3-10. Plot a graph showing the monthly di stribution of norma l precipitation for a
station in your area. Explain briefly the climatological factors causing the observed
distribution.
3- 11. Plot the data of Table 3-1 on logarithmic pap r and determine the equation
of the enveloping straight line. Usc duration in hours.
3-12. Write an equation for thn calorimntri c d termination of snow quality .
4
STREAMFLOW
Most data used by hydrologists serve other purposes in meteorology,

climatology, or other earth ciences. Streamflow data are gathered
primarily by hydrologists and used mainly for hydrologic studies. More
significant to the engineering hydrologist, however, is the fact that streamflow is the dependent variable in most studies, since engineering hydrology
is (1oncerned mainly with estimating rates or volumes of flow or the
chunges in these values resulting from man-made causes.
WATER
TAGE
4-1. Staff gag . River stage i the elevation of the water surface at
a specified station above some arbitrary zero datum. The zero elevation
is sometimes taken as mean sea level but more often it is set slightly
below the point of zero flow in the stream. Be 'ause it i difficult to
make a direct and continuous measurement of the rate of flow in a stream
but relatively simple to obtain a continuous record of water-surface
elevation, the primary field data gathered at a streamflow measurement
station are river stage. Stage data are traIl formed to flow data by
methods discussed in the subsequent s ctions.
The implest way to m asure river stage is by means of a staff gage,
a scale set so that a portion of it is immer d in the water at all times.
The gage may consist of a single vertico.l scale attached to a bridg pier,
piling, wharf, or other structure that extends int the low-water channel
of the stream (Fig. 4-1a). If no suitable structure exists in a location
whioh is acce ible at all stages a sectional staff gage (Fig. 4-1b) may be
used. Short ctiol1s of staff are mounted on available structures or on
specially oonstru ted supports ill such a way that one section is always
accessible. An alternative to the ectional staff is an inclined staff gage
(Fig. 4-1c) wIll h is placed on the slope of the stream bank and graduated
so that the scale read dire tly in verti al depth.
The gage s ale may oonsist of mark painted on an exi. ting structure
or on a wooden plank attaohed to the support. Painted scales are usually
52
STAGE GAGES
53
graduated in feet and tenths. Marlcings ar often in the pattern u ed

on stadia rods for visibility. Enameled metal sections" ith markings
in feet and hundredths are avail ble where particularly accurate stage
data are desir d. If a stream carri s a large am unt of fin s diment or
industrial waste, scale markings may be quickly obliterated. In this
case, a serrated edge on the staff or raised marking symbols may be
helpful.
~
,
FIG. 4-1 . Stoff gages.
"
(0) Simple vertical stoff.
(b) Sectional stoff.
(c) Inclined stoff.
4-2. upended-weight gages. Another type of manual gage is

the suspended-weight gage in which a weight is lowered from a bridge or
other overhead structure until it renches the wnter surface. By subtrn.cting the length of line paid out from the elevation of a fixed reference
point on the bridge, the water-surfl1ce elevation can be determined.
The wire-weight gage (Fig. 4-2) has a drum with a circumference such
that each revolution unwinds 1 ft of wire. A counter records the number
of revolutions of the drum while a fixed reference point indicates hundredths of feet on a scale around the circumference.
4-3. Recording gages. Manual gages are simple and inexpensive
but must be read frequently to define the hydrograph adequately when
STREAMFLOW
stage is changing rapidly. Water-stage recorders, in which the motion

of a float is rec rded on a cho,rt, overcome this diffi 'ulty. In l1 continuous
recorder, the motion of the float moves a p n across a long strip chart.
The chart is usuaJIy 10 in. wide and, at a scale of 2 in. = 1 ft, r presents
a 5-ft change in stage. When the pen reaches the edge of the chart it
reverses direction and records tbe next 5 ft in the oth r direction across
the chart (Fi Y. 4-3a). Tbe chart roll contains enougb paper to operate
for a year at a scale of 2.4 in./ day. Clo k are u ually weight-driven
FIG. 42. Wireweight gage.
(Leupold and Stevens Instrument Co.)
and will run a long as th r is room for the clock weight to drop. Small
0) tric cIo ks that can op rate for 30 days on a flashlight battery have
b en u d.
A short-term r' rd r usually has a chart wrapp d around a drum
whi 'h is rotat d by the float whil the pen is driven at constant speed
parallel to it axis. The circumIer nee of the drUID repres nts any
s I ted ohange in stag . Larger changes are record d beginning again
at the bottom of the chart (Fig. 4-3b).
A float-type water-stage record r requires a shelter house and stilling
well (Fig. 4-4). The stilling well serve to protect the float and c unterweight cab] s from floating debris and, if the intake are properly designed,
suppr ses flu tuatiolls re ulting from urface waves in the stream. Inexpensive tilling wells have been made from timber and corrugated-steel
pipe, while more elaborate tructures are u ually reinforced concrete.
S5
STAGE GAGES
Generally two or more intake pipes are placed from the well into the
om tim s
stream so that at least one will admit water at all times.
the water enters through the opeD bottom of the well. In this cas an
inverted cone may be placed over the bottom of the well to r duce the
size of the opening and suppress the effect of surface wav s. Tho openbottom type of stilling well has the advantage that it is less likely to
-en <Xl -r-
~
It'>
;2:
0::
I~
0::
It'>
ID
,/
-----II
I""'--
~-
/-
--
.-- -- . -
-I-
-r-
-=-~
_ <Xl/
(])
<t
.....
r()
-- --
- ~J
-/- -
\2
\
\
\
8
-f - {; ::-
::.- 0
- -~
7"~
-1
-- I
----
- - --
- <t
-- - - -
(\
1\
\,
DOlle
13 \14 15 16 17
18
(b)
~
(0)
FIG. 4-3. Recording-gage charts: (0) continuous-strip chart and (b) weekly chart.
fill with sediment. If a stilling well with a closed bottom is installed

on a stream with a high sediment load, it is necessary to provide for
removal of the sediment which accumulates in the well. This can be
done by digging the sediment out and removing it in buckets or by providing valves on the intake pipes which may .he closed while the woll
is pumped full of water and then opened to flush out the sediment. It
is customary to install staff gages inside and outside the well. These
staff gages serve to check the performance of the recorder and are read
each time the station is visited.
Recent developments in recording gages include a float recorder which
uses a tubular float and requires only a small-diameter pipe as a stilling
well. Efforts are also being made to develop pressure-sensitive gages
STREAMFLOW
56
Reinforced concrete
shelter --- - ___ _
Recorder float _
"
"
.1'1'- I\,_ . -
,...,-'
I r
I
,.
Stoff gage
~
FIG. 44. A typical waterstage recorder installation.
(U.S. Geological Survey. )
and gages activated by radioisotopes as well as other devices. The

aim of all of these is to eliminate the costly stilling well required with
conventional float-op rated gages. It has proved difficult to develop
gages accurate to 0.01 ft at low stages yet having the ruggedness and
range required to record flood stages.
STAGE GAGES
57
4-4. Cr t- tage gage. A third type of stage-measuring device

is the crest gage, used to obtain a record of flood crests at sites where
recording gages are not installed. A variety of such gages have been
devised, including small floats which rise with the stage but are r strained
at the maximum lev ell and wat r-soluble pl1ints 2 on bridge piers where
they are protected from rain and can
Standard pipe
indicate a definite high-water mark.
The gage used by the U.S. Geological
urvey consists of a length of pipe
(Fig. 4-5) containing a graduated
stick and a small amount of ground
cork. 8 The cork floats as the water
rises and some adheres to the stick ;/'}/;::;at the highest level reached by the
Graduated stoff
fils inside pipe
water. The stick can be removed ,
the crest reading recorded, the cork
wiped off, and the stick replaced
ready for the next rise. Crest gages
provide low-cost, supplementary records of crest stages at locations
where recorders are not justified and
where manually read staff gages are
inadequate.
4-5. Mi cellaneou
tage gages.
Water- or mercury-filled manometers
Holes to adml'l water
are often used to indicate reservoir
Pipe cop
water levels or to actuate recording FIG. 4-5. Crest-stage gage used by the
devices. If flow rate is dependent u.s. Geological Survey.
on the difference in water level, a
single recorder with two floats and two stilling wells may be used to
record water-level differences. Remote recorders in which a system of
selsyn motors is used to transmit water-level information from stream
side to a recorder at a distance are available, as are numerous remote
transmitting telephonic or radio gages. These latter gages use a coding
device which converts stage to a signal which can be transmitted in the
form of a series of impulses which can be countcd, a change in frequency
of oscillation which can be measured, or the time interval required for
I M. H. Collet, Crest-stag Meter for Measuring Static Heads, Civil Eng., Vol. 12,
p. 396, 1942.
J. C. Stevens, Device for Measuring Static Heads, Civil Eng., Vol. 12, p. 461, 1942.
2 F. J. Doran, High Water Gaging, Civil Eng., Vol. 12, pp. 103- 104, 1942 .
. G. E. Ferguson, Gage to Measure Crest Stages of Streams, Civil Eng., Vol. 12,
pp. 570-571, 1942.
STREAMFLOW
58
a sensing element to move from a zero point to the water surface at

constant speed. Such remote recording devices are used primarily for
flood forecasting or reservoir operation. Co t usually precludes their
use for routine data collection.
4-6. SeJection of a s tation ite. If a stream gage i solely to record
water level for flood warning or as an aid to navigation, the prime factor
in its location is accessibility. If the gage is to be used to obtain a record
of discharge, the location should be carefully selected. The relation
betwe n stage and discharge is controlled by the physical features of the
channel downstream from the gage. When the controlling features are
situat d in a short length of channel a section control exists. If the stagedischarge relation is governed by the slope, size, and rouglmess of the
channel over a considerable distance the station is under channel control.
In many cases no single control is effective at all stages but a complex of
controlling elements function as stage varies.
The ideal low-water control is a section control consisting of a rapids
or ritHe. If this control is in rock it will be reasonably permanent and
once calibrated, need be checked only infrequently. Where no such
natural low-water control exists, an artificial control consisting of a low
concrete weir, sometimes with a shallow V notch, may be constructed
to maintain a stable low-water rating. A channel control is more likely
to change with time as a result of scour or deposition of sediment, and
more frequent flow measurements are required to maintain an accurate
stage-discharge relation.
Rapids may also b effective controls at high flows if the slope of the
stream is steep, but where slopes are flat, the section control is likely to
be submerged and ineffective at high flows. High-water controls are
more likely to be channel controls although in some cases the contraction
at a bridge 01' the effect of a dam may control at high stages. It is
advi able to avoid I cations wh re varying backwater from a dam, an
intersecting stream, or tidal action occurs. These situations require
special ratings which tend to be less accurate.
DI CHAnGE
4-7. Curren.t meter. The stage record i transformed to a discharge record by calibration.
ince the control rarely has a regular
shape for which the discharge can be computed, calibration is accomplished by relating field mea urements of discharge with the simultaneous
river stage (Sees. 4-8 to 4-11).
The most common current meter in the United States is the Price
meter (Fig. 4-6), the type used by the U.S. Geological urvey. It
consists of six conical cups rotating about a vertical axis. Electric
conta ts driven by the cups clo e a circuit through a battery and the
CURRENT METERS
wire of the supporting cable to cause a click for each revolution (or each
.fifth revolution ) in headphone worn by the operator. For mea. urements in deep water, the meter is suspended from a cable. Tail van
to keep the meter facing into the urrent and a heavy weight to keep the
meter cable as nearly vertical as pos ible are provided.
p cinl cranes
are available to support the meter over a bridge rail, to simplify handling
of the heavy weights, and to permit m asuring the length of cabl paid
FIG. 4-6. Price current meter and 30-lb C-type sounding weight.
(U.S. Geological Survey. )
out. In shallow water the met r is mounted on a rod, and the observer
wades the stream. A pygmy Price meter has been u ed for measuring
flow at very shallow depths.
The other type of current meter is the propeller type, in which the
rotating element is a propeller turning about a h rizontal axis (Fig. 4-7).
The contacting mechanism of a propeller meter is similar to that of a
Price meter, and similar suspensions are u ed. The vertical-axis meter
has an important advantage in that the bearing supporting the shaft
can be enclosed in inverted cups which trap ail' around the bearings and
prevent entrance of sediment-laden water. The benrings of propeller
meters cannot be so protected and are therefore exposed to damage by
abrasion. On the other hand, the vertical-axis meter has a disadvantage
in that vertical currents or upstream velocity components rotate the
cups in the same direction as the downstream currents. A Price meter
moved vertically in still water will indicate a positive velocity. Hence,
60
STREAMFLO"
the Price meter tends to overestimate the velocity in a str am. If the
measuring section is well chosen with Lbe current flow nearly parallel
to the channel axis and with a minimum of turbulence, the error is
probably not more than 2 per cent. l
Wading rod
Propeller
Foot plate
FIG. 4-7. Propeller-type current meter.
The relation between revolutions per second N of the meter cups and
water velo 'ity f) in feet per se ond i given by an equation of the form
' f)
+ bN
(4-1)
where a is the starting velocity or velocity r quired to overcome m chaniorne

cal friction. For the Pri 'e meter, a is about 0.1 and b about 2.2.
1 M. P. O'Brien and R. G. Folsom, Notes on the Design of Current Meters, Tran8.
Am. GeophY8. Union, Vol. 29, pp. 243- 250, April, 1948.
CURRENT-METER MEASUREMENTS
61
variation in these constants must be expected a a r sult of manufacturon 'equ ntly each meter should
ing limitations and the effect.s of war.
be individually calibrated. This is usually done by mounting th meter
FIG. 4-8. The procedure for current-meter measurement.
on a carriage which moves it through still water. The carriage may run
on rails along a straight channel or may rotate about a centml pivot
in a circular basin. The speed of the carriage is determined by the time
required to travel a known distance.
o
With several runs at various speeds
it is possible to plot a curve showing
I
the relation between meter contacts
Axis of parabolo .....
0.2
per unit time and water speed.
4-8. Current-m ler
mea urements. A discharge measurement
requires the determination of sufficien t

I
point velocities to permit computation
of an average velocity in the stream.
/
0.6
The cross-sectional area multiplied by
u
o
the average velocity gives the total
U:
discharge. The number of velocity
0.8
determinations must be limited to
/
those which can be made within a
reasonable time in terval. This is
especially true if the stage is changing
6
10
8
Velocity in feet per second
rapidly I since it is desirable to complete the measurement with a mini- FIG. 4-9. A typical vertical velocity promum change in stage.
flle in a stream.
The practical procedure involves
dividing the stream into a number of vertical sections (Fig. 4-8). No
section should include more than about 10 per cent of the total flow i
thus 20 to 30 vertical sections are typical. Velocity varies approximately as a parabola (Fig. 4-9) from zero at the channel bottom to a
,g
STREAMFLOW
62
I~
~ I~
ml
::t
~ Jll
~ 'l
~'11
~
I~
,...
'"
I~
...
~~
..
"' ..
~
....~
It t
Ii::::
l:!
I~
!~
l~
.~
~~
"".::~
1'1 ....
~~
I;: ~
...
I ...~
I~
.~
I~
til
I~
~~
I ~ ~
I";
~
In
~~
I ~I ~
<'f40
I'! 40
"
11\
,.,
,~
t")
."
..,
'"
I~
.,
~I ~
~~
~I ~
"!IS!
aJll g
~~
,,!IIQ
~~
N.GQ
,...
~l ~
~I" ~
Oi",
Ito!tI!
..,
~c:
l~
..
:~
"l
l ~l ~
"!IS!
-'"
.~
S~
o 0
1')1'0
C)
...~
~i;;
iii
I~
I ;I~
'f
1<1
......
.~
~~
m
0
I~
:; I ~
~ I"
~I
1
S
(;)
!Ij
,.;
,...
"
..;
I,!
':Ioj
In
iii
Ie
It-
It}
'e;
':t-
til
~~
I'! "1
'"
t;
+- l-
63
CURRENT METER MEASUREMENTS
maximum at (or near) the surfac. On the balSis of many field tests, the
variation for most channels is such that the average of the velooities at
0.2 and 0.8 depth below the surface equals the mean velocity in the
vertical. The velocity at 0.6 depth below the urface is also nearly equal
to the mean in the vertical. The adequacy of the e assumptions for a
particular stream may be tested by making a large number of velocity
determinations in the vertical.
A
I
I
I
I
I
I
- --r-=~True
I
I
I
verlicol
\
WeI line
I
I
Meter
I . /Streom bed
Flow
-\\
\
\\
Weight \
///////')/;1);'7////////T//////T//7/7//7~~/lT/7//;)//
FIG. 411. Position of sounding line in swift water.
Sample current-meter field notes are shown in Fig. 4-10. The determination of the mean velocity in a vertical is as follows:
(a) Measure the total depth of the water by sounding with the meter
cable.
(b) Raise the meter to the 0.8 depth and measure the velocity by
starting a stopwatch on an impulse from the meter and stopping it on
another impulse about 45 sec later. The number of impulses counted
(taking the first as zero) and the elapsed time permit calculation of
velocity from the meter calibration.
(c) Raise the meter to the 0.2 depth and repeat step (b).
In the shallow water near the shore a single velocity determination
at 0.6 depth may be used.
If water velocities are very high, the meter and weight will not hang
vertically below the point of suspension but will be carried downstream
by the current (Fig. 4-11). Under these conditions the length of line
STREAMFLOW
paid out is greater than the true vertical depth, and the meter is higher
than the true 0.2 and 0.8 depths. Very heavy weights are used to
minimize this effect, but if the angle between the line and a vertical
becomes large it is necessary to apply a correction to the measured
depths. 1 The actual correction dep nds on the relative I ngths of line
above and below the water surface, but at a vertical angle of 12 the
error will be about 2 per cent. A slight additional error is introduced if
the current is not normal to the measuring section.
Computation of total discharge is made as follows 2 (Fig. 4-10):
(a) Compute the average velocity in each vertical by averaging the
velocities at 0.2 and 0.8 depths.
(b) Multiply the average velocity in a vertical by the area of a vertical
section extending halfway to adjacent verticals (ABeD, Fig. 4-8). This
area is taken as the measured depth at the vertical (EF) times the width
of the section (AB).
(c) Add the increments of discharge in the several verticals. Incremental discharge in the shore section (GHI, Fig. 4-8) is taken as zero.
Access to the individual verticals of a secti n may be obtained by
wading if the water is shallow. Usually, wading is impractical at high
stCLges, and the meter must be lowered from CLn overheCLd support.
Where possible, bridges are used as th measuring section, if the bridge
is normal to the stream axis and the current essentially parallel to the
str am axis. The measuring section need not be at the same location
as the control section. However, the distance between the sections
should be short enough so that the intervening inflow is not large.
Where no existing bridge is suitable, a special cableway may be used.
The hydrographer rides in a small car suspended beneath the cable and
lowers the meter through an opening in the floor of the car. Where
neither bridge nor cableway is practical, measurements may be made
from a boat. This is far less satisfactory because of the difficulty of
maintaining position during a measurement and because either vertical
or horizontal motion of the boat results in a positive velocity indication
by the Price current meter.
4-9. Stage-discharge relation. Periodic meter measurement of
flow and simultaneous stage observations provide the data for a calibration curve called a rating curve or stage-discharge relation. For most
stations a simple plot of stage vs. discharge (Fig. 4-12) is satisfactory.
Such a curve is approximately parabolic but may show some irregularities
1
D. M. Corbett a.nd others, Stream-gaging Procedure, U.S. Geol. Survey Water-
supply Paper 888, pp. 43-51, 1945.

I This d cribes the pro edure adopt d by tho U..
eologicai Survey about 1950.
See K. B. Young, "A Comparative tudy of Methods for Computation of Discharge,"
u.S. Geological Survey, February, 1950.
65
STAGE-DISCHARGE RELATIONS
if the control changes between low and high flows or if the cross seotion
is irregular.
The adequacy of a rating curve is measured by the scatter of the

measured flows about the mean line. If the control is reasonably
permanent and the slope of the energy line at the station reasonably
con tant for all occurrences of a given stage, a simple rating will suffice.
Discharge In cfs
o
rr'- _
.,.,
Ui
4 -
43
1.2
1.0
004
0,2
20
IA
'-:S 0,8
- g. 0,6
-
10
/'
50
60
f--
;::.:.:;;;'I
40
Low fl ow ra t ing
38
~445
~
........
./
-- -
40
I
4~5'~1
~~\
I_,A
30
_. ...........
\'
,/"
42
4~49
-- --
~
~
High flow rot Ing
'~48
50
1000
600
700
2000
200
300
400
500
3000
Discharge In cfs
I
100
"
Intermediate rating
~""'9
"'~o
,"
\ 48
f--
BOO
Discharge in cfs
FIG. 4-12. A sim ple stage-discharge relation.
If the control is altered by scour or deposition, frequent meter measurements are necessary. Under conditions of shifting control, discharge is
usually estimated by noting the difference between the stage at the time
of a discharge measurement and the stage on the mean rating curve which
hows the same discharge. This difference is applied as a correction
to all stages before entering the rating. If the correction changes between
measurements, a linear change with time is usually assumed.
Individual measurements may deviate from the mean stage-discharge
relation as a result of differences in water-surface slope at the control.
Since velocity head is usually small, the slopes of the water surface and
of the energy line are nearly equal. Differences in slope may be caused
by varying backwater as a result of an obstruction downstream or high
stages in an intersecting stream. If either of these factors is present,
STREAMFLOW
66
the mting curve must include slope as a pa.rameter. 1 The basic approach
is repr sonted by
(8)'" (F)k
Fo
(4-2)
io = So
The equation states that discharge q is proportional to a power of water'lurface slope s. From fluid mechanics the exponent m would be expected
to be H. The fall F is the difference in water-surface elevation between
' 34
o Observed discharge
Adjusted discharge
0.96
30
Q)
Q)
."
..... 22
0
1.00
."
;if?
18
:::IE
c:
7'
1.2
,/1.00
."
,/
00
0
...J
1.06
U2
O'V
~
.: 26
..
..
.e
~/.06
~/.w
.oe
....
0
"0
0
~02
14
;y
,/
o.eo 0
(/)
10
0.61
.06
'('
20
40
0.8
"'"
0.6
.0.4
O.!Jlo-1'fnJ'OO
/"
60
...#'It
1.0
0
80
I
100
~...-
~~
......,..
--
0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1

Observed fall
120
140
160
180
200
DIscharge in 1000 cfs
FIG. 4-13. A slope-stage-discharge rating curve for the Tennessee River at Chattanooga,
Tennessee. (U .S. Geological Survey.)
two fixed sections and is usually measured by two conventional river

gages. Th re is no assurance that the water surface between these
gages is a straight line, i.e., that F jL = s. Consequently the exponent k
need not be H and must be determined empirically.
A slope-stage-discharge relation requires two gages, a base gage and an
auxiliary gage. The gages should be far enough apart so that F is at
least 1 ft to minimize the effect of observational errors. The fall F
applicable to each discharge measurement is determined, and, if the
observed falls do not vary greatly, an average value Fo is selected. Fo
need not be a mathematical average of the falls and for convenience
1 M. C. Boyer, Determining Discharge at Gaging Stations Affected by Variable
Slope, ivtl Eng., Vol. !I, p. 556, 1939.
W. D. 1'itchell, Stag -Fall-Discharge Relations for teady Flow in Prismatic Channels, U.S. Geol. Sttrvey TVateT-Sl~pply Paper 1164, 1954.
STAGE-DISCHARGE RELATIONS
67
should be an even foot. All measurements with values of F = F 0 are

plotted as a simple stage-discharge relation and a curve is fitted (Fig.
4-13). This is the qo curve representing the discharge when F = Fo.
If F ~ F o, the ratio FIFo is plotted against q/ qo on an auxiliary chart.
The discharge at any time can be oomputed by caloulating the ratio
F/ Fo and selecting a value of q/ qo from the auxiliary curve. A value of
qo corresponding to the existing stage is taken from the qo curve and
70
'lii 60
.2!
43"",
~ 30
;5
~
4.5 5.0 5.5
40
f.r
<l>
S 30
01
"L
If
5.1.~
2.~
Normal
fall curve .1 I
I
g'" 20
.S
\,
~40
<l>
~
5.2
60 .S 50
50
5.9....
/'
b
10
I~ O~
5.
r--5.8
13.0
~ ~ qn curve
\ 5.0
/'
1:-4.9
1.2
4.4
C 1.0
'"tY 0.8
Adjusted discharge
Numbers indicate fall
/n feet 1
1
~
1~
.. /
./
-:-
V .,..",.
""
0.6 /
0.4
0 0.2 0.4 0.6 0.8 1.0 1.2
o Measured discharge (qa)
Cofrect~on durv~
- 4.3
Z! \
20
l-/
V
~ ~5.9
1~
1~
(aim I
1~
1~~
Dischorge in 1000 cfs
FIG. 4-14. Normal -fall rating curve for the Ohio River at Metropolis, Illinois.
logical Survey.)
(U.S. Geo-
multiplied by q/qo to give q. If the auxiliary curve plots as a straight

line on logarithmic paper, the slope of the line is k in Eq. (4-2). The
rating just described is known as a constant-fall ral,ing, since the adopted
mean faU F 0 is constant.
In some cases the range of F is large and its variation is related to
stage. In this case a normal-fall rating may be used. This rating is
similar to the constant-fall rating except that the normal faU F" whjch
is used in place of F 0 is taken as a function of stage and defined by a
second auxili ary curve (Fig. 4-14). To compute discharge with a normalfall rating, the actual fall F is computed, a value of F" is read from the
curve, and the ratio F/F n is calculated . The ratio q/ q" can then be
determined and multiplied by q" from the rating curve to give q.
68
STREAMFLOW
Figure 4-15 shows the profile of a flood wave as it moves past a station.
The slope is equal to 8b + 8" where 8b i the slope of the channel bottom
(or the slope of the water surface in uniform flow) and , = dg / u dt. Here
do/dt is the slope of the flood wave expr ssed a a rate of change of stage
with time and tt is the eel rity of movement of the flood wave.
ince 8, is
the variable factor in the slope, a rating can be constructed by plotting
stage against discharge for those measurements for which du/ dt is near
zero. If u is assumed to be constant, a correction curve can then be
established by plotting q/ qc vs. dg / dt, where qc is the discharge corresponding to the existing stage whell dU/ dt = O. Theoretically, dg/ dt is the
FIG. 4-15. The profile of a flood wave.
tangent to the water-stage recorder trace, but in practice it is convenient

to use t:.g, the change in stage during a finite time period, usually 1 hr
but sometimes much longer.
4-10. Exten ion of rating curves. There i no completely satisfactory method for extrapolating a rating curve beyond the highest
measured discharge. It is often assumed that the equation of the rating
curve is
q = k(g - a)b
(4-3)
where a, b, and k are constants for the station. This is the equation of
a parabola in whi h a is approximately the distance between the zero
elevation of the gage and the elevation of zero flow. If the COITect
value of a is determined by trial the equation will plot as a straight line
on logarithmic paper and is easily extended. However, the procedure
includes the assumption of a parabolic rating and cannot account for
any marked change in the hydraulic geometry of the stream at high flows.
Another method 1 of ext uding rating curves is based on the Ch6zy
formula
(4-4)
q = AC Rs
where C is a roughness coefficient, 8 is the slope of the en rgy lin , A
is the cross-sectional area, and R is the hydraulic radius. If C VB iR
I
J. C. Stevens, A Method of Estimating Stream Discharge from a Limited Numb'!!
of Gagings, Eng. New8, pp. 52- 53, July 18, 1907.
69
THE f FFECT OF ICE ON STREAMFLOW
assumed to be onstant for the station and D, the mean depth, is ub tituted for R,
q = kA
Vl5
(4-5)
Kno\\'l1 values of q and A Vl5 at' plotted on a graph and usually define
som thing close to a straight line which may be readily extended. Values
of A VJ5 for stages above the xisting rating can be obLaiued by field
measurement and used with the exLended curve for estimates of q.
A third method of estimating high flows is by application of hydro.ulic
principles. The proeedure is often referr d to as a slope-area computation.
ufficient high-water marks must be located along a reach of channel to
permit determination of the water-surfa e slope at the time of peak.
Cros 'e tions of the channel may be determined by leveling or sounding,
and the area and hydraulic radius calculated. The CMzy-Manning
formula is ordinarily used to compute discharge:
q = 1.49 AR"'sl'i
n
(4-6)
The main source of error in applying Eq. (4-6) is in estimating the roughness coefficientn (Appendix B).
inceq is inversely proportional ton and
the average value of n for natural streams is about 0.035, an error of
0.001 in n represents about 3 per c nt in discharge. Considerable doubt
may exist as to whether the cross section measured after t he flood is the
section whi h exi ted at the time of peak. A stream often scours during
rising stages and redeposits material durillg falling stages (Fig. 4-16).
Under the most favorable conditions, an error of 10 per cent may be
expected in a slope-area estimate of fl ow.
4-11. Th effect of ice on streamflow. When ice covers a stream,
a new friction surface is formed and the stream becomes a closed conduit
with lower discharge because of the d creased hydraulic radius. The
underside of the ice sheet may be extremely rough if ice cakes are tilted
helter-skelter an d then frozen together. Movement of the water under
the ice gradually develops a smooth sulface. If the stage falls, leaving
the ice as a bridge across the stream, the stage-discharge characteristics
return to those of a free stream.
In turbulent str ams the first ice to form is Jrazil ice, small crystals
suspended in the turbulent flow. Frazil ice collecting on rocks on the
stream bed is called anchor ice and may cause a small increase in stage.
If the turbulence is not sufficient to keep the frazil ice mixed in th stream:
it rises to the surface to form sheet ice. Until a complete ice sheet is
formed, small variations in the stage-discharge relation must be expected
from time to time.
70
STREAMFLOW
When ice conditions exist it is necessary to make periodic measurements I

through holes in the ice and to interpolate the discharge between these
measurements in any manner which seems rea anable. The meter
must be moved rapidly from hole to hole and should be kept in the water
at all times (except when being moved) to prevent fr ezing. Fortunately, if the stream is solidly frozen over, the flow is usually small since
there will be little snowmelt or other source of runoff within the tributary
area.
Apnl 18,1952, stage 30.2 fI

Discharge 396,000 cfs
Mean velacdy 8.6 fps
1,1952, stage 14 fI
Discharge 85,000 cfs
Mean velocity 4.2 fps
25,1952, stage 5.1fl
Discharge 32.000 cfs
Mean velocdy 4.3 fps
1000
500
1500
Oislonce in feel
FIG. 4-16. Cross sections of the Missouri River at Omaha, Nebraska, showing the progrl"ssive scour and fllling during the passage of a flood wove.
4-12. Other method of obtaining treanlfiow data. On large

streams the discharge at dams may be determined from calibration of
the spillway, sluiceway, and turbine gates. If a careful record of gate
and turbine operation is maintained, the discharge can be computed
This often requires that the spillway be rated by model tests and the
turbines by salt-velocity 2 measurements. Some progress has been made
in generalized rating curves based on the geometry of the gate structure. 8
On small streams, flow measurement may be made with weirs or
1 W. G. Hoyt, Th Effect of Ice on
tr am Flow, U.S. Gool. Survey Water-supply
Paper 337, pp. 24-30, 1913.
1 C. M. Allen and E. A. Taylor, The Salt Velocity Method of Water Measurem.ent,
Trans. ASME, Vol. 45, p. 285, 1923.
I J . N. Bradley, Rating Curves for Flow Over Drum Gates, Trans. ASCE, Vol. 11 9,
pp. 403-431, 1954.
UNITS
71
flumes.' These devices are commonly rated on the basis of laboratory

calibration although the rating lll11y be checked in place with curr nt
meters. For small streams, a combination of a V-notch weir for low
Bows and a venturi flume for high Bows may be nece ary to assure
accuracy. Large weirs are usually unsatisfactory because deposition of
sediment in the upstream pool changes the discharge characteristics.
Highway culverts have frequently been suggested as Bow-measuring
devices, and in many instances this is feasible. However, the hydraulics
of culverts is quite complex. 2 In Bat terrain, care must be taken to
avoid culverts affected by backwater unless both headwater and tailwater stages are measured. On steep slopes it is nece sary to establish
whether the culvert Bows full with control in the barrel or partially full
with control at the entrance. Temporary changes in discharge capacity
may occur as the result of sediment or debris deposits.
Rough measurements of discharge can be made by timing the speed of
Boats. A surface Boat travels with a velocity which is about 1.2 times
the mean velocity. Obj ects Boating with a greater submerged depth
will travel at speeds closer to the mean velocity in the section. a
INTERPHETATJON OF STUEAMFLOW DATA
4-13. Units. The basic Bow unit in the United States is the cubic
foot per second (also call d second-foot and abbreviated cusec, 01' cfs).
Countries using the metric system usually express Bow in cubic meters
per second. Volume of flow may be expressed in cubic feet or cubic meters
but this leads to very large numbers, and larger volume units are commonly used. In the United tates the second-foot-day, or the volume of
water discharged in 24 hours, with a flow of one cubic foot per second is
widely used. One second-foot-day (sfd) is 24 X 60 X 60 = 86,400 cu ft.
The average Bow in cubic feet per second for any 24-hr period is the
volume of Bow in second-foot-days. The other common unit of volume
is the acre-foot, the volume of water required to covel' an acre to a depth
of one foot. Hence, an acre-foot contains 43,560 cu ft and equals '
0.504 sfd. Within an errol' of 1 pel' cent, one acr'3-foot equals one-half
second-foot-day. In some cases it is convenient to use second-foot-hours
014 sfd) as a volume unit. In countries using the metric system, volume
1 R. L. Parsball, The Parshall Measuring Flume, Colo. Agr. Expt. Sta. Bull. 423,
1936.
2 Culvert Hydraulics, Highway Re8earch Board Rept. I5-B, 1953.
F. T. Mavis, The Hydraulics of Culverts, Penn. State Coli. Eng. Expt. Sla. Bull. 56,
1943.
C. L. Larson and H. M. M:>rris, "Hydraulics of Flow in Culverts," University of
Minnesota, October, 1948.
A float extending irom the surface to mid-depl','n l.ra els with a velocity about 1.1
times the mean velocity.
72
STREAMFLOW
is usually expressed in some multiple of cubic meters, commonly 10,000

cu m which equals 4.09 sfd (Appendix B).
Less common units of flow are the cubic foot per second per sq mi (csm) ,
the inch, and the miner's inch. Cubic feet per second per square mile
is a convenient unit for comparing rates of flow on streams with difT ring
tributary area and is the flow in cubic feet per second divided by the
drainage area in square miles. The inch, the amount of water requir d
to cover the drainage area to a depth of one inch, is a useful unit for
comparing streamflow with the precipitation which caused it. The inch
is a unit of volume only when associated with a specific drainage area.
8
7
.:!!
..,
0
0
.!:
~
0
52
LL
-12
Flow scole in
fOOO cfs for
folded portion
of groph
N
6p
60
6p
60
6p
60
Time in hours
FIG. 4-17. Use of a folded scale to plot extremely high flows.
One inch on one square mile = 640/ 12 acre-ft = 53.3 acre-ft = 26.9 sfd.
The miner's inch is an old unit used in the western United States and
defined as the rate of discharge through an orifice one inch square under
a sp cified head. By statute the miner's inch has been defined between
0.020 and 0.02 cfs, depending on the state (App ndix B). The unit
is no longer used except wher old water rights are specified in terms of
miner's inch s.
It is desirable to treat annual streamflow data in such a way that the
flood season is not divid d between successive years. Vari us water
year have b en used for specinl purposes but the U.S. Geologicnl Survey
uses the period October 1 to eptember 30 for data publication, and this
is the usual water year in the United tates.
4-14. Hydrographs. A hydrograph is a graph of stage or discharge
against time. Many different methods of plotting may be used, depending on the purpose of the chart. Monthly and annual mean or total
flow is used to display the record of past runoff at a station. The characteristics of a particular flood, however, usually cannot be shown successfully by plotting average flow for periods longer than one day. Prefer-
11
CUMB!RLAJID RIVlR BASIH
Cumborland Rtver at Wll11.... burS, 111 .
~~;-~~ ~~~4;~~t!0~s~~=~0 ~ ~~ ~ml::~~U;~ , r~h~~i~;t~~~t;~":n~ri~fm~~.~ d~.m8t..ream from Clear Fork.
Drainage area.--1,673!1q mi.
RocordO nvallable. - -October 1950 to September 1954. Oase-helght reeord. collocted In thll
vIcInity aince 1908 nrc contained 1n report! or U. S . WOllthor Bureau.
~ . --Water-at.ag . rocorder .
Det-um or SagD 18 891.52 rt abovo menn son level, unadJus\ted .
-----Prior to July 2. 1951. w1re -weight. gage at Some 8U.8 and datum.
~ . --"ax1nlum dl.cha~e during yonr( 20,500 cr. Jan. 23 (Sago holghL, 20.38
I!ln~~~5t 1 ~~~1~~' d~.Ct!~S.~6 37~Joog~: ~~tgh~:

III"M.1~~.:;t~~.~~.;.2~~ ~~~.~7i9i~~3:iu rt Jan.
f~~g.
igg1
nl:
heigh , 29.85 rt); mlni-
10,1946 , from graph b cd on U. S.

0 stage of :12.7
rt.
~.~~h:~a~~1:~ea~~ ~Qg~':Rt.h~~orur~u M:~ge 2~uaa~~:~. roached
Re'~~:e;;~~~~~~rg~~:. wa~:~O~:r 0i9&~e:;~8~1~~~1{~Cesp D~3~o~dOd sediment. londa , and wat.(tr

Ratil'\8 table., water Jev 19 S~-5 4 (8ft.8e height, 1.n fe.t., and dl.c hft.rge.
1n cul)lc raet per oond)
Oct. 1 to .rune 18
1 . 64
2.6
'.0
. 0
0
10.0
21.0
. 1
. 0
12
17
29
1.7
1.15
1.8
1.9
2.0
2.'
<5
Jun. 19 \.0 Sept. 30

205
'80
920
2, 420
.,500
21,500
10
1.66
1.7
l.8
30
'5
1..
2.0
13
20
~.--Same ... P1"eced"

ina tabla abov" 2.0 re.
115
,.rr ""..
Dhottarge, 1n cu b1c r .. t P4tr cond, wa ter :rear OCtober 19:53 to SePttlllbar 1954
Da.
Oct.
13
11
1
2
3
e
. 9
10
11
12
21
25
28
11
11
11
11
11
25
26
28
29
171
11
29
28
26
25
26
lOO
79'
008
512
7.'
H' 1'
,."
H
18
17
18
19
20
12
11
9.6
...
28
25
25
25
25
9.0
7.S
S.O
7.0
6.5
2B
>7
'S'
115
2.
27
H
7.0
lli
151
9 .
21
22
2S
25
.,
. 5
29
'0
II
6
10
'l'ata
""an
cr.1II
In .
S15.7
10.2
0 . 0081
0.007
~
1.120
7<6
4,390
', 000
3.UO
'.760
5 , 590
2. 4GO
2,320
2.110
1 ,. 680
1,000
2.870
2.310
1,900
1.750
2,000
30'
270
288
292
"0
, . 540
3,380
2.650
111,770
' . 710
2,030
1121.'0
2 ,030
2 . 130
710
1 . 110
222 10,100
212 10,900
21. "'fO 000
208 ~
202
9,UO
1,>00
1.240
1,lSO
1.200
1.610
2 , 030
1,040
1,720
18,:500
20,000
~
4, 590
2~260
~,170
2,500
2,120
2,220
2,310
~
2 ,490
',BOO
1,950
1 , 920
2,410
1,b30
5,580
',700
' , 150
2,690
2 ,900
_.. __
.......
3' , 151 157,380

1 ,220
5,071
3.03
0.72'
0.76
'.SO
82 , 360
2,745
1.8'
1.8'
1 . 530
....--._-_.
-._-..
M1n
M1n
M!2
2 .410
2,160
Z,110
' ,600
2 ,9BO
3,200
',980
3,580
2 ,7SO
3,030
" 0
.80
ll2
1.310
970
1 ." 0
1.490
l,8!10
3 . 180
2 .620
',2'0
3,UO
3,630
52.
1.950
8,962 16',08'
289
5,293
0 .173
.1.18
0.20
3 . 65
Calend.ar year 1953. Mu

V.ter , .... 1963-54: Mu
7, " 0
4 ,810
113 ,800
3 ,620
4. 100
316
878
1.400
190
169
160
166
11178
170
. CO
.OB
1,030
902
824
1,210
1,070
927
.'2
788
iUgg
12 . 700
18,900
1G,5OO
- 10,:500
.5,330
~6.
72
280
J un.
2 ,~OO
1,910
12.100
.,.
Ma.
2,390
2,OeO
1.820
1,640
1,530
1,500
1, ' 50
111,360
698
626
548
508
...
151
130
l,.'B
SO . 8
0.030
0.03
III
124
121
"TO'
.?
........
13.
Apr.
9 , 700
lZ ,600
1 ,140
157
1<5
""r.
2 , 130
181
.1
57
55
2<
'ab.
110
98
8&
75
6'
9.6
9.6
10
..
Jan.
Nov.
G.1
Htan
6. 1
Ketan
~,"'0
2,150
1, 488
l,SiO
2,"50
1.970
1,560
1,300
1,100
."
'\1&.
July
S.f.2
184
208
222
.20
292
ill
29.
0
H.
121
H2
118
90
70
."
70
1>
11
' 10
.!
i!
68
'n'
......a
1,556
51.9
0.0!1
O.OS
ill.
,.,
ill.
HZ
SS , 912
2,06'
1.23
lo U
18.7'3
826
0 .11'
0.42
6,142
217
0.130
0.15
3,593
10.
0.065
0.08
c rill
1.29
0,889
In.
In.
17
61
108
III
1M
160
IS'
11.
115
100
' BO
6B
Cr..
In'
88
115
1
566
. 20
29.
219
17B
*154
17
H
1<
1>
90
76
8le
722
6>8
580
2BB
2.'
222
28
28
2,
21
lOS
160
598
'GO
>72
112
12
66
105
H5
178
,.,
.2<
.
,.
90
100
98
2"
~
1 . 260
65
.5
22'6
7.
s.pt.
11.
l!Jl
I!.
127
130
2.
.0
105
1>2
151
U8
02
75
17.'3
12.08
FIG. 4-18. Sample page from U.S. Geological Survey Water supply Paper.
73
74
STREAMFLOW
ably, flood bydrographs are plotted by computing instantaneous flow

values from the water-stage recorder chart. A sufficient number of
values should be plotted to indicate adequately all significant changes
in the slope of the hydrograph. The plotting scales will vary with the
problem-cubic feet per second and minutes on the smallest basins to
Instantaneous hydroqraph
(0)
Dote
Instantaneous hydroqroph
J:
u:::
Mean da/ly flow
(0 )
Dote
FIG. 4-19. The relationship between instantaneous and mean doily flows (upper and lower
charts typical of small and large streams, respectively).
thousands of cubic feet per second and hours or days on very large
basins. The shape of the hydrograph is determined by the scales used,
and (in any particular study) it is good practice to use the same scales for
all floods on a given basin. In order to obtain scales adequate for low
flows it is sometimes necessary to use a folded scale for very high flows
(Fig. 4-17) .
4-15. Mean daily flows. Streamflow data are usually publi hed in
t.he form of melUl daily flows (Fig. 4-18). This is the average discharge
rate in cubic feet per second for the period from midnight to midnight.
75
MEAN DAILY FLOWS
Since 1945 the U. . Geological Survey has included the magnitude and
time of occurrence of all significant flood peaks. On large stream this
form of publication is quite satisfactory, but on small streams it leaves
some thin!! to be desired. The picture pre ented by mean daily flows
depends on the chance relation between time of storm occurrence
and fixed clock hours (Fig. 4-19). On large streams the maximum
instantaneous flow may be only slightly higher than the maximum mean
1.0
I 1// I 1/yJA--/
'hif fj_7- 1.Iv V/
/
0.9
~
1;)
08
0.7
,~
J /LI // 'rf
r f
.g 0.5
C>
0,4
'00.3
VI' ",
IV~cFy '~L
-- ~
l,/v
vv
r.oQ_ ,
\ \J
1,2
t--
1.3
..........
)'I'-.1---r1_lL
\
\ Ix r---._
.2
1\ \
'0
a: 0,2
0,\
o
o
1/ / ~~
1/
~ 0,6
"'"
~
0,1
0.2
0.3
r7
Us
0.4
0.5
I--
0.6
0,7
0.8
0,9
\.0
Ratio of succeeding day to maximum dey
FIG. 4-20. Peak discharge and time of crest in relation to doily mean discharge. Labels
on curves indicate clock time of crest on maximum day and ratio of instantaneous peak
Aow to maximum daily discharge. (U.S. Geological Survey. )
daily flow. On small streams the maximum instantaneous flow is usually much greater than the highest mean daily flow. Wherever possible,
the hydrologist should secure copies of the recorder charts and work with
a bydrograph of instantaneous flow when dealing with small basins. If
this is not feasible, Fig. 4-20 may be used as a guide to estimate the
peak flow and time of peak. The figure is an average relationship 1
developed from data for stations all over the United States and cannot
be expected to give exact results. On very small basins where a complete
1
W. B. Langbeiu, Peak Discharge from Daily Records, U.S. Geol. Survey Water
Re80urces Bull., p. 145, Aug. 10, 1944.
76
STREAMFLOW
rise and fall of a storm hydrograph may take only a fraction of a day,
use of the recorder charts is essential if one wishes to study rates of flow.
4-16. Adju t m ent of strea m flow data. The streamflow data publish d by the U .. Geologi al urvey are carefully reviewed and adjusted
for errors resulting from instrumental and observational deficiencies
until they are as accurate a presentation of the flow as it is possible to
make. For a number of reasons, the published flow may not repre ent
the data actually required by the analyst. The location of the station
may have changed during the period of record with a resultant change
in drainage area and, hence, volume and rate of flow. In this case an
adjustment of the record is possible by use of the double-mass curve
( ec. 3-7). The base for the double-mass curve may be either flow at
OIle or more gaging stations that have not been moved or average precipitation at a number of stations in the area. It should be emphasized that
the doubl -mass- 'Ufve method implies 11 relationship of the form q = kP,
which may not be correct if precipitation is used as a base. A more
effective procedure is to develop 0. relationship between precipitation and
runoff and make a double-mass curve of observed streamflow vs. runoff
as estimated from this relation (Sec. 8-13).
Man-made works-storage reservoirs, diversions, levees, etc.-cause
changes in either total flow volume or rate of fl w or both. An analysis
of the effects on the record at a given station requil'es a careful search to
determine the number and size of reservoirs and the number and quantity
of diversions and the date of their construction. Many small diversions
may be unmeasured, and estimates of the flow diverted must be based
on electric-power consumption of pumps, capacity of pump equipment,
duration of pumping, or conduit capacities for gravity diversions.
Diversions for irrigation may he estimated from known irrigable or
irrigated acr age and estimated unit water requirements (Sec. 5-13).
The adjustment of the streamflow r cord for the effect of reservoirs or
diversions on flow volume requires the addition of the net change in
storage and/or the total diversion to the reported total flow. It may
be nec ssary to consider also channel losses and losses by evaporation
from the reservoirs.
Adju tment of hort-perjod or instantaneous flow rates for the effect
of storage or diver ion is a much more complex problem. Levees,
channel improvem nt) and similar works may a1 0 affect flow rate. In
some instances pumping of groundwater has mo.rked1y reduced low
flows, as has also the con truction of stock ponds. l
or recti on for the
effect of storage or diversion on flow rates is made by adding the rate of
change of torage or the diversion rate to the ohs rved flows. In addition,
1 R. C. Culler and H. V. Peterson, Effect of Stock Reservoirs on Runoff in thll
Cheyenne River Basin above Angostura Dam, U.S. Geol. Survey eire. 223, 1953.
77
78
STREAMFLOW
it may be necessary to use storage-routing techniques ( hap. 10) to

correct for the effect of channel torage h twe n the re 'ervoir or diversion
point and the gaging station.
hannel improvement and levee work
affect flow through their effects on the channel storage of the stream.
Unless it is po sible to e tablish "before" and "after" correlations with
some station outside the influence of the channel works, corrections mu t
be made entirely by storage-routing methods.
4-17. Mean annu al runoff. Figure 4-21 is a map of mean annual
runoff in the United tates. In addition to the problems of representing
any highly variable element by i opleths drawn on the basis of limited
data, there are other problems peculiar to treamflow. The record of
flow at a gaging station represents the integrated runoff for the entire
basin above the station. In many areas the production of runoff is not
uniform over the ba in. The average annual runoff of the Missouri
River at Omaha is 10,800,000 aCt'e-ft or 1.2 in. over the 322,800-sq mi
area tributary to this station. The headwater areas produce much
orwin Springs = 15.6 in.)
heavier runoff (Yellowstone River at
whereas other large ar as must average below the mean for Omaha. In
addition, storage and diversion further complicate the picture. Figure
4-21 was prepared by the Geological SurveY,l u ing nIl available information, and repre ents the most reliable map available for the United States.
uch a map is intended to pre ent a genern.l picture of geographical variations in runoff. It cannot po ibly show tine d tail and 'hould be used
only for general information and preliminary studies. It should not be
uch
u ed as a source of information for a pccific de ign problem.
information can be developed more reliably through u e of actual streamflow data and the analytical techniques dis us ed in the chapters which
follow.
4-18. Streamflo w variations. The normal or average values of
runoff serve an important purpo e but they do not disclose all the
pertinent information concerning the bydrology of an area. Especially
significant are the variations of streamflow about this normaL These
variations include the following:
1. Variations in total runoff from year to year.
2. Variations of daily rates of runoff throughout the year.
3. Sea. onal variations in runoff.
To some extent these variations are regional characteri tics, but the
size of the drainage area is invariably a factor. Figure 4-22 shows the
ratio of the maximum annual runoff to the mean annual runoff a a
function of drainage area for a numb r of stations in the Upper Ohio
River Basin having at lea t 20 yr of .record. On the same chart is a
curve showing the ratio of the maximum flood of record to the average
1
W. B. Langbein and others, Annual Runoff in the United States, U.S. fkol . S'Urvell
eire. 52, June, 1949.
RECORD PEAK FLOWS
79
Bow. In the first in tance, little effect of area is evident, but in the
second ell. e substantially higher ratio are as ociated with the maIler
bu ins. The relation hip are not perfect, for ba in hape, geology, and
climatic exposure are also ignificl1nt factors. The map of Fig. 4-23
indicates the variation of these two ratios over the nited tates. To
minimize the effect of are:)', ratios are given for stations having drainage
areas between 1000 and 2000 sq mi. High ratios are generally located
in arid regions with low normal annual runoff.
Area in square miles

FIG. 4-22. Ratio of maximum annual runoff to average annual runoff and maximum flow
of record to mean flow for stations in the Upper Ohio River Basin.
Median monthly runoff for selected stations is shown in Fig. 4-24.

Here again there are significant regional differences. A marked summer
dry season is charapteristic of the Pacific outhwest and is an important
factor in the need for irrigation in that area. In contrast, the low-flow
season in the northern tier of states i during the winter months when
precipitation is largely in the form of snow. In the eastern and southeastern states runoff is more uniformly di tributed throughout the year
as a re 'ult of a more uniform precipitation distribution (Fig. 3-14).
4-19. Record peak flows. Table 4-1 summarizes outstanding flood
peaks in the various parts of the United tates. The flow in cubic feet
per second and cubic feet per econd pe}.' quare mile, date of the peak,
area of the basin, and Myers rating l are given. Stations are selected
1 Myers rating is the ratio of the peak flow in cubic feet per second to 100 VA.
where A is the drainage area in square miles. The Myers rating is sometimes considered to be a regional constant (Sec. 9-10).
,~
o
.t;
.;
'j
,5
~
c::
III
80
81
..
.... ................
........
.......
000
'0
Q
CD CD CD
I!
~:.
'"
!?
00
:!S~~~~~;;;g
--..00._0-0_
~ 's
..( ~
NM....
O'........
'"
..
u
J;
0
'"""a
--
II
0..
.r
u
000
000
"'cr> ....
"'''' .....
'j(
II
...: 0 '"
'0
II
:;
Q,
C)
II)
u
:;:
c:
v'l
Jl
"0
c:
~Z
o
u
~
a
a
.t:
II
II)
:, '"u
,
, E
:'0
,0..
.~
...a
a:
"',.,
=i;;;S!;~l!;
"'0'"
-"N'"'
S~~~~:;:8:R:~~~~~o-.
('04----- -
c
o
'!II
.....
'
"~
83
8~8
~. o. O. "\
<oO<oONM
__ N;-
11
...........
..... _
.."NC:OO
N N~OOCOO"
- _ _ .... _ _
_
-
~::r;;!!~~o~~~
- --
- - N N M C G q - ; _ N..
\t')<OCOOMO-t.nNMM
"ONOMII'lCOO_O_
-MMON_N_
gggogggg8g
-.'''1
g.M."lO.q qO.
M..ot.n.."OCO ..... OON
N
___ _
NM""'N..o"O<o()NN..,.
II
Il.
Q.
;:)
"0
o
>o
...c:
!l
"0
"
:J:
""-"OMO
___ N
__
MM-o."IIIIt
....
"'?
ONO"lllltOMCI)"
OO"lo.COM- _
_ NNMOoQ)CO
NM.. ...;
C"\C"')_"'O_C")...,.
~o .....................................
CO') N
.... '" CD N 0co ........ co
.,
I ......
-0
a
- - - - ................,
N-NO---
co co c:o-o ..... co ..... ....,

M
....... MMC"')MN
"0
... ..,"..,0
--(00"'-"'"
..,-~..oo
--
o~~~~:g~&tl
....
0..
0..
::J I- - - -0
c:
a
D
era
'"o~
u:
...It
oQ)
Il..
...
-0
II
Q.
8
Q)
DI::
eD
II
" "
ccc:
.5
;;
.,
CD
0..
.Sl
01)
~D
cc:
a..
.,
o0 0 0
.,
r:
'u
....D
87
PROBLEMS
in such a. way that the tabulated data. repre ent th se CIl e in which the
peak flow exc ed 75 p r cent of the maximum r orded flow rate for
orresponding drainage area in each region.
BrBLlOGRAPHY
Corbett, D. M., and otllers: tream-gaging Proc durc, U.S. Geol. Suroey Waler-8upply
Paper S88, 1945.
trcam Flow," Wiley, N w York, 1943.
Grover, N. C., and A. W. Harrington:
King, H. W., and E. F. Brater: "Handbook of Hydraulics," 4th d., McGraw-Rill,
New York, 1954.
Linsley, R. K., M. A. Kohler, and J . L. H. Paulhus: "Applied Hydrology," Chap. 9,
pp. 182- 242, McGraw-Hill, New York, 1949.
(l
DATA
OUR E
Th main sources of sl,r amRow data are th Water-supply Paper8 published annually
by the U.S. Geological urvey and the Daily River Sta(Je8 published by the U.S.
Weather Bureau. More detai.lod information may be obtained by writing to itber
of th so agencies or by contacting their field offices located in principal ciLi s throughout the country.
The U. . Corps of Engineers, U.S. Bureau of Reclamation, and U.S. oil Cons rvation Service make occasional observations of flow. Information from tb s sources
can usually be obtained from the nearest field office. Many statcs have agencics which
ma.ke or cooperate in streamflow measurement.
PROBLEM
4-]. Compute the streamflow for tbe mell.Surement data below. Take the meter
rating from 'q . (4-1) with a - 0.1 and b - 2.2. (Note th!1t the rules governing
number of s etion!! have b n ignor d to reduce computation.)
Distance from
bank, ft
Depth, Meter depth,

ft
ft
---
2
4
1
3 .5
5 .2
6.3
11
4.4
13
15
17
2 .2
0 .8
0
Revo- Time,
lutions
seo
0 .6
2 .8
0.7
4 .2
1.0
5 .0
1.3
3 .5
0 .9
10
22
35
28
40
1.3
22
12
0 .5
32
45
28
33
50
55
52
53
58
58
60
45
46
50
49
4.2. The ~able tbat follows measures discharge, base stage, and stage at an auxiliary !J;lIge 2000 ft downstream. Develop a slope-stage-dischargo relationship from
these data. Compute the average error of tb rating, using tL tablllaled data.
What is the esLimated discharge for base and auxiliary stages of 25.00 and 24.20 ft,
respectively?
STREAMFLOW
88
Discharge, Auxiliary stage,
eCs
Ct
Base stage,
ft
14 .02
23.80
17 .70
24 .60
20.40
17 .00
1 . 65
26.40
22 . 20
16 . 20
21 . 10
25.60
23 . 20
2,400
29,600
21,200
85,500
28,200
7,400
34,000
55,000
74,200
9,550
43,500
84,000
93 ,500
13 .00
23 . 25
16 .60
23 . 55
19 . 55
16.40
17.50
25 .70
21 .00
15 .30
20 . 13
24 .60
21.95
4-3. Given below are data for a station rating curve. Extend the relation and
estimate th flow at a stage of 14.5 ft by both the logarithmic and A "';75 methods.
Stag,
ft
Area,
sq ft
Depth,
it
Discharge,
ds
tagc,
ft
Arca,
sq ft
1. 72
3 . 47
4 . 26
5 .61
6.70
7.80
9 . 21
14 .50
2G3
1200
1790
2380
3280
3960
5000
8200
1.5
2 .1
3 .2
4 .G
5.2
5 .7
G. 1
9 .0
1,020
4,900
7,700
10,700
15,100
19,000
25,000
2 .50
4.02
5 .08
5.98
6 .83
8.75
9.90
674
1570
2150
2910
3420
4820
5250
1)
pth,
ft
Discharg ,
cfs
1 .8
2.8
3 .9
4.. 9
5.46 .0
6 .5
2,700
6,600
9,450
13,100
16,100
24,100
27,300
4-4. Wllat volume is represented by 1.43 in. of runoff from a basin of 254 sq mi'(
Giv answer in cubic feet, second-foot-days, and acre-feet.
4-5. Given below are the daily mean flows in cubic feet per second at a gaging station for a period of 5 days. Wllat is th mean flow rate for the period in cubic feet p r
s ond? Wllat is the total disoharge during the period in second-foot-days? AcreI et? If the drainage area is 756 sq mi, what is th runoff volume in inches?
Day . . . ......... . '1
Flow, cis .. ...... . .
700
2
4800
3 \ 4 \_
5
3100
2020
1310
4-6. Using the data of Prob. 4-5 and the relationship of Fig. 4-20, estimate the
p ak rate of flow and the time of peak.
4-7. Obtain a copy 01 a U.S. Geological Survey Water-supply Paper, and for some
str am in your area determine the ma.ximu.m flow of record in cubic feet per second
and in cubic eet p r econd por square mile. The paper will give the mean annual
PROBLEMS
89
How in cubic feet per second. Find the average annual runoff in acre-feet, secondfoot-days, and inches.
4-8. For some stream sel cted by your instructor dctermine the mean flow for
each month on th basis of 10 yr of record. What per cent of the total llJUlual .Bow
occurs each month? Compare th s p ercentages with the normal monthly distribution of pr cipitation. Can you explain the differences?
4-9. For thr basins in your area find the ratio of the maximum peak flow of
record to the average daily flow moo. Is th TO an apparent r lationship , itb drainage area? Can you explain the differences?
4-10. Compare the data in Table 4-1 for the several regions of tbe country. What
regional differences do you sec? Look for such things as date of floods, relativ
magnitude of flood peaks, apparent areal extent of {ioods, etc. Can you explain the
differences?
5
EVAPORATION AND TRANSPIRATION
This chapter discus es that phase of the hydrologic cycle in which

precipitation reaching the earth's surface is returned to the atmosphere
as vapor. Of the precipitation falling earthward, a portion evaporates
efore reaching the ground.
ince the hydrologist measures precipitation just a few feet above the ground, evaporation from raindrops is of
no practical concern. Likewise, evaporation from the oceans lies beyond
his scope of direct interest. Precipitation caught on vegetation (interception) eventually evaporates, and the quantity of water actually
reaching the soil surface is correspondingly redu ed below that observed
in a precipitation gage. Other evaporative mechanisms to be considered
I1re transpiration by plants and evaporation from soil, snow, and free
water surfaces (lakes, reservoirs, treams, and depressions).
Anticipated evaporation is a decisive element in the design of reservoirs
to be constructed in arid regions. Ten re ervoirs the size of Lake Mead
would evaporate virtually the entire flow of the Colorado River in a
normal year, and normal evaporation from Lake Mead alone is equivalent
to almo t one-third of the minimum annual inflow to the reservoir.
Evaporation and transpiration are indi ative of changes in the moisture
deficiency of a basin and, in this capacity, are sometimes u ed to estimate
torm runoff in the preparation of river for ca ts. E timates of these
factors are al 0 used in the determination of water- upply requirements
of proposed irrigation projects.
EVAPORATION
Although there is always continuous exchange of water molecules to

and from the atmosphere, the hydrologic definition of evaporation is
re tricted to the net rate of vapor transport to the atmo phere. This
change in state requires an exchange of approximately 600 cal for each
gram of water evaporated ( ec. 2-9). If the temperature of the surface
is to be maintained, the e large quantities of heat must be supplied by
90
FACTORS CONTROLLING THE EVAPORATION PROCESS
Pl
radiation and conduction from the overlying air or at th expen e of

energy stored b low the surface.
5-1. Factors controlling the evaporation proce . Rate of
evaporation vary, depending on certain meteorological faotors and the
nature of the evaporating surfaoe. Muoh of the en uing disoussion
of meteorological factors is couched ill terms of evaporation from a fre
water surface. The extent to which the material i applicable to other
surfaces of i,ntere t in hydrology i believ d to be self-evident.
Meteorological factors. If natural evaporl1tion i viewed as an energyexchange proce s, it can be demon trated that radiation is by far the
most important single factor and that the term "solar evaporation" is
basically applicable. On the other hand, theory and wind-tunnel experiments have shown that the rate of evaporation from water of sp cified
temperature is proportional to wind speed and is highly depend nt on the
vl1por pres ure of the overlying air. How are these two onclusion to Of.
reconciled ? In essence, it can be said that water temperature i' not
independ nt of wind speed and vapor pre ure. If radiation exchange
and all other meteorological elements were to remain constant over a
shallow lake for an appreciable time, the water temperature and the
evaporation rate would become constant. If the wind speed were then
suddenly doubled, the evaporation rate would al 0 double momentarily.
This increa ed rate of evaporation would imm diately begin to extract
heat from the water at a more rapid rate than it could be replaced by
radiation and conduction. The water liemperature would approach a
new, lower equilibrium value, and evaporation would diminish accordingly. On a long-term basis, a change of 10 per cent in wind speed will
change evaporation only 1 to 3 per cent, depending on other meteorological factor. In deep lakes with capacity for considerable h at storage,
sudden hanges in wind and humidity are more effective for longer periods
of time; heat into or from storage assists in balancing the energy demands.
Thus excessive evaporation during a dry, windy week can, by utilization
of stored energy, reduce evaporation which would otherwise have occurred
in sub equent weeks.
The relative effect of controlling meteorological factors is difficult. t o
evaluate at best, and any conclusions must be qualified in terms of the
time period onsidered. Evaluation of the relative importance of
meteorological factors without an understanding of the subsequent discu sion on energy-budget and mas -transfer equations would be placing
the "cart before the horse." It can be stated, however, that the rate of
evaporation is influenced by solar radiation, air temperature, vapor
pre sure, wind, and possibly atmospheric pressure. Since solar radiation
is an important factor, evaporation varies with latitude, season, time
of day, and sky condition.
92
Nature of the evaporating surface. All surfaces exposed to precipitation,

such as vegetation, buildings, and paved streets, are potentially evaporation surface .
inc the rate of evaporation during rainy periods is
small, the quantity of torm rainfall dispo ed of in thi manner is es 'entially limited to that required to saturate the surface. Although this
evaporation is appreciable on o.n annual basis, it is seldom evaluated
separately but i consid r d part of over-all evaporation and transpiration.
The rate of evaporation from a saturated soil surface is approximately
the same as that from an adjacent water surface of the arne temperature.
As the soil begin to dry, evaporation decreases and its temperature
rises to maintain the energy balance. Eventually, evaporation virtually
oeases since there is no effective mechanism for transporting water from
appreciable depths except when the surface layers are frozen. Thus
the rato of evaporation from soil surfaces is limited by the availability
of water, or the evaporation opportunity.
Evaporation from snow and ice constitutes a special problem since
the melting point (32F) lies within the range of temperatures normally
experienced. Evaporation can occur only when the vapor pressure of
the air is less than that of the snow surface [Eq. (5-9) ], i.e., only when the
dewpoint i lower than the temperature of the snow. Evaporation from
snow is mall relative to that from liquid water (with the same spread
betw en surface and dewpoint. temperature ) because of the curvature
in the aturation-vapor-pressure vs. temperature relation. With asnow
temperature of 30F and a dewpoint of 20F the evaporation rate is only
one-fifth that from a water surface at 80F when the dewpoint is 70F,
with the same wind speed assumed in both ca es. Moreover, it requires
more heat to evaporate snow than water; at 0 the latent heat of sublimation (fu ion plus vaporization) is 677 cal/g and the heat of vaporization is 597. The e and other con iderations lead to the conclusion that,
with temperatures much above fr ezing, the rate of snowmelt must
exceed evaporation. The impression that a chinook,! or foehn, is conducive to xce ive evaporation from the snow cover i fallacious. The
dewpoint increa es downslope, and evaporation from the snow must
ceo. e when the dewpoint ri es to 32F. Reasonable a umptions yield
an upper limit of about 0.2 in. (water equivalent) per day for evaporation
from a snow urface.
Akin to the above discussion concerning the nature of the surface is
the effect of water quality on evaporation rate. The eff ct of salinity,
or dissolved solid, is brought about by the r duced vapor pressure of the
solution. The vapol' pressure of ea water (35,000 ppm di solved alts)
is about 2 per cent les than that of pure water at the arne temperature.
The reduction in evaporation is less than that indicated by the change in
1
A chinook is a downslope wind in which the air is warmed by adiabatic heating.
WATER-BUDGET DETERMINATIONS OF RESERVOIR EVAPORATION
93
vapor pre ure [Eq. (5-9)] beau with reduc d vaporation th re is an

incr ase in water tempern.ture which partially offs t the vapor-pr ure
reduction. 1 Even in the co. e of ea water, th reduction in vaporation
i never in exces of a few p r ceut (over an extend d p riod of time), so
that salinity effects can be neglected in the estimation of res rvoir
evaporation.
5-2. Water-budg t determination
of r . ervoir evaporation.
The direct measurement of vapomtion under field conditions is not

feasible, at least not in the s nse that on i abl to measure river stage,
di charg ,etc. As a consequence, a variety of techniques hav b n
derived for determining, or estimating, vapor tran port from water
surfaces. The mo t obvious approach involves the maintenance of a
water budget. As uming that storage S, surface inflow I, surface outflow
0, subsurface s epage Oq, and precipitation P 'an be measured, evaporation E can b computed from
(5-1)
This approach is simple in theory, but appli cation rarely produces r liable
results since all errors in m asuring outflow, inflow, and change in storage
are reflected directly in the computed evaporation.
Of tho e factors required, seepage i usually the most difficult to
evaluate since it must be estimated indirectly from measurements of
ground water levels, permeability, tc. If seepage approaches or exceed,
evaporation, reliable evaporation determinations by this method are
quite unlikely. Recent studies indicate that both se page and evaporation can be e timated by simultaneous solution of the water-budget and
mass-tran fer equations. 2 If a stage-seepage relation can be derived,
the water budget can be applied on a continuing basi..
The determination of rainfall generally does not represent a major
ob tacle except po sibly where a lake is surrounded by very rugged
terrain or where precipitation is r latively much gr ater than evaporation.
Snowfall, on the other halld, often renders the water budget totally
unreliable for short periods of time. A lake tends to trap blowing snow,
and the quantity added to a small lake may be several times as great
as that observed in the gage.
Water-stage recorders are sufficiently precise for determining the
storage changes provided that the stage-area relationship is accurately
established. Variations in bank storage are sometimes an important
source of error in monthly computations but can usually be neglected in
1 G. E. Harh ck, Jr., The Effcct of Salinity on Evaporation, U.S . Gool. Survey
Profess. Paper 272-A, 1955.
2 W. B. Langbein, C. a Hains, and R. C. Culler, Hydrology of Stock-water Reser'l"oirs in Arizona, U.S. Geol. Survey Cire. 110, 1951.
94
EVAPOR.ATlON AND TR.ANSPIRATlON
estimates of annual evaporation.

imilarly, expansion or contraction of
stored water with large temperature changes can introduce appreciable
errors. At Lake Hefner, Oklahoma, J corrections as large a 0.4 in./month
were required as a result of this elTect.
The relative effect of errors in the surface-inflow and outflow terms
varies considerably from lake to lake, depending upon the extent of
un gaged areas, the rc1in.bility of the rating curves, and the relative magnitude of flows with r spect to evaporation. Determinations of streamflow to within 5 per cent are normally considered" excellent, 11 and corresponding evaporation errors may be expected in an off-channel reservoir
without appreciable outflow. If the quantity of water passing through
a reservoir is large in comparison with evaporation losses, water-budget
results are of questionable accuracy.
Under somewhat idealized conditions, it was found that daily evaporation from Lake Hefner, Oklahoma, could be reliably computed from a
water budget; results were considered to be within 5 per cent one-third
of the time and within 10 per cent two-thirds of the time. It hould be
emphasized that Lake Hefner was sele'ted, after a survey of more than
100 lakes and reservoirs 2 in the West, as one of the three or four best
meeting water-budget requirements. The requirements are not nearly
so stringent for estimates of annual or mean annual reservoir evaporation,
and reasonable estimates can be made for a number of reservoir in the
United States.
5-3. Energy-budget determination
of reservoir evaporation.
The energy-budget approach, like the water budget, employs a continuity

equation and solves for evaporation as the re idual requir d to maintain
a balance. Although the continuity equation in thi case i one of
energy, an approximate water budget is required as well, since inBow,
outflow, and storage of water repr sent energy values which must be
considered in conj unction with the resp ctive temperature .3 Application of the energy budget has be n attempted by numerous investigators,4 with cases selected so as to minimize the effect of terms that could
Dot be evaluated. However, the Lake Hefner experiment is believed
to constitute the first test of the method with adequate control.
e Bibliography at th end of this chapter.
sO. E. Harbeck and others, Utility of Selocted Western Lakes and Reservoirs for
Water-loss Studies, U.S. Geol. Survey Circ. 103, March, 1951.
a E. R. Anderson, L. J. Anderson, and J. J . Marciano, A Review of Evaporation
Th ory and D v lopment of Instrumentation, U.S. Navy Elect)'onics Lab. Rept. 159,
pp. 41-42, February, 1950.
G. F. McEwen, Results of Evaporation Studies, Scripps lnst. Ocea.nog. Tech. Ser.,
Vol. 2, pp. 401-415, 1930.
B. Richardson, Evaporation as a Function of Insolation, Trans. ASCE, Vol. 95,
pp. 996- 1019, 1931.
ENERGY-BUDGET DETERMINATIONS OF RESERVOIR EVAPORATION
95
The energy budget for a lake or re ervoir may be expre s d as

Q. - Qr - Qb - Q" - Q.
= QB - Q.
(5-2)
where Q. is sun and sky short-wave radiation incident at the , ater

surface; Qr, reflected hort-wave radiation; Qb, net en rgy 10 t by th
water body through exch::mge of long-wave radiation with the atmosph re;
Q", ensible-heat transfer (conduction) to the atmo phere; Q., energy
used for evaporation; QB, the increase in energy tored in the water body;
Q., net energy advected 1 into the water body; and all are in calori per
square centimeter. Letting H. repr ent the latent heat of vaporization
and R the ratio of heat 10 s by conduction to heat 10 s by evaporation
(Bowen ratio), Eq. (5-2) becomes
+ Q. + R)
E = Q. - Qr - Qb
pH.(1
Q,
(5-3)
where E is the evaporation in centimeters and p is the density of water.

The Bowen rati0 2 can be computed from the equation
= 0.61 T. -
T" _]!_
e, - e" 1000
(5-4)
where p is the atmosph ric pre sure; T" and e", the temperature and
vapor pre sure of the air, respectively; T., the water-surface temp ratur ;
e., the saturation vapor pressure corre ponding to T.; and all tempern.tures
and pressures are in degrees centigrade and millibar..
ensible-heat transfer cannot be readily observed or computed, and the
Bowen ratio was conceived as a means of eliminating this term from the
energy-budget equation. The validity of the constant in Eq. (5-4) has
been the subject of much di cussion.3 Bowen found limiting values of
0.58 and 0.66, depending on the stability of the atmosphere, and concluded that 0.61 was applicable under normal atmo pheric conditions.
Using an independent approach, Pritchard (unpublished notes) derived
value of 0.57 and 0.66 for smooth and rough surfaces, respectively. At
Lake Hefner, Oklahoma, monthly values of the ratio [computed from
Eq. (5-4)] were found to vary from - ~ in February to X in November,
while the annual value was -0.03. It is obvious that one need not be
concerned over variations in the constant of Eq. (5-4) for annual computations. If the correct value is assumed to have been one of the limits
1 Net energy content of intlowing and outflowing water is termed advected energy.
11. S. Bowen, The Ratio of Heat Losee by Conduction and by Evaporation from
Any Water Surfac PhY8. Rev., Vol. 27, pp. 779-787, 1926.
a E. R. Anderson, L . .T. Anderson, and J . J. Marciano, A Review of Evaporation
Theory and Dev lopment of Instrumentation, U.S. N a.vy Electronics La.b. Rept. 159,
pp. 43-48, February, ]950.
determined by Bowen, the extreme error in monthly evaporation at

Lake Hefner would be only about 4 per cent.
In the application of Eq. (5-3), it is important that the radiation
exchange be accurately evaluated. Mo t routine observations from
established networks provide mea urements of incident short-wave radiation only, and prior to the development of radiometers,l radiation
exchange was nece arily estimated from empirical relations. 2 Radiometers can be designed to measure either the total incoming or net
radiation- the algebraic sum of incident and reflected sun and sky shortwave radiation, incident and reflected atmospheric long-wave radiation,
and long-wave radiation emitted by the water body. Another promising
approach to the determination of net radiation involves the maintenance
of an energy budget for an insulated pan, known as a Cummings radiation
integrator. 8 The assumption is made that incident and reflected solar
and long-wave radiation are the same for the pan as for an adjacent lake.
Having the sum of these items for the pan and computing the long-wave
radiation from the lake on the basi of its urface temperature, net radiation for the lake can be determined. Whereas net radiation is an independent fllctor in computing lake evaporation, it is the dependent factor
in the cllse of the integrator, plln evaporation being one of the observed
clements.
The energy advection and torage terms (Q. - Qo) of Eq. (5-3) are
omput d from an approximate water budget and temperatures of the
respective water volumes. Equation (5-1) can be written as
(5-5)
Variations in density are neglected and all terms arc expr s d in cubic
centimeters. The energy content p r gram of water is the product of
its specific heat and temperature. Assuming unity for the values of
density and specific heat,
Q. - Qo
(IT r + PT p
OTo - OuTu - ETB + 8 11\ - 8 2'1'2)
(5-6)
where 7'r, T p , etc., are the temperatures (0 ) of tb respective volumes

1 R. V. Dunkle et al., Non elective Radiometers for Hemispb rical Irradiation and
Net Radiation Interchange Measurements, Univ. Calif. Eng. Dept. Thermal Radiation
Project Rept. 9, October, 1949.
V. E. Suomi, M. Franssila, and N. F. Islitzer, An Improved Net-radiation Instrument, Am. Meteorol. Soc. J., Vol. 11, pp. 276-2 2, 1954.
t H. L. Penman, Natural Evaporation from Open Water, Bar Soil, and Grass,
Proc. Royal Soc. (London): A, Vol. 193, pp. 120-145, April, 1948.
O. E. Harbeck, Cummings Radiation Integrator, in Water-loss Investigations,
Vol. 1, Lake H fncr Studies, U.S. Gwl. Survey Profess. Paper 269, pp. 120-126, 1954
(reprint of U.S. Gwl. Survey Circ. 229, 1952).
MASS-TRANSFER DETERMINATIONS OF RESERVOIR EVAPORATION
97
of water, and the surface area of the lake A i introduced to convert

energies to unit of calories p r square centimeter. Equation (5-5)
should be balanced before solving Eq . (5-6), although approximat value
of the individual terms will uffice. The temperature of pr cipitation
can be taken as the wet-bulb temperature, s page temperatur a thn.t
of the water in the lowe t levels of the lake, and TEas the lake-surface
temperature. Advected energy and change in energy torage tend to
balance for most lakes, particularly over long period of time, and ar
frequently assumed to cancel when considering ann ual or m an annual
evaporation.
5-4. Mass- t ran fer determination
of re ervoir evaporation.
The theoretical development of turbulent-transport equation has followed two basic approache : the di 'continuou , or mixing length, concept
introduced by Prandtl and clunidt and the continuou mixing concept
of Taylor. An exten ive physical and mathematical review l of the two
approaches was prepared in advance of the Lake Hefner experiment, and
s number of the equations were tested at Lake Hefner and Lake Mead. 2
Eq uations derived by Sv rdrup and Sutton gave good results at Lak
Hefner but were con idered inadequate when applied to Lake Mead.
There is reason to believe, however, that the Thornthwaite-Holzman
equation would give sati factory results with instrumentation meeting
the exacting r quirements. A complete discussion of mas -transf r
theory is beyond the scope of this text, and, in view of the present status
of the approach, only the Thornthwaite-Holzman eq uation is presented.
Assuming an adiabatic atmosphere and logarithmic distribution of wind
speed and moisture in the vertical, the derived equation is
E = 833k 2(e l - e2)(v2 - VI)
(7' + 459.4) log. (,z2/ZI)2
where E is the evaporation in inche per hour, lc is von Karman's constant
(0.4), e is the vapor pressure in inch s of mercury, v is the wind speed in
miles per hour, and T is the mean temperature in degrees Fahrenheit of
the layer between the lower level Z I and upper level Z2. Since computed
evaporation is proportional to the small differences in wind and vapor
pressure at two levels near the surface, in trument requirements are not
easily satisfied under field conditions.
Numerous empirical formula 3 have been derived which express evapoIE. R. Anderson, L. J . Anderson, and J. J. Marciano, A Review of Evaporation
Th oryand Development of Instrumentation, U.S. Navy ElectroniC8 Lab. Rept. 159,
pp. 3-37, February, 1950.
II Water-loss Investigations, Lake Mead Studies, U.S. Oeol. Survey ProfeBs. Paper
298, pp. 29-37, 1958.
a R. K. Linsley, M. A. Kobler, andJ. L. H. Paulhus, It Applied Hydrology," pp. 165169, McGraw-Hill, New York, 1949.
98
ration as (l. function of atmospheric lements and which parallel the masstransfer approach in 'orne respects, The Meyer formula l for evaporation
in inches p r day is
E = c(e. - ea )
(1+ Ivo)
(5-8)
where e, and ea are the vapor pre sure of the water surface and overrunning air (in, of Hg) ann vi the wind pe d (mph). Meyer states that
the co fficient c has a value of about 0,36 when the formula is applied to
daily data for an ordinary lake, provided that the wind and humidity
observations are about 25 ft above the surface,
Several empirical equations have been derived from the data collected
at Lake Hefner:
E = 0,OO304(e, - e2)v,
E = 0,00241 (e. - es)vs
E = 0,00270(e. - e2)v,
(e2 and v, over lake)

(es and Vs over lake)
(e2 over lake, v, upwind) .
(5-9)
(5-10)
(5-11)
where E is lake evaporation in inches per day, vapor pres. ure .is in inches
of m rcury, wind i in miles p r day, and numerical subscripts de ignate
heights above the surface in meter, Equation (5-9) was found to yield
excellent results for Lake Mead, and there i good reason to believe that
it is of general applicability. Equation (5-10) yielded a satisfactory
value of annual evaporation for Lake Mead, but there was a seasonal
bias which appeared to b ofl'elated with stability.
Vapor pre sure of th air increase downwind across an open-water
urface; hence mass-transfer concepts have led to the belief that point
evaporation decrcases downwind. Sutton2 concluded that average depth
of evaporation from a circular water surface is proportional to the -0.11
power of its diameter under adiabatic conditions, and this functional
relation has been v rifled in wind-tunnel experiments. 3 The theory
assumes that water temperature and wind are unchanging downwind, and
this condition tends to prevail in a wind tunnel where solar radiation is
not a factor. Observations show that wind speed increases downwind
from the leading dge of a lake, however, and a con ideration of energy
con ervation requires that any imm diate reduction in evaporation rate
1 A. F. Meyer, "Evaporation from Lakes and Res rvoirs," Minnesota Resources
Commission, t. Palll, Minn., 1942.
2 E. R. Anderson, L. J. Anderson, and J . J. Marciano, A Review of Evaporation
Theory and Devclopment of Instrum<'.ntation, U.s. Navy Electronics Lab. Rept. 159,
p. 10, Febrllary, 1950.
H. Lettau and F. Dorffcl, Dor WassordampfUb rgang von oin r Nasseu Platte an
Stromende L\lft, Ann. H1Idrographie 11. Maritimen Meteorol., Vol. 64, pp , 342, 504.
1936.
ESTIMATION OF RESERVOIR EVAPORATION FROM PAN EVAPORATION
99
brought about by deeren ed vapor-pre sure gradi nt be f How d by an

increl1 e in water temperature. Although xp rimental data are pr s ntly
in ufficient to determine th magnitude of the " ize eff ct," Eq . (5-9)
and (5-10) can be applied to lak s ranging up to everal hundred square
miles in area without appreciable error in this re pect.
5-5. Estimation of re ervoir evapora tion from pan evaporation
and related meteorological data. The pan is undoubted ly th rno t
widely used evaporation in trument today, and it application in hydrologic de ign and operation i of long standing. Although criticism of the
pan approach may be justified on theor tical ground, the ratio of annual
lake-to-pan evaporation is quite con i tent, year by year, and does not
vary excessively from region to region.
Pan observations. There are three types of expo ures mployed for
pan in taUations-- unken, floating, and urface--and divergent views
persist a to the be t exposure. Burying the pan tend to eliminate
objectionable boundary effects, such a radiation on the side wall and
heat exchange between the atmo phere and the pan itself, but create
observational problems.
unken pans collect more trash; they are
difficult to install, clean, and repair; leak are not easily detected; and
height of vegetation adjacent to the pan is quite critical. Moreover,
appreciable heat exchange do take place between the pan and the
soil, depending on such factors as soil type, moisture content, and
vegetative cover. Rather than attempt the necessary observations
required to adju t for heat exchange with the soil, it would appear
advisable to use a large pan to minimize the relative effect.
The evaporation from a pan floating in a lake more nearly approximate
evaporation from the lake than that from an on-shore in tallation.
Observational difficulties are prevalent with floating pans-splashing
frequently renders the data unreliable--and installation and operational
expense is exce sive. Relatively f w such installations are now in existence, and, consequently, floating pans are not considered in the subsequent discussion.
Pans exposed above ground experience greater evaporation than sunken
pans, primarily becau e of the radiant energy intercepted by the side
waU. Moreover, en ible-heat transfer through the walls results in
geographical (climatological) variations in lake-to-pan ratios. While
both the e deficiencies can be eliminated by insulating the pan, this
approach is unduly expensive. It would appear more economical to
adjust data collected from an uninsulated pan. The principal advantages
of surface expo ur are economy and eo. e of installation, operation, and
maintenance.
Of the various sunken pans used, only three gained prominence in the
United States: the Young screened pan, the Colorado pan, and the
100
Bureau of Plant Industry (BPI) pan. The Young panl is 2 it in diameter, 3 ft deep, and i covered with J4-in.-mesh hardware cloth. The
screen modifie the pan coefficient (lake evaporation divided by pan
evaporation) to near unity, on an average, but the small ize of the pan
leads to an unstable coefficient, and the over-all effect of screening may be
adverse. The olorado pan is 3 ft squar and 18 in. deep. The BPI
FIG. 5-1. Class A evaporation station. Numbered items are as fallows: (1) thermam , ter
(2) evaporation pan and anemometer, (3) standard 8-inch nanrecarding precipitation gage, and (4) weighing-type recording precipitation gage. (U.S. Weather Bureau. )
~helter,
pan, 6 ft in diameter by 2 ft deep, provides by far the best index to lake

evaporation because of it size.
The standard Weather Bureau Class A pan is the most widely used
evaporation pan in the nited Stnte ; in 1956 records were published for
about 350 stations. It is of unpainted galvanized iron 4 ft in diameter
by 10 in. deep and is exposed on a wood frame in order that air may
circulate beneath the pan (Fig. 5-1) . It is filled to a depth of 8 in., and
instructions 2 require that it be refilled when the depth has fallen to 7 in.
Water-surface level is measured daily with a hook gage in a stilling well,
and evaporation is computed as the difference between observed levels,
adj u ted for any precipitation measured in a standard rain gage. At
many new installations, water is added each day to bring the level up to
I A. A. Young, Some Receot Evaporation Investigations, Trans. Am. Geoll,hys.
nion, VoL 28, pp. 279- 284, April, 19-17.
I Instructions for
limatologicll.l Observers, U.S. Weather Bur. Cire. B, 10th ed.,
November, 1952.
ESTIMATION OF RESERVOIR EVAPORATION FROM PAN EVAPORATIO N
101
a fixed point in the stilling well. Thi method as ure pc pcr water
level at all times.
Pan evaporation and meteorological factor. Many attempt hrw b Cll
made to derive reliabl r lation. betwe n pan evaporation and meteorological factors.l Obviou purposes to be served by uch relation ar a
follows:
1. To increa e our knowl dge of the phenomenon.
2. To e timate mis ing pan records (pans are 11 t operat d during
winter in areas where ice cover would 0 'cur much of the tim, und records
for days with snow or heavy rain ar frequently falla iou ).
3. To derive estimated data for tatioos at which pan ob rvations are
not made.
4. To test the reliability and repre entativene of ob erved data.
5. To aid the study of lake-pan relations.
ome of the relations which have been developed involve the sub titution
of air temperature for water temperature with a resultant sea onal and
geographic bias. Pan evaporation can be reliably estimated from empirical functions involving wind, dewpoint, and wat r temperature, but such
relations are of little practical value; wh n water temperature is observed,
pan evaporation is al 0 measured. Penman 2 has shown that the need
for water-temperature observations can be eliminated, however, through
simultaneous solution of an empirical rna s-transfer equation and one
expressing an energy balance. A swning the change in heu,t ..torage of
the water body and the heat conductcd through the containing walls to
be negligible, Penman derived the equation
+1 -y (Q,A + -yEa)
(5-12)
where A is the slope of the saturation-vapor-press~re vs. temperature

curve at the air temp rature T a; Ea is the evaporation given by an empirical mass-tran fer equation, assuming water temperature T, = Ta; Q" is
the net radiant-energy exchange expressed in the same units as evaporation E; and -y is defined by the Bowen ratio equation [Eq. (5-4)]
R =-y
T. - Ta
e. - ea
(5-13)
The Penman approach was applied to the composite record from a number of tations over th United tate to derive (Appendix A) the coaxial
1 R. K. Linsley, M. A. Kohler, and J. L. H. Paulhus, "Applied Hydrology," pp. 168169, McGraw-Hill, New York, 1949.
I H. L. Penman, Natural Evaporation from Open Water, Bare Soil, and Grass,
('roc. R,qyal Soc. (London): A, Vol. 193, pp. 120-145, 1948.
102
1\
1\ \ j\ ,\, 1\ .\ 1\\
,\ 1\' l\ \1\ ,\/\

\1\ ,\1\ l\ \
,\ 1\' l\ \1\ , \1\
\\ \ j\ \ 1\ \ \
1\-\ 1\
1\
1\ 1\ 1\
i
.;. I-- 1--11
-i I-- 1--11
-! I-- 1--1-
-! I-- I--jl
\. \ 1\\ .\ \1\ \1\
- I--
1\ 1\ \ \
..\ 1\\1\ \1\ \ \ -I
11\ 111\ \.'1
\ 1\' \ :\\1\ ,,~\
#
111\ '\ ~
-I~....
- ~\ \\ l-\ rt._ '\ \
'\~ \ \\ \ ~\ \ \1\
'. R ~.,,~
.\ \1\ \ \\ \ r-.\
"- "- "- ' . '- ~
-.I' \ 1\ \ \ 1\ \~
""- ~
1'.
i". r-...
/"--.
I".,\ \1\ \ 1\' \ 1\\
t'--- "r-. :--.. r-.... ~ r-.....
\.' \ ri \
r--- t- r-.~ t-.. t- ~ t--- h: r-: ~ t;::: ~
\ 1\ \ 1\
"
""
'" "
I--
'" .. '"
~
r-I-
I--
I\~.~!\~.) 1\ \
1\ 1\ \
\ 1\ \
'\ 1'\ 1
\ 1\ l\ ~\
'\
\!\' \\
\, ,,\ 1\ 1\\
1'\ \.
. ~ f"- Ill\. ~ l\\ \\
~:
::--. 1'\ 11\ t\\ \\\
~
l'\ ~\
""~ ~ \.\ I\'\'
"'-i ~ N
\
1\
I~ \
!'.
!- ~
":"."':,...
"h:
t'-. '" f"- 111\ 1\
i't--l'-,
~"t.
f""" l ~
t::-::~ 1::\ ~ 11\\
""
~~ ~ ~ ~
1\\)\\
;{
~ ~ ~ ~ ~
1\'
11\\ ~
I--
\J\N\\~~
1-
t-I--
~:"'l"'iiIla
~
.. -.!....!.uw;.,.......
" ""
~ ~
~
~
~
,I
I~
103
relation 1 shown in Fig. 5-2. It wa found that the value of 'Y for the
Class A pan ;s about two and one-half times the the or tical valu ,
because of the fact that evaporation i re tri ted to the water urfa e
while ensible-heat transfer al 0 occurs through the pan wall. To
compute pan evaporation, Ea i fir t d termin d from the upper lefthand chart of Fig. 5-2 by entering with value of air temperature, dewpoint" and wind. Equation (5-12) is solved by entering t.he other t.hrec
cw-ve families of Fig. 5-2 with air temperature, solar radiation, E a, and
air temperature. Although the correlation wa based on daily data,
TABLE 5- 1. V a r iation of Cla ss A Pa n Eva po ration with Rela ted Factors
Cale
Na.
Ta
(OF)
% errar /
FO change
in Ta
Td
(OF)
% errar/
Q,
O
F change
(Iy / day)
in Td
% errrx /
% errar /
v
E
% change (ml / day) % change (In. / day)
In Q,
in v
--_.
-1
2
3
4
5
6
7
8
9
10
I 919184
66
45
91
91
84
66
45
1.8
2. 3
3. 8
6.0
6.3
1.8
2.3
4.4
6.2
6.2
41
63
75
55
28
41
63
75
55
28
0.4
0.8
2.7
4.0
2.9
0.3
1.0
3.1
4.6
3.1
700
700
600
300
250
700
700
600
300
250
0.7
0.8
0. 9
0.6
0.3
0.6
0.7
0.8
0.5
0.3
50
50
SO
SO
SO
100
100
100
100
100
0.2
0. 1
0.1
0.2
0.3
0.3
0.2
0.2
0.4
0.5
0.51
0.46
0.28
0.12
0.09
0.60
0.52
0.3 1
0. 15
0.11
experience has shown that only minor errors result when monthly
evaporation (i.e., mean daily value for th month) is computed from
monthly averages of th daily values of the parameters.
There are only a limited number of locations where all the data required
for application of Fig. 5-2 are available; solar radiation is ob erved at only
about 60 stations in the United States. There are reasonably relil1bl
means of estimating this factor,2 however, and the other required elements
can usually be estimated accW'l1tely enough to provide adequate value
of monthly or normal monthly evaporation. Table 5-1 shows the magnitude of evaporation error resulting from error in each of the variou
factors, for selected meteorological situations. In view of the high
degree of correlation found to exist between computed and ob erved
1 M. A. Kohler, T. J. Nordenson, and W. E. Fox, Evaporation from Pans and
Lakes, U.S. Weather Bur. Re8earch Paper 38, May, 1955.
I R. W. Hamon, L. L. Weiss, and W. T. Wilson, Insola.tion as an Empirical Function of Daily Sunshine Duration, Monthly Weather Rev., Vol. 82, pp. 141- 146, June,
1954.
Sigmund Fritz a.nd T. H. MacDonald, Average Solar Radiation in the United
States, Heating and Ventilating, Vol. 46, pp. 61- 64, JulY. 1949.
1()4
evaporation, Table 5-1 is also indicative of the relative influence of the

various elements under typical meteorological situations.
It will be noted from Eqs. (5-4) and (5-13) that 'Y is directly propOltional to pressure and thus might be expected to cause a bin in computed
evaporation when Fig. 5-2 i applied through a wide range of elevation.
Although high- levation data are notably spar e, analysis of limited data
TABLE 5-2. Summary of Pan Coefficients (Lake-to-pan Ratios) as
Determined by Various Investigators
location
Vears of
record
Basis of
coefficient
Closs
A
8Plt
(sunken)
Colorado
(sunken)
Screenedt
(sunken)
- - --lake Hodges, Calif ...

Newell Reservoir, Canada
Ft. Collins, Colo. (85-ft dlam.)
Denver, Colo. .......
Denver, Colo. ....
Milford Expt. Sta., Utah .
Fullerton, Calif. ...........
lake Elsinore, Calif .......
Ft. Mcintosh, Texas .......
Red Bluff Reservoir, Texas ..
lake Okeechobee, Fla ....
lake Hefner, Okla . ....
Lake Mead, Arlz.-Nev .
1919-21
1919-25
1926- 28
1915-16
1916
1926-27
1936- 39
1939-41
1950
1939- 47
1940-46
1950-51
1952-53
June-Oct.
May- Sept.
April-Nov.
Mean annual
June-Oct.
May-Ocl.
Mean annual
Mean annual
Feb.- Dec.
Mean annual
Mean annual
Mean annual
Mean annual
....
....
0.70
0.67
....
0.67
0.77
0.77
0.72
0.68
0.81
0.69
0.60,[
. ...
....
0.96
0.95
0.79
0.94
0.94
....
....
. ...
0.91
0.89
....
0.9B
0.98
0 . 89
0.98
0 . 83
0.91
. ...
* For detailed Information on the derivation of the.e data, consult the lake Hefner and lake Mead
reports (see Bibliography).
t Bureau of Plant Industry pan, 6 ft in diameter and 2 ft deep.
t Pan 2 ft In diameter and 3 ft deep.
Computations based on assumption that evaporation from a sunken pan 12 ft in diameter and
3 ft deep [s equivalent to that from a lake for periods during which advected energy and stored
energy In the lake may be neglected.
~ for a pan at representative site.
for nIt Lak

ity, Utah, and Grand Junction, Colorado, indicates that
the relation can be used for elevations up to 5000 ft, mean sea level,
without appreciable error.
Pan coefficients. W~tter-budget, energy-budget, and rna s-transfer
techniques can be used to estimate evaporation from exi ting re ervoirs
and lakes. However, the e method are not directly applicao;e to
de ign problems, in' water-temperature data are required for their
use. Virtually all estimates of re ervoir evaporation, both for design
and operation, have been made by applying n coefficient (Table 5-2) to
observed or derived pan evaporation. Although too few determinations
have been made to appraise the approach accurately, assuming an annual
Class A pan coefficient of 0.70 for the lake included in the table would
lOS
result in a maximum difference of 12 per cent. Part-year coeffi ients

are more variable Lecau e n rgy storag in the lake can b appr iably
differ nt at the beginning and end of the period. Changes in heat
storage cause pronoun 'ed variation in monthly coefficients.
0.9 r--y-----r----,r---r---.---r----,---~--___, 0.9
0.7
f---l-- - - - t -
0.6 f---+---
0.5
i
I
1--+-----:7""':......t- - - I1--t------l 0.1

1
I
I
Elevation =1,000 ff above ms/
32 40
60
Elevation =10,000 ff above ms/
80
100 32 40
60
Lake water temperature, OF
80
FIG. 53. Portion of advected energy (into a lake) utilized for evapora tion.
0
100
(U.S. W eather
Bureau.)
Effects of advected energy on lake and pan. Observations demonstrate

that the sensible-heat transfer across the pan walls can be appreciable
for the Class A pan and that it may flow in either direction, depending
upon the pan and air temperatures. Since heat transfer through the
106
bottom of the re ervoir is essentially zero, the pan data require adj u tment.
imjlarly, the pan does not account for advection to and energy
storage in the reservoir.
Not all energy advected into a lake is dis ipated through increa ed
evaporation.
ince the increased energy re ults in higher water-surface
temperatures, radiation 10 sand ensible-heat exchange are al 0 affected.
It has been, hown that the portion a of advected energy utilized in the
~9
0.4
Q9
.I
...
I
I
Elevation = 1,000 ft above ms/
Elevation: fO,OOO f t above msl
100 32 40
60
1003
Pan water t emperatu re, of

FIG. 5 4. Portion of advected energy (into a Class A pan) utilized for evaporation.
Weather Bureau.)
llI.s.
evaporation process i approximately that given by Fig. 5-3.

ince a
varies with atmo pheric pr ure, two relation are shown, one for an
elevation of 1000 ft and the other for 10,000 ft, mean sea level. Since
advection can be ith r into or out of the lake and heat storage may
ith l' in rease or d cren during the p riod under consideration, an
adju tment should b mad for the difference between these two terms
(net adv ction). Thu , for compari on with pan evaporation, observed
lake evaporation should be adj u t d by adding the quantity a(Qc - Q.),
wh r Q6 is th chang in energy torage during th period and Q. is
the advection into th lake, both in equivalent inches of evaporation
( ee. 5-3).
Annual n t advection is u ually mull and may generally be neglected
in the d ign of r ervoir . If utflow i to be primarily from a low
107
level in a deep reservoir, however, this may not be the cn. . Av rag
n t adve tion at Lake Mead i equival nt to about on foot of evaporation
per year.
FIG. 5 5. Lake evaporation a s a function of Class A pan evaporation and he at transfe r

through the pan. (U.S. Weather Bureau.)
Figure 5-4, derived in much the same manner as Fig. 5-3, shows the
relative eff ct a" of net advection on pan evapomtion. Adv ction by
means or water added to the pa.n is normally unimportant, but advection
of sensible heat through the pan wall is sufficient to produce moderate
variation in the pan coefficient under varying climatic regimes. From the
Bowen-ratio concept and the empirical relation of Fig. 5-2, a function
can be derived for estimating heat transfer through the pan from observations of air and water temperature, wind movement, and atmospheric
108
pressure. Thi i th underlying relation of the chart in 'ig. 5-5 which

further ~\. sumes a 0.7 pan coefficient when air and water temperatures
are qual. Lake evaporation computed from the chart should be adjusted
for any appreciable net advection to the lake.
"1'heorctical" pan concept. Reliable data are notably limited, but
such data as are available indicate that the ratio of lake to Class A pan
evaporation is, for practical purpo e , 0.70, provided that
1. Net advection to the lake is negligible.
2. Net transfer of sensible heat through the pan is negligible.
3. Pan exposure is representative.
Figure 5-2 yields timate of evaporation from a Class A pan with
its boundo.ry los e ; that is, 'Y i two and one-half time the theoretical
value. If the theoretical value of 'Y i ub tituted into the relation,
computed values of evaporation hould corre pond to those which would
be observed in a "theoretical" pan which has the radiation characteristics .
of the Jas' A pan but which permits no s n ible-heat transfer through
th pan. On the basis of the data now available, it is evident that the
annual coefficient for this" theoretical" pan is near 0.70 and is essentially
indcp lldent of climatic variation. Figure 5-6 was derived from Fig. 5-2
by sub tituting the theoretical value of 'Y and assuming a pan coefficient
of 0.70, to provide a convenient mean of estimating reservoir evaporation.
Although Lhe relation is strictly applicable only for daily data, use of
monthly averages will, in general, cause no appreciable bias. Adjustment should be made, however, for adve tion to the reservoir.
5-6. ummary and appraisal of technique for e timating
re ervoir evaporation. When conclitions are such that satisfactory
results cannot be obtained by applying a water budget, evaporation from

an existing l' ervoir can be determined by eith r the empirical masstransfer or the energy-budget approach. Both instrumentation and
maintenanc of observations are xpensive for these two approaches,
and their wid pread u 'e may not be economically fea ible for some time.
The operation of a pan station (near the reservoir, but not so close as to
be materially affected by it) is relatively inexpensive and should provide
satisfactory estimate of annual reservoir evaporation (Fig. 5-5).
orne
reliability would be gained if supplem ntary data were obtained for
estimating net reservoir advection, but this item is frequently not of
great importance. If monthly or sea ono.l di tribution of annual evaporation i consid red necessary, net advection to the re ervoir must be
evn.luatod. Figure 5-6 cn.n 0,1 0 be applied whero the required data (or
reliable e timat s th reof) ar available.
For ro orvoir design studi ,Fig. 5-5 may be applied if data are available from a repres ntative pan. If such data are not available, Fig. 5-6
is believed to constitute the best approach. Re ult should be checked
by applying what appears to be an appropriate coefficient to data from
0.2
04
0.6
08
1.0
,..
12
~~"t
90
18
:il
,u
90
I
80
70
O.9
60
0.8 .S
I
J
,~
70
v-.,.
.oj
~
II':
'6
0.5
30
0.056
Iv
rl
~I'
0.052
J....
20
50
fl
40
30
,.~
I;~
).1
-I-
14~\~
:~'\(
0.028
0.024
cl'
...
o
~Id
0.020
0.016
Not': Eo lines ore bl]slJ(/ on y . 0.0105

0.012
0.008
0. I 5
0.20
0.25
0.30
0.35
0.40
0.45
Dolly lakl evoporollon In Inchn
D.5O
D.55
0.&0
-o.ocn
QOO2
000<'
aD06
0.008
0.010
0.012
0.014
0.016
0.018
0.020
D.022
0.024
0.026
FIG. 5-6. lake evaporation a. a function of solar radiation. air temperature, dewpolnt, and wind movement-the latter "'r.. elements at standard heIGhts for !tie CIaa It. Itat1on.
.~.
0.028
0.030
0,,6
(U.S. Weotfrer
INCREASED WATER SUPPLIES THROUGH REDUCED EVAPORATION
109
the most nearly repre entative pan (or generalized map uch a Fig. 5-7).
There is eldom ju tifiable rea on, however for con tru ting a major
reservoir prior to th colle tion of at lea t 1 or 2 yr of pan data nt the
site for checking purpo es.
In reservoir de ign, the engine r i I' ally conc rn d with the incr a d
10 at the reservoir site r ulting from the con truction f the dam, i.. ,
reservoir evaporation Ie evapotranspiration und r natural condition .
Thus in humid areas con truction of a dam cau. e only 11 nominal inc rea e
in water loss. If reliable streamflow data ar available at the dam 'ita
and immediately above the reservoir both b for and after construction,
the re ulting average increa d loss can be det rmined by dir ct computation, provided that climatic condition are comparabl for the two
periods. 1
5-7. Increa sed water supplie through r educed evapora t ion.
Any steps which can be taken to reduce reservoir evaporation per unit
of storage provide a corresponding increa e in usable water supply.
Selecting the site and design yielding a minimum of reservoir area per
unit of storage is advantageous.
mall municipal reservoirs are sometimes covered to minimize the losses from evaporation. 2 Natural or constructed windbreaks around the reservoir have been advocated, but
this approach is obviously ineffective. A 25 per cent reduction in wind
speed will seldom reduce evaporn.tion by more than 5 p r cent, and wind
reductions of this magnitude are not feasible over re ervoirs of appreciable
size.
In deep reservoirs, summer water-surface temperature may be 30 FO
or more above that near the bottom. If the dam is designed to permit
di charge of warm urface water, evaporation will b less than if colder
water is discharged. The use of monomolecular films on the wuter
surface offer another possibility for evaporation reduction. It ha. been
demonstrated that extremely small amounts of cetyl alcohol will reduce
pan evaporation by a ' much as 40 per cent under some meteorological
condition .. 3 The feasibility of its use 011 major re ervoirs is still to be
determined.
1
W. B. Langbein, Research on Evaporation from Lakes and Reservoirs, TraWl.
Intern. A8soc. Hydrolo(JY, Brussels Assembly, pp.
11, 1951.
'Putting a Sun Shade on Five Acres of Water, En(J. News-Rerol'd, Vol. 156, p. 47,
May 31, 1956.
s W. W. Mansfield, Influence of Monolayers on the Natural Rate of Evaporation
of Water, Nature, Vol. 175, p. 247, 1955.
"Experiments of the Use of Cetyl Alcohol to Reduce Evaporation Losses from
Free Water Surfaces," Interim Report January- October, 1955, East African Meteorological Department, ovember, 1955.
Pilot Stud ies fo r Reservoir Eva.poration Control- Interim Results with Cetyl
Alcohol (Sept. 9, to Nov. 4, 1955), U.S. Bur. Reclamation Spec. I nve8t. Memo. 55-15,
Nov. 18, 1955.
.,..;
~
.S
.2t;
....
Q.
.,~
c
Q.
<
110
111
FACTORS AFFECTING TRANSPIRATION
TRA
P I RATI O
Only minute portion of the water ab orb d by the root sy tems of
plants remain in the plant tis ue ; virtually all is di harg d to the

atmo phere a vapor through the proce known a transpiration. This
proces con'titutes an important pha of the hydrol gi cycle ince it
is the principal mechanism by which the preci pitation falling on land
area is returned to the atmosphere. In tudying the water balan e of
a drainage basin, it is u ually found impracticable to epamte vaporation
and transpiration, and the practicing engineer therefor treo.t the two
factors as a single item. Nevertheless, a knowledge of each proces i.
required to assure that the techniques employed are con is tent with
physical reality.
5-8. Factor affecting transpiration . The meteorological factors
which influence evaporation are also active in the transpiration process.
The importance of solar radiation i amplified since the rate of movement
of water within the plant depends upon sunlight; transpiration and plant
growth are both clo ely related to radiation received. Thus while 75
to 90 per cent of daily 'oil evaporation occur between sunrise and
sun et,l about 95 per cent of the daily transpiration occurs during th
daylight hours.2 Optimum temperature and radiation for maximum
growth vary with plant species, but all activity virtually ceo, es when
the temperature drop to near 40F.
Transpiration is limited by the rate at which moisture becomes available to the plants. Although there is little doubt that the rate of soil
evaporation, under fixed meteorological conditions, decrel1ses qUl1siexponentially with time, divergent views per ist with respect to transpiration. s It is believed that the controversy and apparent discrepancies
can be attributed to the varied methods of deriving supporting data and
the nondescript terminology used to describe the results. Some inv stigators believe that transpiration is independent of available moisture
until it has receded to the wilting pojnt (moisture content at which
permanent wilting of plants occurs) while others assume that transpiration is roughly proportional to the moi ture remaining in the oil and
available to the plants. The field capacity is defined as the amount of
water h Id in the soil after excess gravitational water has drained, and
the range in soil moisture from field capacity to wilting point (available
water) i a measure of the maximum quantity of water available for
1 H. Landsberg, "Physical Climatology," p. 136, Pennsylvania State University,
University Park, Pa., 1941.
J C. H. Lee, Transpiration and Total Evaporation, Chap. VIII, p. 280, i.n O. E.
Meinzer (ed.), "Hydrology," McGraw-Hill, New York, 1942.
3 F. J . Veibmeyer and A. H. Hendrickson, Does Transpiration Decrease as the Soil
Moisture Decreases, Tran8. Am. GOOphY8. Union, VoL 36, pp. 425-448, June, 1955.
112
plant use without rcpleni hment (Sec. 6-3). Available water varies
with soil type, ranging from about 0.5 in. per foot of depth for sand to
2 in. or more per foot of depth for clay loams.
Penman 1 states that oil type, crop type, and root range are of minor
importance in the determination of transpiration rates immediately
following wetting.
n the other hand, Blaney 2 reports that annual
transpiration (including evaporation and with adequate irrigation) in
the an Fernando Valley, alifornia, is 47 in. for alfalfa and only 24 in.
for orange groves. This difference may result in large part from (1)
excess watering of the alfalfa, (2) in ufficient watering of the groves, and
(3) less than 100 per cent vegetal cover in the orange groves. Whether
or not soil type, crop type, and root range appreciably affect transpiration
rates during periods of fully adequate water supply, all three factors
become important as drying proceeds, since they all affect the water
available for plant growth.
That tran piration is c]o ely related to the rate of plant growth is
we]) established. Thus there is pronounced seasonal and annual variation in addition to the diurnal cycle. Transpiration is restricted to the
growing ea on, and the stage of dey lopment i an important factor.
5-9. Measurement of transpiration. It is not pos. ible to measure
tran piration ]0. s from an appreciahle a1' a under natural conditions, and,
hen 'e, determination ar restricted to studies of small sample under
laboratory conditions. One method involves placing one or more potted
plants in a closed container and computing tran pi1'ation as the increase
in moi ture content of the confined space. Most mea 'urements are
made with a phytomete1', a large vessel filled with soil in which one or more
plant are rooted. The only escape of moisture is by tran pi1'ation (the
soil surface j sealed to prevent evaporation), which can be determined by
weighing the plant and container at desired interval of time. By
providing aeration and additional water, a phytometer tudy may be
carried through the entire life cycle of a plant. It is virtually impossible
to simulate natural condition and therefore the results of phytometer
observations n.re mo tly o[ a ad mic interest to the hydrologist. They
ean be on id r d to on titut little more than an index to water use by
a crop under field conditions.
EVAl'OTRAN PInATION
In tudying the hydrologic balance for an area, one is usually concerned

only with the total evaporation (evapotranspiration or consumptive use),
1 H. L. Penman, Natural Evaporation from
pen Water, .Bare Soil, and Grass,
Proc. Roy. Soc. (London): A, Vol. 193, pp. 120-145, April, 1948.
H. L. Penman, Estimating Evaporation, Trans. Am. Geophys. Union, Vol. 37,
pp. 43-46, February, 1956.
I H. F. Blaney, Di cussion of paper by H. L. Penman, Trans. Am. Geophys. Union,
Vol. 37. Pl'. 46-48, February. 1$1,jt:..
WATER-8UCGET DETERMINATION OF MEAN BASIN EVAPOTRANSPIRATION
113
the evaporation from all water, oil, now, ice, v getativ , and oth r
surface plus transpiration. Although attempt has b n mad in the
pa t to draw fine di tinction between the terms evapotranspiration and
consumptive u e, they are now g n rally considered yuon mou . I
Potential evapotranspiration i defin d 2 a - the vl1potran piration that
would 0 cur were th re an ad qua moi -tura supply at all times.
There are numerou approache to stimntiol1 of vapotranspirntion
and potential evapotrn.nspiration, non of which i generally applicabl
for all purposes. The type of data required dep nds on the intcnd d u 'e.
In some hydrologic studies, mean basin evapotran piration i requir d,
while in other cases we are interested in water use of a particular 'rop
cover or the change in water u e resulting from changed y getal cover.
5-10. Water-budget determination of mean ba in evapotranspiration . A uming that storage and all items of inflow and outflow
except evapotranspiration can be measured, the volume of water (u ually

expressed in incha depth) required to balan e the continuity equation
for a basin repre 'cnts evapotranspiration. Among other thiugs, th
reliability of a wat>r-budget computation hinges largely 11 the time
increments con idered, As a rule, normal annual evapotrn,nspiration
can be reliably computed as th difference between long-timc averages
of precipitation and streamflow, since the change in storage over a long
per.iod of years is incon equcntiaP Any deficiencies in such computation are usually attributable either to inadequate pr 'ipitation or runoff data or to subterranean flow ioto or out of the ba in. E timates of
annual evapotran piration are uhje t to appreciable nors if change in
storage are neglected, except wher moisture storage in a basin i nearly
the same on a given date each year. Generally it is ne essary to evaluate
soil-moisture, groundwater, and unace storage at the beginning of each
year.
The water-budget method can also bc applied to short time periods 4
(Fig. 5-8), Over () in. of rain fell in a 3-day period, re luting in the rise
of June 21, and 4.60 in. more fell by June 29. The runoff (Jun e 26- 30)
produced by the second storm was 2.37 in. If it is assumed that the
soil was equally near saturation at thc end of rain on June 2] and 29,
the evapotranspiration during the period wns 2.23 :1l1. (4.60 - 2.37), or
0.28 in. / day. The error in computations for periods as short as a week
1 H. F. Blaney, Consumptive Us' of Water, 'i'rans. AS 'E, Vol. 117, pp. 949-973,
1952.
s C. W. Thorntbwaite, An Approach toward a Rational Claesifi 'ation of Climate,
Goooraph. Rev., Vol. 38, pp. 55-94, 194 .
a C. E, Knox and T. J . ordcnsoll, Average Annual Runoff and Precipitation in the
New England-New York Area, U.S. Gool. Survey Hydrologic lnveat. Atla, HA-7,
undated.
W. E. Fox, "Computation of Potential and Actual Evapotranspiration," U.S.
Weather Bureau (processed), 1956.
can be appreciable, and it is advisable to use longer periods when feasible.

If the computations llre arried through July 16, the estimn.ted evapo~
transpiration averages about 0.23 in./ day, undoubtedly a more realistic
value. The computation mu t be ba ed on total runoff ( ec. 7-5).
This procedure is best adapted to regions where d pth to groundwater is
relatively small and precipitation is evenly distributed throughout the
yel1r. Although evapotran piration estimate derived in the manner
100
90
6.20"
~o=
4_4:J-
FL
'-- 475"
I .t::.
:::
...
eo r--
.5
2S:
70 f - ~
u
60
.s 50
~
..c;
:J!
is
40
r-Rv/lOff-_ f+- Runoff = 1.60"- -
Runoff: 3:01"-
2.37"
"
30
20
'\
10
0
,__
rrV 1\
June
1\
.9 20 2. 22 23 24 25 2S 27 28 29 30
r-I
234
/ \
5
Is-
./
8
9 10 11 12 13 14 15 16
July
FIG. 5-8. Derivation of short-period evapotranspiration estimates.
described must be geared to the fortuitous occurrence of large storms or

relatively wet periods, sufficient d terminations can be made from many
years of record to d fine the s asonal di tribution (Fig. 5-9). If the
resulting curve is to represent n rmal annual evapotranspiration, the
amputations mu t be carried on 11 continuous basis, since omitting dry
periods will bias the 1'0 ults. If the curve is to represent potential
evapotranspiration, computations should be made only for those periods
during which potential conditions exi ted.
5-11. Field-plot determination of evapotranspiration. Application of a water budget to field plots produces sati factory results only
under ideal conditions which are rarely attained. Precise measurement
of percolation is not po ible, and consequent errors tend to be accumulative. If the groundwater table lies at great depth, accretions may be
inconsequential. Under these circum tances, soil-moisture measure-
FIELD -PLOT DETERMINATIONS OF EVAPOTRANSPIRATION
115
ments become the prin ipal ource of rror--random in nature but of su h

magnitude a to preclude computation of short-period eytl.potl'nnspiration.
The energy hudget can be applied to the d termination of vapotranspiration from a field plot much as for a lake ( ec. 5-3). In tead of
being concerned with heat storage in a mil. of water, however, that tor d
in the 'oil profile mu t be computed.
p cific heat of 'oil varic from
about 0.2 to O. cal/cc, depending upon moi tur content and soil type,
and 0 one must know the specifi heat as welill,s t mp ratuf throughout
FIG. 5-9. M ean evapotransp iration curve derived throug h ana lysis of rainfa ll and stream-
flow data.
the profile. In applying Bowen's ratio [Eq. (5-4)] to a lake, temperature

and vapor pres ure of the surface are used. Measuring the vapor pressure
of a vegetated surface pre ents quite another problem, and it becomes
necessary to mea ure the temperature and vapor-pressure gradients
between two levels above the surface. Advected energy, in terms of
water-budget items, is usually small and customarily neglected.
The Thornthwaite-Holzman-type mil. s-transfer equation [Eq . (5-7)] is
particularly suited for evapotranspiration measurement. It ha frequently been u ed for the purpo e,l although there is orne question as
to whether it has been adequately tested. Instrumental requirements
are not easily atisfied under field conditions, since computed evapotran1 C. W. Thornthwaite and B. Holzman, Measurement of Evaporation from Land
and Water Surfaces, U.S. Dept. Agr. Tech. BuU 817, 1942.
N. E. Rider, Evaporation from an Oat Field, Quart. :. Roy . Meteor"l. Soc ., VoL
LXXX, pp. 198- 211, April, 1954.
116
spiratioll is proportional to the difference in measured wind and vapor

pre sure at two level near the urface under investigation.
5-12. Determination of potential evapotranspiration. AU the
technique discus. ed in ecs. 5-10 and 5-11 are also uited to determination of potential evapotranspiration, the only added requirement being
that the area under ob ervation have sufficient water at all time. Many
ob rvations of potential evapotranspiration are now being made 1 in
grass-covered soil containers, variously known as tanks, evapotranspirometers, and lysimeters. The fir t two terms customarily refer to containers with sealed bottoms, while there has been an attempt to restrict
the word ly irneter to containers with pervious bottoms. Evapotranspiration is computed by maintaining a water budget for the container.
Like evaporation pans, evapotranspirometers provide only an index
to potential evapotranspiration. Accordingly, instrumental and operational standardization is of extreme importance. In summarizing the
results of world-wide ob ervations, Mather states:
The evapotranspirometer, when properly operated, i.e., when watered sufficiently so that ther is no moisture deficiency and no appreciable moisture surplus in the soil of the tank, and when xposed homog neously within a protective
buffer area of the prop r size to eliminate th eff ct of moisture advection, is an
instrum nt which should give reasonably reHabl values of potential evapotranspiration. Gr at care must be taken in the operation of the instrument, and standardiz d soil, vegetation, cultivation, and watering practices must be maintain d
on the tanks in order to insur comparabl r suits from ooe installation to another.
Reliable tank observations of natural evapotranspiration (when

appreciably less than potential) are seldom attained, since it is virtually
impossible to maintain comparable soil moisture and vegetal cover in
and adja ent to the tank under su h condition. There i orious doubt
that lysimeters (with pervious bottoms) cn.n be expected to provid
satisfa tory results, unless provision is made to apply a suction force at
the bottom comparable to that in the natural soil profile. 2
5-13. E timating potential evapotranspiration from meteorological data. Several empirical techniques have been developed for
estimating potential evapotranspiration from readily available climatological data and latitude (duration of possible sunshine). Thornthwaite 3 has derived f1 somewhat involved procedure using only temper1 J. R. Math r (ed.), The Mea lIr ment of Pot ntial Evapotranspiration, Publ. in
Climatolouy, Vol. VIT, No. l, Johns Hopkins Ulliv rsity Laborll.tory of Climatology,
eabrook, .J., 1954.
L. L. Harrold and F. R. Dreibelbis, Agricultural Hydrology as Evaluated by Monolith Lysimeters, U.S. Dept. AUr. Tech. Bull. 1050, 1951.
'C. W. Thornthwaite, Discussion of paper by L. L. Harrold and F. R. Dreibelbis.
Tran8. Am. GecphY8. Union, Vol. 26, pp. 292- 297, etober, 1945.
a C. W. Thornthwaite, An Approach toward a Rational Classification of Climate,
ESTIMATING POTENTIAL EVAPOTRANSPIRATION FROM METEROLOG/CAL DATA
11 7
ature and duration of po sible un hine. Elan y's approach I involve.'

th same two factors but was de igned priml1rily to transpo ob rv d
con umptive-use data for irrigated a!' a to other 10 aliti on th ba
of derived coefficient.
sing average yearly data, Lowry and John 011 2
found high correlation b tween on umptive us and tlCcumulat d
degree days during the growing en. on. PI'OC dure whi'h r lyon
temperature a the sole index to heat supply at a particular latitude and
which neglect cloudine , humidity, wind, and oth r factor are ubject
to rather large errors under adverse circum tl1l1ce. Vario.tions in solar
ro.diation at a given latitude and temperature onstitute the gr ate t
source of error.
Potential evapotranspiration and evaporation from a thin fr -watel'
surface are affected by the same meteorological factors: radiation, humidity, wind, and temperature.
ven though the relative importance of
these factors may be appreciably different with respect to the two
phenomena, free-water evaporn.tion hould be a better index to potential
evaporation than is air temperature. Fritz 3 has observed that the
albedo of the eastern United tates i within a few per cent of that of a
fresh-water surface. Blaney 4 finds that the consumptive u e of alfalfa
is 0.66 that from a Class A pan or nearly equal to lake evaporation,
assuming a pan coefficient of 0.7. Penman 6 found annual potential
evapotranspiration in southeastern England to be 0.75 of the evaporation computed from Eq. (5-12) with E" estimated from an empirical
equation like Eq. (5-8) and Q... estimated from Brunt's equations. s
Thus, his finding is that potential evapotranspiration is about 0.75
of lake evaporation. This conclusion is dependent on hi as umption
that evapotranspiration proceeds at the potential rate until the available moisture is deplet d to the wilting point throughout the root zone.
H he had assumed that evapotranspiration rates decreased a. soil moisture
decreases, his estimate of potential evapotranspiration would have been
Geouraph. Rev., Vol. 38, pp. 55- 94, 1948.
J. R. Mather (ed.), The Measurement of Potential Evapotranspiration, Publ. in
Clirnatology, Vol. VII, o. 1, .Johns Hopkins University Laboratory of Climatology,
Seabrook, N.J., 1954.
I H. F. Blaney, Consumptive Use of Wat r, TraWl. ASCE, Vol. 117, pp. 949-973,
1952.
I R. L. Lowry and A. F. Johnson, Consumptive Us of Wat r for Agriculture, TraWl.
ASCE, Vol. 107, p. 1252, 1942.
S. Frit1., The Albedo of the Ground and Atmosphere, Bull. Am. MeteoTol. Soc.,
Vol. 29, pp. 303-312, June, 1948.
H. F. Blan y, Discussion of paper by H . L. Penman, Estimating Evaporation,
Trans. Am. Geophys. Union, Vol. 37, pp. 4G-48, February, 1956.
t H. L. Penman, Natural Evaporation from Op n Water, Bare
ii, and Grass,
Proc. Royal Soc. (London): A, Vol. 193, pp. 120- 145,194 .
I D. Brunt, "Physical and Dynamical Meteorology," Cambridjte, Lannon, op. 136,
144,1939.
118
higher. There is also indication that Penman overcompntes free-water

vaporation 1 a compar d with Figs. 5-5 and 5-6. Much work needs to
be don , but th eviden e suggest that annual lake evaporation and
poi ntial evapotranspiration are approximately equal.
SOIl-moisture uh'ltzotion, 4.0
Moisture
deficiency,
1.0 incll
oct.
Nov,
Dec.
Jon.
Feb,
Mar,
Apr.
May
June
July
Aug,
Sept.
FIG. 5 10. March of normal precipitation and potential evapotranspiration at College

Park, Maryland. (After Thornthwaite.)
5-14. Moisture-d 6ciency accounting. The normal monthly potential evapotran. piration o.nd PI' cipitation for College Park, Maryland,
aro plotted in Fig. 5-10. Potential evapotmn piration exceeds precipitation beginning in late May, and by late eptember the accumulated
ince th estimated maximum possible oil-moisture
difference i 5.0 in.
deficiency is 4.0 in., actual evapotranspiration during the summer is
limited to precipitation plus 4.0 in. of moisture depletion. By early
December the 4.0 in. of oil moi tUl'e has been replaced by excess precipitation, and a continued excc s until late May produces 12.2 in. of runoff.
Although thi illustration is schematic in some respects, the annual
magnitudes of the variou factors are realistic. Thus while there is an
exc ss of precipitation in an average year (40,75 in. compared with
29.45 in. of potential evapotran piration), its di. tribution is such that a
deficiency d velops in August and eptemb r.
uroerous attempt have been made to apply the bookkeeping approach
to maintain a continuing accounting of evapotran piration and soilmOl tur deficiency. 2 In most ca es it has been assumed that evapoPenman actually treated a small sunken pan as a "free-water Burfa e."
C. W. Tbornthwaite, The Moisture Factor in Climate, Trans. Am. Geoph1l6.
Union, Vol. 27, pp. 41-48, February, ]046.
C. W. Thornthwaite, Contribution to Report of tb Committee on Transpiration
1
BIBLIOGRAPHY
119
transpiration is at the potential rate at all time wh n the soil-moistur

deficiency is Ie s than some predetermined value (root-zone constant)
repre nting the available moi ture (3 to 5 in.) within the root zone.
With deficiencies in exce s of this limiting value, it i a suroed that
evapotranspiration either ceases or drop to some small fixed proportion
of the potential rate.
As indicated in ec. 5-8, considerable difference of opinion exists over
the effect of a moisture deficiency on the rate of tran piration; som
investigators claim to have demonstrated that transpiration continues
at an undiminished rate until moisture content throughout the root
zone drops to the wilting point, while oth rs in ist that the rate is approximately proportional to the moisture content throughout this range. Be
this as it may, the actual evapotranspiration from a field plot and that
from a natural basin differ in many respects. In a natural basin of
appreciable size it can be said that:
1. There are numerous types of vegetation at varying stages of
development and having roots extending to various depths.
2. Differences in slope and aspect are reflected in evapotranspiration
rates so that, all other things being equal, the wilting point should be
reached sooner in some n.r as than others.
3. There are areas of bare soil where evaporation decreases with
decreasing moisture content. Also, evaporation from other surfaces
(buildings, streets, etc.) decreases with time subsequent to minfall.
4. Depression storage is a sourco of free-water evaporation. The
area of such water surfaces decreases as evaporation and infiltration
proceed subsequent to rainfall.
When these and other facts are considered, it eems obvious that the
rate at which moisture is lost from a natural basin tends to be proportional
to the quantity of moisture available, that evapotranspiration is in this
sense a decay function, possibly similar in form to Eq. (8-2).
BIBLIOGRAPHY
Anderson, E. R., L. J. Anderson, and J. J. Marciano: A Review of Evaporation Theory
and Development of Instrumentation, U.S. Navy Electronic8 Lab. Rept. 159,
February, 1950.
Blaney, H. F., and others: Consumptive Use of Water- A Symposium, Trans . ASCE,
Vol. 117, pp. 949-1023, 1952.
and Evaporation, 1943- 1944, 'l'rans. Am. GwphlJ8. Union, Vol. 25, Part 5, pp. 886-893,
1944.
H. L. Penman, The Dependence of Transpiration on Weather and Soil Conditions,
J. Soil Sci., Vol. 1, pp. 74- 9, 1949.
W. B. Langbein, Monthly Evapotranspiration Losses from Natural Drainage
Basins, Trans. Am. GecphY8. Union, Vol. 23, Part 2, pp. 604-614, 1042.
M. A. Kohler, Computation of Evaporation and Evapotranspiration from Meteorological Observations, U.S. Weather Bur. Rept., 1957 (mimeographed ).
120
Hydr logy Handbook, ASCE Manual 28, 1949.

Kohl r, M. A., T. J. Nordeo80n, and W. E. Fox: Evaporation from Pans and Lakes,
U.S. Weather Bur. Re8earch, Paper 38, May, 1955.
Mathor, J. R. (ed.): The Me8.8ur mcnt of Potential Evapotranspiration, Pu1.Jl. in
Climatology, Vol. VII, o. 1, Johns Hopkins University Laboratory of Climatology, Seabrook, N.J., J 954.
Tbornthwaite, . W., and J . R. Matber: The Water Balance, Publ. in Climatology,
Vol. VIII, No.1, Drexel Institute Laboratory of Climatology, Centerton, N.J.,
1955.
Water-IoS8 Investigations, Vol. 1, Lake Hefner Studies, U.S. Gool. Survey Profes8.
Paper 269,1954 (r print of U.S. Gool. Survey Circ. 220, 1952).
Water-loss Investigations, Lake Mead Studies, U.S. Gool. Survey Profes8. Paper 298,
1958.
OURCE OF DATA
Regular observations of pan evaporation arc summarized in the Weather Bureau 's
monthly publication Climatological Data. Summarized data on water 1088 by evaporation and transpiration may be found III the following publications:
Lowry, R. L., and A. F. Johnson: Consumptive Use of Water for Agriculture, Tran8.
ASCE, Vol. 107, pp. 1243- 1302, 1942.
Williams, C. R., and others: Natural Water Loss in elected Drainage Basins, U.S.
Geol. S'urvey Water-supply Paper 846, 1940.
Young, A. A., and H. F. Blaney: Use of Water by Native Vegetation, Calif. D-iv. of
Water Resource8 Bull. 50, 1942.
PROBLEM
5-1. A variety of units appear in the literatur on vaporation, and familiarity with
tho relevant conversion factors is therofore desirable. As an exercise, rewrite Eq.
(5-4) for a.pplication with temperatures in d grees Fahrenheit and pressur's in millimoters of mercury. Rewrite Eq. (5-7) converting E to centimeters per hour, e to
millibars, v to centimeters per s cond, and T to degrees centigrade.
5-2. Entering the upper left-hand r latioo of Fig. 5-2 or Fig. 5-6 with evaporation,
Whid movement, and dewpoint in reverse of the ord r indicated provides an estimate
of surface-water t mperature in the pan (on the axis labeled "air temperature ") . In
this manner, com put water temperature for each of the 10 met orological conditions
enumerat d in l'able 5-1.
cscrib those conditions which result in a water ternperatur high r than that of tbe overlying air, and vice versa.
5-3. Using Fig. 5-6, compute lake evaporation (thus neg! cting adv ction and
changes in heat storage) for ach set of data prcsented in Table 5-1. Also compute the
pan coefficient for ach cas. Is the range of coefficient displayed here indicative
of the range in annual oefficients to be expected over til Unit d States? Whyi'
5-4.. All indication of th groBs variation of lake evaporation with elevation can be
obtained from Fig. 5-6 for specified conditions if values of tb ' respective parameters
are known for two elevations. Considering the data in Table 5-1 to constitute a
seri s of sea-level obs rvations, compute tbe evaporation at 3000 ft, mean sea level,
usir\g the following elevation gradients (per 1000 ft): air temperature, - 3 FO; dowpoint, -1 FO; wind, 10 per cent; and radiation, 2 per cent. If tbese results are compared with thos of Prob. 5-3, is tbe "elev(ltion effect" reasonably constant for these
select d circumstances? Discuss.
5-5. Given Lho data tabulated below, compute monthly and annual lake evaporation from Fig. 5-6. Assuming that a proposed reservoir will experience this computed
amount of evaporation per year (and noting tbat pr cipitation Ie B runoff is natural
PROBLEMS
121
evapotranspiration), what would be the net anticipated 10 from tho r crvoir per
acre of surface area? On the basis of the comput d lake vaporation and that provided for a Class A pan, compute monthly, mean monthly, and annual pan eo ffi ients.
Dis uss qualitatively the effects of heat storage in a d p re crvoir on th doriv d
monthly pan coeffi(ients.
Air tcmperature,
Dewpoint,
57
46
37
34
36
44.
46
35
28
25
24
30
90
290
200
150
150
230
92
310
54
39
83
65
72
77
75
69
51
60
65
64
58
66
50
44
41
44
Sum .. . . .
. ..
...
Mean ... .
56
44
Period
Oct .. . .. .
Nov ... . ..
Dec ......
Jan . .. ...
Feb ... .. .
Mar . . .. .
Apr .. .. . .
May .....
June .... .
July ... . ..
Aug ......
Sept .... .
OF
OF
Wind, Radiation,
mi/day
ly / day
52
59
68
81
Pan eYllpPrecipitaoration,
tiou, in.
in .
Runoff,
ill .
2.7
2.4
3.1
3 .4
2 .9
3 .5
3 .6
3 .7
3 .9
4 .0
4.6
3 .6
0 .8
400
480
510
500
440
370
3 .2
1.7
1.0
0 .8
1.0
2.8
4 .8
6 .0
6.4
7.1
5 .8
4 .2
...
. ..
44.8
41.4
15 . 1
70
336
1.1
1.3
1.6
1 .7
1.8
1.7
1 .4
1.2
0 .8
1.0
0.7
5-6. Using the precipitation and potential evapotranspirntion data tabulated below,
develop a chart similar to Fig. 5-10 (assume moisture capacity of 4.0 in.). What arc
the annual moisture excess and annual moisture deficiency? During what period of
the year is a need for supplemental irrigation most apt to develop? What geographical section of the United States is typified by the derived hart?
Month . . . . .. .. .... .. . Oct. Nov. Dec. Jan. Feb. Mar. Apr. May June July Aug. Sept.
-- -- -- -- -- -- -- - - - - - -- -- -- - -- - ------
- - - - --
Precipitation . .. . .. . . . . 1.2 2.4 4 .2 5.2 4 .5 3 .7 1.5 0.8 0 .3 0 . 1 0 . 1 0 .6

Potential evapotranspiration . . .. .... . . . .. . 2.5 1.8 1.0 0.9 1.2 1.8 2.3 2 .8 3 . 2 3.5 3.2 2.9
5-7. Using data published in Water-8upply Paper8 and Climatological Data, plot
hydrograpbs of mean daily flow and bar charts of daily precipitation (several years)
for a selected small basin. The basin and period should be selected to include reasonably saturated conditions on a number of occasions. Compute mean daily evapotranspiration for several periods delineated by times of assumed basin saturation
(Fig. 5-8). Whieh of the periods analyzed do you believe to be indicative of "potential" conditions?
6
GROUNDWATER
In 1950 about 25 billion gallon of water 1 was pumped each day from
groundwater in the llited tates. This quantity was about one-sixth
of tb total water us in the country ex 'lu ' ive of hydroelectric-power
gen ration. Becau e groundwater is relatively free of pollution it is
espe ially useful for dom tic purpo e', particularly for isolated farm8
and small town. In the arid regions of the West, groundwater i often
the only source of wat r for irrigation. Groundwater temperature
remain relatively low during the summer, and large quantities are u ed
for air conditioning in the East and South.
A ide from its gr at economic importance, groundwater is al 0 an
important pha e of the hydrologic cycle. Most p renlitlll treams derive
the gr ater part of their flow from groundwater, while in arid regions
much of the surface str amRow percolates to the groundwater. The
occun nce and movement of ubsurface ,Yater are nece arily intimately
related to geological structure, and a knowledge of geology is prerequisite to a thorough comprehension of groundwater hydrology. This
chapter stres e th.e hydrologic aspects of groundwater and presume
only elementary knowl dg of geology.
6-1. The occurrence of subsurface water. Figure 6-1 is a schematic cros section of the upper portion of th earth's ru t. Near the
urface in the zone oj aeration pOl' spaces contain both air and water.
Water in the zone of aeration i known a suspended or vadose water, or
oil moisture. The thickne s of the zon of a ration varies from practically zero in wamplands to s veraI hundred feet in arid regions with
substantial relief.
Below th zon' of aeration the pores are filled with "ateI'. This is the
zone of saturation, 01' the groundwater. At the surface parating the e
zones, called the water table or phreatic surface, the hy.:iro tatic pre. sure
in the groundwater is atmospheric. The zone of saturation may extend
1 C. L. McGuinn ss, The Water Situation in the Unit d States wi'oh Special Rere!'
ence to Ground Water, U.S. Geol. Survey Cire. 114, p. 95, 1951
122
SOIL- WATER RELATIONSHIPS
123
to considerable depth but, a depth incr asc , tb w ighl of Lbe ov dying

material tends to 10 the pore paces, and rehttively littl \Vat r is
found at depth greater than 2000 ft.
cca. ionally 10 oj zone of atu,atioD exi t a perched groundwater
above a locally impervious tratum
ometim a body of groundwater
Transpiration (Xerophytes)
"Spring
Influent
stream
Transpiration (Phreatophytes)
AQuIfer
--
Groundwater
Confining layer
(Aquiclude)
FIG. 6- 1. Schematic cross section showing the occurre nce of groundwater.
is overlain by an impervious stratum to form confined or artesian waler.

Confined groundwater is usually under pressure becau e of th weight of
the overlying soil and the hydrostatic head. If a well penetrates the
confining layer, water will rise to the piezometric level, the artc ian
equivalent of the water table. If the piezometric surface is above ground
level, the welt discharges as a flowing well.
MOl TURE IN THE ZONE OF AEUATIO
6-2. Soil-water relationships. Moisture in the zone of aeration

may be present as gravity water in tran it in the larger pore spaces, as
capillary water in the smaller pOl' s (Fig. 6-2), as hygroscopic moisture
adhering in a thin film to th oil grains, and as water vapor. Gravity
water is a tran ient state. After a rain, water may move downward in the
iarger pore, but this water must either be di p r d into capillary
pores or pa s through the zone of aeration to the groundwater or to a
!!tream channel. Hygroscopic water, on the other hand, is held by
molecular attraction and is not normally removed from the soil under
GROUNDWATER
12..
u ual climatic conditions. The important variable element of soil

moisture is, th reiore, the capillary water.
If a oil-filled tube is placed with its lower end in a container of water,
some water will move up into the 'oil. The rate of upward movement.
becomes progressively smaller with
time, until it eventually approaches
zero. Mea urement of the moi ture
content of the soil at various level
will show that moisture in the soil
column decreases with height above
the water surface (Fig. 6-3). If a
sample of soil is saturated with water
FIG. 6-2. The occurrence of co pillory and then subjected to sUCce ively
moisture in soil.
greater negative pressures and the
moisture content is noted after each change in pre sure, a similar curve
(Fig. 6-4) results.
Buckingham 1 first proposed characterizing soil-moisture phenomena
on the basis of en rgy r lationships. He introduced the concept of
5011 filleo tube
Free water surface

10
20
50
Moisture content. per cent
f iG. 6-3. Moisture-content vs. height curves for typicol soils.
(After Buckingham.)
capillary potential to describe the attraction of oil for water. With a

free water surface taken as reference, capillary potential i defined a the
work r quired to move a unit ma of water from the reference plane to
any point in the soil column. Thu, capillary potential is the potential
energy per unit mass of water. By definition, capillary potential is
l E. Buckingha.m, Studies on the Movement of Soil Moisture, U.S. Dept. Agr. Bu.r.
Soils Bull. 38, 1907.
SOIL -MOISTURE EQUILIBRIUM POINTS
125
nega.tive since water will move upward by capillarity without external

work. It 'an be shown that capillary potential", i r latcd to th acceleration of gravity g and height abov datum y (negative) by the equation
'" =
(6-1)
gy
Curves such as Fig. 6-3 provid a ba i for relating capillary potential

and moisture content for a particular soil.
chofield 1 uggested the
term pF to repre ent the common logarithm of the capillary potential
in centimeters of water. This is analogous to the use of pH in chemistry.
10POO,000
1,000,000
...
20 100,000
\\
J:
'0
E
'"~
10,000
'\
';;;
c:
1,000
vWiltln9 point
~~
~~
/~
.2
'"
'0
~~
~q_
""'"'-..
10
, ~.
100
Field copocify
L~
10
,~
~\
20
30
40
Moisture content, per cent
FIG. 6-4. Moisture-te nsion curves fo r a typical soil.
50
(After Schofi eld.)
6-3. E qu ilibrium points. Visualizing the several states of water

in soil, early soil scientists tried to define the limits of these states by
"equilibrium points." Figures 6-3 and 6-4 indic te that no clear-cut
boundaries exist, but the equilibrium points are convenient for discussing
soil moisture. The two of greate t interest to the hydrologist are field
capacity and wilting point. Field capacity is defined as the moisture
content of soil after gravity drainage is complete.
olman 2 has shown
1 R. K. Schofield, The pF of the Water in Soil, 7'rans. Third Intern. Cvngr. Soil Sci.,
Vol. 2, pp. 37-48, 1935.
2 E. A. Colman, A Laboratory Procedure for Determining the Field Capacity of
Soils, Soil Sci., Vol. 63, p. 277, 1947.
GROUNDWATER
126
that field capacity is es entin-lly the water retained in soil ubjected to a

tension of ~ atm. V ihmeyer and Hendrickson I found that the moisture
equivalent, the wat r retained in 0. oil .. ample % in. deep after beir\g
centrifuged for 30 min at a 'pe d equivaJent to a force of l000g, was also
nearly the field capacity of fine-gru.ined soil .
The wilting point repre ent the soil moisture at the time that plants
cannot extract water from the soil. It iii the moi ture held at a tension
equivalent to the 0 motic pre. sure exerted by the plant roots. For
many years the standard method of finding the wilting point was to
gl'OW sunflow r eedlings in oil samples. Recent te ts indicate that it
i represented by the moisture content at a ten ion of about 15 atm.
The difference b tween the moisture content at field capacity and at
wilting point i called available moisture. It represents th useful storage
capacity of the soil ( ecs. 5-8 and 5-14) and the maximum water available
to plants. Typical valu s of moisture content at field capacity and
wilting point and available moisture are given in Table 6-1.
TABLE 61. Typical Moisture Values for Various Soil Types
Field
capacity
Wilting
point
Available
water
Soil type
Speciflc
weight,
Ib/ cu ft
Per cent dry weight of soH
Sand ....
Sandy loem .
Loom . ..... . ..
Silt loom .......
Cloy loom .........
Cloy ........ ....
Peot .......
5
12
19
22
2.4
36
140
2
5
10
13
15
20
75
3
7
9
9
9
16
65
95
90
85
80
80
75
20
6-4. Mea urement of soil moisture. The conventional method of

measuring soil moisture is the laboratory determination of weight of
water lost when a ample is oven-dried. This i slow and is not adapted
to continuou mea urement. A tensiometer (Fig. 6-5) consists of a
porous ceramic cup buried in the soil. The cup is filled with water and
connected to a manometer. If the soil is dry, water leaves the cup,
'r ating negative pre sure which i indicated by the manometer. When
the ten ion exc eds 1 atm the sy t m is no longer workable. Thu a
IF. J. Vcihmeycr and A. H. Hendri kson, The Moisture-Equivalent as
of the Field Capacity of Soils, Soil Sci., Vol. 32, pp. 181-193, 1931.
9.
Mcasurl'
MEASUREMENT Of SOIL MOISTURE
127
ten iometerl is atisfactory for moisture content from slightly blow

field capacity to aturation.
N urnerou indirect mea ur ments of moi tura content have b 1.1
devi ed, including men urem nt of gamma-radiation p n trntion, 2 h ut
transmis ion,3 and el ctrical r i tivity. R si tivity' method ar mo t
widely used. A pair of el ctrod s mbedd d in a porous di lectric ar
buried in the soil with wires running to a surfa 'e terminal. Plaster of
paris, nylon, and Fibergla have
been u ed a di lectrics. The resistance between the I ctrod
varies with the moi ture content
of the dielectric which is as umed
to be in moi ture equilibrium with
Water
the soil. Re istance is u ually
measured with an alternatingcurrent bridge so as to avoid polarization of the element. The
manometer
resi tance element must be carefully installed in close contact with
the soil and with minimum soil
disturbance.
Soil moisture i related to resistance by calibration. This may be
done by embedding the block in a
pan of soil and noting the change
in resistance as the oil dries.
Nonuniform drying of the soil FIG. 6-5. Schem a tic drawing of a sim pl e
sample, difficulty of imulating the te nsiometer.
in-place condition of the soil, and
lack of representativity of the sample make this method of calibration
unsatisfactory. A more r alistic, although seemingly less accurate, calibration can be obtained by taking numerous soil samples around the site
where the block is placed. The moisture content of the e samples is related
to the resi tance reading at the time the sample was taken. This
calibration makes the resistance element an index to the moisture content
1 L. A. Richards, M. B. Russell, and O. R. Neal, Further Developments on Apparatus for Field Moisture Studi 5, Proc. Soil Sci. Soc. Am., Vol. 2, pp. 55-64, 1937.
I D. J. Belcher, T. R. Cuykendall, and H . S. Sack, The Measurem nt of Soil Mois!,ure and Density by Neutron and Gamma-ray Scattering, U.S. Civil Aeronaut.
A.dmin. Tech. Develop. and Evaluation Center Rept. 127, 1950.
a C. F . haw and L. D. Baver, Heat Conductivity as an I ndex of Soil Moisture, J.
Am. Soc. AUron., Vol. 31, pp. 8 6- 801, 1930.
W. M. Broadfoot and others, Some Field, Laboratory and om e Pro dur 8 for
Soil-moisture Measurement, outhern Forest Expt. Sta. Occa8ional Pap r J35, 1954.
GROUNDWATER
128
of the soil zone encompassing the sample sites. Since the elements actu
ally measure oil-moisture tension they have about the ame re istance at
field capacity in any type of soil. The same is true at the wilting point.
6-5. Movement of soil moisture. Infiltration is the movement of
water through the soil surface into the soil and is distinguished from
percolation which is movement of water through the soil. When water
is first applied to the soil surface, gravity water begins to move down
through the larger soil opening while the smaller surface pores take in
water by capillarity. The downward-moving gravity water is also taken
in by capillary pores. As the capillary pores at the surface are tilled
and the intake capacity is reduced, the infiltration rate decreases. In
homogeneous soil, as pores at lower levels are filled, infiltration decrea es
gradually until the zone of aeration i saturated. Normally, the soil
is stratified and frequently the subsoil layers are less permeable than the
surface soil. In this case, the infiltration rate i eventually limited to the
rate of percolation through the least pervious subsoil stratum.
Two cases of infiltration must be recognized. Infiltration from rainfall is distinguished by very shallow depths of water on the soil surface
but is extensive over In.rge areas. Quantities of water infiltrated are
usually very small (a few inches per day maximum) and rarely sufficient
to saturate a great depth of soil. At the termination of rain, gravity
water remaining in the soil continues to move downward and, at the
same time, i taken up in capillary por spaces. Usually the infiltrated
wat r is distributed within the upper few feet of soil, with little or no
contribution to groundwater unless the soil is highly permeable or the
zone of aeration very thin. Infiltration from rainfall is discussed in
greater detail in Sec. 8-2.
For irrigation and artificial recharge of the groundwater (Sec. 6-16),
water is ponded to a con iderable depth over limited area for long periods
of time. The aim of re harge operation is to saturate the soil down to
the wat~r table. Under these condition the time variation of infiltration i complex, with temporary increa es in rate superimposed on a
gradually declining trend. Escape of soil air around the infiltration
basin, bacterial action, changes in water temp rature, changes in soil
structur , and many oth r factor appear to influence the e variations.
Movement of moistur within the soil is governed by the moisture
potential following the equation
i)A
q = -K",-
ax
(6-2)
where q is the flow per unit time through unit area normal to the dir ction of flow, x is distance along the line of flow, Kw is condu tivity, and
A is potential. After gravity water has left the soil, the principal
129
AQUIFERS
component of total potential is the capillary potential. Equation (6-2)

states that 1l0w is from a region of high potential to 11 region of lower
potential. Quantitative determination of the conductivity is difficult
although it has been shown to increa e with moisture content and decrea e
with pore ize. Thus capillary movement decrea es as oil dri and is
least in fine-grained soil. Fortunately, a Qualitative wlderstanding of
the e phenomena ii normally sufficient for engineering hydrology.
Transport of water vapor in the soil is controlled by tempcl'I1ture
differences. Vapor movement is from high temperature (high vapor
pressure) to low temperature. Vapor transport is an important factor in moisture movement when the moi ture content is lower d to
the point where capillary moisture is di continuous.
nder thi condition, however, moistur -content and temp rature gradi nt ar u ua,lly
so small thl:l.t the quantity of moisture moved is negligible. When the
surface oil is frozen , the vapor-pressure gradient is upward and i
accentuated by the lower vapor pressure of ice relative to water at the
same tempcrature. Thus when frozen soil thaws, its moisture content
may be greater than at the time of freezing.
onversely, during umm r,
vapor-pres ure gradients would be downward were it not for evaporation
and transpiration .
MOISTUllE IN THE ZONE OF SATURATION
Within th zone of aturation all pore spaces are filled with water, and
the different states of moisture, moisture tension, etc., are of little
concern. Interest is centered on the amount of water present, the amount
which can be removed, and the movement of this water.
6-6. Aquifer. A geologic formation which ontains water and transmits it from one point to another in quantities sufficient to permit
economic development is called an aquifer. In contra t, an aquiclude
is a formation which contains water but cannot transmit it rapidly
enough to furnish a significant supply to a well or spring. An aquifuge
has no interconnected openings and cannot hold or transmit water.
The ratio of the pore volume to the total volume of the formation is
called porosity. The original porosity of a material is that which existed
at the time the material wa formed. Secondary porosity re ults from
fractures and solution channels.
econdary porosity cannot be measured without an impossibly large
sample. Original porosity is u ually measured by oven-drying an undisturbed sample and weighing it. It is then saturated with some liquid and
weighed again. Finally, the saturated sample is immersed in the same
liquid and the weight of displaced liquid i noted. The weight of
liquid required to saturate the ample divid d by the weight of liquid
displaced is the porosity as a decimal. If the material is fine-grained,
130
GROUNDWATER
the liquid may have to be forced into the sample under pressure to assure
complete saturation.
High porosity does not neces arily indicate a productive aquifer, ince
much of the water may be retained in small pore spaces under capillary
tension as the material is dewatered. The Ilpecific yield of an aquifer
is the ratio of the water whi h will drain freely from the material to the
total volume of the formation and must always be less than the porosity.
The relation between specific yield and porosity is dependent on th
size of the particles in the formation. Specific yield of a fine-grained
aquifer will be small wberea coarse-grained material will yield a greater
amount of its contained water. Table 6-2 lists approximate average
TABLE 6-2. Approximate Average Porosity, Specific Yield, and
Permeability of Various Materials
PorOSity, Speciflc yield ,
Material
Cloy .... . .
Sond .......... . ...
Grovel . . ... ......... . .
Groveland sand ..............
Sondstone ............ .. . . .. .. ....
Dense limestone and shale .......
Quartzite, granite . . .. ...
45
35
25
20
15
5
1
25
22
16
8
2
0.5
Permeability
[K p , Eq. (6-4) ),
gpd/ sq ft
1
800
5000
2000
700
1
0.1
valu of poro ity and specific yield for some typical materials. Large
variations from th se average values mu t be expected. Note that clay,
although having a high poro ity, has a very low specific yield. Sand and
gravel which make up most of the more productive aquifers in the United
States will yield about 80 per cent of their total water content.
6-7. Movement of groundwater. In 1856 Darcy confirmed the
applicability of principles of fluid flow in capillary tubes, developed several
year earlier by Hagen and Poisseulc, to tbe flow of water in permeable
media. Darcy's law is
v = ks
(6-3)
where v is the velo ity of flow, s is the slope of the hydraulic gradient,
and k is a coefficient having the units of v (usually ft/ day). The di charge q is the product of area A and velocity. The effective area is the
gross area times the porosity p of the media. Hence
q = 7.48lcpAs = KpAs
where 7.48 converts to gallon when the oth r terms are in feet.
(6-4)
The
131
DETERMINA TlON OF PERMEABILITY
coefficient of permeability Kp is u uaUy xpres ed in .Meinzer units, the

discharge in gallons per day through an ar a of one square foot under a
gradient of one foot p r foot at 60F (Table G-2). Value. nt oth I'temp rature can be found by multiplying by the rn.tio of th kinematic viscosities, i.e.,
K = K Veo
(6-5)
pr
p VT
It is convenient to use the transmissibility T to repre ent the flow in

gallon per day through a se tion 1 ft wide and the thickne s of the
aquifer under a unit head (slope of 1 tI ft) .
(6-6)
wh re Y i the saLurated thicklle ' of the aquifer.

Eq. (6-4) b comes
q = TBs
With thi coefficient

(6-7)
where B is the width of the aquifer.

Note that the equations of groundwater flow are analogous to the
electrical equation
(6-8)
where the current i is equivalent to q, the voltage E i comparable to s,

and the reciprocal of resistance R is equivalent to permeability. This
similarity is sometimes u ed to advantage in electrical models of groundwater-flow problems.
6-8. Determination of permeability. Laboratory measurements
of permeability are made with instruments called permeamelers (Fig. 6-6).
A sample of the material is ubjected to water under a known head, and
the flow through the sample in a known time is measured. Such tests
have limited practical value because of the difficulty of placing samples of
unconsolidated materials into the permeameter in their natural state and
the uncertainty as to whether a sample is truly representative of the
aquifer. Flow in solution cavities or rock fractures and the effect of
large boulders in gravel aquifers cannot be duplicated in a permeameter.
The earliest field techniques for determining permeability involved
introducing salt into the aquifer at one well and timing its movement to a
downstream well. l Fluorescein dye,2 detectable at a concentration of
0.03 ppm by the unaided eye and in concentrations as low as 0.0001 ppm
1 C. S. Slichter, Field Measurements of Rate of Movement of Underground Water,
U.S. Gecl. Survey Waler-84pply Paper 140, 1905.
I R. B. Dole, Use of Fluorescein in
tudy of Underground Wat ra, U.S. Gecl. Survey
Water-supply Paper 160, pp. 73- 85, 1906.
GROUNDWATER
'32
under ultraviolet light, has also been u ed as a tracer. More recently,

radioactive materials have been tested. Tracer techniques have encountered numerous difficulties. l
hemi 'a1 reaction betw en the tracer
elements and the formation ometimes occur. Because of diffu ion,
te ts must be conducted over hort distances in ord r to have detectable
concentrations at the downstream well, and even then it is difficult to
ql
Kp = - = As Allh
fiG. 6-6. A simple upword -Aow permeometer.
determine a representative time of arrival. Tracers are mo t useful for

determining path of flow a , for example, when it is nece 8ary to locate
a ource of pollution.
Today permeability is most commonly determin d by pumping tests.
By using the principles of w 11 hydrauli s ( ec. ()-12) it is possible to
stimate average permeability of an aquifer for a large distance around
the te t well.
6-9. ource of groundwater. Almost all groundwater is meteoric
water derived fx-om precipitation. Connate water, present in the rock at
its formation and fx-equently highly saline, is found in some areas.
Juvenile water, formed chemically within the earth and brought to the
surface in intrusive rocks, occurs in small quantities. Connate and
juvenile water are sometimes important source of unde irable minerals
1 W. J. Kaufman and G. T. Orlob, An Evaluation of Ground-water Tracers, Tram.
Am. GeophY8. Union, Vol. 37, pp. 297- 306, June, 1956.
DISCHARGE OF GROUNDWATER
133
in the groundwater. Groundwater in the an Joaquin Valley, alifornia,

contain consid rable boron brought to the urfac from great depth.
Water from precipitation r ache the groundwater by the proces of
infiltration and by per oln.tiOll from treams and lak .
ir ct percolat ion cannot yield large quantiti of groundwater except where the soil
is highly permeable or the wat r table i clo. e to the urface. Large
quanti tie of dire t percolation 0 cur through the p rmeabl ba altic
lava of northern alifornia, eastern regon, outh rn Idaho, and
Hawaii, and in the southern Appalachian region where thin soil overlies
cavernous lime tone.
In the southern part of the Central Valley of alifornia groundwater
is hundreds of feet below the surface, and little or no recharge from rain
can occur. In such case most of the recharge i from stream chann Is.
treams contributing to the groundwater are called influent streams.
uch streams frequently go dry during prolonged dry spells when percolation ab orbs all the available flow.
tream are rarely influent throughout their entire length. Often the chann I cros 'es strata of varying
permeability, with most of the percolation 10 s in short reaches of high
permeability.
onsiderable percolation often occurs from stream channels crossing coarse gravels in an alluvial fan . In areas of artesian
groundwater, the overlying aquiclude prevents appreciable direct
recharge; the recharge area may be far removed from the artesian
area.
6-10. Discharge of groundwater. Without interference by man, a
groundwat r basin fills with water and discharges its excess by several
routes.
treams inter ecting the water table and receiving flow from
the groundwater are known as ejfi1.tenl strea,ms. Perennial streams ar
generally effluent through at least a portion of their length and may flow
in impermeable formations elsewhere so that there is little or no loss by
seepage.
Where an aquifer intersects the earth's surface, a spring may form.
There may be a concentrated flow constitut,ing the headwater source of
a small stream or merely effiuent seepage which evaporates from the
ground surface. Figure 6-7 illustrates several types of springs. The
flow of most springs is small and usually of little hydrologic significance,
although even a small spring may provide water for a single farmstead.
Meinzer l clas. Hied springs from first to eighth magnitude with respect to
flow. First-magnitude springs discharge 100 cfs or more while eighthmagnitude springs have a flow Ie. s than 1 pt/ min. According to Meinzer 2
10. E. Meinzer, Outline of Ground-water Hydrology, U.S. Geol. SUnJell Wateraupply Paper 4.94, 1923.
'0. E. Meinzer, Large Springs in the United States, U.S. Geol. SUnJey Water-Iupply
Paper 557, 1927.
134
GROUNDWATER
there are 65 first-magnitude springs in the United tate -38 in voicanic

rocks of California, Oregon, and Idaho; 24 in limestone in the Ozarks,
the Balcones Fault area of Texas, and in Florida; and 3 sandstone springs
in Montana. The Fontaine de Vauclu e in France has a discharge often
exceeding 4000 cfs. It is the world's largest spring and i in a limestone
formation.
Where the water table is close to the uriace, groundwater may be
discharged by direct evaporation or by transpiration from the capillary
Spring
-----~~\
--
Woter toble
Water hole or dimple spring
Anticlinal spring
Perched spring
Spring from solution channel
FIG. 6-7. Types of springs.
fringe. Plants delhing their water from groundwater, called phreatophytes, often have root yst-ems extending to depths of 40 ft or more.
This invisible evapotranspiration loss may be quite large. At a rate of
3 ft/yr the loss would be 1920 acre-ft/sq mi/yr.
The various dlannels of groundwater discharge may be viewed as
spillways of the groundwater reservoir. When groundwater is high,
discharge through the natural spillways tend to maintain a balance
between inflow and outflow. During dry periods the natural discharge
is reduced as groundwater levels fall and outflow may even cease.
Artesian aquifers may not reflect this natural balance as rapidly as watertable aquifer , but su tained drought will decrease water levels in the
recharge area and decl'ea e discharge from the aquifer.
6-11. Equilibrium hydraulics of well. Figure 6-8 shows a well
in a homogeneous aquifer of infinite extent with initially horizontal water
table. For flow to occur to the well ther must be a gradient toward the
well. The resulting water-table form is called a cone Qf depression. If
EQUILIBRIUM HYDRAULICS OF WELLS
135
the decren. e in water level at the well (drawdown) is small with respect
to the total thickne of the aquifer und if the well completely pen trate
the aquifer, the streamlines of flow to th well may b a wned to be
horizontal. In thi co. e an approximate formula relating well di charge
and aquifer characteristic may be d rived.
GroUnd surface
Original water table
I
I
~.~I--+-~--+-~I~
I
FIG. 6-8. Deflnition sketch and flow net for equilib rium flow conditions at a we ll.
Flow toward the well through a cylindrical surface at radius Xl must

equal the di charge of the well, and from Dar y's law [Eq . (6-4)]
q
dy
27rxyKp dx
(6-9)
where 27rxy j the area of the cylinder and dy / dx the slope of the water
table. Integrating with respect to x from rl to r2 and y from hi to h2 yield
q = 7rJ(p(h I 2 - h 22)
log. (rl/r2)
(6-10)
where h is the height of the water table above th~ base of the aquifer at
di tance r from the pumped well.
Thi equation was first proposed by Dupuit in 1863 and subsequently
modified by Thiem 1 in 1906. It suffers from the restrictive assumptions
required for its derivation and is now rarely u ed. A further serious
restriction is the fact that, because of the low velocities of groundwater
flow, true equilibrium conditions occur only after a very long period of
pumping at constant rate.
I
G. Thiem. "Hydrologische Methoden," J . M. Gebhardt's Verlag, Leipzig, 1906.
136
GROUNDWATER
6-12. onequilibrium hydraulics of well. During the initial

period of pumping from a new well, much of the di charge is derived from
storage in the portion of the aquifer unwatered as the one of depr ssion
develops. Equilibrium analysis indicates a permeability which is too
high be au only part of the discharge comes from flow through the
aquifer to the well. This leads to an overestimate of the potential yield
of the well.
In 1935 Thei 1 presented a formul a derived from a consideration of the
1"0/1
to'
m4
m'
06
r-------------~------------~------------~
I 100
Values of 1/
r---~~-----------4_0~.I~----------~t.qo
' - - - - - - ' - - - - - - - - - . l . . - - - - - - - - - l O.t
FIG. 6-9. Use of the Theis method for solution of a well problem.
u = 0.4,
W(u) = 0.7,
Z = 3.4,
Match point coordinates
r 2/t = 5.3 X 10 4
heat-flow analogy which account for the effect of time and the storage
characteristic of the aquifer. Hi formula i
Zr = -q47rT
f"-e-" du
u
(6-11)
wh re Zr is the drawdown in an observation well at distance r from the

pumped well, 7' is transmissibility, and u i given by
U
1.87r2S.
Tt
(6-12)
In Eq_. (6-12) t is th tim in days since pumping began, and S. is the

1 C. V. Theis, The Relation betw en the Lowering of the Piezometric Surface and
the Rate and Duration of Discharge of a Well Uijing Ground-wat r torage, TraWl.
Am. (Jeoph1l8. Union, Vol. 16, pp. 519-524, 1935.
NONEQUIL/BRIUM HYDRAULICS OF WELLS
137
storage con tant of the aquifer, or the volume of water removed from a
column of aquifer 1 ft quare when the water table or piezometri urface
is lowered 1 ft . For water-table aquifer it is e ntially the sp cific
yield. The integral in Eq. (6-11) is ommonly writt n as W(u) , called
t he" well function of u." It may be evalu ated from the eries
W(u)
- 0.5772 - log. tt
+u
u2
- -221
u
+ 331
-. ..
(6-13)
Values of W(u ) for variou values of u are given in Table 6-3 adapt d
from a more complete tabl by Weuzel. 1
Equation (6-11) i u ually solved graphically by fir t plotting a "type
curve" of u vs. W(u) on logarithmic paper (Fig. 6-9). From Eq. (G-12),
r2
T
- = - -- u
(6-14)
t
1.87S c
TABLE 6-3. Values of W(u) for Various Values of u
(After W enzel)
u
1.0
2.0
3.0
I 4.0
5.0
6.0
7.0
8.0
9.0
- -Xl
Xl0- 1
X l0- 1
Xl0- a
X l0- ~
X 10- '
X l0- '
X l0- 1
X 10- 3
X l0- '
X l0- IO
X 10- 11
Xl0- 12
X l0- 13
X l0- l<
Xl0- u
0 . 219 0.049 0. 013 0. 0038 0.00114 0.00036 0.00012 0.000038 0.000012

0.37
1.82 1.22 0.91 0. 70
0. 56
0.45
0.31
0.26
2.15
4.04 3.35 2.96 2. 68
2. 48
2.30
2.03
1.92
4 .73
6.33 5 . 64 5 . 23 4.95
4. 54
4.39
4 . 26
4. 14
8 . 63 7.94 7 . 53 7.25
6 . 69
6.55
6.44
7 .02
6.84
10 . 95 10.24 9.84 9.55
9 . 14
8.99
8.86
8 .74
9 . 33
11.29
11.16
13.24 12.55 12.14 11.85 11 . 63
11.45
11.04
15 . 54 14 . 85 14 . 44 14.15 13 . 93
13.75
13.60
13 .46
13.34
17.84 17.15 16 . 74 16.46 16 . 23
16.05
15.90
15 . 76
15.65
18 .20
18 .07
17 . 95
18 . 35
20.15 19 . 45 19 . 05 18 .76 18 . 54
22 . 45 21 .76 21.35 21.06 20.84
20 . 66
20.50
20 . 37
20.25
22.81
22 . 55
22.67
24 . 75 24.06 23.65 23 . 36 23.14
22 . 96
24.97
25 . 26
25.11
24.86
27.05 26.36 25.95 25.66 25.44
27.16
27.41
27.28
27.56
29.36 28.66 28.26 27.97 27 . 75
29.87
29.71
29.58
29 . 46
31.66 30.97 30.56 30 . 27 30.05
31.76
32.17
32.02
31.88
33 . 96 33.27 32.86 32 . 58 32.35
If q is onRtant, Eq. (6-11 ) indicates t hat Z. equals n. con tant times

W(u). Thus a curve of r2/t vs. Z. should be similar to the type curve
of u vs. HI(u) . After t he field observations are plotted, t he two curves
are superimposed with their axe parallel and adj usted until some portions
of the two curves coincide. The coordi nates of a common point taken
from the region where the curves coincide are used to solve for T and Se,
using Eqs. (6-11) and (6-14). Values of Zr and r2/ t may come from one
I L. K. Wenzel, Methods for D termining the Perm ability of Water-bearing
Materials, U.S. Geol. Survey Water-8upply Paper 887, 1942.
GROUNDWATER
138
well with various values of t, from several wells with different values of
r, or acorn bination of both.
When u is small, the term of Eq. ()-13) following log. u are mall and
may be neglected. Equation (6-12) indicates that u will be small when
t is large, and in this case a modificd solution of the Theis method i
possible l by writing
2.3q
t2
T = 41r t.Z loglo t;.
(6-15)
where t.Z is the change in drawdown between times tl and t2.

0
,~"r/"-:
r-.....
-- - -- -
, """-:
"-
1'\
::
.5
'"
------
[\.
0
0
;:::
0
10
c:::.
:::::
II
'"
<l
"'-\
''''~
12
14
---
Observed dolo
c:
"C
~
17
The
10
20
'\
40
60
80 100
Time since sforl of pumping in hr
FIG. 6-10. Use of the modifie d Theis method.
drawdown Z is plotted on an arithmetic scale against time t on a logllr

rithmic cale (Fig. 6-10). If t.Z is taken as the change in drawdown
during one log cycle, loglo (t2/tl) = 1 and T is easily determined from
Eq. (6-15). When Z = 0,
(6-16)
where to is the intercept (in day) obtained if the strn.ight-line portion of
the curve is extended to Z = O.
1
E. Jacob, DrulVdowD Test to D termine the Effectivo Radius of Artesian WillI,
Trans . A CE, Vol. 112, pp. ]047- 1070, 1947.
139
BOUNDARY EFFECTS
As in the Thiem equation, Theis as ume parallel tr amlin ,i .. ,

small drawdown and full pen tration of the well. Whil Thei ndjust
for the eff t of torage in the aquifer, he do s a sume in tl1nLan ous
unwateriug of the aquifer material a the water table drop. Th
conditions are rea onably well atisfied in arte ian aquifer. Howev r,
the procedure hould be used with caution in thin or poorly p rmeabl
water-table aquifer.
6-13. Boundary effects. The a sumption of a symmetrical cone of
depre sion implies a homogeneou aquifer of great extent.
uch an ideal
aquifer is rarely encountered, although in many ca e th condition i
approximated closely enougb for reasonable accuracy. Wh n s v ral
wells are 10 e together, their cone of depression may overlap or interfere
Ground surface
water table
FIG. 6-1 1. Effect of interference between wells.
and the water table appears as in Fig. 6-11. Where the con of depression overlap, the drawdown at a point is the sum of the drawdowns
caused by the individual wells. The two-dimensional analysis is greatly
simplified, but it should be evident that, when wells are located too close
together, the flow from the wells is impaired and the drawdowns iucren ed.
Figure 6-J2 shows an aquifer with a positive boundary in the form of
an intersecting surface stream . The gradient from the stream 1,0 the
well causes influent seepage from the stream. If the tr amftow is more
than the seepage, so that flow continues in the stream, the cone of depression of the well must coin 'ide with the water surface in the stream. A
rigorous analysis would requir that the channel be the full depth of the
aquifer to avoid vertical flow components. However, if the well i ' not
too close to the stream no serious error is introduced if tbis condition is
not satisfied. The method of image devised by Lord Kelvin for electrostatic theory is a convenient way to treat boundary problems. An
image well is assumed to have all the properties of the real well but to be
located on the oppo ite side of the stream and at the same distance from
it as the real well. Since the stream adds water to the aquifer, the image
well is assumed to be a r charg well, i.e., one that adds water to the
o.quifer. It cone of depr ion is the same as that of the real well but is
GROUNDWATER
140
inverted (Fig. 6-12). The resultant cone of depres ion for the real well
i found by ubtracting the dmwdown caused by th imag well from that
caus d by the real well (assuming no boundary). The corrected water
tab! between the real w il and the stream is, therefore, higher than without the effect of the str am. At the tream, the two drawdowns are
equal and the !let drawdown is zero. Thus the image well satisfies the
conditions first set up for this problem.
Image well----f
.....
....... ..-1
I
I
Cone of depression wif/lO/lf sfreom

Modified cone of depression
FIG. 6- 12. Image w ell sim ulating the effect of seepage from a stream on water levels
adjacent to a pumped well.
Negative boundaries, i.e., faults and similar structures across which no

groundwater is transmitted, can be analyzed in a similar manner. More
complicated boundary problems require judiciou
lection of multiple
image wells. Mo t geologic boundaries are neither abrupt nor straight.
nless the aquifer is very small relative to the zone of influence of the
well, the assumption of sharp discontinuitie is not erious. Relaxation
methods and electrical and membrane analogies have also been used in
the solution of boundary problems.
THE POTENTIAL OF A GR
NDWATER RE ERVOIR
A basic problem in engineering groundwater tudies i t he question of

the permissible rate of withdrawal from a groundwater basin. This
quantity, commonly called the safe yield, is defin d by Meinzer1 as:
. . . the rate at which water can be withdrawn for human use without depleting
the supply to such an extent that withdrawal at this rate is no l0nger economically
fea ibJe.
Many other definition of safe yield have been suggested and alternative
terms su h as ustaiued yield, feasible rate of withdrawal, and optimum
yield have been propo ed. The concept of afe yield has been mi used
10. E. Meinzer, Outline of Ground-wate Hydrology, with Definitions, U.S. Gwl.
Survey Waler-8upply Paper 494, p. 55, 1923.
SAFE YIELD
141
and hn receiv d on iderable criticism. Kazmann 1 ha ugge t d that

it be abandoned becllu e of it fr quent interpretation a a permanent
limitation on the permi ible withdrawal. It i important, th I' fore,
that the afe yield be I' cognized a a quantity determined for a p iii
set of controlling condition and ubje t to chang a a resu! of changing
economic or phy ical condition. It hould also be recognized that the
concept can be applied only to a compl te groundwater unit. The po sible withdrawaJ from a singJe weU or group of well in a field is affected by
a variety of factor such as size, con tru tion, and pacing of well a
well a by any control on the flow of groundwater toward the parti u!ar
field.
6-14. Safe yiel d. The safe yield of a groundwater ba in is governed
by many factor, one of the most important being the quantity of water
available. Thi hydrologic limitation is often expressed by the equation
(6-17)
where G is safe yield; P, precipitation on the area tributary to the aquifer;
Qs, surface streamflow from the same area; E T , evapotranspiration; Qu,
net groundwater inflow to the area; 13.80 , change in groundwat r storage;
and 13.8., change in surface storage. If the equation is evaluat d on a
mean annual basis, 13.8, will usually be zero.
With the exception of precipitation, all terms of Eq. (6-17) are subject
to artificial change and G may be computed only by assuming the conditions regarding each item. Artificial-recharge operations can reduce Qs.
Irrigation diversion from influ nt streams may increa.e evapotran piration. Lowering of the water table by pumping may in Tease groundwater inflow (or reduce groundwater outflow) and may make otherwise
effluent streams into influent streams.
The permanent withdrawal of groundwater from storage is called
mining, the term being used in the same sen e as for mineral resource .
If the torage in the aquifer is small, exccs ive mining may be di astrous
to any economy dependent on the aquifer for water. On the other hand,
many large groundwater basins contain vast reserves of water, and
planned withdrawal of this water at a rate that can be u tained over a
long period may be a wi e u e of this re ource. The annual increment of
mined water, 68 a of Eq. (6-17), increases the safe yield. Thus Eq. (6-17)
cannot properly be considered an equilibrium equation or olved in terms
of mean annual values. It can be solved correctly only on the basis of
specified assumptions for a stated period of years.
The factor which control the assumptions on which Eq. (6-17) is
solved are primarily economic. The feasibility of artificial recharge or
1 R. G. Kazmann, "Safe Yield" in Ground-water Development, Reality or Illusion ?, J. I rrigation and Drainage Div. ASCE, Vol. 82, ovcmber,1956. See also di~
cussion by McGuinness, Ferris, and Kramsky, Vol. 82, May, 1957.
GROUNDWATER
142
surface diver ion is usually determined by economic. If water levelR

in' th aquifer are lowered, pumping co ts are increased. Theoretically.
there is a water-tabl elevation at which pumping costs equal the value
of the water pumped and below which water levels should not be lower d.
Practically, the increased co t i often pa , ed on to the ultimate con umer
and the minimum level is never attained. Exce sive lowering of the
water table may result in ntamillation of the groundwater by inflow of
Ground surfoce
Salt groundwater
(0) Natural equilibrium
Pumped well
Salt groundwater
(b) Effect of pumping
FIG. 6. \3. Saltwater- freshwater relations adiacent to a coastline.
undesirable water. Thi hazard is especially prevalent near seacoasts

where ea-water intrusion ( e .6-15) may occur. A imilar problem may
d velop wherever an aquif r i adjacent to a source of saline groundwater.
Tran mi ibility of an aquifer may al 0 place a limit on safe yield.
Although Eq. (6-17) may indi ate a potentially large draft, this can be
realiz d only if th aq uifer i capable of tru.nsmitting the water from the
source ar a t the w 11 at a rate whi h i high enough to u tain the draft.
Thi probl m i e pecially likely to develop in long arte ian aquifers.
6-15. ca-wat r intrusion.
ince a water (specific gravity about
1.025) is heavi r than fresh water, the groundwater under a uniformly
permeable ci r ular i land would appear as shown in Fig. 6-13. The lens
of fresh water floating on alt water is known as a Ghyben-Herzberg lens,
SEA-WATER INTRUSION
143
afLer tile codiscoverer of the principle. About 40 ft of fresh water is

required below sea I v I for each foot of fr h water above a level to
maintain hydro tatic quilibrium. If a cone of depre ion is formed
about a well in the fresh water, alt water may rise 40 ft for each foot of
drawdown. Becau
of thi, horizontal kimming wells with mall
imilar intru ion can
drawdown are common in the Pa ific island.
deve10p along any coa tline if the water tabl 01' piezomeiri 1 vel drops
ufliciently low.
The be t means of controlling salt-water intru ion is to manage the
aquifer 0 that adequate water-table or piezometric levels are maintain d
FIG. 6-14. Contamination of deep aquifers from above.
at the coast. This requires a continuous flow of fresh water into the ocean.
l'his flow is part of Qu, the net groundwater exchange of the aquifer. It
has been suggested that an impermeable barrier might be constructed
across the eaward end of the aquifer to prevent intrusion without loss
of fresh water.
uch a barrier will b a major construction project if
ever attempted. An alternative barrier is a water mound formed by
injection of fresh water into wells near the coast. If the injection is
seaward of the salt-water front some ea water will be trapped in the
aquifer, but this will not be seriou if sufficient fresh water xists to
dilute it. Injection of treated sewage effiuent has proved effective in
southern alifornia.
So.lt water sometimes enters an aquifer through damaged or corroded
well casings. If a well passes through an aquifer containin r unde 'irable
water or through salt water into an underlying fresh-water aquifer, th salt
water can enter through the leakyca ingand drop down toth fre 'h water. t
J acob 2 has pointed out that several aquifer may exist at a coastline
(Fig. 6-14). Development of the upper aquifer fir t permits salt-water
intrusion, and leakage through the underlying aquiclud results in
I R. R. Bennett, Ground-water Situation at Baltimore Badly Confused, Water
Work8 Eng., Vol. 98, pp. 891, 904-907, August, 1945.
2 C. E. Jacob, Full Utilization of Groundwater Reservoirs, Trans. Am. Gcophya.
Union, Vol. 38, p. 417, June, 1957.
GROUND WATER
contamination of the d eper aquifers. In such a case withdrawal should

begin from the lower aquifer .
6-16. Artificial recharg. The safe yield of an aquifer can be
increa ed if additional water can be introduced into it. In some case 1
floodwaters are retained in surface r servo irs until they can be released
at rates a t or below the per 'olation capacity of the stream channels.
In other instances floodwaters are diverted to preading ground where
they are pond d until they infiltro.te. Infiltration rates have generally
been les than 5 ft/ day, and con iderable effort2 has been exerted to find
ways to increase infiltration rates by furrowing the spreading area, use
of chemicals or organic additives, etc.
uter 3 has reported rates as high
as 75 ft/ day in a recharge pit near Peoria, Illinois. The problems of
infiltration are most complex, and the developm nt of a spreading
program should be preceded by careful site. urveys and t sts.
Where recharge by spreading is not feasible because of low permeability
or lack of adequate area for spreading grounds, recharg by wells may be
attempted. The hydraulics of the operation is similar to that of a di charge well except that a groundwater mound build up around the well
and the rate of recharge i dependent on the rate at which this water
moves away from the well. Recharge wells are usually succe. ful when
the water i reasonably clean, but suspended solid or bacterial slimes
may clog the well screen or the adjacent aquifer. Storm water should
be settI d to remove suspended matter before recharging. The capacity
of a single recharge well is generally mall and the cost relatively high
compared with spreading unle producing wells are used for recharge on a
sea onal basis. This may be prohibited by health authorities unless the
water i chlorinated, because pollutants may be injected directly into
the aquifer.
To oft's t the large withdrawal of water for air conditioning on Long
Island, groundwater users are requir d to return an equal amount of
water to the aquifer. This is u ually done through 1'e harge wells. An
interesting r ult has be n a teady incr ase in groundwiLter temperature' resulting from the recharge of warmer water than that which was
withdrawn.
1 H. Conkling, Utilization of Ground-water
torage in Stream-system Developm nt, 7'rans . AS E, Vol. 111, pp. 275-354, 1946.
F. B. Laverty, Correlating Flood ontrol and Wat r Supply, 'l'rans. ASCE, Vol.
111, pp. 1127- 1174, 1946.
IL. Schiff, Water Spreading for torage Underground, Agr. Eng., Vol. 55, pp. 794BOO,1954.
I M. Suter, The Peoria Re harge Pit: Its Development and Results, J. Irrigation
and Drainage Div. AS E, Vol. 82, November, 1956.
'M. L. Brashears, Ground-water Temp rature on Long Island, N. Y., as AiIected
by Recharge of Warm Water, Econ. Geol., Vol. 36, pp. 811- 828, D comb~r, 1941.
ARTESIAN AQ'.I!FERS
145
A well con tructed clo e to a stream in rea e the per oln.tion to groundwater and is a form of artifi ial recharge. Becaus the water i filt red
a8 it move through the ground, the weU become a natw'al filter plant,
allowing development of pur wat r from [1 polluted stl' am at minimum
expense for treatment. The proc ss, known a induced percolation, 1
has been exten ivelyemployed along the Ohio River. Horizontal infiltration galleries may be constructed adjo.cent to a stream for the same
purpose.
6-17. Arte ian aquifers . The evaluation of the potential of an
arte ian aquifer involves some special factors. Th 'onfining strata are
commonly assumed to be watertight. If the p rmeabilityoftheaquiclude
is one Meinzer unit and the hydraulic gradient i unity, the daily seepage
would amount to about 28 mgd (80 acre-ft) / q mi. A quantity of this
magnitude would be quite significant in the groundwater excho.nge of an
aquifer. Hantu h 2 has demonstrated a procedure which accounts for
uch leakage in the analysis of pumping tests on artesian aquifers.
Artesian aquifers demonstrate considerable compressibility. This is
evident from numerous cn. es where fluctuations in tide level, barometric
pressure, or even the superimposed load of train are reflected in fluctuations of water level in wells penetrating the aquifer. If the pressure in
an artesian aquifer is relieved 10 'ally by removal of water, compre sion f
the aquifer may result with subsidence of the ground above it.
ucIt
subsidence has been observed 3 in areas subject to heavy withdrawal of
groundwater, with ground-surface elevations declining more than to ft.
The observation are not completely explained but it seems likely that
much of the compression is in the olays of the overlying aquicludes.
Aside from the disrupting effects of the surface subsidence, the phenomenon suggests that pumping tests on such aquifers may be misleading
because of the flow derived from storage as a result of the compression.
Although the small fluctuations appeal' to exhibit ela tic behavior, there
is no evidence that the ground levels in regions of pronounced subsidence
will be recovered if the aquifers are repressurized.
6-18. Tim effects in groundwater. Flow rates in the groundwater
are normally extremely slow, and considerable time may be involved in
groundwater phenomena. A criti 'allowering of th water table adjacent
to a coast may not bring immediate salt-water intrusion because of the
I R. G. Kazmann, Induced Infiltration Supplies Most Productive Well Field,
Civil Eng., Vol. 16, pp. 544-546, December, 1946.
1M. S. Hantush, Analysis of Data from Pumping Tests in Leaky Aquifers, Trans .
.tim. Gwphys. Union, Vol. 37, pp. 702-714, December, 1956.
IJ. F. Poland and G. H. Davis, Subsidence of the. Land Surface in the TulareWasco (Delano) and Los Banos-Kettleman City Area, San Joaquin Valley, California,
'frans. Am. Geophys. Union, Vol. 37. pp. 287- 296. June, 1956.
GROUND WArE,.,
1.t6
time r"~C"'( uired for the salt water to move inland. Werner l suggests that
several hundred years might be requir d for a sudden increa e in water
level in the recharge area of an extenr;ivc arte ian aquifer to be tran,
mitted through the aquifer. Ja ob 2 found that water level on Long
Island w re related to an IT tive precipitation which wa th um of the
rainfall for the previou 25 yr, each weighted by a factor which decl'ea ed
with time. McDonald and Langbein 3 found long-term fluctuation s in
streamflow in the olumbia Basin which they believe are related to
groundwater flu ctuution. Thu, in interpreting groundwater data it is
important to give full weight to the influ nce of time. Observed variations in groundwater I vels must be correctly related to causal factor 01'
serious misconceptions D111Y result.
BIBLlOGRAPIIY
Ferds, J. G.: Ground Water, Chap. VII, in C. O. Wisler and E. F. Brater, "flydrology," pp. 198- 272, Wiley, New York, 1949.
Hubbert, M. K.: The Theory of Ground-water Motion, J . Geol., Vol. 48, pp . 785- 1)44,
1940.
Jacob, C. E.: Flow of Ground Water, Chap. V, in H. Rouse (ed.), "Engiuooring

Hydraulics," pp. 321- 386, Wil y, New York, 1950.
Meinz r, O. E.: Ground Water, Chap. X, in "Hydrology," Vol. IX, Physios of the
Earth Series, pp. 385-476, McGraw-Hill,
IV YOI'k, 1942 (reprinted Dovel'
l'ubli 'ati DS , New York, 1949).
Muskat, M.: "The Flow of llomogeneous Fluids through Porous M dia," Edwards,
Ann Arbor, Mich., 1046.
Thomas, 1:1. E.: "The Conservation of Ground Water," McGraw-Hill, New York,
1951.
Tolman, C. F.: "Ground Water," McGraw-Hili, New York, 1937.

DATA
OURCE
Tho primnry source of groundwater da.t a is the bull tins Water Levels and Artesian
Pressures in the Unit d Stat s which are published annually as Wa.ler-supply Pa.pers
of the U.S. G ologi al urvey. Many stntes also publish groundwater data.
)ROBLEMS
6-1. An undlsturb d rock sample has an oven-dry weight of 652.47 g. After

saturation with kerosen.e ita weight is 731.51 g. It is th n immersed in keros nand
displa os 300.66 g. What is th porosity of tIle sampl ?
6-2. At tation A th water-table elevation is 642 ft above sea level and at B the
elevation is 629 ft. Th stations ar 1100 {t apart. The aquifer has a perme\\bility
1 P. W. Werner, Notes on Flow-time Effeots in the Gr at Artesian Aquifers of the
Earth, 'J'ran8. Am. Geophys. Union, Vol. 27, pp. 6 7- 708, Oetober, 1946.
t
. E. Jacob, Correlation of Ground-w!~t r Levels and Pr cipitation on Long
Island, N w York, 7'ran8. Am. Geophys. Union, Vol. 24, Part 2, pp. 564-580, H143.
and Vol. 25, Part 6, pp. 92 939, 1944.
I C. C. McDonald and W. B. Langbein, Trends in Runoff in the Paci60 NorthweAt,
Trans. Am. Geophys. Union, Vol. 29. pp. 387-397, June, 1948.
PROBLEMS
147
of 300 Meinzer units and a por sity of 14 p r c nt. 'W hat i the a('tual velocity of
Bow in the aquiler?
6-3. Ii tLe root zone in clay-loam soil is 3 It Lhi 'k, what qUl1l1tity of avnilabl
e Tab! 6-1.
moisture (in inches d pth ) should it hold?
6-4. A Boil sampl ha a oeffi i Ilt of p rmeability of 250 MoiHZ r units. What
would b it:.! permeability at 50 F?
6-5. A 12-in.-diameter well p n trates SO ft below th" static water tltbl . Aftor
24 hr of pumping at 1100 gpm the water I vel in a test well at 320 ft is lowered] .77 ft
and in a well 110 ft away the drawdown i 3.65 ft. What is th transmis ibility of
the aquiler'{ Use Eq. (6-10).
6-6, The time-drawdown data for an observation well 296 ft from a pumped well
(500 ~pm) ~r tabu lated below. Find th e transmissibility and storago constant of
the aquifer. Use the Theis method .
Time, hr
Drawdown, it
Time, hr
Drawdown, ft
1.9
2 .1
2.4
2 .9
3 .7
4 .9
7 .3
0 . 28
0 .30
0 .37
0.42
0 .50
0 .61
0 .S2
9 .8
12 .2
14 .7
16 .3
]S.4
21.0
24 .4
1.09
1.25
1.40
1.50
1.60
] .70
1.80
6-7. Tabulated below are the time-drawdown data for an observation well 150 ft
from a well pumped at 350 gpm. Find the transmissibility and storage constant by
the modified Theis method.
Time, hr ...... ...... . . . .. . . '1
Drawdown, ft ...... ....... . . .
l.S I~I~I~1
1.S
2 .4
3 .6
4 .3
] .0
5 .8
54 .0
S. l
6-8. A well 250 ft deep is planned in an aquifer having a transmissibili ty of ]0,000

gpd per foot width and a storage eoeffi cient of 0.010. The weU is expeded to yield
500 gpm and will be 12 in. in diameter. If th static water level is 50 ft below the
ground surfa e, estimate the pumping lift at the end of 1 yr and 3 yr' of operation.
6-9. After pumping a new 12-in . well for 24. hr at 150 gpm, th e drawdowns in a
number of nearby observa.tion wells are as given I elow . Find the storage co ffi'ient
and transmissibility of the aquifer.
Well nUID ber ...... . .. .
2
5
1
4
6
7
3
- - - - - -- -1 - - -1- - - - - - - - - - - - - - - - - Distance, ft .. .. . . , . . . .
100
141
190
200
23
347
400
--------11---1--- ---------------
Drawdown, It. . . . . . . . .
10 .5
7 .5
6 .2
4.0
2.4
1.4
0.6
6-10. An IS-in . well is in an aquif r with a transmissibility of SOOO gpd p r foot

width Imd 11 storage coefficient of 0.07. What pumping rate can be a.dopted 80 that
th maximum drawdown after 2 yr will not exceed 20 it?
6-] 1. A 24-in. well is in an aquifer with a transmissibility of 10,000 gpd per foot
width and a "torage coefficient of 0.05. Draw the profile of the cone of deprcssion
after 1 yr of pumping at 500 gpm. If a fault is located 1000 ft from this well, what
'Would be the profile of the eone of depression?
148
GROUNDWATER
6-12. The well of Prob. 6-11 is 800 It from a stream which flows all year. How
much is the drawdown mjdway between the well and the stream decreased because
of this seepage?
6-13. Using data from the Water-8u.pply Paper8 or other source, find out what you
can about tbe trend of groundwater levels in your area. What explanation can you
see for the obs rved trends? What is the source of the groundwater? Is an overdraft of the available supply indicated? Are there any possible ways of improving
th yield?
6-14. For a. basin selected by your instructor, make an estimato of the safe yield.
assuming no change in pr sent conditions.
7
CHARACTERISTICS OF THE HYDROGRAPH
The water which con titutcs streamflow may reach the stream channel
by any of ev ral paths from the point where it fir t reaches the earth a
precipitation. Some water flow over the soil surface a surfac runoff and
reaches the stream soon after it OCCUlT nce as rainfall. Other water
infiltrates through the soil surface and flows beneath the urface to th
stream. This water moves mor slowly thl1n t he surface runoff and contributes to the ustained flow of the stream during periods of dry wenther.
In hydrologic studies involving rate of flow in streams it is usually
necessary to distinguish between these components of total flow. The
fir t step in such studies is often the division of the observed hydrograph
of streamflow into component , preliminary to the analy is of the relation
between rainfall and runoff ( hap. 8), the determination of th ho.racteristic shape of hydrographs for a basin ( hap. 9), or a study of drought
conditions ( ec. 11-10).
7-1. The component of runoff. The actual route followed by a
specific water particl from the time it reaches the ground until it enters
a stream channel is devious. It is convenient to visualiz three main
routes of travel: overland flow, interflow, and groundwater flow.
Overland flow or surface runoff is that water which travel' over the
ground surface to a channel. The word" channel " a u ed here r fers
to any of the depre sions which may carry small rivulet of water in
turbulent flow during and for a short while after a rain.
uch channels
are numerou , and the distance water must travel as overland flow is
relatively hort, rar ly more than 200 ft and commonly much less.
Therefore overland flow soon reaches a channel and, if it occurs in sufficient quantity, i an important element in the formation of flood peaks.
Th amount of urface runoff may, however, be quite small, for surface
flow over a permeable soil surface can occur only when the rainfall rate
exceeds the infiltration capacity ( hap. 8). In many small and modernte
storms, surface runoff may occur only from impermeable surfaces within
the basin or from precipitation whjch falls directly on the water surface
149
150
of the basin. Except in urban areas, the total of impermeable area and
water !olurfa i usually a small part of the ba in area. Henc , surface
runoff is an important factor in streamflow only a the result of heavy or
high-inten ity rains.
Some of the water whi h inflltrate the soil surface may move laterally
through the upper soil layers until it enter a stream channel. This
water, called interfiow or subsurface flow , moves more lowly than the
surface runoff and reaches the streams somewhat later. The proportion
of th total runoff which occurs as interflow is depend ent on the geology
of the basin. A thin soil over overlying rock or a hardpan or plowbed a
short distance b low the soil surface favor ' sub tantial quantities of
interflow, whereas uniformly permeable soil en ourag s downward percolation to groundwater. Although traveling more slowly than overland
flow, interflow may be much larger in quantity, especially in storms of
moderate intensity, and hence may be the principal source of water for
the smaller ris s of streamflow. There is evidence that interflow in the
Coweeta Forest, North arolina, may account for 85 per cent of the total
runoff. 1
Some precipitation may percolate downward until it reache the water
table (Chap. 6). This groundwater accretion may eventually discha rge
into the streams as groundwater flow, al 0 alled base flow and dry-weather
flow, if the water table intersects the stream channels of the ba in. In
this v nt the stream are said to be effluent. The groundwater ontrihution to streamflow cannot fluctuate rapidly b ause of th d viou
path followed and th lower velocity. In som r gions more than 2 yr is
r quired 2 for the effect of a given accretion to groundwater to be dischal'g d into the str am .
Basin having permeable surface soil' and large, effluent groundwater
bodj s show 'u tained high flow throughout the year, with a relatively
small ratio between flood Bow and men.n flow. Ba ins with urface soil
of low permeability or influent groundwater bodies will have higher
ratio of peak to average flow and very low or zero flows between .flood .
Hydrographs for [Lcll type of basin ar shown in Fig. 7-1. Hat reek
drains volcanic t rro,in with a large groundwn.ter contribution, while the
Santa Yn z Riv r is influent throughout most of its length.
The distinction which hav b en drawn between th three component.
of flow are arbitrary and to orne degree, artificial. Water may start
out as surface runoff infiltrate from the sh et of overland flow, n.nd compI te it trip to the 'tream as interflow. On the oth r hand, interflow
1
R A. Hertzl'T, Engineering Asp cts of the Influence of Fore til on Mountain
Streams, Civil Eng., VoL. 9, pp. 487- 4 !), 1939.
C. C. McDonald and W. B. Langh in, Trends in Runoff in the Paoific Northwest,

7'rans. Am. Geophys. Union, Vol. 29, pp. 387-397, June, 1948.
151
STREAMFLOW RECESSIONS
may come to the urface where a relatively imperviou strntum inter'ects a hillside and finish it journey to th stl' am a overland flow.
de cription of interflow i imilar in Dlany way to what ,va
"per hed groundwater " in hap. 6.
rtainly what is d crib d u.s
interBow varie from groundwater only in pe d of travel. In lim ton
terrain, groundwater frequently mov at r Intiv ly high v 10 ities as
turbulent flow through olution chann 1 and Iractur in th lim stone.
7
I
I
I
I
I
_lnstontaneolJs peok = 33,000 cfs
I-Sanla YnBz R. nr SonIa Ynez, Calif:

Waler year 1951, total flow 124,000 ac. -ff
.!!! 5
u
o
o
Q
.~ 4 1-- .
~
;;:
HoI Cr. 01 Hot Cr., Calif.

Water yeor 1951,10101 flow ~ lll,OOO a\ft
~~ . .\. 1
\
Oct.
Nov.
Dec.
Jon.
Inston'1neolJs peal< =350 cfs
Feb.
Mar.
Apr.
May
T
June
July
Aug.
Sept.
fiG. 7-1. Comparison of hyd rographs from two streoms of differing geologic choracteristics.
Streams in lime Lone country often exhibit a high ratio of f1ood-p ak

flows to average flow, a condition characteristic of streams having small
groundwater contribution. In such terrain, groundwater Bow actually
has orne of the characteristics ascrib d to interflow. In practice, therefore, it is customary to consider the total flow to be divid d into only two
parts : storm, or direct, runoff and base flow. The distinction is actually
on the basis of time of arrival in the stream rather than on the path
followed. Direct runoff is presumed to consist of overland flow and a
substantial portion of the interfiow, whereas base flow is considered to be
largely groundwater.
7-2. Streamflow reces ions. A typical hydrograph reSUlting from
an isolated period of rainfall is shown in F ig. 7-2. The hydrograph
consists of a rising limb, crest segment. and falling limb or recession. T he
CHARACTERISTICS OF THE HYDRO GRAPH
152
shape of the rising limb is influenced mainly by the character of the storm
which caused the rise. Storm factor are discus ed in detail in Chap. 9.
The point of inflection on the falling ide of the hydrograph is commonly
assumed to mark the time at which surface inflow to the channel system
ceases. Thereafter, the recession cw-ve represents withdrawal of water
from storage within the ba in. The shape of the recession is largely
independent of the characteristics of the storm causing th rise. On
large basins subject to runoff-producing rainfall over only a part of the
400
r-Rislilg limb.+Crest. f--Rec6SS1On

on
U 300
0
0
52
.=
., 200
oJ:;
on
is
100
L\
/
16
17
18
19
March,1936
Drainage area =
5936 sqmi
1\""-
I---
20
21
fiG. 7-2. A typical hydrograph showing the nomenclature for its components.
River at Shepherdstown, West Virginia.)
iPotomac
basin, the reeession may vary from storm to storm, depending on the
particular area of runoff generation. If rainfall occurs while the recession
from a previous storm is in progress, the recession will naturally be
distorted. However, the recession curve for a basin is a useful tool in
hydrology.
The rece sion curve, ometime called a depletion curve because it
represents depletion from storag , is described by a characteristic depletion equation
(7-1)
wh re qo is the flow at any time, ql i the flow one time unit lat.er, and Kr
i a recession con tant which is less than unity. Equation (7-1) can be
written in the more general form as
(7-2)
where g. is the flow t time unit after qo. The time unit is frequently
taken as 24 hr although on small basin a horter unit may be necessary.
The numerical value of Kr depends on the time unit elected. Integrating
153
Eq. (7-2) and remembering that the volume of water discharged durin
i.ime dt is q dt and i equal to the clecrea e in storage -d ' during the
same interval, the torage , remaining in the ba in n,t time t is
s, =
_ _ q_,_
(7-3)
log. 1(,
Equation (7-2) will plot as a straight line on emilogarithmic paper

with q on the logarithmic scale. If the rece slon of a stream rise i~
fOO801==
~~~~~~
60 ~--~--~\
~--~--~---+----~--~--~----+---~
40 ~--+---~~~--~--+---4---~---+--~----r-~
~
'"
20
0
0
e
.S:
10
""<;'"
.r:.
'"
i5
"\ '\
./
~6'A~
7 1
7 IT
Surface runoff p,~

/uJL
s ..!!.
in!!..!
' fe<!..lr"!.{Jla'!.!'w~-1 _ _-I
'"
"~,> \
4r-'-lr--b-<'X=--k:::---+----+---+----+--I
-..........
..l~ --~~_ t\.
K,=o.91
~ --~I'-
1 '\ ) ~ ""-,
S"f,:~::J,f;_\\-~~~
_-=-G~.:..a;:.:un~d.:w..:. :. :al.:..erTr..e.:. :. :ce:.s.:. sl.:;
:. :'an.:..--+---1
J"
/1
16
I
17
I
18
'~;
19
~
20
21
22
23
24
25
March,1936
FIG. 7 -3. Semilogarithmic plotting of a hydrograph, showing method of recession analysis.
plotted on semilogarithmic paper (Fig. 7-3) the result is usually not a

straight line but a curve with gradually decreasing slope (i.e., increasing
values of 1(,). The reason for this is that the water is coming from three
different types of storage-stream channels, surface soil, and the groundwater-each having different lag characteristics. Barnes l suggests that,
the recession can be approximated by three straight lines on a semilogarithmic plot. Often, however, the transition from one line to the
next is so gradual that it is difficult to select the points of change in slope.
The slope of the last portion of t he recession should represent the charI
B. S. Barnes, Discussion of Analysis of Runoff Characteristl.)s,
Vol. 105. p 106. 1940
Tran~.
ASCIG,
154
acteristic K, for groundwater since, presumably, both interflow and

surface runoff have cea ed. By projecting this slope backward in time
(Fig. 7-3) and replotting the difference between the projected line and
the total hydrograph, a reces ion which for a time con ists largely of
in terBow is obtained. With the slope applicable to interBow thus determined, t he process can be repeated to establish the recession characteristics of surface runoff.
The technique described above represents a degree of refinement rarely
nece sary for engine ring problems. The ba 'e-flow rece sion curve i
most frequently us d, and one method for developing such a curve is to
700
600
Indlviduol storm recessions
:!
500
u
.!: 400
300
Time in days
FIG. 7-4. Construction of a composite recession curve.
piece togeth r ctions of rece 'ion from various torms until a composite
curv (Fig. 7-4) is obtained which covers the nece sary range of flow
rates. A template onforming to this curve is often convenient.
A rece sion curve may al 0 be developed 1 by plotting values of qo
against q, some fixed time t later (Fig. 7-5) . If Eq. (7-1) were strictly
corr ct, the plotted data would indicate a straight line. Normally,
however, a curve indicating a gradual change in the value of K, results.
This curve becomes asymptotic to a 45 line as q approaches zero.
The methods illustrn.ted in Figs. 7-4 and 7-5 may be u ed to construct
recessions for ba Bow, direct runoff, or total flow. For a base-flow
recession, data hould be selected from periods several days after the
peak of a flood 0 that it is reasonably certain that no direct runoff i
included. After the base-Bow roce sion has been established, it may be
projected back under the hydrograph immediately following a flood
peak, and the difference between the projected ba e flow and the total
hydrograph used to develop a direct-runoff reces ion curve. When the
method of Fig. 7-5 is used, it is customary to draw the base-flow curve
1
W. B. Langbein, Some Cha.nnel Storage and Unit Hydrogra.ph Studies, Trani.
Am. Geophllc. Un.ion, Vol. 21, Part 2, pp. 620-627, 1940.
155
to envelop the plotted dat a on the right. The argument for this is that
such a curve repre nts the slowe t rece ion (high K r) and that points
deviating to the left may include direct, runoff. By a imilar argum nt
it is common to envelop the daia for the dir at-runoff recession on the left.
45
40
35
30
....
til
0
0
25
I:::r
20
iL:
15
10
_4
10
o
o
/
/
"
... Direct runoff

recession
/
~
~
~
~
/00
$2
.=
JI
fV
!v
"
"
""
"
""
"
"
~"
15
20
10
Flow 24 hours loter in 1000
25
30
~ fs
FIG. 7-5. Recession curves in tne form q o vs. q l for tne American River at Fair O aks,
California.
A total-Bow reces ion will show more variation from storm to storm
because of the varying quantities of direct and base flow. Total-flow
recessions can be derived by either of the methods just described. At
high Bows when the relative base-flow contribution is small, the t otal-flow
recession approaches the direct-runoff recession.
CHARACTERISTICS OF THE H"fDROGRAPH
156
7-3. H ydrograph separation. Becau e of the differing characteri tics of direct and groundwater runoff, it is common practice to divide the
hydrograph into two components as a basis for subsequent analysis.
Thi proces is known as hydrograph separation or hydrograph analysis.
Since there is no real basis for di tingui bing between the direct and
groundwater flow in the stream at any instant and since the definitions
SJ
Y
6dO
Jdoys
kOf t ....
....
'f0 . . SI'""Abo
. ht_ -utr!2
__ :..::..-=..-:...-:...-_-:. - - - - - - - - - - -
",,""
Too long
DaysFIG. 7-6. Selection of the time base for the surface-runoff hydrog raph.
of the e tw components are relatively arbitrary, the method of separation

is usually equally arbitrary.
Practically, the method of eparation hould be such that the time
base of direct runoff remains relatively con. tant from storm to storm.
This is usually provided by terminating the direct runoff at a fixed time
after the peak of the hydrograph. As a rule of thumb, the time in days
N may be approximated by
N = AO.2
(7-4)
where A is the drainage area in square miles. However, N is probably
better determined by inspection of a number of hydrograph , keeping in
mind tha.t the total time ba.se should not be excessively long and the rise
HYDROGRAPH SEPARATION
157
of the groundwater hould not be too great. Figure 7-6 illu trat
orne
rea on able and unreasonable a umptions regarding N.
The most widely u d eparation pI' edure con i t of ext nding th
ecession existing prior to the storm to a pint under the peak of the
hydrogmph (AB, Fig. 7-7). From this point a straight line is drawn to
the hydrograph at a point N days after the peak. The reasoning behind
this procedure i that, as the stream rises, there i flow from the stream
Days
FIG. 7-7. Some sim pl e base-flow sepa ra tion procedures.
into the banks (Fig. 7-8). Hence, the base flow should decrease until
stages in the stream begin to drop and the bank storage returns to the
channel. While there is some support for thi reasoning, there is no
justification for assuming that the decrease in ba e flow conforms to the
usual rece sion. Actually, if the increment of bank storage is greater
than the inflow from the groundwater, the base flow i effectively negative.
Hence, this procedure is quite arbitrary and no better than line AC (Fig.
7-7), which is simply a straight line from the point of rise to the hydrograph N days after the peak. The diff renee in volume of base flow
by these two method ' is quite small and probably unimportant as long
as one method is u ed con istently.
A third method of separation is illu trated by line ADE (Fig. 7-7).
This line is COD ' tructed by projecting the reces ion of the groundwater
after the storm back under the hydrograph to a point under the inflection
158
point of the falling limb. An arbitrary rising limb is sketched from the
point of ri 'e of the hydrograph to conne t with the project d ba e-flow
r cession. This type of separation may have some advantages where
groundwater is relatively large in quantity and reaches the stream fairly
rapidly, as in limestone terrain.
7 -4. Analysis of complex hydrographs. The discu sion of hydrograph separation in Sec. 7-3 a sumed an isolated streamflow event without
subsequent rainfall until after the direct runoff had left the basin. This
type of event i ea ier to analyze than the complex hydrograpb re ulting
Low flow conditions
High woter conditions
FIG. 78. Bonk-storage variations during a flood .
from two or more clo ely spaced bur ts of rainfall (Fig. 7-9). Often,
however, analysis of the more complex cases cannot be avoided. In
the e case it is nece sary to separate the runoff cau ed by the individual
bursts of rainfall in addition to separating di.rect runoff from ba e flow.
If a simple base-flow separation such as line ABC or AC of Fig. 7-7 is
to be used, tbe divi ion between bursts of rain is usually accompli hed
by projecti.ng the small segment of recession between peaks, U 'ing a
total-flow recession curve for the basin (line AB, Fig. 7-9). The baseflow separation is then completed by drawing CDB and EF. The
direct runoff for the two periods of rain is given by the shaded areas
marked I and II. Note that a eparation of this type is impracticable
unless tbere are two clearly defined p aks with a short segment of recession following the fir ..t penk.
A separation of the typ illu trated by ADE in Fig. 7-7 can also be used
for compl x torms.' The ba e-flow reces ion DE of the second rise (Fig.
7-10) is constructed by projecting the ba e-flow rece sion backward.
The ba e flow during the period of reces ion after the first peak can be
1 R. K. Linsloy and W. C. Ackennann, A M thod of Predicting the Runoff from
Rainfall, 'l'ran8. ASCE. Vol. 107, pp. 825-835, 1942.
ANAL YSIS OF COMPLEX HYDROGRAPHS
159
I-RoinI
TimeFIG. 7-9. Separation of complell hydrographs, using recession curves.
t
31:
o
Li:
Time_
FIG. 7-10. Separation of complell hydrographs, using qo
VS. ql
curves.
determined by noting that

(7-5)
where liqd is the change in direct runoff in a unit of time, liqb is the change
in ba e flow during the same time unit, and liq is the change in total
flow during the period. If Eq. (7-1) applies to both direct and base
160
flow, then
(7-6)
where Kd and Kb are th rur ct- and ba -flow re e sion coefficients
and q, is the total flow at the end of the period. Remembering that
qd, = qo - qb" then
(7-7)
Even though the r
ssion constants cannot be determined, a graphical
r lation between go, gl, and qb, can be derived by trial, u ing the rece ion
'U 4
o
o
.. :3
.5
.<:
i5
10 11 12 13 14 15 16 17
Dole
FIG. 7-11. Determination of total-runoff volume.
curves. To do this, a value of qb, is estimated and the corresponding

value of /::"qb determined from the r ces ion. Since Aq is known, /::"qd can
be computed from Eq. (7-5). This value of Aqd should conform to that
indicated by the direct-runoff recession curve for a value of qd, = qo - qb,.
If it does not, a second trial is indicated. A relation between /::"qd and
qd, is helpful and can ea ily be constructed from the recession curve.
After the position of the groundwater rece sion has been established,
the balance of the groundwater hydl'ograph is drawn to conform to these
values. The dire t runoff for th two storms can then be eparated by
extending the total recession b tween the two until it intersects the corresponding ba e-flow curve (extended).
7-5. Determination of total runoff. In some types of analysis
( ee Sec. 5-10) there i ne d to determine the total streamflow (direct
runoff plus groundwater) r sulting from a particular storm or group of
storms. This can be done by computing the total volume of flow 0 curring during a period beginning and ending with the same discharge and
encompa sing the ri e under consideration, provided that groundwater-
PROBLEMS
161
rece sion conditions prevail at both times. In Fig. 7-11 the ar a under
the hydrograph between times A and A', or B llnd B ' , repr nt th
total runoff from the torm causing the fir t ris.
imilariy, that betw n
times 0 and 0 ' constitute the total runoff PI' du d by th last tbre
storms. If groundwater-rece ion condition do not pI' vail at th b ginning and end of the ri e under tudy, the l' cion mu t be extend d as
described earlier. Proof of the foregoing i found in the fact that, from
continuity, total runoff for the period AA' must equal the ob rved
streamflow plus any change in storage b tw -en A and At. But from
Eq. (7-3) the storage is a fun tioll of the flow, and ince th flows at A
and A' are equal, the change in storn.ge mu t be zero. Hence th ob 'erved
streamflow must equal total runoff.
BlBLlOCllAI'UY
Linsley, R.. K., M. A, Kohler, and J . L. H. PauUlus : II Applied Hydrology," Chap. 15,
pp. 387-444, McGraw-llill, New York, 1949.
PltOBLEM
7-1. Using thc sample pag!' from a Water-8upply Paper which appears as Fig. 4-18,
construct direct- and basc-flow recession curves of tbe type illustrated by Fig. 7-5.
Are the values of Kr constant for these curves? What are th avemge valu s of Kr
for direct and base flow? Using an average valu of K r , find til volume of groundwater storage when the flow is 1100 cfs.
7-2. Tabulated below arc ordinates at 24-hr intervals for a hydrograph. Assuming
these ordinates arc for th basin analyzed in Prob. 7-1, separate the base flow from
the direct runoff by each of the three methods illustrated in Fig. 7-7 . Computo the
volume of direct runoff in each case.
Time, days
Flow, cia
Tim , days
Flow, cCB
1
2
3
4.
5
6
7
2,340
34,300
25,000
14,000
8,060
5,740
4,300
8
9
10
11
12
13
14
3,230
2,760
2,300
2,060
1,770
1,520
1,320
7-3. Plot the data of Prob. 7-2 on semilogarithmic paper and determine r cession
for surface runoff, interfiow, and growldwater now. What volum of each
'It the three (,OTTlpononts is present?
I~onstants
8
RUNOFF RELATIONS
Precipitation is tbe primary factor determining streamflow. The

sequence of vents is such that runoff lags behind the precipitation producing it, the amount of the lag depending upon the characteristics of th~
drainage area. Reliable rainfall-runoff relations and techniques for
di tributing rUlloff through time are the basis of efficient design and
operation of hydraulic projects. They are ulso important tools in river
forecasting.
TIlE PHENOMENA OF RUNOFF
Much of the rain falling during the first

part of a torm is stor d on the vegetal cover as interception and in surface
puddle as depression storage. As rain continues, the oil surface becomes
cover d with a film of water, known a surface detention, and Bow begins
downslope toward an e tabli hed surface channel. En route to a channel,
the water is de ignated as overland flow and, upon entering a channel, it
becomes surface runoff. That part of storm precipitation which does
not appear either a infiltration or as surface runoff during or immediately
following the storm i surface retention. In other words, surfa 'e retention
includes inter eption, deprc ion torage, and evaporation during the
torm but does not include that water which is temporarily stored en
route to th streams.
Although the effect of veg tal cover is unimportant in major floods,
interception by some types of cover may be a considerable portion of the
annual rainfall. Interception storage capacity is usually satisfied early
in a storm so that a large p rentage of the rain in the numerous small
storms is intercepted. After the veg tation is saturated, interception
would case were it not for the fact that appreciable water may evaporate
from the normous wett d urfa 'e of the foliage. Once interception
storag i fill d, the amount of wat r r aching the oil surface is equal to
tb rainfall J
vaporation from the vegetation. Inter eption storage
capacity is reduc d as wind sp d increases, but the rate of evaporation
8-1. Surface retention.
It-2
SURFACE RETENTION
163
is increased.
Apparently, high wind peed tend to augment total

inter eption dUl'ing a long torm and to decrease it for a short torm.
Extensive experim Ilta! interception data hav b ell ac wnulat d by
numerous inve tigators,l but evaluation and application of th data to
specific problem are made difficult because of varied experim ntal
technique employed. Data for forest cover, relatively more plentiful
than tho e for crop and other low-level vegetation, are u ually obtained
by placing several rain gages on the ground under the canopy and comparing the average of their catch with that in the open.
ometime th
gages (interceptometers) are placed at random in an attempt to m a ur
average interception for an area. In other Cll es, the int rceptometers
are placed at carefully el ctcd point undel' the tr e crown to m IlSUI'
interception on its projected area. In either co. e, appli 'ation of th
data requires de toiled knowledge of the cover density over the area of
interest. Most data for fore t have been collected by placing int rceptometers free of underbrush, gra s, etc., and little is known of total
interception as required in most hydrologic problems.
Trimble and Weitzman 2 found that mixed hardwood about 50 yr old
and typical of considerable area in the southern Appalachian Mountains
intercepts about 20 per cent of the rainfall both in summer and winter.
No measurements were made of flow down the tree trunk., but it is
likely that net interception is nearer 18 per cent for storms with rainfall
in the order of 0.5 in. Qualitatively, it call be said that annual interception by a well-developed forest canopy is about 10 to 20 per cent of
the rainfall and that the storage capacity of the canopy ranges from
0.03 to 0.06 in.
Horton 8 derived a series of empirical formulas for estimating interception (per storm) by various type of vegetal cover. Applying these
formulas to I-in. storm. ' and a .'uming normal cover density give the
following values of interception:
1 R. K, Linsl y, M. A. Kohler, and J. L. H. Paulhus, "Applied Hydrology," pp.
263- 268, McGraw-Hili, Tew York, 1949.
p, B. Rowe and T. M. Hendrix, Interception of Rain and Snow by Sc ond-growth
Ponderosa Pine, Tran8. Am. GeophY8. Union, Vol. 32, pp, 903- 908, D cember, 1951.
W. M. Johnson, The Interception of Rain and Snow by a Forest of Young Pond rosa Pine, Trans. Am. Geophys. Union, Vol. 23, Part 2, pp, 566-570, 1942.
J. Kittredge, H. J, Loughcad, and A. Mazurak, Interception and Stem Flow in a
Pine Plantation, J. Fore8try, Vol. 39, pp. 505-522, June, 1941.
N, L. Stoltenberg and C. W. Lauritzen, Interception Storage of Rainfall by Corn
Plants, Trans . Am. GeophYB. Union, Vol. 31, pp. 443-44.8, April, 1948.
2 G. R. Trimble, Jr., and S. Weitzman, Eff ct of a Hardwood
orest Canopy on
Rainfall Intensities, Tran8. Am. GeophY8. Union, Vol. 35, pp. 226-234, April, 1954.
I R. E. Horton, Rainfall Interc ption, Monthly Weather Rev., Vol. 47, pp. 603-623,
September, 1919.
1604
RUNOFF RELATIONS
Cr p
orn ........... . .... . . .
COtton ......... . ...... .
Tobacco ............... .
mall grains . . ......... .
Meadow gra ' .. ....... .
AHalfa .... . . .. ........ .
High t,
ft,
4
4
3
1
Inter 'eption,
in.
0 .03
0 .33
0 .07
0 .16
0 .08
O. ll
Rainwater retained in puddle, ditches, and other depres. ions in the

soil surface is termed depression storage. The e depression vary widely
in area and depth ; their size depend to a considerable degree on the
definition of a depressi on. As soon as rainfall intensity exceeds thfl
infiltration capacity ( c. -2), the rainfall excess begins to fill surface
depressions. Each d pression has its own capacity and, when filled,
further inflow is baJanc d by outflow plus inflltration and evaporation.
Depressions of various sizes ar both superimpo 'ed and interconnected.
Almost immediately aft r th beginning of rainfall exces , the smalie t
depressions become filled and overland flow begins. Most of this water
in turn ruls larger d pre' ion , but some of it follows an unobstructed
path to the t,'eam channel. Thi 'hain of events continues, with successiv 1y larger portions of overland flow contributing to the stream .
Water held in depr sions at the end of rain is eit her evaporated or
absorbed by the 'oil through infiltration.
VirtuaUy nothing can be said of the magnitude of depression storage,
since m aningful observations cannot be easily obtained, and values
are highly dep ndent on definition of terms. Individual depressions of
appreciable area relative t o the drainage basin under consideration are
usually call d blind drainage and ex luded from hydrologic analysis.
The remaining dep!, sion stomge i usually lumped with interception and
treat d as initial loss with re pect to storm runoff.
revertheles ,
depres ion storage may be of considel'a bl magnitude and may play an
important role in th hydrologi cyl.
tock ponds, terraces, and contour farming all tend to moderate th flood hydrograph by increasing
depres ion storage, whil land leveling and drainage reduce depre sion
storage, Because mo t basin have some very large depressions, it is
likely that the depres ion-stomge capacity of the basin is never completely filled.
8-2. Infiltration. Infiltration i the pas age of water through the
soil surface into the soil. Although a di. tinction is made between infiltra
tion and percolation, the movement of water within the soil, the two
phenomena are clo ly related since infiltration cannot continue unim-
INFILTRATION
165
peded unless percolation remove infiltered water from th urhe soil.

The oil i permeat d by noncapillary channel through which gmvity
water flows downward toward the groundwater, following the path of
least resistance.
apillalY forces continuou ly divert gravity water into
capillary pore spaces, 0 that the quantity of gravity water pas ing
',uccessively lower horizons is steadily dimini hed. Thi lead to increa ing re istance to gravity flow in the surface layer and a d rasing rate of
infiltration as a storm progr e. The rn.te of infiltration in th early
FIG. 8-1 . Simple separation of inflltratian and surface runoff, using hourly rainfall data.
estimated retention, and an infil tration-capacity curve.
phases of a storm is less if the capillary pores are filled from a previous
storm.
There is a maximum rate at which water can enter the soil at a particular point under a given set of conditions; thi rate is called the infiltration
capacity. The actual infiltration rate Ji e'luals the infiltration capacity
Ip only when the supply rate i, (rainfall intensity Ie rate of retention)
equals or exceeds 111' Theoretical concepts presume that actual infiltration rates equal the upply rate when i, ~ I" and are otherwi e at the
capacity rate (Fig. -1). The value of 11' is at a maximum fo at th
beginning of a torm and approaches a low, constant rate fe as the soil
profile becomes saturated. The limiting value is controUed by subsoil
permeability. Horton 1 found that infiltration-capacity curves approxiI
R. E. Horton, The Role of Infiltration in the Hydrologic Cycle, Tran8. Am.

Union, Vol. 14, pp. 446-460, 1933.
Geophl/8.
166
RUNOFF RELATIONS
mat the form

(8-1)
where e is th N apierian base, k i an empirical constant, and t is tim

from beginning of rainfall. The equation i applicable only when i. ;::: fT>
throughout the storm.
In{iltration capacity depends on many factors such as soil type,
moisture content, organic matter, vegetative cover, and season. Of the
soil chara teristics affecting infiltration, noncapillary porosity is perhaps
the most important. Poro ity determines storage capacity and also
3.0
,,
2.5
...
.t:.
~ 2.0
"-
.E
.~ 1.5
~.=
1.0
0.5
I',
"'-
..._
\ - r- __
'"
WeI
- Cecl'l sandy loom

- - Houston block loom
15
I"'--
30
...... ..._
~I
._
Iniliol
45
._
--- --
..._ ..!;zll/o/
60
75
90
-- -105
120
135
Time. minutes
FIG. 8-2. Comparative inflltration rates during initial and wet runs.
and Musgrave.}
(After Free, Browning,
affects 1'e istance to flow. Thus, infiltration tends to increas with

porosity. An incr ase in organic matter also re ults in increa d infiltration capacity, largely b cause of a corr 'ponding change in poro ity.l
Figur 8-2 demonstrate the effect of initial moisture content and the
variations to b expected from oil to soil. The effect of vegetation on
infiltration capacity is difficult to determine, for it al 0 influence interception. Nevertheless, vegeto1 cover does increase infiltration as compared with barren soil becau (1) it retards surface flow, giving th water
additional time to enter the soil; (2) the root y tems make the soil
more pervious' and (3) the foliage hicld the soil from raindrop impact
and reduce rain packing of the surface soil.
Mo t data on infiltration rates are derived from infiltrometer tests. 2
1 G. R. Fr e, O. M. Browning, and G. W. Musgrave, Relative Infiltration and
Rclat d Physical Characteristics of Certain Soils, U.S. Dept. Agr. Tech. Bull. 729,
).940.
t
Hydrology Handbook, ASCE Manual 28, pp. 47-51, 1949.
THE RUNOFF CYCLE
167
An infiltrometer i a tube, or other boundary, de igned to i olat a

of soil. The efT ctive area vari s from Ie than on quar foot to vera!
hundred square feet. Although many of the earlier te t w r mnd by
flooding the infiltl'ometer, this method i no long r recornmend d and ha
been supplanted by sprinkling technique. 1 ince it is impo ibl to
mea ure directly the quantity of water penetrating the soil urface,
jnfiltration is computed by a suming it to equal the cliff renee b tween
water applied and mea ured. urface runoff. In addition to the difficulties
inherent in simulating raindrop ize and v 10 'ity of fall with sprinld rs,
experiments using artificial rainfall have oth I' features tending to cause
higher infiltration rates in te ts than und I' natural conditions.
Attempts have been made to del'ive infiltration data by analyzing rainfall and runoff data from small drainage ba. ins with homogeneou ' soils,2
and Horton, 3 h rman,4 and others have present d techniqu , for driving
average" equivalent" infiltration-capacity data from record of heterogeneou basins. On natural basjl1s allowance must be made for subsurfa e flow and surface retention.
ince pre ent method of hydrograph
analysis do not permit reliable s paration of. ul'face and ubsurface flow,
it is extremely difficult to obtain reliable infiltration data by analysis of
data from natural bllsins.
8~3. The runoff cycle. Tho runoff cycle is th> descriptive term
applied to that portion of the hydrologic cyc! between incid nt precipitation over land areas and subs quent discharge of thi water through
stream channels or evapotranspiration. Hoyt 5 has prese nt d a comprehensive description of the hydrologic phenomena occurring at . elected
times during the runoff cycle by con 'id ring an idealized cros. . cction of
a basin. In the discu ion here, it appears more instructive to consider
the time variations of thc hydrologic factor during an extensive storm
on a relatively dry ba in.
Figure 8-3 show schematically the dispo ition of a steady, moderateintensity rainfall. The dotted area of the figure r presents the portion
of total precipitation which eventually becomes streamflow measured at
the ba in outlet. Channel precipitation is the only increment of stream1 Tests to determine infiltration rates during irrigation are customarily made by
flooding.
I A. W. Zingg, The D termjnation of Infiltration Rates on Small Agricultural
Watersheds, Tran8. Am. GeophY8. Union, Vol. 24, Part 2, pp . 476- 4 0, 1943.
I R. E. Horton, Determination of Infiltration Capacity for Large Drainage Basins,
Trans. Am. Geophys. Union, Vol. 18, pp. 371- 385, 1937.
L. K, Sherman, Comparison of F-curves Derived by the Methods of harp and
Holtan and of Sherman and Mayer, Trans. Am. Geophys. Union, Vol. 24, Pa.rt 2,
pp. 46()-467, 1943.
5 W. G. Hoyt, An Outline of the Runoff Cycle, Penn. State Coil. School of Eng., Tech.
B1Lll. 27, pp. 57-{)7, 1942.
RUNOFF RELATIONS
168
flow during the initial period of rainfall. A the treo.m ri e, their total
Burfa c area increases and, con equently, the volume rate of channel
precipitation increo. es.
The rate of interception is high at the beginning of rain, especiaUy
during summer and with den e vegetal cover. How vcr, the available
stomge cl1pacity i.' depleted rather quickly, so that the interception rate
decrea es to that r quired to r place water evaporated from the vegetation. Most of the interception is eventually retum d to the atmo phere
by evaporation.
Time from beginning of rainfall
FIG. 8-3. Schematic diagram of the disposition of storm rainfall.
The rate at which depre ion storage is fill d by incident precipitation

decrea 'es rapidly from a high initial val ue a the smaller depre sions
be ome filled. It approaches zero at a relatively high value of totalstorm rainfall, depending on land slope, ba in ar a, and other factors.
While depre ion storage is shown as a total 10 to streamflow in the
simplified figure, a portion of thj water does infiltrate and reach the
stream as groundwater or interflow. The remainder is returned to the
atmosphere through evaporation.
Except in very intense torm, the greater portion of the soil-moisture
d ficiency is sati tied b fore appreciable surface runoIT takes place.
However, som of the rain 0 cuning late in the storm undoubtedly
be omes soil moi ture, since th downward movement of this water is
relatively slow.
Water infiltrating the soil surface and not retained as soil moisture
either moves to the stream as inter flow or pen trates to the water table
and ventuaUy r aches the stream a groundwater. There is a difference
of opinion on the quantitative division of this water between interfiow
INITIAL MOISTURE CONDITIONS
169
and groundwater. Figure 8-3 shows a gradual decrease in the contribution to interBow but this may not be true in every case.
The rate of surface runoff start at zero, incr ase slowly at fir t and
then more rapidly, eventually o.pproaching a relatively con tant percentage of the rainfall rate. Both the per entage and the rate of runoff
are dependent upon rainfall intensity. B cau e the e 'act division
between surface runoff and inter80w is uncertain, the details of the diagram are not precise in this respect.
Figure 8-3 illu trate only one of an infinite number of possible cases.
A change in rainfall intensity would change the relative magnitude of all
the factors. Further complications are introduced by varying rainfall
intensity during the storm or by occurrence of now or frozen ground.
To appreciate further the complexity of the proce s in a natural basin,
remember that all the factor of Fig. 8-3 vary from point to point within
the basin dUl'ing a storm. Neverth Ie ,the foregoing description 'hould
aid in understanding the relative tim variations of hydrologic phenomena
which are important in considering the runoff relations discuss d later
in the chapter.
ESTIMATING TIlE VOLUME OF STO RM RUNOFF
8-4. R ainfall v . torm runoff. Until recently, the estimation of

runoff as a percentage of storm rainfall was standard practice. The
applicable coefficient varies from near zero for a small storm to a relatively
high value for a major storm and is dependent upon initial moisture
conditions. Although the experienced hydrologi t may be rea on ably
successful in selecting the proper coefficient for each storm, the method is
extremely subjective, and its use cannot be advocated for estimating
the runoff from specific torms. The coefficient method is still widely
used in the design of storm-drainage and small water-control structures
(Sec. 9-10) and is the basis of design required by many city and county
codes.
The variation in runoff coefficient with magnitude of storm can be
taken into account by graphical correlation (Sec. A-I) of rainfall and
runoff (Fig. 8-4).
uch relations are typically curved, indicating an
increasing percentage of runoff with incI'ea:ing rainfall. Only a moderate degree of correlation can be expected, however, since variations in
initial moi ture conditions are not considered.
8-5. I n itial m oisture condit ions . The quantity of runoff produced by a storm depend ' on (1) the moisture deficiency of the basin at
the onset of rain and (2) the storm characteri tics, such as rainfall
amount, int nsity, and duration. The storm characteristics can be
determined from an adequate network of precipitation gage, but the
direct determination of moisture conditions throughout the basin at the
RUNOFF RELATIONS
170
beginning of the storm is not feasible. While reliable observations of

soil moisture can be made at a point, three-dimensional measurements
are required in a medium recognized for its marked physical di continuities. Also, any complete accountability of moisture within a basin must
include consideration of depression and interception storage.
In humid and subhumid areas where streams flow continuously, groundwnter discharge at the beginning of the storm has been found to be a good
index to initial moi. ture conditions (Fig. 8-5).
ince recent rains affect
3~--'---~----.----r--~----.----r---.----.----r----ro~-'
.,
II>
-5 2
.:
.:
...'0
c:
~
E
(;
v;
Storm precipitation in inches

FIG. 8-4. Rainfall -runoff
(817 sq mil.
relation
for
Monocacy
River
at Jug Bridg e, Maryland
current moisture deficiency, even though they have no effect on streamflow, this index should be supplemented by a weighted index of the rainfaU for several days preceding.
Pan-evaporation data are ometimes u ed in the computation of a
moisture index. In a study of the Valley River, North Carolina, Lin ley
and Ackermann 1 found that field-moisture deficiency at any time was
approximately equal to 90 per cent of the total lass A pan evaporation
since the ground was last aturated, Ie s any additions made to field
moisture by intervening rains. Bnsin-accounting techniques (Sec. 5-14)
applied on a daily basis provide a reasonably accurate estimate of moisture
deficiency which can be u ed a an index to runoff. 2 This approach is
laborious but should yield excellent results.
The most common index is ba ed on antecedent precipitation. The
1 R. K. Linsley and W. C. Ackermann, Method of Pr dieting the Runoff from Rainfall, Tran8. ASCE, VoL 107, pp. 825-846, 1942.
1M. A. Kohler, Computation of Evaporation and Evapotranspirati n from Meteorological bScfVatiotl6, U.S. Weather Bur. Rqlt" Hl57 (m im('ograpilf'd ,l.
171
INITIAL MOISTURE CONDITIONS
rate at which moi ture is depleted from a particular hn in under p cified

meteorological condition i roughly proportional to the amount in
storage ( ec. 5-14). In other words, the soil moisture should decr a e
logarithmically (Appendi'{ B) with time during periods of no precipitation, 1
( -2)
where 10 is the initial value of antecedent-precipitation index, It is the

1.6 ,.----,.--....---.---r--,--,.....,.-".--r-r--7T-_,.----,.----,
'"
-5
.s
1.0
.S
E
E
0.6 I---- r - - + _ Ul
Storm rainfall In Inches
FIG. 8-5. Rainfall-runoff relation for Susquehanna River at Towanda, Pennsylvania, using
groundwater flow as a parameter (7797 sq mil.
reduced value t days later, and lc is a reces ion factor ranging normally
between 0.85 nnd 0.98. Letting t equal unity,
11
kla
(8-3)
Thus, the index for any day is equal to that of the previous day multiplied
by the factor lc (Appendix B). If rain occurs on any day, the amount of
rain is added to the index (Fig. 8-6). Since storm runoff does not add
to the residual moisture of the basin, an index of precipitation minus
runoff, i.e., basin recharge, should be more satisfactory than the precipitation index alone. Commonly, however, the minor improvements go.ined
do not justify the added computation.
Equation (8-2) a sumes that the daily depletion of soil moisture
1 M. A. Kohler and R. K. Linsley, Predicting the Runoff from Storm Rainfall,
U.S. Weather Bur. Research Paper 34, 1951.
RUNOFF RELATIONS
172
(primarily evapotranspiration) is
10 - II
= 10 (1
- k)
(8-4)
ince actual evapotranspiration is a function of the potential value and

the available moisture (1 0), k bould be a function of potential evapotranspiration. The variation in potential evapotranspiration is largely
seasonal, and Eq. (8-2) has be n found to be reasonably satisfactory when
Gr-.--r~~--T-~-r~--r-~~-T~r-~.-~-.--r-'-I
Days
FIG. 8-6. Variation of antec:edent index. with doily precipitation.
used jointly with calendar date ( ec. -7). There is an added advantage
in using the date as a parameter becau e it also reflects variations in
surface conditions related to farming practices, vegetation, etc.
The value of tho index on any day theoretically depends on precipitation over an infinite antecedent period, but if a reasonable initial
value is assumed, the computed index will closely approach the true
value within a few we ks. The index value applicable to a particular
storm is taken as that at the beginning of the first day of rain. Thus a
value of 1.8 in. would be used for the storm of the 9th and 10th in Fig. 8-6.
8-6. Storm analy i. In any statistical correlation, it is extremely
important that th basic data be a consistent and reliable as po ible.
M tbods of torm analy i hould be rigorou and objective. Only t hat
storm rainfall wbi h produced t he runoff being considered should be
COAXIAL RELATIONS FOR TO TAL STORM RUNOFF O R BASIN RECHARGE
173
included. Small how r oc urring aft l' the hydrogrnph ho.d tarted
to r cede should not be included if they had littl eff t upon the a.mount
of runoff.
imilarly, showers occurring before the main storm should be
excluded from the storm rainfall and included in th ante d nt-pr ipitation iud x. Long, complex storm ' should be epamted into a many
short storm period as po sible by hydrograph analy is. Averag basin
rainfall for the elected storm period is computed by one of the method
described in hap. 3.
Runoff al 0 depends upon rainfall intensity, but for ba ins of 100 q
mi or more, an average int n ity as refl cted by amount and dumtion
is u ually adequate. In thi ca e duration can be estimated with ufficient accuracy from 6-hr rainfall data. All obj ctive rule is pI' ferabl ,
such as "the urn in hours of those 6-hr period with more than 0.2 in.
of rain plus one-half th intervening p riods with Ie s than 0.2 in."
Although experimental infiltration data indicate rates commonly in
excess of 0.1 in./ hr, relations such as Fig. 8-8 consistently show the
effect of duration on storm runoff to be in the order of 0.01 in. / hr. The
difference is largely caused by intercorrelations and the inclusion of
interflow with surface runoff.
Surface runoff is the re idual after interception, depression storage,
and infiltration are taken from rainfall. H nce, it is not uncommon to
use the difference between rainfall and runoff (basin recharge) as the
dependent variable in correlations. The definition of r harge is contingent on that us d for runoff. The contribution to groundwater may
be treated as part of total runoff' or it may be considered a groundwater
recharge if one is interested in estimating only storm runoff.
8-7. Coaxial relations for total storm runoff or basin rech arge.
If storm characteristic and ba in conditions are to be adequately r presented in a runoff relation, the correlation must include a number of
indep ndent factors as well as the dependent variable (either runoff or
recharge). The coaxial method (Sec. A-3) of graphical correlation is
particularly suited for this work. 2
To illustrate, a sume that a relation for estimating basin recharge is
desired, using antecedent precipitation, date (or week number) , and rainfall amount and duration as parameters. Values of each of these parameters are compiled for 50 or more storms. A three-variable relation is
developed first (Fig. 8-7, cbart A) by (1) plotting antecedent precipitation vs. recharge, (2) labeling the points with week number, and (3)
fitting a smooth family of curves repre enting the various weeks.
hart
1 R. K. Linsley and J. B. Franzini, "Elements of Hydraulic Engineering," p. 35,
McGraw-Hill, New York, 1955.
1M. A. Kohler and R. K. Linsley, Predicting the Runoff from Storm Rainfall,
U.S. Weather Bur. Research Paper 34, )951.
174
RUNOFF RELATIONS
B is placed with its horizontal scal matching that of chart A to facilitate

plotting. Point are plotted in chart B with ob erved recharge as
ordinate and r charge computed from chart A as ab 'cissa, and the e
points are label d with duration. A family of mooth curve i drawn to
represent the effect of duration on recharge. Charts A and B together
A
1/ I
'-f~
II I Ti
I I~ I
/ '~
ij
./il3
,~'I ~I- zt~
l~
/
/~
_MD
~/
: /
. f-i
V1 ~ ~/1
I
iI
~VIJ_~~-.l.l'.i....J,_,--o
~ ' /11
;/
.V
!Y'
.~ r ---
1Jf
I
I
I/ .
--
1/
III
j1
II
-- -- -- -~i
oL,--,_j_._,--._ - . cow.UTEoJ
FIG. 8-7. Basin-recharge relation for the Monocacy River at Jug Bridge, Maryland.
(AFter U.S. Weather Bureau.)
are a graphical relation for estimating recharge from antecedent index,

week, and duration .
torm precipitation is th n introduced (chart C) by
(1) plotting recharge computed from charts A and B against observed
recharge, (2) labeling the points with rainfall amount, and (3) fitting a
harts A, B, and con titute the first approximation
family of curves.
to the desired relation. Chart D indicates the over-all accuracy of the
derived charts.
Since the parameters are intercorrelated and the first charts are devel-
COAXIAL RELATIONS FOR TOTAL STORM RUNOFF OR BASIN RECHARGE
175
oped independently of factor sub quently introduced, r vi ion of the

charts may improve the over-all relation. In other word , the proce '
is one of succe 'ive approximations. To check the week curv ,th
other curve families are a sumed to be correct and the adju ted abs is a
for a point in chart A is determined by ntering charts B o.nd C in r v l'S
order with ob erved recharge, rainfall amount and duration. The
ordinate for the adjusted point is the ob 'erved ant cedent-precipitation
index. In other words, the week eurve should be revised to fit thi
adjusted point if the relation i to yield a computed recharge qual to the
ob erved recharge. The econd (and subsequent) approximations for
duration and rainfall are made in the same manner. In ach co. e the
points are plotted by entering the hart equence from both end with
ob 'erved values to determine the adju ted coordinate '.
Although the method presented in the previou paragraph i general
and can be used as described, certain modifications simplify the procedure
and require fewer approximation.
ince storm rainfall is extremely
important, the first plotting of chart A may show little correlation, and
the con truction of the curves will be diffi ult. However, an important
advantage in having the rainfall parameter in the last chart is that the
po sibility of computing runoff in excess of rainfall or of computing
negative values of runoff is eliminated. Moreover, the arrangement of
Fig. 8-7 results in the determination of a unified index of moisture conditions in the fir t chart, which is a decided advantage in river forecasting.
If the plotting of chart A is limited to storms ho.ving rainfall within a
specified class interval (2 to 4 in., for example), the construction of the
curves i simplified, provided that there are sufficient do.ta. Only
limited data are required since the general curvature and convergence
are always as shown in the example. The relations are quite similar
throughout a geographic region, and chart A and B for one basin may
be u ed as a first approximation for another basin in t he area.
Figure 8-7 shows that enol'S for points with negligible runoff are considerably magnified when routed back through the chart sequence to
derive second-approximation curves. Hence, low-runoff points may
well be omitted from this tep.
When basin recharge for a storm has been computed from Fig. 8-7,
the runoff is estimated by subtractir.g recharge from observed rainfall.
Since both rainfall and recharge appear on chart C, this subtraction can
be eliminated by reorientation of the curves to read runoff directly
(Fig. 8-8). The charts can also be superimposed (Fig. 8-9) to conserve
space without reducing scale.
To use the runoff volume computed from a relation such as Fig. 8-8
with a unit hydrograph (Chap. 9) to synthe ize an extended-storm
hydrograph requires that runoff increments be e timated for successive
RUNOFF RELATIONS
176
time periods. This is done by computing runoff from accumulated

rainfall up to the end of ea h p riod and subtracting these successive
values to obtain increments.
ince it is impo 'sible to segregate the water pas ing a gaging station
according to the portion of the ba in in which it feli, tatistical runoff
1'1
1 -;;;
II i
U 11i~ ~
~ / IU/_. ~
I ~4 !:::~
..~- /J..J..-"_~r
/JtL-IJj/~'
~
-uy~
y~1-1
./
....-::;:::V
4
-
<--
lz
~f-:_:j ~~
~f7
~ ~,Tr1~
1 7~~T~V-~'
Z
'j
STORM RUNOFF (Inch es)
1/1 / /
/~EJ";' .. -~-..y
~~17
--.1_ / :s.' I/ ~"D /
I
__
~.:~ ~B7
~' ~7j
~r~ (/~
~_I-
/I
__ ~Q/
/I ~~X/
~~~
~ :)/1
V.J /V
~\Q
I _/ /r_ -/_~t,;
4 1~ltl
f
'j
kfp_"
I / / /
~Q"+ l/
/
6
1/ / / j
{/
/ /r
/ I
FIG. 8-8. Storm-runoff relation for Monocacy River at Jug Bridge, Maryland,
U.S. Weather Bureau.)
(After
r lations must be based on basin averages of the parameters. Unfortunately, 3. r lation which i based on storms of uuiform areal distribution
will yield runoff values which are too low when applied to storms with
extremely uneven distributions. This can be demonstrated by computing
the runoff for 4, 6, and 8 in. of rainfall, a uming all other factors to
remain fixed. While 6 i the average of 4 and 8, the average runoff from
the 4- and &-\u. ra.infalls is not equal to that from a 6-in, rain, An uneven
COAXIAL RELATIONS FOR INCREMENTAL STORM RUNOFF
177
di tribution of antecedent pr cipitation produce similar r ults. Runoff relation ba ed on uniform areal ondition can prop rly b us d to
compute the runoff in the vicinity of nch rainfall station, and the averag
of these runoff values will, in g n ral, more nearly approach the obs rved
runoff from the ba in when either th torm or the ant c d nt precipitation i quite variable.
S-8. C OB ial relations for increm n tal storm ru noff. When
relations of the type shown in Fig. 8-8 are appli d to very small ba in ,
there is a decided tendency to undercornpute th p ak of the larger
floods. These rare event are often Lhe result of high-intensity, shortduration storms which cannot be reproduced without taking into account
the time di tribution of the rainfall. This difficulty can be ov r orne l
by revi ing the curve to permit stima(,es of hourly increments of runoff
from corre ponding hourly rainfall.
The chief difficulty in developing an hourly (inc.rem nt) relation lies
in the fa (, t,hat one cannot determine incr ment. of runoff for storms of
record by hydrograph analysi. Thus, valu . of the dep ndent. fac tor
in the correlation cannot he obtained directly from the streamflow
record. In practice, the runoff iner ments mu -t first be estimated from
a total storm relation (Fig. 8-8) as described in ec. 8-7. The correlation
proce permits succe ive revi ion of the derived runoff increments,
always maintaining the correct total-storm runoff. The derived in r ments can be checked qualitatively by application of a unit hydrograph
(S c. 9-8).
Still another problem is the derivation of a soi l-moisture index which
is representative of condition. at any time during a storm . Although not
as ensitive as might be desired, a relation such as that of chart A,
Fig. 8-7, is reasonably sat,isfa tory, if storm precipitation up to the
beginning of the time increment under consideration is included in the
antecedent-precipitation index.
To illustrate, as ume that a unit bydl'ograph and a relation such as
Fig. 8- are applicable to a small basin and that an in 'remental relation
such as Fig. 8-9 is desired. The analysis would pro eed a follow:
1. For each of a number of short .torms compute the accumulated
storm runoff, hour by hour, with Fig. -8. Obtain hourly increments by
subtraction of successive accumulated values, and adju t these increments
to yield the correct total-storm runoff.
2. Compute storm hydrographs by applying the increments just
derived to t he l-hr unit hydrograph ( ec.9-8). If hydrograph reproduction is not sati factory, adjust in remental runoff values as required.
3. Derive the antecedent-precipitll.tion index for each hour by adding
1 J. F. Miller and J. L. H. Paulhus, Rainfall-Runoff Relation for
'l'ranB. Am. Geophys. Union, Vol. 38, pp. 216-218, April, 1957.
mall Basins,
178
RUNOFF RELATIONS
the storm rainfall up to the de ignated hour to the initial value of the
index for th Lorm.
4. Entering Fig. _0 1 with the antecedent index and week number,
plot at the corrc ponding hourly runoff, and label the point with hourly
rainfall. Points hould be plotted so that tho e for each storm are readily
discernible (color 01' ymbol).
ince hourly in rement are les reliable
2.0
I,
HOURLY
II
14
ANTECEDENT
RUNOFF (INCHES)
1.2
1.0
0 .1
0 .'
INDEX
(INCHES)
FIG. 89. Hourlyrunoff relation for Little Falls Branch at Bethesda, Maryland.
U.S. Weather Bureau.)
(After
than total-storm runoff, the analyst may make compensating change

in the points for a giYen storm to obtain better agreement among torms.
5. onstl'uct hourly-rainfall curve to fit the plotted data.
6. Using the first-approximation curves, compute hourly-runoff value
by entering with antecedent index, week, and rainfall, and adju teach
set of values to the correct storm total.
7. To check the week curves, enter the relation with hourly runolT
(step 6) and rainfall, plot at the orresponding antecedent index, and
label each point with the week number. Revi e the week curves as
indicated by the plotted data.
1 The curve families of this relniion hav been superimposed for illustrative purposes. In derivation they should be separated to avoid confusion.
THE INFILTRATION APPROACH TO RUNOFF
179
8. If the week curve require revision (step 7) , the fir t-approximation

rainfall curves should be checked. Enter the relation with antee d nt
index and week, plot at the corr sponding runoff (step 6), lab 1 points
with rainfall, and revise the curves as required.
E:~.'perience with relations of this type i limited, but r ult are en ouraging. The development process outlined above is cumber orne and
time-consuming. With one such relation, however, its modification to
fit another basin in the same hydrologic region is achieved rath r quickly.
8-9 . The in filtration approach to runoff. Basically, the infiltration approach is quite simple. The urface runoff resulting from a given
torm is equal to that portion of the rainfall which is not dispo ed of
through (1) interception and depre sion torage, (2) evaporation during
the storm, and (3) infiltration. If it ms 1 and 2 are invariable Ot' insignificant or can be assigned rea 'onable value, one necd be concerned only
with rainfall, infiltration, and runoff. In the simplest case, where the
supply rate i. i ' at or in excess of the infiltration capacity, surface runoff
is equivalent to the storm rainfall les~ rmrface retention and the area
under the capacity curve. The computations are equally direct if i. is
continuou ly below i1' at the end of the storm (Fig. 8-1).
Natural rain of varying intensity and intermittently below the infiltration capacity results in a distortion of the capacity curve which is, by
definition, applicable only when i. ;:::: i1" When i, = 0, capacity must
increase with time, and the capacity may either increase or decrease with
time when 0 < i. < i.". It is often as umed that the infiltration capacity
at any time during a storm is determined by the mass infiltration which
has occurred up to that time. If rain begins at a low rat and the rainfall during the first hour is one-half the infiltration capacity, the capacity
at the end of the hour would be taken as that at about 72 hr on the applicable capacity curve.
The infiltration approach is frequently applied in deriving the design
discharge for airport drainage and similar projects when the drainage
area is homogeneous and small, but it i rarely satisfactory in operational
problems requiring reliable hydrograph prediction on a continuing basis.
Ev n when applicable, it can be used to compute only one component of
flow-surface runoff. Subsurface flow must be computed independently,
and this flow may constitute a major portion of the flood flow in some
areas. The infiltration concept can be applied to the rational computation of surface runoff from a natural heterogeneou ba in only when the
area is subdivided with respect to infiltration and storm characteristics
and an applicable capacity curve i available for each subarea. The difficulties in applying the theory have led to the use of infiltration indices I
1 H. L. Cook, The Infiltration Approach to the Calculation of Surface Runoff,
Trana. Am. GeophYB. Union, Vol. 27, pp. 726-747, October, 1946.
RUNOFF RELATIONS
180
which must be correlated with factors indicative of initial moi ture

conditions.
8-10. Infiltration indice. The index is defined as that rate of
rainfall above which the rainfall volume equal the runoff volume (Fig.
8-10).
iDce repre' nts the combined effect of infiltration, inter-
T ime
FIG. 810. Schematic diagram showing the meoning of the inde)(.
ceptioD, and depression storage, it is equivalent to basin re harge divided

by the duration of rainfall, provided that i ~ throughout the storm.
The W index is the av rage infiltration rate during the time rainfall
inten ity exceed ' th capacity rat ; i.e.,
W = -
= -t1 (P
- Q - S)
(8-5)
where F is total infiltration, t is time during which rainfall intensity

exceeds infiltration capacity, P is total precipitation corresponding to t,
Q is surface runoff, and S i the effectiv urface retention. The W index
is essentially equal to the index minus the average rate of retention by
interception and depre ion storage. While the segregation of infiltration and r tention would seem to be a refinement, the task of estimating
the retention rate is such that combining it with infiltration is probably
equally satisfactory.
With v ry wet conditions, when the infiltration capacity is at a minimum and the retention rate is very low, the values of Wand are almo t
PHYSICS OF SNOWMELT
181
equal.
nder the e conditions, the W index be orne the W min ind X by
definition. Thi index is u ed principally in tudie of maximum flood
potential.
Values of the indices are ea ily derived from rainfa,lI and discharg
data, using an objective method of hydrograph eparatioll. To apply
the indice the procedure i rever ed; runoff i computed from a select d
index value.
ince the variability of Wand cI> is comparable with that
of runoff, correlation are till requir db tween the index tLnd aut cedent
conditions. Neverth 1 , the approach i u ed to advantage in the
design of project when the as umption of minimum infiltration is
appropriate. I
ESTfMATI G
NOWMELT It
OFF
8-11. Phy ics of snowmelt. The basic problem of the hydrologi t

in regard to snow is twofold, First, hc is concerned with e timating the
rates of snowmelt from his knowledge of the heat supplied and, second,
he must determine bow the res ulting melt water, p rhaps in combination
with concurrent rainfall, "ill affect streamflow. This section treats the
physical aspects of snowmelt; practical solutions to snowmelt problems
are covered in ec. 8-12.
From a physical point of view, the snowmelt and evaporation processes
are quite similar, Both are thermodynamic processes and can be
treated by the energy-budget approach. For each inch of water melted
from snow at 32F, heat must be supplied in the amo unt of 750 Btu/ q ft
or 203 caljsq cm. Snowmelt computations are simplified by the fact
that the surface is always 32F during melt and, at the same time, are
made more complex by variation in albedo, the percentage of solar and
diffu e radiation reflected by the surface. The energy for snowmelt is
derived from (1) net radiation, (2) conduction and convective tran fer
of 'en ible heat from the overlying air, (3) conden ation of water vapor
from the overlying air, (4) conduction from the underlying soil, and (5)
heat supplied by incident rainfall.
Components of radiation exchange are discu sed in Sec. 5-3 in connection with evaporation, Although only 5 to 10 per cent of incident
hort-wave radiation is reflected by a fre water surface, 0 to 90 per
cent is reflected by a clean dry snow surfa e. As snow ages, it alb do
drops to 50 per cent or less becau e of changes in crystalline structure,
density, and amount of dirt on the surface, Snow radiates e entially
as a black body, and the outgoing long-wave radiation (32F) during a
24-hr period is equivalent to about 3,3 in. of negative melt. The net
loss by long-wave radiation is materially Ie than this amount since the
1 P. Leatham and H. S. Riesbol, Infiltration and Detention Tests as Related to
Spillway Design Floods, Tram, Am, GeophYB, Union, Vol. 31, pp. 234-242, April, 1950,
RUNOFF RELATIONS
~82
atmosphere reradiates a portion back to the earth. Net long-wave

radiation depends primarily on air and now temperatures, atmospheric
vapor pressure, and extent and type of loud cover. With clear skie
n t 10. by long-wave radiation i qual to about 0.8 in. of melt per day
wh n d wpoint and air and snow temperatures are near 32F. Numerous empirical formula have b en derived to estimate this factor l and
net all-wave radiation, but they are not particularly satisfactory. Radiometer observations of net radiation could provide the needed data if an
adequate network of stations was established.
Heat exchange between a snow pack and the atmo phere is also effected
by the pro sse of conduction, convection, condensation, and evaporation. Although it i readily shown that conduction in still air is very
mall, convective exchange can be an important factor. 2 If the dewpoint of the air is above 32F, condensation on the snow occurs, with
con equ nt release of latent heat. If the dewpoint is les than the snowsud ace temp rature, the vapor pressur of the snow i greater than that
of th air nnd evaporation occurs. The rate of transfer of ensible heat
(conve tion) is proportional to the tempemture difference between the
air To and the now T" while vapor transport i . proportional to the vaporpressure gradient e" - e,. Both proces es arc proportional to wind
velocity v. Since the latent heat of vaporization is about 7.5 times the
latent heat of fusion, condensation of 1 in. of water vapor on the snow
surface produc
.5 in. of water, in luding onden ate. The two
proces e can b de crib d by similar eqmLtion
MaoDV'~'IOD = kr(To - 32)v
M coDdenI = k.(e o - 6.11)v
(8-6)
( -7)
where 1\1 i m lt, k is an exchange coefficient, and 6.11 is the vapor

PI' sure (millibar) at 32F.
Temp rl1ture beneath the oil surface do not vary a much or as
rapidly as surface temperatures, and th re is usually an increase of
temperature with depth during the winter and early spring. This results
in a transfer of heat from the soil to th snow pack above. While the
heat tran fer is small on a daily basis, it may be equivalent to several
inches of melt during an entire eason- sufficient to keep the soil saturated
and to permit a prompt response of streamflow when melting from other
causes take place.
IE. R. Anderson, Energy-budget tudj s, Water-loss Inv Btigations, Vol. 1, Lake
Befn r tudi s, U.S. Geol. Sllrvey Profe8s. Paper 269 (reprint of U.S. Geol. Survey
eire. 229), pp. 71- 119, 1954.
" now Hydrology, ummary Report of the now Investigations," pp. 145-166,
North Pacific Division, U.S. Corps of Engineers, Portland, Ore., June, 1956.
I P . Light, Analysis of High Rates of Snow Melting, Trans. Am. Geophys. Union,
Vol. 22. Part I, pp. 195- 205,1941. .....
ESTIMATING SNOWMELT RATES AND CONSEQUENT RUNOFF
183
Raindrop temperatures corre pond clo ely to th surfa e wet-bulb

temperature. As the drop enter a now pack, their temp ratur i
reduced to 32F and an equival nt amount of heat is imparted to th
snow. The melt from rain i given by
= P(T", - 32)
/I{
'01.
144
(8- )
where P i the rainfall in inche , T withe wet-bulb temperatur (OF),

and 144 is the latent heat of fu ion in Btu per pound. Thus the heat
available in 1 in. of rain at 50F will melt about '8 in . of wat r from th
snow. There has been a tenden y to ov remphasiz the importanc of
rainfall melt, perhap h 'aus warm rains are arcompanied by high
humidity and temp rature and oft n by mod rat, \'0 , trong wind .
From a theoretical point of view, it would app ar that Sllowm It
computations could be made by applying the energy-hudg t approach,
and thi is basically true for snowmelt at a point. On 11 natuml basin,
howc\'er, tbe hydrologist is confronted with numerou complications
which have not yet h en .'uccessfully treated. Among the more obviou
complications are the etTect of variations in elevation, lope, a p ct,
forest cover, and albedo on radiation , temperntur , wind, and other k y
factors in snowmelt. The heat required to melt snow is dependent on
its thermal quality ( ec. 3-1 :~) which may vary widely within a basin.
Finally, since the hydrologist is concerned with the runotT from snowmelt,
water retain d in the snow pack i not etTective. Limited data indicate
that snow can retain from 2 to 5 per cent liquid water by wight. Th se
complexities and the scarcity of pertinent data have made the th or tical
approach generally inapplicl1ble for operational u e. This does not
preclude u e of these concepts in studying upper limits of snowmelt for
design purpo es.
8-12. Estitn a t ing s nowmelt rates a n d consequ e n t runoff. Of
all the pertinent meteorological factors, air temperature is hy far the
mo, t reliable index to snowm It. J n fact, it ,0 completely reflects
radiation, wind , and humidity that residual errors are not materially
correlated with the e factors. Since snowmelt does not 0 cur with
temperatures appreciably below freezing, the data are commonly converted to degree days or degree hours above some base, usually 32F.
A day with a mean temperature of 52F represents 20 degree days above
32F. If the minimum temperature for the day is above freezing, then
there are 20 X 24 or 4 0 d gree hours. On the other hand, there are
actually more than 480 degre hours in a day with a mean temperature of 52F if the temperature drops below 32F for a portion of th
day. Degr e-day computat.ions are usually ba d on dry-bulb temperature readings becau e uch data are readily available, but hetter
RUNOFF RELATIONS
184
correlation is sometime found hetween snowm It and wet-bulb r adings.

Degree days are sometime computed from a base of 30F for convenience.
In some cases base temperature as high a 40F have been used on the
l.rgument that the shelter temperature is higher than that at the now
surface. In a few instances degree day have been omputed from maximum rather than mean temperatur . Since any value that is u ed is
only an index to the available heat, it is probable that little real difference
in utility exists b tween the various methods of computing degree days.
If)
2.5 -
~
u
.!:
-g
'0
~
i3
u
10
<t
0.5
10
15
20
25
30
Accumulated degree - days (32 0 F bose)
since lost freezing lemperatures
FIG. 8-11 . Temperature-snowmelt relation for Pennsylvania.
35
(AFter Snyder.)
nowro It relations are generally determin d by correlating runoff

with degl' e day and possibly other factor. Sometimes the degree-day
factor, the ratio of nowmelt (01' runoff) to the concurrent number of
degree days, is tr ated as the dependent variable. As originally conceived, the degr e-day factor was expected to be relatively constant,
but actual variation are often of such magnitude that li ttle advantag i.
gained from its use. Direct use of degree day in a correlation provides
greater flexibility. Average degree-day factors for a melt period range
between 0.06 and 0.15 in. per degree day, if the basin is completely
snow-covered. The factor in rea es with time after the beginning of
melt if computed from str amflow, largely because of the detention of
arly melt water in the now and soil. As the snow cover becomes patchy,
the factor again decreases (Fig. 8-]3).
lope, aspect, and degree of
forest cover cause yu.riation in degree-day factor from basin to basin
ESTIMATING SNOWMELT RATES AND CONSEQUENT RUNOFF
185
and easonally in mountainou tert'ain. In tudi s of w tern P nn ylvania, Hortonl found the factor to vary from 0.09 on thinly co or d
basins to 0.06 on heavily fore ted area.
Figure -11 bow a snowmelt relation d rived for the usquebanna
River Bo. in. 2 This relation tend to account for the ripening and gradual
di appearance of the snow pack, but it provides estimates of snowm It,
~
o 65
co
(\j
I()
Li.J
60
c:
o
~
o
55
.,
~ 50
Weighted degree-days (32 of bose and lapse rate
ot 3 F.0 per 1000 t tl
FIG. 8- 12. Deg ree- day curves for the Tuolu mne River above Don Pedro Dam, California.
not runoff. Runoff must be computed by the u e of a rainfall-runoff

relation or similar technique.
imilar relations have been used in t he
plains area with moderate success.
In many basins of high relief, winter conditions are typified by semipermanent snow cover in the headwaters while the lower portions of t he
basins have only intermittent snow cover. The lower limit of snow cover
(snow line) i a dynamic featu re, moving up and down the slope with
passing time. In practice, the snow line i often consider d as a contour
of levation, a reasonably realistic assumption following a storm. As
melting proceeds, however, snow cover recedes more rapidly on southerly
1 R. E. Ho rton, In1iltration and Runoff during the Snow Melting Season, with
Forest Cover, Trans . Am. Geophys. Union, Vol. 26, Part 1, pp. 59-68, 1945.
sF. F. Snyder, "Cooperative Hydrologic Investigations," Part II, Commonwealth
of Pennsylvania, Harrisburg, Pa., 1939 (mimeo.) .
186
RUNOFF RELATIONS
and barren slopes, and the snow line can be defined only as an average
ince surface air temperature is an
elevation of the lower limit of snow.
inverse function of elevation, the rate of snowmelt decreases with elevation. If the freezing isotherm is below the snow line, there is no melting within the basin. Thu, temperature at an index station must
be considered in conjunction with the extent of snow cover in estimating
snowmelt.
Analysi of mean daily surface air temperature in mountain regions
has shown that temperature drops
about
3 to 5 F O per lOOO-ft increase
,
0.16
in elevation. Given the tempera~
,"
ture at an index station, an area~
.g 0.14
elevation curve for the basin, and
8. 0.12
the average snow~line elevation,
~
one can compute the area subject
.~ 0.10
to melting on the basis of an as.5
::; a08
sumed rate of change of temperature with elevation. The sum
~ 0.06
};DA, where D is the degree days
~
"0
for each small increment of elevat 0.04
5.
tion above the snow line and A is
~ 0.02
the
per cent of basin area in each
0---- :
elevation
zone, gives a weighted
Jan. Feb. Mar. Apr. May June July
of
degree
days over the meltvalue
FIG. 8- 13. Degree-day-factor curve for the
zono.
The
weighted
degree days
ing
Tuolumne River above Don Pedro Dam,
are
thus
a
function
of
snow-line
eleCalifornia.
vation and index temperature (Fig.
8-12). Multiplying the degree days so determined by a degree-day
factor gives the snowmelt in inches over the whole basin. In some highelevation basins of the West, the degree-day factor shows a systematic
variation with date during the melting eason (Fig. 8-13).1 It is important that relations such as tho e of Figs. 8-12 and 8-13 be derived and used
as parts of an integrated procedure for synthesizing the hydrograph. The
degree-day-factor data of Fig. 8-13 are directly dependent upon the
method of deriving degree-day values and the technique for computing
snowmelt from the discharge hydrograph.
SEA ONAL- AND ANNUAL-RUNOFF RELATIONS
The hydrologist is frequently called upon to estimate seasonal and

annual Bow volumes, either to ynthesize a streamflow record for de ign
or for an operational forecast. In estimating flow volumes for long
I R. K. Linsley, A Simple Procedure for the Da.y-to-day Forecasting of Runoff from.
Snowmelt, Tram. Am. Geophys. Union, Vol. 24, Part 3, pp. 62-67, 1943.
SEASONAL- AND ANNUAL-RUNOFF RELATIONS
187
periods, variations in antecedent ondition ar not critical, and th re is

u 'ually no need to distingui h between groundwater and dir at runoff.
Precipitation is the principal indep ndent factor to be considered. Inasmuch as the meteorologist cannot y t make reliable predictions of precipitation far in advance seasonal- or annual-runoff forecll tare f a ihle
only where snow accumulate during the winter and i follow d by a
characteristic spring thaw. In the nited tate u h conditions xist
35
0
30
I/)
Q)
l{"
25
.c
o~
.S
:: 20
'0
c
C
.... 15
0
~
0
.
.
>~
'0 10
l/oo
rv
8..t1
POo~
",
20
25
30
35
40
45
50
55
Mean basin water - year precipitation in inches
FIG. 8-1". Relation between annual precipitation and runoff for the Me rrimack River above
lawrence, Massachusetts (4460 sq mil.
in varying degree throughout the mountainous regions of the West and

in the north-central and northeastern states.
8-13. Precipitation-runoff relations. A simple plotting of annual
precipitation vS. annual runoff will often di playa high degree of correlation (Fig. 8-14), particularly in area where the major portion of the
precipitation falls in the winter months. Numerous refinements can
be introduced if a moderate increase in reliability ju tifies the additional
effort required in analysis. 1 First, there is the que tion of cOllsi tent
data throughout the period of record under tudy. Ob 'erved runoff
should be adju ted for storage hanges in reservoirs. In addition, construction of reservoirs during t he record period may increase evaporation
1 M. A. Kohler and R. K. Linsley, Recent Developments in Water Supply Forecactmg, TraM. Am. Gwphys. Union, Vol. 30, pp. 427- 436, June, 1949.
188
RUNOFF RELATIONS
losses to an extent requiring consideration. Irrigation water use ha

also been 3. variabl factor iu many ba ins. Finally, changes in ob erV!1tional techniques, discharge section, gage location, and other factor'i
have contributed to discharge records which in some cases show defini
time tr Jlds unexplained by precipitation. Testing precipitation and
runoff datu by double-mas analysis (Chap . 3 and 4) may di close
inconsi ten 'ie which 'an b eliminated through appropriate adjustments.
Figure 8-14 is based on averages of precipitation observed at 30 stations
in and near the basin. If record are available for only a few station "
application of -tation weights may improve results. Summer precipitation, and a portion of that occurring in the fall and spring, is in the
form of rain and fall ' on bare ground. Consequently, a sizable portion
of ach storm's pr cipitation goes to basin recharge. Winter snows
accumulate as a snow pack, and the portion of each increment utilized
as recharge is relatively small. The effectiveness of precipitation in
producing streamflow thus depends on when it occurs, and the application
of monthly weights will improve the reliability of the relation. Station
and monthly weights can be derived through least-squares analysis.!
Th high intereorrelation of precipitation ob erved at stations within a
limited area may result in regression coefficients (weights) which overemphasize the relative value of the better stations, and, if so, the result.
hould b tempered toward equal weighting. Derived monthly weights
may tend to be erratic, particularly if based on a short record, but they
can be adjusted to provide a smooth transition throughout the year.
There i often a substantial lag between precipitation and the ub equ nt discharge of that portion which recharge the groundwater.
Where groundwater i an appreciable part of the total flow, introduction
of pr cipitation during the previolls year as an additional parameter
may improve the r liability of the relation.
treamflow during the
previous y ar and flow during one or more winter months2 have also
been u ed as indice of groundwater carry-over.
8-14. Use of n ow urvcy. The application of snow-survey data
to the preparation of water- upply forecasts is appealing because of the
rather simple relation envi ion d. If the sea onal flow results primarily
from melting of a mountain snow pack, mea urement of the water in th
snow pack prior to tbe beginning of melt hould indicate the volume of
runoff to b expe ted. Hence, for many years snow surveys were made
near the end of the snow-accumulation sea on, usually about April 1.
While surveys are now made monthly during the snow season at most
cour es, the April 1 urvey i till considered the be t index to ea onal
M. Ezekiel, "Methods of Correlation Analysis," 2d ed., Wiley, New York, 1941.
IE. L. Peck, Low Winter StreamBow as an Index to the Short- and Long-term
Carryover from Previous Years, Proc. Westem Snow Con!., pp. 41-48, 1954.
1
II SE OF SNO W SURVEYS
PRECIP ITATION WEIGHTS

SNOW-COURSE
Station
Month
WEIGHTS
0.05 Gibson Dam 0.05 Sep. 0.8 Marias Pass
0.30 Kalispell
0.09 Oct. 0.2 Cottle Queen
0.05 Ovando
0.1 .3 Nov.
0.05 Polson
0.16 Dec.
0.30 St. Ignatius 0.16 Jon.
0.10 Summit
0.1 5 Feb.
0.05 West Glacier 0.12 Mar.
0.10 Whitefish
0.08 Apr.
0.04 May
0.02 Jun.
189
60
--,
+-
'+-
'"c:
.2
.-
E
. 2
--'0
--o
:=
...
0
1:
Q)
>-
':;
i5
>
Q)
rJ
:;;
Q)
+-
15
Q)
30
J;
+-
III
...
Q.
<{
20
FIG. 8-15. Annual-runoff relation for South Fork, Flathead River, above Hungry Horse
Reservoir, Montano. (After U.S. Weather Bureau.)
runoff. Becau e of drifting, variation in winter melt from point to

point, etc., snow cover is u ua,Uy more variable than precipitation within
a given area. It is not C 'onomi ally feasibJ to tak a sufficient number
of samples to d termine dir ctly the volume of W:1ter tored in the pack,
and the surveys constitute nothing more t han an index. A 'cordingly,
sa.mples must be taken at the same points and in a consistent manner
190
RUNOFF RELATIONS
each year.
are must be taken to avoid cha.nges in exposure brought
about by forest fires, growth of timber and underbrush, etc.
Although th re is good correlation between snow-survey data and
sea. onal runolI, it i now recognized tha.t reliable water- upply forecasts
cannot be made from snow survey alone. l Runoff subsequent to the
urvcys is al 0 dep ndent upon (1) groundwat r storage, (2) antecedent
soil-moi ture deficiency, and (3) precipitation during the runoff period.
It has been found 2 that snow-survey data ean best b treated as an
independent measure of winter precipitation in a multiple correlation
(Fig. 8-15).
BIBLIOGRAPHY
Hydrology Handbook, ASCE Manual 28, pp. 33- 63, 78- 80, 1949.
Linsley, R. I{., M. A. Kohler, and J . L. H. Pl1ulhus: "Applied Hydrology," pp. ]33w York, 1949.
143, 309- 315, 405-44.3, 650-(j55, McGraw-Hill,
" now Hydrology, ummary R port of Snow Investigations," North Pacific Division, U .. Corps of Engin ers, Portland, Ore., Jun e, 1956.
trnu8s, F . A.: Forecasting Water upply through now urveys, J. Am. Water
Works A88oc., Vol. 46, pp. 853-863, 1954.
PROBLEM
8-1. Drive 0. mean rainfall vs. runoff curve (Fig. 8-4) from the data shown in the
table on page 191, for the Ramapo Riv r at Pompton Lakes, ew J ersey (drainage
area - 160 sq mil. Comput the average error of the relation and the bias for the
tabull\tcd summer (May- Ootober) and winter (November- April ) storms.
8-2. Derive a mean rainfall vs. recharg curve from the data provided in Prob . 8-1.
Compute th avemge error of the relation for the tabulated storm s.
S-3. Using the data provided in Prob. 8-1, derive 0. relation similar to that shown in
ompute th average error of th r('la tioll for
Fig. 8-7 for estimo.ting bo.sin recharge.
the tabulat d storms. Conv rt th d rived r lation to on for computing storm runoft (like Fig. -8).
8-4. Compare the runoff for several assumed storms as computed from Figs. 8-8
and 8-9. Under what circumstances does Fig. 8-9 yield appreciably more runofI?
8-5. Comput() tbe and TV indices for the storm depicted in Fig. 8-1.
IJ-6. Find the contribution to r 'charge and runoff (combined) f r !Io day on which
n t radiation ~ 150 cal/sq om; convectiv traDsfer to tho snow paok = 75 cal /sq ern ;
I J. C. Marr, Effe t of Soil-priming by Fall Precipitation on Spring Runoff- Upper
Snak Basin, Trans. Am. Geophys. Union, Vol. 20, Part 1, pp. 106- 109, 1939.
A. R. Croft, orne Factors That InRu nce th Accuraoy of Water-supply For casting in the Intermountain Region, Trans . Am. GeophY8. Union, Vol. 27, pp. 375- 388,
June, 1946.
"Review of Procedur s for Forecasting Inflow to Hungry Horae Reservoir, Montana," 'Vater Management Subcommittee, Columbia Basin Interagoncy Committee,
June, 1953.
I J. Hannaford, Multiple-graphical Correlation for Water Supply Forecasting,
Proc. W 8tem Snow Conj., pp. 26-32, 11)56.
M. A. 1 ohler, Water-supply For asting Dcvclopm nts, Proc. Western Snow Coni.,
pp.62- 6 , 11)57.
PROBLEMS
Storm
No.
1
2
3
4.
5
6
7
8
9
10
11
12
13
14
15
)6
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
191
DaLe
3/15/40
4/22/40
5/ 17/40
9/ 2/40
10/ 3/40
12/17/40
2/ S/4 1
4/ 7/ 41
12/25/41
2/ 1/42
2/1 / 42
3/ 4/42
S/10/ 42
8/ 18/4.2
9/28/4.2
2/ 7/43
6/ 2/43
11 / ]0/ 43
3/ 8/44
4/25/44
1/ 2/45
7/23/45
S/ 7/45
11 /30/45
5/28/46
9/25/46
6/ 9/47
5/14/48
5/30/48
7/ 14 /4. 8
12/31/48
11 /26/50
3/31/ 51
10/ 8/51
3/12/52
4/ 6/52
6/ 2/ 52
8/ 17/52
9/ 2/52
12/ 4/52
3/12/ 53
3/23/53
Antecedent- Wek Storm

torm
torm
Ba in
precipitation
of
duration, pr cipitation, rUDofT, recharge,
year
index
hr
In.
in.
in .
0 .90
2.20
0 .58
2.70
1.00
0 .75
0 .50
0 .29
1.20
0 ,65
1.03
0 .45
1.26
3 ,29
1.00
1.06
1.07
2 ,05
0,45
1 ,57
1.25
5 .45
2.56
1.42
1. 44
1,03
1. 27
1.11
1.56
1.00
0 .70
1.12
1.00
0 .40
0 .81
0 .89
2 . 17
2 ,66
1.20
2 , 11
1.01
2 .33
II
16
20
35
40
51
6
14
52
5
7
9
32
33
39
6
22
45
10
17
1
30
32
48
22
39
23
20
22
28
52
48
13
41
11
14
22
33
35
49
11
12
12
18
9
12
12
12
Hi
20
15
9
12
I)
15
6
18
18
9
21
12
21
6
12
9
24
15
12
10
27
6
9
36
15
36
18
12
18
15
15
6
12
36
12
2 .03
1 .95
1.50
1 .SI
1.07
1. 13
1.95
1. 30
1 ,60
1.06
0 ,79
1.83
2 .87
2 .81
3 .58
0 ,67
1,47
2 ,33
0 ,95
1,94
1.30
3 .55
1.61
2 .25
2 .90
2 . 27
1.54
2 .27
1.18
1.83
4 .67
3 .30
5 .25
1.85
2 .83
3 .10
4 , 10
3 ,08
3 ,94
1.50
3 .50
2 .00
1 .23
1.05
0 .46
0 .33
0 .10
0 .33
0 .99
0 .53
0 .53
0 .30
0 ,22
0 ,89
0 ,49
1.06
) ,01
0 . 17
0 .30
0 .92
0 ,50
1 . 10
0 ,74
2, 39
0.42
0 .92
1.44
0 ,43
0 ,31
0 ,71
0 ,34
0 ,22
2.77
1.12
3.51
0 . 11
2.09
1.55
2 ,07
0 .83
1.11
0 .71
2 , 15
).08
0 .80
0 .90
1.04.
1.4.8
0 .97
0.80
0 .96
0 .77
1.07
0 .76
0 .57
0 .94
2 .38
1. 75
2.57
0 .50
1.17
1.41
0 ,45
0 .84.
0 .56
1.16
1.19
1.33
1.46
1.81
1. 23
1. 56
0 ,84
1 ,61
1. 90
2 . 18
1 ,74
1. 74
0 . 74
1.55
2,03
2 ,25
2 ,83
0 .79
1 ,35
0 .92
192
RUNOFF RELATIONS
condensation ... 0.02 in. depth i rainfall (at 45F) = 1.00 iu. i a.nd wa.ter equivalect of
the snow pack "" 1.35 in.
8-7. Giv n maximum and minimum daily temperatures of 55F and 35F, respectively, and assuming temperature to follow a sine curve, compute the degree days and
degree hours for a 24-111' period using a base of 32F. Rep at the computations with
temp ratures of 45F and 25F. Comm nt on the results.
8-8. Assuming the temp ratute-index station to be at an elevation of 1000 ft and
th variation of t mperature with elevation to be 4 FO per 1000 ft, derive a r lation
similar to Fig. 8-12 for a basin with the following area-elevation cha1'a.cteristics.
Elevation,
ft
Per cent
of area
Elevation,
ft
Per cent
of area
1000
2000
3000
4000
0
5
30
60
5000
6000
7000
BOOO
80
90
96
100
Compute weighted degree days for ;udex-fltation temperatures of GO and BOoF with
;mow-line elevations of 3000 and 5000 ft, respectively.
8-9. For a basin selected by the instructor, compile the neeessary data. and derive
8. precipitation-runoff relation of the type shown in Fig. 8-14.
9
HYDROGRAPHS OF RUNOFF
In the de ign of spillway, storm-drainage work, and flood-control

projects an e timate of probable p ak flows i neces ary. Wh re ad quate
stream.B.ow record are available, the answer may be found by use of
frequencyanalysi ( hap. 11). More commonly, however, it is necessary
to synthesize the design-flood hydrograph by other means.
The earliest method for estimating peak flow wa the use of empirical
ome
formulas involving various physical characteristics of the basin.
of the e formulas are discu ed in ec. 9-10, mainly to point out their
inadequacy as engineering tools. In 1932, Sherman introduced the unit
hydrograph ( ec. 9-2) as the first basic tool for estimating hydrograph
shape. Today the unit hydrograph is the mainstay of the flood hydrologist, but attention i also bing directed toward flood-routing methods
( ec. 10-13) which offer pro p 'ct of greater flexibility and accuracy than
the unit hydrograph in synthesizing the outflow hydrograph from a basin.
Chapter 8 presents m thods of estimating the volume of water that
would be deliver d to a stream as the result of rain or melting snow. It
is the purpo 'e of this chapter to explain how this runoff volume i transform d into a hydrograph of flow at a point in the stream system. For
rea ons that will be developed, the method' which follow are generally
applicable only to relatively small drainage basins, u ually under 5000
sq mi and preferably much smaller.
9-1. The elemental hyd.r ograph. If a small, imp rvious area is
subjected to rainfall at a constant rate, the resultmg runoff hydrograph
will appear much a in Fig. 9-1. Since she t flow over the surface cannot
occur without a finite depth of water on the surface, some of the rainfall
goes into temporary storage, or surface detention. At any in tant the
quantity of water in such storage is equal to the difference between total
inflow to the area (rain) and total outflow from the area. When equilibrium has been reached the rate of outflow equals the rate of inflow
(point C) and the volume in detention is ABC. The water is in constant
motion, and any given element of rainfall may pass through the system
193
194
HYDROGRAPHS OF RUNOFF'
in a fairly short time, but the volume difference between inflow and outflow remains constant.
Whcn rainfall ends (point D) there is no further inflow to su tain the
the detention, and rate of outflow and detention volume decrea e. The
outflow follows a typical recession with flow decreasing at a decreasing
rate; that is, d2q/dt2 is negative. This is usually accentuated by an
exponential relation between storage and discharge.
Theoretically, an infinite time is required for both the ri ing portion
of the hydrograph to reach equilibrium and for the rece ion to return
to zero. Practi ally, both the rising and falling curves approach their
FIG. 9-1. The elemental hydrograph.
limits rapidly. The results of experimental study of the hydrographs

from small areas are discu scd in ec. 9-9.
9-2. The unit-hydrograph concept. The bydrograph of outflow
from a small basin is the sum of the elemental hydrographs from all the
subarea of the basin, modill d by the effect of transit time through the
basin and storage in the stream channel. Since the physical characteristics of the basin-shape, size, slope, etc.-are constant, one might
expect considerable similarity in the shape of hydrographs from storms of
similar rainfall characteristics. This i the es ence of the unit hydrograph
as proposed by Sherman. 1 The unit hydrograph is a typical hydrograph
for the basin. It is called a "unit hydrograph" because, for convenience,
the runoff volume und l' the hydrograph is commonly adjusted to 1.00 in.
It would be wrong to imply that one typical hydrograph would suffice
for any basin. Although the physical characteristics of the ba in remain
relatively constant, the variable characteristics of storms cause variations
in the shape of the resulting hydrograph. The storm characteristics
1 L. K. Sherman, Streamflow from RainiaU by the Unit-graph Method , Eng. Ne'lll3Record, Vol. 108, pp. 501- 505, 1932.
THE UNIT-HYDROGRAPH CONCEPT
195
are rainfall duration,. time-intensity pattern, areal distribution of rninfall,

nnd amount of runofT. Th ir flects ar di eu d below,
Duration of min.
inc the unit hydrogmph always contain 1 in. f
runoff, increasing the duration of th rainfall lengthens the time ba e
and lower the unit-hydrograph peak. A parate unit hydrogrn.ph is
theoretically nece ary for each pos ibl duration of rn.in. A tually,
the effect of small differences in duration i not large, and a tolern.nc of
25 per cent from the establi hed duration is ordinarily ac eptable.
Further, a unit hydrograph for a short duration of rainfall can be u ed to
develop hydrographs for torm of longer duration ( ec. 9-5). Thu a
few unit hydrographs for short durations will serve all requirements.
Time-intensity pattern. If one attempted to derive a separate unit
hydrograph for each pos ible time-in ten ity pattern an infinite number
of unit hydrographs would be required. Practically, unit hydr graphs
can be based only on an as 'umption of uniform intensity of runoff,
However, large variations in rain ioten ity (and henc runoff rate) during
a storm are reflected in the shape of the resulting hydrograph. The
time scale of intensity variations that are critical d pends mainly on
ba in size. Bursts of minfnllla -ting only a f w minutes may cause clearly
defined peaks in the bydrograph from a basin of a few acre , while on
basins of several hundred sq uare miles only changes in storm intensity
lasting for hours will cau e di stinguishable eiTects on the hydrograph.
If tbe unit hydrograpb. for a basin are applicable to storm of sborter
duration than the critical time for the basin, hydrogrn.phs of longer storms
can be synthesized quite ea 'ily (Sec. 9-). A basic duration of about
one-fourth of the basin lag ( c. 9-6) i generally satisfactory.
Areal distribution of runoff. The areal pattern of runoff can cause
variations in hydrogl'aph shape. If the area of high runoff is near the
basin outlet, a rapid rise, sharp peak, and rapid recession usually result.
Higber runoff in the up tream portion of the basin produces a slow ri e
and recession and a lower, broader peak. Unit hydrographs have been
develop d for specific runoff patterns, e.g., heavy upstream, uniform,
or heavy down tream. This is not wholly satisfactory because of tbe subjectivity of clas ification. A better solution i to apply the unit-hydrograph method only to basins small enough 0 that the usual areal variations will not be great enough to cause major changes in bydrograph
hape. The limiting basin size is fixed by the accuracy desired and
regional climatic characteri. tics. Genemlly, however, unit hydrographs
hould not be used for basin much over 2000 sq mi unless reduced
accuracy is acceptable. What has been 'aid does not apply to rainfall
uch rainfall ' patterns are
variations r.aused by topographic controls.
relatively fixed characteri ti 'S of the basin. It is departures from the
normal pattern that cause trouble.
HYOROGRAPHS OF RUNOF, :
196
A mount of runoff. A basic part of the unit-hydro graph idea is that the
ordinates of flow are proportional to the volume of runoff for all torms
of a given duration. This i predicl1ted on the a sumption that the time
bases of all hydrographs re citing from storms of a given duration are
constant. Actually, from the character of recession curves it is evident
that the duration of the reces ion is a function of peak flow. Practically,
.; 15 ~------~------~--~L-~------~----~
e'
o
..c
u
II>
'6
1 10~------~-----n~------~------~----~
~
II>
<3
10
15
20
25
Computed peok discharge in 1000 cfs
FIG. 9-2. Relation between observed and computed peak flow for Appalachee River necr
Bvckhead, Georgia (436 sq mil.
it appears that th assumptions of a fixed time ba e and ordinates proportional to runoff volume are adequl1te for engineering purpo es. There
is evidence that the unit-bydrograph peak is generally somewhat higher
for extreme flood than for average floods. Because of limited data on
extreme floods, there is little quantitative evidence on this point.
Hydrologists frequently increa e peak flows from 5 to 15 per cent when
making estimates of very extreme floods. Figure 9-2 shows a relation
between ob rved flood p aks and those computed from a unit hydrograph developed from moderate floods.
In the light of the foregoing discussion, the unit hydrograph can be
defined as the hydrograph of one inch of direct runoff from a storm of specified duration. For a storm of tbe same dUl'atioll but with a different
197
THE UNITHYDROGRAPH CONCEPT
amount of runoff, the hydro graph of dire t runoff can be eo"p cted to
have the sam time ba e as the unit hydrogra.ph and ordinates of flow
proportional to the runoff volume. Th duration a sign d to n. unit
hydrograph bould be the duration of rainfall producing ignificant
runoff. This hould be determined by inspection of hourly-rainfall
uri
Dote Hour Totol 80.. Dlrec:t u.G. oter
How flow R.D. ord. or
10
Drainage area =40 SQ mi
Fib. 0600
16
.t1
u
0
11
!2
0800
1000
1200
1400
1600
1800
2000
2200
2400
0200
Q.4OO
.S
0600
0800
1000
1200
1400
1600
Q)
.t:
01)
i5 4
500
seoo
9200
0100
1800
6600
5e50
4100
4000
3!00
2100
2300
1950
1650
1400
1200
1000
800
500
450
400
400
'ISO
450
500
550
600
800
600
650
650
100
100
150
150
800
Totol
5150
8800
9100
1350
61eO
5050
4150
!400
2100
2100
1650
1500
950
TOO
450
250
0
59850
1120
191e
2110
1600
1340
1100
900
140
590
460
360
280
210
150
100
50
4
$
8
10
12
14
16
18
20
22
24
26
28
30
32
34
------ --- __ _~O!!. f!!>!- - -- - - - - --0600

01)
2000
1200
Feb. 17
1800
"0
E., 1500
_g'" 1000
:;:
500
i5
18
.36
Hours from be91nnlng of runoff
FIG. 93. Development of a unit hydrograph.
data. The characteri tic hydrograph may be given as the percentage

of the total runoff 0 'curring in successive short increments of time. In
this form it i known as a distribution graph.1 Because the period averages define the hydrograph less explicitly than instantaneous flows, one
need not be quite so careful in the use of the distribution graph as for
the unit hydrograph. Distribution graphs are useful where estimates of
M. Bernard, An. Approach to Determinate Stream Flow, 'Pram. ASCE, Vol. 100,
347, 1935.
J).
HYDRO GRAPHS OF RUNOFF
198
flow volume are more important than detailed rates of flow, as, for example, in analyzing the inflow to a reservoir.
9-3. Derivation of the unit hydrograph. The unit hydrograph is
best derived from the hydrograph of a storm of reasonably uniform
intensity, duration of desired) ngth, and a runoff volume near or greater
than 1.0 in. The first step (Fig. 9-3) is to separate the base flow from
60~---.----~-----r----.-----.-----~---.-----'
50
/\\ _.,./Average peak
! .
I
4!!
U 40
0
0
Curve defined by arithmetIc average
Original uni t hydrogrophs
.:
QI
30
.&:.
.!!!
0
20
10
12
24
36
48
60
72
84
Time tn hours
FIG. 9-4. Construction of an average unit hydrograph.
t he direct runoff ( hap. 7). The volume of direct runoff is then determined, and the ordinates of the direct-runoff hydrograph are divided
by the observed runoff in inches. The adjusted ordinates form a unit
hydro graph for the sp cifi d torm duration.
A unit hydl'ograph derived from a single st rm may be in errol', and it
is desirable to average tho unit hydrograph from several storms of the
same duration. Thi should not b an arithm tic average of concurrent
coordinates, since if peak do not occur at the same tim , the averag
peak will be lower than many of the individual p ak. The proper
procedure is to compute the average of the peak Bows and times to peak.
The average unit hydrograph is th n sketched to 'onform to the shap
of the other graph, passing through the comput,ed avera,ge peak, a.ad
having a volume of 1.00 in. (Fig. 9-4).
9-4. Derivation of unit hydro graph from complex storms. The
simple approach outlin d in Sec. 9-3 often proves inadequate because
199
DERIVATION OF UNIT HYDROGRAPH
no such id a1 storm is available in the record. It i then nece ary to

develop the unit hydrograph from a compl x torm. If the individual
bur ts of rain in the torm re ult in well-defin d p aks, it i possible to
separate the hydrograph of the everal bur t ( c. 7-4) and us the
hydrographs a indep ndent storms. If th re ulting unit hydrograph
are averaged error in the eparation are minimized.
3500
3000
~
<.>
2500
.S
.,
2000
<.>
'"
:;;
c:
1500
t;
~
i5
1000
500
48
0
Time in hours
FIG. 9- 5. Unit hydrographs from a compl ex storm.
If the hyd rograph does not lend itself to separation (Fig. 9-5), the
analy i b gins with estimates of the direct-runoff volumes Ql, Q2, . . . ,
Q" in successive periods during the storm. The equation for any ordinate
of the total hydrograph q.. in terms of runoff Q and unit-hydl'ograph
ordinates U is
qn = Q"U 1
+ Qn-
U2
+ Q,,- 2U a + ... + QIU..
(9-1)
The first ordinate ql = Q1U I , and since Ql is known (or estimated) VI

can be computed. The seconli ordinate is then
(9-2)
200
HYDROGRAPHS O F RUNOFF
The only unknown in this equation is U2 which can be computed. All

the ordinates can be determined in a similar way.
Although the procedure ju 't outlined may seem simple, there are
numerous difficulties. Each computation depends 011 all preceding
computations for values of U. Errors in estimating the runoff increments, errors in the observed streamflow, and variations in intensity
and areal distribution of rainfall during the several storm periods can
lead to cumulative errors which may become large in the latter portion
of the unit hydrograph. Large negative ordinates sometimes develop.
A unit hydrograph can also be developed by successive approximations.
A unit hydrograph is assumed and used to reconstruct the storm hydrograph (Sec. 9-8). If the reconstructed hydrograph does not agree with
the observed hydrograph, the assum d unit hydrograph is modified and
tp.e process repeated until a unit hydrograph which seems to give the
best fit is d termined.
A more el gant but laborious method IS the use of least squares, a
statistical technique to find the constants a and b in an equation of the
form
q = a + blQl + b2Q2 + b3Q3 + . .. + b,Qi
(9-3)
Equations (9-1) and (9-3) are simiJar. Theoretically, for the unit hydrograph, a should be zero and th b's equivalent to the unit-hydrograph
ordinates U. The least-square technique would use data from a number
of flood events for which value of q and Q are established to develop a
set of average values of U. More detailed explanations can be found
elsewbere. 1
9-5. Unit h ydrographs for variou d urations. If a unit hydrograph for duration t hr is added to itself lagged by t hr (Fig. 9-6), the
resulting hydrograph repre ents the hydro graph for 2 in. of runoff in 2t
hr. If the ordinates of this graph are divided by 2, the result is a unit
hydrograph for duration 2t hr. The final graph represents the flow from
1 in. of runoff generated at uniform intensity of 1/ 2t in. / hr in 2t hr. Thi
simple example illustrates the ea e with which a unit hydro graph for a
short duration can be converted to a unit hydrograph for any multiple
of the original duration.
Con truction of a short-duration unit hydrograph from a longer
duration can be accomplished by the m thods described in ec. 9-4.
A more convenient technique for conver ion to either a shorter or longer
dUl'8.tion j the S-curve, or summation-curve, method. The S curve is
the hydrograph that would result from an infinite series of runoff incre1 R. K. Linsley, M. A. Koh ler, a.nd J . L. H. Paulhus, "AppHed Hydrology," pp. 448449, McGraw-Hill, New York, 1949,
W. M. Snyder, Hydrograph Analysis by Method of Least Squares, Proc. ASCE,

Vol. 81. Sepa.ra.te 793, September, 1955.
201
UNIT HYDROGRAPHS FOR VARIOUS DURATIONS
ment of 1 in. in t hr. Thus, ea.ch curv applic to a specific duratioll

within which each inch of runoff i g n rat d. Th S curve is con -truct d
by adding together a erie of unit hydrograph, ach lagged t hI' with
re pect to the preceding one (Fig. 9-7). If the time base of the unit
hydrograph is T hI', th n a continuou rainfall producing 1 in. of run ff
7
6
Sum of two 6-lIr unit
lIydrogroplls offset 611r
5
<II
V
0
0
0
c
Q)
0
.r:;
(.J
<II
is
2
48
Time in hours
FIG. 96. Construction of a unit hydrograph for duration 2t.
every t hr would d velop a constant outflow at the end of T hr. Thus

only T / t unit hydrographs need be combined to produce all
curve
which should reach equilibrium at flow q.:
q. =
24 X 26.9 X A
645.6A
= - -t-
(9-4)
where A is drainage area in square mil s, t is duration in hours, 24 is the

number of hour in a day, and 26.9 is the number of second-foot-days in
1 in. of runoff from 1 sq mi.
Commonly, the S curve tends to fluctuate about the equilibrium flow.
This means that the initial unit hydrograph do s not actually represent
runoff at a uniform rate for t hr. If a uniform rate of runoff is applied to
202
a basin, equilibrium flow at the rate given by Eq. (9-4) must eventually
develop . If the actual effective duration of runoff a 'ocia-ted with the
original unit hydrograph is not t hr, the summation process results in a
runoff diagram with either periodic gaps or periodic increases to a rate
of 2 in. in t hr (Fig. 9-8). Thus the S curve serves as an approximate
45
40
35
30
:?
t.>
S-curve
0
0
$2 25
/" .....
.=
eo
0
..:
'"
III
i5
15
I
/
10
'I
I'
12
18
\
\
\.
\
\
\
\
"-
\.
........
\
\
\.
"-
....... .......
24
30
Time in hours
""'
\
\
'", .........
\
\
r,
\
\
\
\
...... ,
\
\
'\
\.
20
'" '\
\
\
\
Qj
--
v-' '\
- , "-
\.
"-
'"'"
".......
....... ........ -...

36
"-
.......
"-
----_
.................
42
48
FIG. 9-7. Graphical illustration of the S curve of Table 9-1.
check on th assumed duration of effective rainfall for the unit hydrograph. A duration which re ults in minimum fluctuation of the S curve
ote, however, that fluctuation of the S curve
can be found by trial.
can also result from nonuniform runoff generation during the t hr, unusual
areal di tribution of rain, or errors in the basic data. For this rea on, the
S curve can indicate only an approximate duration.
Construction of an S 'curve does not require tabulating and adding
Tit unit hydrographs ,'lith successive lags of t hr. Table 9-1 illu trates
the computation of an S curve, starting with an initial unit hydrograph
SYNTHETIC UNIT HYDROGRAPHS
203
for which t = 6 hr. For the fir t 6 hr the unit hydrogrnph and S curve
are identical ( 01. 2 and 4) . The -curve additions ( 01. 3) are the
ordinates of the curv set ahead 6 hr.
inc an -curve ordinate i the
sum of all the wlit-hydrograph ordinates falling at that tim , combining
the '-curve additions with the initial unit hydrograph is the same as
adding all previous unit hydrographs.
The difference between two curves with their initial point displaced
by t' hr gives a hydrograph for the new duration t' hr.
ince the curve
If duration is actually 1511r tllis is file actual rUnoff
12
18
Time in hours
24
30
36
FIG. 98. Influence of nonuniform rote of runoff on the S curvo.
represents runoff production at a rate of 1 in. in t hr, the runoff volume

represented by this new hydrograph will be t'l t in. Thus the ordinates
of the unit hydrogl'aph for t' hr are computed by multiplying the S-curve
differences by the ratio ti t'. The computations for a 2-hr unit hydrograph are illustrated in Table 9-1.
9-6. ynthetic unit hydrograph. OnJy a relatively small number
of the streams in the United tates are gaged.
nit hydrographs can
be d rived as described in the previous sections only if records are
available. Hence some means of deriving unit bydrographs for ungaged
basins is necessary. This requires a relation between the physical
geometry of the basin and the resulting hydrographs. Three approaches
hUv-e been u ed: formula relating hydrographs to basin characteristics,
transpo ition of unit hydrographs from one basin to another ( ec. 9-7),
and storage routing ( ec. 10-13).
Most attempts to derive formula for the unit hydrograph have been
aim d at determining time of peak, peak flow, and time ba e. These
items plus the fact that the volume must equal 1.00 in. permit sketching
of the complete hydrograph. The key item in most studies has been the
TABLE 9-1. Application of S-curve Method
Time,
hr
6hr unit
graph
5-curve
additions
5-curve
Col. (2)
(3)
Logged
S-curve
Col.
(4) - (5)
2-hr unit
graph
--
(1)
(2)
(3)
(4)
(5)
(6)
(7)
0
2
4
6
8
10
12
.4
16
18
20
22
14
26
28
30
32
34
36
38
40
42
44
46
48
0
400
1,400
3,100
5,400
8,600
12,600
15,400
14,600
11,800
9,900
8,400
7,200
6,000
5,100
4,200
3,400
2,700
2,100
1,600
1,100
700
400
200
0
0
400
1,000
1,700
2,700
4,200
5,700
5,500
3,400
2,900
2,400*
2,000
1,800*
1,600
1,400
1,200
1,000
800
600
400
200
100
0
0
1,200
3,000
5,100
8,100
12,600
17,100
16,500
10,200
8,700
7,200
6,000
5,400
4,800
4,200
3,600
3,000
2,400
1,800
1,200
600
300
0
......
......
......
0
400
1,400
3,100
5,800
10,000
15,700
21,200
24,600
27,500
31,100
33,000
34,700
37,100
38,100
38,900
40,500
40,800
41,000
41,500
41,900
42,000
0
400
1,400
3,100
5,800
10,000
15,700
21,200
24,600
27,500
31,100
33,000
34,700
37,100
38,100
38,900
40,500
40,800
41,000
41,500
41,900
42,000
42,000
42,100
42,100
.....
0
400
1,400
3,100
5,800
10,000
15,700
21,200
24,600
27,500
31,100
33,000
34,700
37,100
38,100
38,900
40,500
40,800
41,000
41,500
41,900
42,000
42,000
42,100
Adjusted value.
basin lag, most frequently defined as the time from the centroid of rainfall to the hydrograph peak.l The first ynthetic-unit-hydrograph procedure was pres nted by Snyder. 2 In a study of basin in the Appala"hian Mountains region, he found that the basin lag tp (in hr) could be
expre sed by
(9-5)
I Lag i al 0 d fined as the time diff rence between tIl centroid of rainfall and the
centroid of runoff. This defLnition i more rigorolLs, hilt the one in the text is simpler
to apply.
2 F. F. Snyder, Synthotic ~nit Hydrographs, 1'rans. Am. GeophY8. Union, Vol. 19.
Part 1, pp. 447-454, 1938.
205
SYNTHETIC UNIT HYDRO GRAPHS
where L is the length of the main stream from outlet to divide in mil
and Lc is the distance from the outlet to a point on the stream near t
the centroid of the basin. The product LL. i a measure of the ize and
shape of the basin. The coefficient C, varied from 1. to 2.2, with ome
indication of lower values for ba ins with steep r lope.
Before an equation lor peak flow can be written, a standard duration
of rain tr must be ad pt d, and nyder took tr = tp / 5.5. For rains of
this duration he found that the unit-hydrograph peak gp was given by
qll
= 640C"A
(9-6)
tp
where A is the drainage area in square miles, C1I i a coefficient ranging

from 0.56 to 0.69, and 640 is a conversion factor to give g" in cubic feet
per second.
Snyder adopted the time base T (in day) of the hydrograph as
T = 3+3~
24
~~)
The constants of Eq. (9-7) are fixed by the procedur used to separate
base flow from direct runoff. Equations (9-5) to (9-7) define the three
factors necessary to construct the unit hydrograph for duration tr For
any other duration tn the lag is
t
PR
= tp
+ tn 4-
tr
(9-8)
and this modified lag tps is uscd in Eqs. (9-6) and (9-7).
Taylor and chwarz, J using 20 basins in the North and Middle Atlantic
States, found an expression for lag as
(9-9)
where the exponent m i.' given by
0.212
= (LL c)O . 36
(9-10)
and the coeffici nt C, is

(9-11)
where s is the weighted slope of the channel.

everal measure of ba in
size and shape were tested, but the term LLc with an exponent very nearly
equal to nyder's value was found be t . The place of slope in the coefficient C, was also qualitatively indicated by Snyd r.
1 A. B. Taylor and H. E. Schwarz, Unit-hydrograph Lag and Peak Flow Related
to Baam Cbaractcriati 's, Trans. Am. Geophys. Union, Vol. 33, pp. 235-246, April, 1952.
206
Taylor and chwarz also derived an expression for the unit-hydrograph peak in the arne form as Eq. (9-9) but with more complex expressions for the co ffici nt and exponent. They give a nomogram for solving
the r lationship.
nyder's synthctic-unit-hydrograph formulas have been tried elsewb're in the country with varying au ce s. The coefficients Ct and Cf'
.!!
'e
1000
600
"-
600
.."
0
0"
400
8.
..,
I
100
60
...
60
40
Q.
0.
'0
If'.'\. I' P,aA' Q

~- e---t
"
W:~./.Q
~~
_" I
""'-
0
U
.5
...
.c
.....
:...
"
..
'0
E!'
:.:
..
-.. ~ ~
. . . "- ~~
....
Observed value of W50

Observed value of W,5
10
6
6
0.2
r8 '~~
rf
.'\....
;~~i.
,g
ao
~
20
...'6
Time in hours
SCHE MATIC DIAGRAM
'"
.t:
.t:
50% Q
'c
:>
'"
~~~ '"~o
200
..
.c
.
"-
"" ~
Q;
.:!!
"
0.4
0.6 0.6 1
6 10
20
40
60 60100
Width of unit hydrogroph in hours
FIG. 9-9. Unit-hyd rog raph width at 50 and 75 per cent of peak flow.
Engineers. )
(U.S. Corps of
are found to vary con iderably.l Th best way to use his equations is to
derive values of Ct and Cf' from unit hydrographs for gaged basins of
similar characteristics as the problem area and apply these coefficients
to the ungag d stream. The procedure thus become' a means of transpo ing unit-hydrograph characteristics from one basin to another.
Figure 9-9 pI' nts the results of an analysis by the Corps of Engincers 2
of the width of unit hydrogmphs (in h r) at 50 and 75 per cent of the peak
1 R. K. Linsley, Application of Synthotic Unit-graphs in the W stern Mountains
tates, Trans. Am. Geaphys. Union, Vol. 24, Part 2, pp. 5 0-5 7, 1943.
2 "Engineering Manual fo r
ivil Works," Part II, hap. 5, fice of Chief of Engineer , D partment of Army, April, 1946.
TRANSPOSING UNIT HYDROGRAPHS
207
Bow. This provide a supplementary guide for sketching the unit

hydrograph.
. 9-6 a gen ral
9-7. Transpo ing unit hydrograpb. From
expres ion for basin lag might be exp ct d to take the form
tp =
C' (~D"
(9-12)
If known value of lag are plott d again t LLc/ Va on logarithmic paper

Drainage
area,sq ml
Drainage
area, sq ml
to.
1. Son Gabriel River at Son Gabriel dam

162.0
LIve Oak Creek at Live Oak dam
2.3
2. West Fork Son Gabriel River at Cogswell dam 40.4 11 . Tujunoo Creek at BIO TuJunoo dam No. I
B1.4
3. Santo Anita Creek at Santo Anita dam
10.8 12. East Fullerton Creek at Fullerton dam
3.1
16.2 13. Los Angel", River at Sepulveda dam
152.0
" . Son Dimas Cree~ oj Son Olma. dam
5. Eaton Wash at Eaton Wash dam
9 . 5 14. Pacoima Wa sh at Pacoima dam
27.B
6. Son Antonio Creek near Claremont
16.9 15. Alhambra Wash above Short Street
14.0
7. Santo Claro River near SouQus
355.0 16. Broadway drain above Raymond dike
2.5
8. Temecula Creek at Pouba Canyon
16B.0 17. Bollana Creek 01 Sawtelle Blvd.
8B .6
9. Santo Margarita River near Fallbrook
645.0 lB. Son Jose Creek at Workman Mill Rood bridoe 81.3
10
'":;
o
J8
Lag =U? ( LLe

I/R
I/og curve for ts
I ~ Ifoo tl1ill drainage area I
;{.
J
5
( L'L e
Lag= 0 .72
sl/R
~/O:; IR
"2
)OJ8
.s::.
-=
1.0
g'0.6
...J 0.4
......
[.....---'
0 ,2
O. 1
9_;Z F
Log curve for mountain drainage area -:-t
I--v
CD
g8
00
0
0
.....-
16
11
v
N
II
IJ
~.-JI.:-c1
r-
:tt1....
".....-
=
( L ' L t.J8
Log =0.35 --;vI-
Log curve for volley

drainage area
III
I 1111111
=t
o
o
o
o
.,.
LL c
Sl/Z
FIG. 9 10. Relationship between basin log (beginning of rain to

characteristics. (U.S. Corps of Engineers.)
Jq df =
Q / 2) and basin
(Fig. 9-10) the resulting plot should define a straight line, provid d that
the values are taken from basins of similar hydrologic chara teri tics,
that i , Ct = const. A relation such a Fig. 9-10 offers a means of estimating basin lag. The peak flow and shape of the unit hydrograph may be
estimated by use of a plot relating qp to tp or by use of dimension] s
hydrographsl such as Fig. 9-11. The dimensionless form eliminates the
effect of basin size and much of the effect of ba in shape. The similarity
f)f the several graphs of Fig. 9-11 reflects a considerable similarity in flood
I G. G. Cornmon , Flood Hydrographs, Civil Eng., Vol. 12, p. 571, 1942.
H. M. Williams, Di cussion of" Military Airfields, " Trans. ASCE, Vol. llO, p. 820,
1945.
208
1.0
l' ~
0.8
0.6
l'.
Ii
~,SOH Conservation Service
h
/iii
0.4
0.2
.\
Ai;j
[}I'/
Commons
r,-Wi/liams
.....
~ ,""
.
r---.
'-
__
........_.
~ 1-- __
'-
o
o
----
2
flip
FIG. 9- 11 . Some dimensionl ess unit hydrographs.
0
c;
21.0-
;E0.5c
'0 0 -
S?
.s:.,
cO
en
0::
0
0
'"
'0
..c 3
. en
is
\..
""
"""
.Estimated:--... --.._
0600
o
0500
o
0800 850
1100 1330 1920
1400 1010 3000
0
1100 770 2280
2000 570 1730 2560
2300 410 1280 4000
0200 260 920 3030
0500 140 S80 2320
50 310 1710
0800
1100
o 110 1230
1400
o 780
1700
420
2000
150
2300
0
300
1.150
3540
4300
3330
5140
5980
4510
3350
2390
1660
1110
760
500
360
"
!c:::_f!!_w_\_ ___ ~::::---=- '- -._--- 1200
300
300
290
290
280
280
290
300
310
320
320
330
340
350
360
"Unit hydrogroph ordinate
t - _.......___
Total
Dote Hour
:::t .5-
1800 . 2400
0600
1200
1800
~---------4
2400
0600
Hours
FIG. 9- 12. Use of a unit hydrog r aph to synthesize a st reamflow hy drog raph.
hydrographs from various regions.

nfortunately, the relationships
between the hydrograph shape and the characteristics of the basin are
so complex that no completely successful relationships have been derived. l
I C. G. EdSall, Paramet ra for Relating Unit Hydrographa to Watershed Characteristics, 7'rans. Am. Geophys. Union, Vol. 32, pp. 591- 596, August, 1951.
THE HYDROGRAPH OF OVERLAND FLOW
209
9-8. Application of th unit h ydrograph. The application of a

unit hydrograph has becn d mOl1strated in ec. 9-5 on the basi of uniform
produ tion of runoff. Figure 9-12 illu trates the use of a 3-hr unit
hydrograph to synthesize th storm hydrograph from a serie of rainfall
periods, with varying intensity. The
"""
increments of runoff for successive
"_ f v
3-hr period are computed u ing run0
f
Slope = s
off relations ( hap. 8). The hydro- 7lM'Mti'M"i~MM"'MM""i1ML""'Ii''''7iJ(\aI""
graph of direct runoff resulting from FIG. 9.13. Deflnition sketch for laminar
each 3-hr increment is given by multi- sheet flow.
plying the unit hydro graph by the
period runoff. The total hydrograph is the sum of all the incremental
hydrograph and estimated base flow.
9-9. The hydrograph of overland flo w. Although the d pth of
flow in the overland sheet is quite mall , the quantity of water t mporarily
detained in this sh eet (surface detention) is r lutivcly great. It is gencrally
assumed that overland flow is laminar (Fig. 9-13) . Hence,
pg(D - y)s =
iJ.
dv
dy
(9-13)
where p is density, g is gravity, and iJ. is absolute vi cosity. The

as umption is made that the lope is 0 small that the sine and tangent
are equal. Since iJ./ p is equal to lcincmatic viscosity v,
dv =
(IS
(D - y) dy
(9-14)
Integrating and noting that v = 0 when y = 0,

V
(IS (
y2)
= -;
yD -"2
(9-15)
Integrating from y = 0 to Y = D and dividing by D, the mean velo 'ity is

(9-16)
and the discharge per unit width is v"J) or
q = bD3
(9-17)
where b is a coefficient involving lope and viscosity.

The most extensive experiments on overland flow are those of Izzard. l
'C. F. Izzard, Hydraulics of Runoff from D veloped Surfaces, Proc. Highway
Research Board, Vol. 26, pp. 129-150, 1946.
R. K. Linsley, M. A. Kohler, and J. L. H. Paulhus, "Applied Hydrology," pp. 275282, McGraw-Hill, New York, 1949.
HYDRO GRAPHS OF RUNOFF
210
lIe made tests 011 long flumes at various lopes and with various urfaces.
His te LS showed that the time to equilibrium is
2V.
60q.
(9-18)
where t. is defined as the time in minute when flow is 97 per cent of the
supply rate and V. is the volume of water in sW'face detention at equilibrium. From a strip of unit width the equilibrium flow q. is
q.
iL
(9-19)
43200
,
where i is the rainfall rate (or the rate of rainfall excess if the surface is
pervious) and Lithe distance of overland flow. The constant 43,200
gives q. in cubic feet per second when i is in inches per hour. Sub tituting average depth on the strip V./ L for outflow depth, Eq. (9-17)
becomes
V. = kq.'/d
L
(9-20)
ubstituting Eq. (9-19) for q. gives
V.
kL"'i'/d
35.1
(9-21)
where V. is the volume of detention (cu ft) on the strip at equilibrium.

Experimentally k was found to be given by
k
0.0007i
S~i
+c
(9-22)
where s is the surface slope and the retardance coefficient r. is as given in

Table 9-2.
TABLE 9-2. Retardance Coefficient c in Eq. (9-22)
Very Imooth asphalt pavement .. .. .. "
Tar and land pavement .. .. .
Crushed-slate rooflng paper.. .. .... . .
Concrete ............................
Tar and gravel pavement...... . . .
Closely clipped .od ...................
Dense bluegrass turf.... ..
0.007
0.0075
0.0082
0.012
0 . 017
0.046
0.060
Izzard found that the form of the overland-flow hydrograph could be

presented as a dimensionless graph (Fig. 9-14). With t. and q. known,
the ql q. curve permits plotting of the rising limb of the overland-flow
hydrograph. The dimensionle s recession curvp- of Fig. 9-14b defines
the shape of th receding linlb. At any time to after the end of rain, the
FLOOD FORMULAS
211
factor {3 is
{3 - 60g.l..
(9-23)
Vo
where Vo is the det ntion given by Eqs. (9-20) and (9-22), taking i = O.
9-10. Flood formulas. Many empirical formulas have b en preented for estimating flood peaks from small basins. For the most part
these formulas are mi leading and unsound, and they are discussed here
(0)
0.9
0.7
0.6
/
\
0.5
t:>o
(b)
0.8
.
~
f..-
II
0.4
0.3
\
\
0.2
0.1
.-/
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
tlte
r"-.
--
0.5 1.0 1.5 2'.0 2.5 3,0 3.5 4.0 4.5
/3
FIG. 9 -14. Izzard's dimensionless hydrograph of overland flow.
only in a general way.

function of basin area:
One type of formula makes th peak flow a

qp
= bAm
(9-24)
Fuller suggested m = 0.8, Myers2 propo 'ed m = 0.5, and Fanning used
= 5/ 6. The Talbot formula, which gives crOS6- ectional area of the
waterway rather than peak flow, us 6 the exponent 0.75. The differences
in the e exponents is 6uffi cient evidence that the formula is inadequate.
The coefficient b must incorporate the parameter of a rainfall-runoff
relation, frequ ency curve, and the factors which control hydrograph
shape. Only luck will permit the selection of the correct value of b for
a basin. Formulas of this type should never b used for engin ering
design.
Another group of formulas is typified by that of Burkli-Ziegler,3
l
qp
1
J
= Aci
.J
(9-25)
W. E. Fuller, Flood Flows, Trans. ASCE, Vol. 77, pp. 564-694, 1914.
C. S. Jarvis, Flood-flow Characteristics, Trans. ASCE, Vol. 89, pp, 985- ]032, 1926.
"The Greatest Discharge of Municipal Sewers," Zurich, ]8 O.
212
where i is the expe ted average rainfall in inches per hour, 8 is the average
slope of th watershed in feet per 1000 ft, A is the area in acres, and gp
is in cubic fe t per s condo The McMath! formula is of the arne form
except that the fifth root is used. The coefficient c in these formulas
is almost as inclusive as b in Eq. (9-24). Land slope is recognized as a
factor in basin torage, and the rainfall factor is introduced. Equations
of this type have no plaee in modern engineering design.
The roost widely u ed design equation for small basins is given the
misleading name It rational formula,"
qll
(9-26)
= CiA
where qll is in acre-inches per hour, i is average rainfall intensity in

inches per hour for a duration equal to the time of concentration of
the basin, and A is area in acre. With an error of less than 1 per 'ent,
flow rate in inches per hour per acre i equal to flow rate in cubic feet
per second. The quation states that the rate of runoff equals the rate
of supply (rainfall excess) if the rain lasts long enough to permit the
entire area to contribute. This assumption is literally true for an area
of a few square feet and may not involve serious error for basins up to a
few acres in area if the time of concentration is correct. For larger
areas, the coefficient C not only expre " es the proportion of the total
rainfall which runs off but al 0 the effect of overland flow and channel
storage on the peak. The coefficient is the ratio of the t.heoretical
absolute peak (100 per cent runoff and no storage) to the actual peak.
The problem of estimating C is no les than for any other formula.
Whether C is defined as qp!iA or as total runoff divided by total rainfall,
experience shows that its value varies widely from storm to storm.
Frequency is presumably accounted for by choosing i of known retu1'D
period. If C and i are statistically independ nt, C should have a return
period of uuity (1 yr) if the return period of the computed qp is to be the
TABLE 9-3. Values of Coefficient
Type of Area
C in Eq. (9-26)
Value of C
flat residential, 30% impervious area. . . . . . . . . . . . . . . . .

Moderately steep residential, 50 % Impervious .........
Built-up area, 70 % impervious .. ... ... ............. ..
Flat cultivated land, open sandy soil. . . . . . . . . . . . . . . . ..
Rolling cultivated land, cloy-loom soil. . . . . . . . . . . . . . . . .
Hill land , forested, clay-loam soll.. ... . . . . . . . . . . . . . . ..
0 . 40
0.65
0 . 80
0 . 20 t
0. sot
O. sot
... W. W. Horner and F. L. Flynt, Relation between Rainfall and Runoff from Small Urban Areas,
Tran . ASCf, Vol. 10 I, p. 140, 1936.
t M. Bernard, Discussion of Run-off--Rotionol Runoff Formulas, by R. L. Gregory and C. E. Arnold,
Tron,. ASCf, Vol. 96, p. 1038, 1932.
1 R. E. McMath, Det rmination of the Size of Sewers, Trans. ASCE, Vol. 16, pp.
183- J90, 1 7.
PROBLEMS
213
same a that for i. In practice, C i related only to typ of terrain without r gard to frequency (Tabl 9-3). Of all the flood formula , th
rational formula has the advantage that its physical meaning j rea,'onably
clear. However, it should be u ed with extl'eme caution since it does
not adequately recognize all the complication ' of the runoff proce s.
BLBLlOGRA PliY
Hoyt, W. G., and others: tudies of R lations of Rainfall and Runoff in the United
States, U.S. Geol. Survey Water-supply Paper 772, 193G.
Jarvis, larence S.: Floods, Chap. ' 1 G, in . E. Meinz r (ed .), "Hydrology,"
Jarvis, C. S., and others: F loods in the United tat,cs, Magnit lid and Frt'qllency,
U.S. Geol. Survey WatcNupply Paper 77 J , 1936.
Langbein, W. n., et al.: Topographic haract.eristics of Drl1inllgp Bn8ill~, U.S. Oeol.
Survey Water-8upply Paper 968-C, 1947.
Linsley, R. K., M. A. Kohler, and J . L. H. Paulhus: "Applied Hydrology," bap. 1.1,
Sherman, L. K.: The Unit Hydrograph, Chap. XI E, in O. E . Meinzer (ed .), "Hydrology," McGraw-Hill, New York, 1942.
Williams, G. R.: Hydrology, Chap. IV, in Hunter Rouse (ed.), "Engin ering Hydraulics," Wiley, New York, 1950.
PROBLEMS
9-1. Neglecting storage and assuming a linear rise and reccssion of the I mental
hydrograph, sketch the outflow hydrograph from a basin in the sbape of a GO sector
of a circle with outflow at th apex. Assume tmvel time proportional to distan ce and
rainfall duration equal to time of conl'entration.
9-2. Repeat Prob. 9-1 for a semicircle with outflow at the mid-point of the boundary diam'ter. What change in shape would result if the rainfall (runoff) were to
occur at unit rate for half tbe duration and illl'ce tilTlc~ that rate for tbe remainder'?
9-3. How would the hydrograph of Prob. 9-2 be affected if runoff were to oc;cur
only from the outer half of the area?
9-4. Given below are the observed flolVs from 8. storm of 3-hr duration on I~ stream
with a drainage area of 122 sq lUi. Derive the lInjt hydrogmph . Assum consto.ut.
base flow - 600 cis.
Rour
3
6
A.~{. . . . . . . . . . . .. .
A.M . . .... . ... ...
A.M . . . . . . . . . . . . . .
Noon . ............. ..
31'.M . . . . . . . . . . ...
61'.M . . . . . . . . . . . . . ..
9 P.M . ... . . . ........

Midnight ..... ........
Day 1
Day 2
600
550
6000
9500
SOOO
7000
6100
5300
4600
4000
3500
3100
2700
2400
2100
1900
Day 3
._--
I
I
1700
1500
1300
1100
900
800
700
600
214
9- 5. Given below ar three unit bydrographs d rived from separate storms on a

smaU basin. All ar believed to have resulted from 4-hr rains. Find the average
unit hydrograph. Drainage area = 3.5 sq mi.
HoUl'S
Storm 1
Storm 2
0
1
2
3
4
5
0
110
365
500
390
310
235
175
130
95
65
40
22
10
5
0
0
25
125
358
465
405
305
220
170
130
90
60
35
20
8
0
7
8
9
10
11
12
13
14
15
I Storm 3
0
16
58
173
337
440
400
285
215
165
122
90
60
35
16
0
9-6. The hydrograph tabulated below result d from three successive 6-h .. periods of
rainfall having runoffs estimated as 0.6, 1.2, and 0.9 in., respectively. Using the
method illustrated by Eq. (9-1), lind the 6-hr unit hydrograph for this basin . Drainage area - 58 sq mi. Base flow has been subtracted.
Time, hr
0
3
6
9
12
15
1
21
24
27
30
33
36
Flow, cfs
0
750
2800
2830
6620
4320
6450
3140
1950
930
310
DO
0
9-7 . As an illustrat.ion of the eff ct of minor errors on the computation of Prob. 9-6,
repeat, using th same figures except for the ordinate at 6 br which is changed to
2600 crs.
9-8. Using storm 2 of Prob. 9-5, construct the S curve and find the 2- and 6-hr
unit hydrographs. Smooth the S curve as required.
PROBLEMS
215
9-9. Given b low is the 4-hr unit bydrograph for a basin of 4 sq rni.
the S curve and find the 2- and 6-hr unit hydrographs.
Time, hr
Flow, cf
Time, hr
Flow, cfs
0
1
2
3
4
5
6
7
8
9
10
0
400
2500
44.00
6000
7000
6100
5200
4500
3800
3200
11
12
13
14
15
16
J7
18
19
20
21
2700
2200
1800
1400
1100
800
600
400
200
Construct
]00
0
9-10. Using the unit hydrograph of storm 3, Prob. 9-5, find the peak flow resulting
from four successive 4-hr periods of ra.infall producing 0.35, 0.87, 1.39, and 0.77 in. of
runoff, respectively. Ignore base flow .
9-11. For a drainage basin selected by your instructor, derive a synthetic unit
hydrograph using nyder's method with Ct - 2.0 and C1I - 0.62.
9-12. A basin of 139 sq mi has L = 16 mi, L .... 6 mi. Using Snyder's metbod,
coefficients as in Prob. 9-11, and tn = 3 hr, find the unit hydrograph.
9-13. Using actual streamflow data for a basin assigned by your instructor, fmd
the unit hydrogra.ph for a storm. What values of Ct and CJ) for Snyder's metbod are
indicated by the data?
9-14. A parking lot 150 ft long in tho direction of the slope and 80 ft wide has a tar
and gravel pavement on a slope of 0.0025. Assuming a uniform rainfall intensity of
2.75 in./br for 30 min, construct the outflow hydrograph, using Izzard's method.
9-15. A city lot 200 ft deep and 100 ft wide bas a slope of 0.005 toward the streot.
The street is 60 ft wide and bas a 6-in. crown. Assuming a rainfall intensity of 1.8
in./hI, c "" 0.040 for the lot and 0.007 for the street, and a rainfall duration of 60 miD,
find the peak flow into the gutter. What will be the peak flow if the rainfall duration
is 10 min?
....,.
: /
A
r'"
.'
v"
... V
.'.'
,.
_",..
.,
".~,;
,,
~.....
-I
~)
C"
.""' ....._~ ..
"II
'II'
10
STREAMFLOW ROUTING
As the discharge in a channel inc rea es, stage also increases and with
the volume of water in temporary storage in the channel. Durin!!.
the falling portion of a flood an equal volume of water must b released
from storage. As a result, a flood wave moving down a channel appeare
to have its time base lengthened and (if volume remains constant) it
crest low red. The flood wave is said to be attenuated. Streamflow
routing is the technique u d in hydrology to compute the effect of channel
storage on the shape and movement of a flood wave.
j~
FIG. 10-1. Definition sketch for analysis of a monoclinal rising wave.
Giv n the flow at an upstream point, routing may be used to computo

the flow at a downstr am point. The principles of routing apply also
to computation of the eff ct of a reservoir on the shape of a flood wave.
Hydraulic storage occur not only in chann Is and re ervoirs but also
as water flowing over the ground surface. Hence, 'torage i effective at
the very inception of the flood wave, and routing technique may be u 'ed
to compute the hydrogl'aph which will result from a pecified pattern of
rainfall exces .
10-1. Wave movement. One of the simplest wave forms is the
monoclinal rising wave in a uniform channe1.
uch a wave (Fig. 10-1)
consist!:' of an initial steady flow, a period of uniformly increasing flow,
and a continuing steady flow at the higher rate.
uperimposing on this
wave system a velocity equal and opposite to the wave celerity u causes
216
WAVE MOVEMENT
217
the wave configuration to become stationary, and a steady Bow q' take
place from right to left, with the velocities shown. This Bow, known a
the overrun, is
q' = (u - vl)A I = (u - v2)A 2
(10-1)
where A is the oross- ectional area of the channel.
the wave celerity gives
olving Eq. (10-1) for

(10-2)
The celerity of a monoclinal wave is thus a function of the area-discharge

relation for the stream (Fig. 10-2).
Since velocity usually increases with
stage, area-discharge curves are
usually concave upward. The
slopes of the secants OA and OB
represent the water velocities at
sections 1 and 2, respectively
(VI = qt! Al = tan 01), while the
~
slope of the secant AB represents ..:ou
the wave celerity [Eq. (10-2)]. It C'"
may be concluded that (1) wave
celerity is greater than the water
velocity in most channels; (2) for
a given peak Bow, the wave having the highest initial flow will
travel fastest; and (3) for a wave
of very small height
Area
OJ
(10-3)
FIG. 10-2. Typical area-discharge relation

for a stream and its influence on wave
celerity.
where B is the channel width.

Equation (10-3) is known as Seddon's law after the man l who first demonstrated its validity on the Mississippi River.
From the Cbezy formula for flow in a. wide, open channel (assuming
depth equal to hydraulic radius),
(10-4)
and
."
q = Av = vBy =
CByHs~S
(10-5)
= ~ Bv
(10-6)
Differentiating,
dq = ~ CByY,8~i
dy
2
1
J. Seddon, River Hydraulics, Trans. ASCE, Vol. 43, pp. 217-229, 1900.
STREAMFLOW ROUTING
218
Substit uting Eq. (10-6) in Eq. (10-3),

3
= ZV
(10-7)
The derived ratio between water velocity and wav eel rity depends on
channel shape and the flow formula used. Value shown in Table 10-1
may be us dar ugh guides for e timates of wave celerity.
TABLE 10 1. Theoretical Ratio between Wave Celerity and Water
Velocity for Typical Sections
Sho p.
Manning
Chezy
Triangle . . . .. .
Wide rectangle . . . .. .
Wid e parabola . . .
1. 33
1.67
1. ....
1. 25
1.50
1. 33
A second type of wave is shown in Fig. 10-3. Tbis is an abrupt wave,

and the igur illustrates conditions 1 sec after the instantaneous opening
~.---
--+-I",
~q
FIG. 10 3. Deflnition ske tch for the analysis of an abrupt translatory wave.
of the gate. The volume of wat r entering the channel in tbis time is
q2 = A 2v2 (area acfd) .
The il1crea d volume abhg is
(10-8)
ubstituting Av = q,
V2
= (Alvl
+Au 2
(10-9)
A1u) A2
The volume dfjg has been accelerated from
VI
t o V2 by the force F:
(10-10)
WAVES IN NATURAL CHANNELS
219
where w is the p cHic wight of water.

ince F 0.1
in pres ure on A I and A 2,
F = WA21/2 - wA l1/1
equal the difT r nce

(10-11)
where 1/ is the depth to cent r of gravity of the section. Equating

Eqs. (10-10) and (10-11), inserting V2 from Eq. (10-9), and solving for u
give
(10-12)
In unit width of rectangular channel we may substitut D = A and

D/ 2 = 1/. Hence
u
= VI
For very small wave height,
~~~: (D2 + D 1)
DI "'" D 2
u =
VI
(10-13)
and
ViJ5
(10-14)
Equation (10-12) is a general equation applying to any channel. Equation (1 0-13) applies only to rectangular channels, and Eq. (10-14) only
to waves of very small height in rectangular channels. Abrupt translatory waves occur as tidal bores in many estuaries, as surges in power
canals and tailraces, as seiches in lakes, and occasionally as flood waves
caused by intense, small-area storms.
10-2. Wave in natural channel. Controlled experiments l in
flumes of regular cross section have confirmed the equations developed
in Sec. 10-1. Reasonable checks have also been obtained in natural
streams where the effect of 10 '0.1 inflow is negligible, as in eddon's work
on the Lower Mississippi and Wilkinson' s2 study of waves downstream
from TVA dams. Equation (10-14) gives very good estimates of the
celerity of impulse waves in still water.
Simple mathemati '0.1 treatment of flood waves is n cessarily limited to
uniform channels with fairly regular cross section. The hydrologist
must deal with nonuniform channel of complex section with nonuniform
slope and varying roughness. The formulas of S c. 10-1 apply to waves
generated at a point on a channel, but most flood waves are generated
by nonuniform lat ral inflow along all the channels of the stream system.
Thus natural flood waves are considerably more complex than the
1 R. E. Horton, Channel Waves Subj ect Chiefly to Momentum Control, Permanent
Intern. AS8oc. Navigation ongr. Bull. 27, 1939.
E. E. Moots, A Study in F lood Waves, Univ. I()'IJ)a Studies Eng. Bull. 14, 1938.
A. choklitsch, Dam Break Waves, Math. natl~rw. KlaB8e, Vol. 126, pp. 1489- 1514,
1917.
I J. H. Wilkinson, Translatory Waves in Natural Channels, Tran8 . ASCE, Vol. 72.
pp. 1203- )236, 1945.
STREAMFLOW ROUTING
220
simplified cases which yield to mathematical analysis and experimental

7erification.
Theoretical treatm nt is useful in studi s of surges in canals, impulse
waves in till water (including s iches and tides), and waves released
from dams. Waves in natural channels are usually analyzed by use
of the continuity equation. Natural flood waves are generally intermediate between pure translation as de eribed in ee. 10-1 and pond age
which would occur in a re ervoir or lake. Figure 10-4 hows an example
13
"
12
11
10
.l!!
("I
0
0
~
.S
Q)
eo
0
,;
("I
OIl
I- Brtdgeport,
Neb.
_..........,
FIG.
1\
I
I
I
/
Mdt.
'\-Lisco, Neb.
'\..
6
5
I"
\ /
'"
North Platte River _

June 1935
"
......_ r---....... -......
- --
__./
Noon
June 2
Mdt.
Noon
June 3
Mdt.
..........
Noon
June 4
10 ." Ex ample of t ra nslatory wave movement, North Platte River b etween Bridgeport
and Lisco, Nebraska.
of a flood wave moving with nearly pure translation , i.e., little change in
hape. Figure 10-5 illustrates the great modifications which can occur
when a flood wave moves through a reservoir in which discharge is a function of the quantity of water in storage. Momentum fore predominate
in pure translatory waves, and such waves have relatively short time
bases compared with the dimension of the system in whieh they move.
Most natural waves move under friction control and have time bases
considerably exceeding the dimensions of the stream system.
10-3. The storage equation. Routing i the solution of the storage
equation which is an expre ion of continuity;
d
1 - 0 =dt
(10-15)
68
1- 0 = -
(10-16)
or, more commonly,
THE STORAGE EQUATI O N
221
where I is inflow rate, 0 i outflow rate, and is storage, all for a sp cHic
reach of a stream. Equation (10-16) i exact, but it application to
practical probl m involves approximation. To provid a form more
convenient for u e, it j commonly a umed that the average of th Bows
at the beginning and ending of a hort time period t (routing period) qual
30
25
~
L\
20
.,
15
0'"
-_
"..-
5
I
--
r\ T ~\
~'
13
EnqlewOOd, Ohio
10
0
.c;
Stillwater River
January 1937
~Pleosonf HilI, Ohio
0
0
.:
Storage in the Engllewood I

retard in g basin of the Miami
Conservancy District
14
15
16
18
17
19
1-- ...... -
If 1\- \
,-~
-...._)
20
21
22
Dote
FIG. 10-5. Reduction of discha rge throug h reservoir action, Stillwater River, Ohio.
the average flow during the period. Using subscripts 1 and 2 to represent
the beginning and end of the period, respectively,
11
+ 12 t
2
_ 01
+O
t =- S2 - S1
(10-17)
Most storage-routing methods are based on Eq. (10-] 7). It is assumed

that II, 1 2, 0 1, and 1 are known and O2 and S2 must be determined.
Since there are two unknowns, a second relation between storage and
flow is needed to complete a solutioll. The major difficulties in storage
routing are involved in this latter relation.
The assumption that (11 + 1 2)/2 = 1 implies that the hydl'ograph is
a straight line during the routing period t. Thus the controlling factor
in selecting the routing period is that it be sufficiently short so that this
assumption is not seriously violated. The routing period should never
be greater than the time of travel through the reach, for, if it were, it
would be po sible for the wave crest to pa s completely through the reach
during a routing period. If the routing period is shorter than is really
necessary, the work of routing is increased since the same computations
222
STREAMFLOW ROUTING
are required for each routing period. Generally a routing period between
one-half and one-third of the time of trav 1 will work quite weil.
10-4. Determination of storage. B fore a relation b tween storage and flow can be establish d it is nee ary to determine the volume
of water in the stream at various times. The obvious method for finding
storage is to compute volumes in the chann I from cros sections by use
of the prismoidal formula. Th water surface is usually a umed to be
level between cross sections. Total storage in the reach for any given
flow conditions is the sum of the storage increments betw en succes ive
cro s sections. For the summation, the elevation in any subreach is
Computed backwater profile
FIG. 10-6. Computation of reach storage from channel cross sections.
the elevation indicated by a backwater curve for the mid-point of the

subreach (Fig. 10-6). This method requires extensive surveys to provide
adequate oross sections and computation of water-surface profiles for
many unsteady, nonuniform flow situations in order to represent the
range of conditions expected. The method is difficult and relatively
costly and is used where no alternative is possible. It would be used,
for example, to compute storage in a reach in which channel alteration
or levee construction is planned, since conditions after construction would
be quite unlike those existing before construction.
Storage-elevation curve for re ervoirs are usually computed by planimetering the area enclosed within successive contours on a topographic
map. The measured area multiplied by the contour interval gives the
increment of volume from the mid-point of one contour interval to the
mid-point of the next highest interval. A level water surface is usually
assumed, a condition which is satisfied in mo t reservoirs. In reservoirs
with relatively small cross-sectional area, the water surface may be far
from level when large flows occur (Fig. 10-7). Under such conditions
a computation similar to that described above for natural channels must
be used.
The common method of determining storage in a reach of natural
DETERMINATION OF STORAGE
223
channel is to us{; Eq. (10-16) or (10-17) with ob erved flows. Figure

10-8 shows the inflow and outflow hydrographs for a r ach of riv r.
When inflow exceeds outflow, AS is po itive, and wh n outflow exc d
inflow, flS is negative. Since routing involve only flS, absolute storage
volumes are not n essary and the point of zero storage can b taken
arbitrarily. Thus, the storage at any time is the urn of th po itive
and negative storage increments since the sleeted zero point. The
computation is illustrated in the figure.
-l:!
"
568
<::)
u; 566
E
C1)
>
564
.0
c:
>
C1)
q;~
..
"" ~
% <;
562
<::)
560
~ i3
~
0
\OJ
558
556,
0
es
!
15(;000
49. 00 ,
10
20
30
40
50
60
70
80
Miles above Wheeler Dam
FIG. 10-7. Proflles of the water surface in Wheeler Reservoir on the Tennessee River.
(Data from TVA.)
One of the most annoying problems in flood routing is the treatment

of the local inflow which enters the reach between the inflow and outflow
stations. If the local inflow enters mainly near the upstream end of the
reach, it is usually added to the main-stream inflow to obtain total inflow.
If local inflow occurs primarily near the downstream end of the reach,
it may be 8ubtracted from the outflow before storage is computed. In
this case, the main-stream flow is routed through the reach and the local
inflow added after routing is complete. Between these two extremes
lie many possibilities of combining various percentages of the local with
the main-stream inflow before routing and adding the remainder to the
outflow after routing. If the local inflow is relatively small compared
with the main-stream inflow, any reasonable treatment consistently
applied should give satisfactory results. If local inflow is large, consideration should be given to reducing the length of the reach.
STREAMFLOW ROUTING
22..
The total volume of ungaged local inflow can be found by subtracting

measured outflow from measured inflow for a period beginning and ending
at about the same low flow, that is, t:..S = O. The time distribution of
the ungaged local inflow is usually assumed to agree with the observed
flows in a small tributary similar in size and character to the typical
streams of the ungag d area. This procedure throws all errors of flow
measurement into the ungaged local inflow, and the resulting flows may
7
nme
3I6p
3/Mdl
\
/ t v-'
Gain to
storage
\
/
f\ ,
F:::.d/
o
4/N
60
6p
3S50
2S~0
3300
4200 3350
8~0
5130
4/MdI
8430
5/ 60
9 280
2950 3700 - 750
8530
5IN
2100 3680 -1580
~Outf/ow
6950
5/6p
1470 3100 -1630

5/MdI
5320
1000 2450 -1 450
3870
S/So
740 2000 -12S0
S/N
2SIO
SOO IS50 1050
"""",
~ .......___
Noon
~ 5 00 18 ~0
5950
Md!
1480
4/Sp
\t
J V
1/ /'
0
2500 1020 1480
s"
from
' \. ,~
storage
660 660
4/So
- /nflow
.S
-0/- f
0
560 700 - 140
Md!
60
Noon
6p
Mdt
"
60
I
Noon
S/Sp
1560
530 1300 -770
~ 6/Md!
6p
790
"
._1--
.......... ....._
Mdt
60
Noon
6p
Mdt
FIG. 10-8. Calculation5 of channel storage from Inflow and outflow hydrographs.
not appear altogcth r rea onable. If influent seepage is large, the computed un gaged inflow may be negative.
10-5. Routing in a sim.ple re ervoir. A reservoir in which the
discharge is a function of water-surface elevation offers the simplest of
all routing situations. Such a reservoir may have ungated sluiceways
and/ or an uncontrolled spillway. Reservoirs having sluiceway or spillway gates may be treated as simple reservoirs if the gates remain at
fixed openings. The known data on the reservoir are the elevationstorage curve and the elevation-discharge curve (Fig. 10-9) . Equation
(10-17) may be transformed 1 to
I
R. D. Goodrich, Rapid Calculation of Reservoir Discharge, Civil Eng., Vol. 1,
pp. 4.17- 418, 1931.
ROUTING IN .A SIMPLE RESERVOIR
11
+ 12 + e~l
225
2~1 + o.
0 1) =
(10-18)
Solution of Eq. (10-1 ) requir s a routing curve ho\ing 2 It + 0 vs. 0

(Fig. 10-9). All terms on th left-hand side of the quation are known ,
and a value of 28 2/ t + O2 can be computed. The corresponding valu'.J
35
30
AI
OJ
~,. ~
t :~ dOY'
~~ + 0 : [ 40 + 100 '
1
,
,
Wi
~
c:
0
'"
W
--
./
40
20
100
200
300
40 0
500
120
'"""o
~/
1/
40
60 ~
'>' /
10
6'
cS
100
~1
15
1
/
84iV
Elevotion of splllwoy cresl
.S
15:>
.!\.'
20
ng
[ I
..:::
flow of 100 cfs the storoge is 185 si d.
T7
I /
25
600
70 0
800
900
1000
Sioroge in s td, disc harge and ( 2/ +0) in cts
FIG. 10-9. Routing curves for a typical reservoir.
of O2 can be d termin d from the routing curve. The computatiolJ

then repeated for succeeding routing periods. Table 10-2 mustratef..
a typical 'olution. It should be noted that 281t - 0 is easily computed
as (281t + 0) - 20 .
18
TABLE 10-2. Routing with the 28 f t
-Dol.
Hour
25, _ 0, d.
I, d.
2
3
Noon
Midnight
Noon
Midnight
Noon
Midnight
20
50
100
120
80
40
+0
Curve of Fig. 10-9
25 2
+ 0, cf.
O,ck
500
540
658
798
832
162
470
508
578
632
642
620
I
NOTE. Dare: available at .tart of routing .hown in boldface.
16
16
40
83
95
7!
-- -
226
STREAMFLO W ROUTING
10-6. Routing in a gated re ervoir. Routing in a re ervoir with

gated outlets dep nds on the method of operation. A general equation
is obtained by modifying Eq. (10-17) to
II
+2 12 t _
01
+2 O
t - Out
= 82
(10-19)
where 0 is uncontrolled outflow and On is regulated outflow.
If 0 is
Gales
apen
12
VI
ClI
V
~10
.~
~ 8
.c
o
.!:
c
.2
~ClI
OJ 4
(5
&.
1000
2000
3000
4000
5000
Storage in sId, outflow ond
6000
7000
8000
f + 0 in cIs
FIG. 10-10. Routing curves for a gated reservoir.
zero, Eq. (10-19) b comes
It -
ant
+8
(10-20)
82
which can be readily solv d for 8 2 and re ervoir elevati

zero, the r uting equation becomes
11
+ 12 - 20n + e~l
0 1)
(2~2 + 0
~.
2)
If 0 is not
(10-21)
The solution of Eq. (10-21) is identical with that outlined in ec. 10-5
xcept for the inclusion of On.
227
ROUTING IN RIVER CHANNELS
If the gates are et at fixed opening 0 that th di charge i a. function

of head, the solution require a family of 2 It + 0 curve for variou
gate openings (Fig. 10-10). The routing method i the sam a that
illustrated in ec. 10-5 except that the curve appropriate to th xi ting
gate opening i used each time.
5
...
x=O
0
0
2 3
.:
./'
~ 2
I
=
+
....
V 1/
~V
""
0.5
1.0
1.5
x=o.fO
..",
V~
2.0
'7
~
1900
K= 2000 =0.95 day
2.5
0.5
1.5
2.0
2.5
S in 1000 sId
FIG. 10-11. Determination of the Muskingum storage constants [Eq. (10-23)).
S .in 1000 sfd
1.0
3.0
Negative wedge
sfarage
prism storage
prism storage
FIG. 10 12. Some possible water-surfoce proflles during the passage of a flood wave.
10-7. Routing in river channels. Routing in natural river channels is complicat d by th fact that torage is not a function of outflow
alone. This is illustrated when the storage computed in Fig. 10-8 i
plotted against simultaneous outflow. The resulting curve is usually
a wide loop indicating greater storage for a given outflow during ri ing
stages than during falling (Fig. 10-11). The cau e is obvious if on considers the backwater profiles existing at various times during the paosage
of the flood wave (Fig. 10-12). The storage beneath a line parallel to the
.,tream bed is called prism storage; between this line and the actual profile,
wedge storage. During rising stages a considerable volume of wedge
228
STREAMFLOW ROUTING
storage may exist before any large increase in outflow occurs. During
falling stages, inflow drops more rapidly than outflow and the wedgestorage volume becomes negative. Routing in streams requires a storage
relationship which adequately represents the wedge storage. This i~
usually done by including inflow as a parameter in the storage equation.
10-8. Streamflow routing: analytical method. One expression
for storage in a reach of a stream is
b
a
S = - [xImln
+ (1
- x)Omln]
(10-22)
where a and n are con tants from the mean stage-discharge relation for
the reach, q = ag", and band m are constants in the mean stage-storage'
relation for the reach, S = bg m In a uniform rectangular channel, storage would vary with the first power of stage (m = 1) and discharge would
vary as the % power (Manning formula). In a natural channel with
overbank flood plains the exponent n may approach or become less than
unity. The constant x expre es the relative importance of inflow and
outflow in determining storage. For a simple reservoir, x = 0 (inflow
has no effect), while if inflow and outflow were equally effective, x would
be 0.5. For most streams, x is between 0 and 0.3 with a mean value
near 0.2.
The Muskingum method l assumes that mi n = 1 and lets bl a = K.
Equation (10-22) then becomes
S = K[xI
+ (1
- x)O]
(10-23)
The constant K, known as the storage constant, is the ratio of storage to

discharge and has the dimension of time. It is approximately equal to
the travel time through the reach and, in the absence of better data, is
som times estimated in this way. If flow data on previous floods are
available, K and x are determined by plotting S vs. xl + (1 - x)O for
various value of x (Fig. 10-11). The best value of x is that which causes
the data to plot most nearly as a single-valued curve. The Muskingum
meth d assumes thi curve is a straight line with slope ~: equivalent to K.
Th unit of K depend on units of flow and storage. If storage is in
second-foot-days and flow in cubic feet per second, K is in days.
If Eq. (10-23) is substituted for S in Eq. (10-17) and like terms are
collected, the resulting equation reduces to
(10-24)
1 G. T. McCarthy, The Unit Hydrograph a.nd Flood Routing, presented at confereno of orth Atlantic Division, U.S. Corps of Engineers, June, 1938 (see aJso "Engine ring Construction-Flood Control," pp. 147-156, The Engineer Schooi, Ft.
Belvoir. Va., 1940).
SEMIGRAPHICAL METHODS
wher
Co
Cl
C2
Kx - 0.5t
+ 0.5t
Kx + 0.5l
= K - Kx + 0.5t
K - Kx - 0.5l
= K - Kx + 0.5t
(10-24a)
= - K - Kx
(10-24b)
(10-24c)
Combining Eqs. (10-24, a, b, and c)

Co
+ CI + C2
(lO-24d)
In these equations t is the routing period in the arne tim unit as K.

With K, x, and t establi hed , values Co, CI, and C2 can be comput d. The
routing operation is simply a solution of Eq. (1 0-24) with the O2 of one
routing period becoming the 0 1 of the succeed ing period. Table 10-3
illustrates a typical computation.
TABLE 10-3. Applicotion of the Muskingum Method
Date
Hour
I, d.
col ,
c]/1
- - - - - -- -_. - - I
6 A.M.
Noon
6 P.M.
Midnight
6 A.M.
Noon
6 P.M.
10
30
68
50
40
31
23
0, d.
C, OI
...
....
....
3.7
S.4
6.2
5.0
3.S
2.9
3.5
10 . 6
20(.0
17.7
14 .1
10.9
5.2
6.5
13 . 3
22.7
23 . 7
- -10
21.8
12 . 4
25.5
43 .5
45 . 4
41.6
35.6
NOTE Based On K = 11 hr, t - 6 hr, " ... 0.13; hence, co = 0.124, c] "" 0.353, and
Values known at the beginning of routing are shown in boldface.
Cl . .
0.523.
ince most routing procedures involve computation of cumulative

storage, the ouflow at any time can be determined only by routing from
t he last known value of outflow. An expression for O. can be written
from Eq. (1 0-24) as
04
col 4
+ cJa + c2cola + C1C J 2 +
ince C2 is Ie s than unity,

coefficients gives
C2 3
c2
colz
+ C2 Cdt + C
2
01
(10-25)
will usually be negligible, and combining

(10-26)
Equation (10-26) proyjdes a means of computing outflow at any time

if the preceding inflows are known.
10-9. Streamfiow routing: emigraphicaJ methods. The Muskingum method assumes that K is constant at all flows. While thi8
STREAMFLOW ROUTING
230
aRSumption is generally adequate, in some cases the storage-flow relation

is so curv d that an alternative method mu t be found. If torage is
expr ased as a function of aI + 0, Eq. (10-18) can be written
(10-27)
A curve r lating 28/ t + aI + 0 and aI + 0 is required. All terms on
the left are known, and a value of 28 2/t + aI 2 + O2 can be computed and
Time
FIG. 10-13. A graphical routing method.
the corresponding value of aI 2 + O2 read from the curve. Since 12 is

known, O2 can be computed.
Many solutions of the semigraphical type are possible using any
storage function which reduces the data to a single-valued curve. All
solutions are based on a transformation of Eq. (10-18) to fit the ele ted
storage function. Equation (10-27) should be considered as a typical
example rather than as a specific method.
10-10. Streamflow routing : graphical m e thod. A variety of
graphioal methods foc solving the routing equation have been suggested.
ften graphical methods are Ie s convenient than other methods unless
many repetitions of the operation are expected. Graphical method are
the ba is for a number of meohanical routing devices.
The Muskingum storage equation with x = 0 may be expressed as
d8 = KdO
dt
dt
(10-28)
GRAPHICAL METHODS
231
From Eq. (10-15), I - 0 = d /dt. Combining thi with Eq. {lO-2 )

gives
1-0
dO
(10-29)
= dt
--r
Equation (10-29) can be u ed as the ba is for a very simple graphical

routing method. 1 Given an inflow hydrograph (Fig. 10-13), a straight
Time
FIG. 10- 14. G raphica l d ete rmination of th e storage fa ctor K.
line connecting O2 with a point K time units to the right of I2 indicates

the slope of the outflow hydrograph at O2 No routing period is involved,
and the slope of the outflow hydrograph can b defined at any desired
point. Note that K need not be constant but can be expressed as a function of O. The procedure is therefore suitable for routing through an
ungated reservoir for which a K curve (dS / dO vs. 0) can be constructed.
It is not neces ary to determine K by the procedure described in ec.
10-8. Instead, K may be found by reversing the routing procedure
described above. A straight line conforming to the slope of the outflow
hydrograph at any time t is projected to a discharge value equal to the
inflow at that time. The time difference between the inflow and this
projection is K (Fig. 10-]4).
The graphical proc dure de cribed above assumes pure reservoir action
(x = 0), and the peak of the outflow hydrograph must fall on the receding
limb of the inflow hydrograph. A graphical construction introducing
the equivalent of the Muskingum x could be derived, but a simpler
I W. T. \Vilson, A Graphical Flood-routing Method, Tran8. Am. GeophY8. Union,
Vol. 21, Part 3, pp. 893- 898, ] 941.
232
STREAMFLOW ROUTING
solution is available. The factor x may be viewed as a measure of the

translatory component of the wave motion. Figure 10-15 show that,
x=0.2
x =0.4
,,
,, [
'I ..,\,
-...... ...'"'.,\,
.,
... '"
\
" ' :"

...
'~
... ~
'"'-,
Time
FIG. 10- 1S. Effect of changes in the Muskingum x on the outflow hydrograph.
with a constant K, the translation of the hydrograph increases a. x

increases. Thus the elIect of inoreasing x can be introduced by lagging
the inflow hydrograph. If th lag
TL is constant, it is immaterial
wh ther the inflow is lag-ged and
then routed or the routed flow is
lagg d . A completely flexible procedure would utilize both variable
.,
K and 'i'L as functions of flow.
'"
o
.c
u
ince, with no translation, the outVI
Ci
flow peak would fall on the inflow
recession, a measure of TL is the
time difference between the outflow
peak and the occurrence of an equal
flow during the rece sion of inflow
(Fig. 10-16). Values of TL can be
Time
determined
from the hydrographs of
FIG. 10-16. Determination of the lag Tj,.
several historic flood s and plotted
as a TL vs. I curve (Fig. 10-17). Using the historic data, inflow is lagged
a cording to the TL vs. I curve, and a K vs. 0 curve is constructed from
the lagged inflow and ob 'erved outflow as de cribed earlier. 1
1 M. A. Kohler, Mochanical Analogs Aid Graphical Flood Routing, J. Hydra1J.tic.
Dill. ASCE, Vol. 84, April, 1958.
233
ROUTING AIDS
10-11. trcamflow routing: stati~tical m thods. Equation

(10-24) and (10-26) are linear, multivariable equations. The constants
in the equations could be e tabli h d by conventional least-square
analy i u ing data from historic flood. Anoth r approach is the use
of the coaxial graphical-correlation technique. Figure 10-18 illustrates
uch a graphical correlation u 'ing an equation such as Eq. (10-26)
involving only inflow values. A similar relationship might have been
developed using the conventional parameter (11,12,0 1,02).
7
-\
........
6
Observed
inflow--........,
I> (f/\~
'i. vsI .....
\'...-rLOqqed inflow
1\
f--K
A/ \
I
I It::Yj
I
.E
- :1 /.
--_
,
,
"
\\. /ROUfed outflow

\
20
",
~' ' , '" ' .....
\SfrOiq/1fedqe
.J--..i
10
~ ................ --,_
....
........
............
30
~ KvsO
40
Time in hours
50
60
70
80
\\
o
1i. ond
\~
10
20
I< in hours
FIG. 10-17. Graphical flood routing with a variable lag and storage factar.
10-12. Routing aid. Routing is one of the few operations in

hydrology which lends itself to computational aid, and numerous such
devices have been developed. These aids can be ciassified as slide rules or
nomograms for solving the routing equations, mechanical devices which
perform the routing operation graphically, and analogs which simulate
hydraulic wave travel with some other medium.
Since Eq. (10-18) involves only addition of terms, this can be easily
accomplished with a slide rule (Fig. 10-19). Scales are laid off in terms
, f the factor in the equation but marked in terms of inflow and outflow
by u e of the relation between outflow and storage. Nomograms may
be developed to serve the same p urpo e. More complicated scale
arrangements are po sible to handle special situations.
STREAMFLOW ROUTING
~ 300r---~----~~~~+---'_--~
'"
~-----+-------r------~~~~+-----~ 300 ~
o
o
.s
~----~-------1200 ~
L-----~----~----~------~----~O
FIG. 10-18. Coaxial graphical.routing relation.

closely to K = 12 hr, x = 0.2.
o
60
With t = 12 hr, the curves correspond
Oufflow at end of period, O2 (1000 cfs)

10
20
30
40
50 40 30 20 10
Average inflow (1000 cfs)
~I,;I2 1+
50
10 20
Outflow at beginning of period, 0, (1000 cfsl
$, - O~f ~
FIG. 10-19. A simple routing slide rule using modifled form of Eq. (10-18).
are marked at distances from origin equal to corresponding values of S
Outflow values
Ot/ 2.
235
ROUTING AIDS
Mechanical routing d vices are eith r linkag

imilar to a pantograph
or rolling devices which perform a. graphical-routing olution m chanically. In the pantograph type the operator u uaUy follows a storage
curve and inflow hydrograph with pointers whil a. third pointer traces
the outflow hydrograph.
uch d vices usually require con idera.ble
modification to adapt them from one reach to another. A rolling device,
the Harkness flood router, is shown in Fig. 10-20. This devi e u e a
30
!t-o.f ----- K
- - --+-I
50
40
Inflow
~
30
31:
20
Offset
10
O~----------~----------------L_
_____
Time
FIG. 10-20. The Ha rkness flood router.
solution similar to that described in Sec. 10-10. The constant [( is set

along bar AB and the pencil P traces the outflow graph, so that PC is
equivalent to the straightedge of Fig. 10-17. Because of mechanical
problems, the outflow hydrograph is offset vertically by a fixed distance
so that the outflow pencil and the inflow tracer will not be at the sarno
point when the hydrograpbs cross. A more flexible mecharucal analog
has since been developed 1 which can be used for problems involving variable K and 'I'L,
The most convenient analog computer is electrical. In a simple
circuit having a capacitor of capacitance C, the charge stored in the
1 M. A. Kohler, Mechanical Analogs Aid Graphical Flood Routing, J . HydrauliC8
DiIJ. ASCE, Vol. 84, April, 1958.
236
STREAMFLOW ROUTING
capacitor is equal to CE, where E i the voltage drop aero

ince E = Ri, the product of resistance and current,
=
the capacitor.
CRi
(10-30)
which is equivalent to the hydrologic equation S = Kq. A circuit such

as that given in Fig. 10-21 can be shown to have characteristics such
that storage in the two capacitors conforms to Eq. (10-23), when I is
the inflow current and 0 is the outflow current.1 The inflow hydrograph
R3
FIG. 10-21. Circuit diagram for an electronic flood-routing device.
Bureau.)
(U.S.
Weather
is drawn on the chart of potentiometer PI and th inflow current controlled

so that the p n of the potentiom ter follow this trace.
ince the storage
in the system is analogous to that of a stream, the current at potentiometer Po represents outflow. Value of K and x for the system are adjusted
by changing the r sistances R l , R 2, and Rs or chart speed. Since the
analog solves the differential form of the storag equation [Eq. (10-15)],
it is free of the errors introduced by use of finite routing p riods. Values
of K and x for a reach may be found by trial and error rather than by the
analysis presented in ec. 10-8.
1 R. 1(. Linsley, L. W. Foskett, and M. A. Kohl r, Electronic Device Spe ds Flood
Routing, Eng. New8-Record, Vol. 141, No. 26, pp. 64-66, Dec. 23, 1948.
237
DERIVING BASIN OUTFLOW BY ROUl'ING
10-13. Driving ba in outilow h y routin.g. Th shape of t he

bydrograph from a basin i dependent on the travel time through the
basin and on the hap and storage characteri tic of th ba in.
onsidering exce. ' rainfall (runoff) to be inflow and the hydrograph to be
outflow, the probl m i analogous to to rag routing. The similarity
of Eq . (9-1) and (J 0-26) show that the ullit hydrograph it eli i ba i ally
a set of average routing coefficients.
r--
7
'" 6
.!:!
's:
f!:'
~4
co
___ ___ Isochrones
\<1
3r--
0~~--2~~3--4L-~5--~6
Time
in hours
FIG. 10-22. Derivatian af th e time-area diagram for a basin.
The nature of the problem suggests the u e of " lag and route" methods
( ec. 10-10). Inflow may be lagged by dividing the basin into zones by
iso hrones of travel time from the outlet. The area between isochrones
is then mea ured and a time-area diagram (Fig. 10-22) is plotted. This
diagram may be viewed as inflow to a hypothetical reservoir with storage
characteristics equivalent to tho'e of the basin and located at t he basin
outlet. Thus routing the time-area diagram by the Muskingum method
( ec. 10-8) with x = 0 yields the outflow hydrograph after adjustment
for units. Because of the method of constructing the time-area diagram.
such a hydrograph would be the result of an instantaneous rainfall
(duration = 0 hr), and it is called an instantaneous unit hydrograph. It
can be converted to a unit hydrograph for any duration t by averaging
ordinates t units of time apart and plotting the average at the end of the
period (Fig. 10-23).
The technique outlined above need not be limited to deriving unit
hydrographs. For a storm of duration equal to the interval between
STREAMFLOW ROUTING
238
isoohrones, the average runoff may be estimated for each time zone and
expr ed in cubic feet per econd. 1 The resulting time-runn('f diagram
is then routed through storage to give the actual outflow hydrograph.
If rain lasts for several time periods, the time-runoff diagrams are lagged
and superimposed (Fig. 10-24), and the summation is routed. The
method accounts for time-intensity variations and areal distribution of
i'ainfall, two factors which the unit hydrograph cannot readily consider.
For this r ason the routing approach can be applied to much larger
basins than can the simple unit-hydrograph approach.
InstanfalllOUS unIt
\
hydrograph
\~f-lIr
.....
........
unit hydrogropll
..........
........
-_ -__
Time
fiG. 10-23. Converting an instantaneous unit hydrograph to one of finite duration.
Horton 2 suggested that for a storm of reasonably uniform rainfall

intensity the virtual channel-inflow graph" would be approximately
triangular. He assumed that the shape of this triangle (Fig. 10-25) is
fixed by the following:
1. Inflow and outflow must begin at the same time.
2. The inflow hydrograph must pass through the cre t of the outflow
hydrograph (reservoir conditions).
3. Inflow ends at the point of contraflexure on the recession limb of the
outflow hydrograph.
4. Volumes of inflow and outflow must be equal.
Using these rul S, Horton demonstrated that the hydrograph could be
(I
1 q in oubic feet p r second = (24 X 26.9AQ)/t = 646AQ/t, where A is in square

miles, Q is in inches, and t is in hours.
I R. E. Horton, Virtual Cha.nnel-inflow Graphs, Trans. Am. Gwphys. Union, Vol. 22,
Part 3, pp. 811 - 820, 1941, and Flood-crest Reduction by Chann l-storage, Trans. A.m.
GOOphY8. Union, Vol. 22, Part 3, pp. 820- 835, 1941.
239
DERIVING BASIN OUTFLOW BY ROUTING
o
c
.,
"'0
o'"
'"
..J
Time
FIG. 10-24. Time-runoff diogram for a long storm.
Virtual chanMI inflow graph
o
.c
..
u
is
Time
FIG. 10-25. Virtual channel-inflow graph.
2AO
STREAMFLOW ROUTING
accurately reproduced by routing the triangular inftow hydrograph

through storage.
The difficulty with the use of the routing t chnique lies in determining
the routing constant K and the lag through the basin T L. Generally,
a trial-and-error approach bas b en used in which variou values are
tried until a combinati n that gives a good fit to historical floods i
found. Although laborious, this procedure is sati factory for gaged
streams but obviously unsuited to ungaged basins.
~
125
tOh~1
"E 120
Q;
...
115
ir
~ 110
E
~
105
..,
"" 100
.
v;'"
V
",..-J
/'-9h r
~8hr
tl
?hr
Approx. time
of trovel
6hr
'" 95
U
70
75
80
85
90
95
100
65
Crest stoge of Socromento River ot Bulle City, Col if.
FIG. 10-26. Simple gage relatian for the Sacramento River from Ord Ferry to Butte City,
California. (U.S. Wealher Bureau.)
An estimate of K may be obtained from data on the r cession of flow
for a basin. From Eqs. (7-3) and (10-23), assuming x = 0,
1
K= - -log. Kr
(10-31)
where Kr is the recession constant for the stream.

Clark! described the derivation of a unit hydrograph by routing the
time-area diagram of a basin as explained above. He suggested that
K (in hI') is given by
(10-32)
where L is the length of the main stream in miles, s is the mean channel
slope, and c varies from about 0.8 to 2.2. Linsley in a discussion of
Clark's paper suggested the formula
10. O. Olark, Storage a.nd the Unit Hydrograph, Trans. ASCE, Vol. 110, pp. 1419J4&8. 10.....
GAGE RELATIONS
241
K = bL
v'8
(10-33)
where A is the drainage area in square miles and b varies from about 0.04
to 0.08 for the ba ins tested.
It is often assumed that the lag T L = K, but there is vid nce that this
is not necessarily true. 1 One possible estimate of T L is the basin lag
f-6hr-j
/"_'~ Totol flow
....., , '
,_....
\
\
FIG. 10-27. Effect of timing of flood p eaks on reach outflow.
tp1 using any of the methods of Sec. 9-0 or 9-7. It should be noted that
basin lag can be observed quite readily by noting the time of peak and
comparing this with the time of rain, even though the basin is ungaged.
10-14. Gage relation. A discussion of wave travel and routing
would be incomplete without brief mention of a simple empirical solution
which is often quite uccessful. Gage relations are graphs correlating an
observed stage or di charge at an upstream tation with th resulting
stage or discharge at a downstream station. Gage r lation are roost
effective when dS/ dt = 0, i.e., at crest (Fig. 10-26). If a l' lation like
that in Fig. 10-26 is to be reliable, the quantity of local inflow between
the stations in each flood mu t bear a fixed relation to the reach inflow at
the upstream station.
ince such a proportional relation is unlikely,
gage relations are most effective when the local inflow is relatively small
compared with the main-stream inflow. It is also necessary that the
peak of the local inflow bear a fixed time relation to the peak of the mainstream inflow. If a slight difference in time of occurrence can cause a
considerable difference in the resulting outflow (Fig. 10-27), gage relations
will not be successful. ThuB gage relations are most useful on large
streams where local inflow is small with respect to main-stream flow and
rates of chang of flow are relatively low.
I
Research in progress at Stanford University.
STREAMFLOW ROUTING
242
More complex gage relations can be con tructed so as to account for

variable tributary inflow (Fig. 1O~28). It is al 0 po ibl to derive charts
for routing in terms of stage. I Stage relations and stage routing are
18o.----r----r----.----~--~----~--~
Note: Average lime of /ravel from Benl

Creek fa Scottsville is 8 flours
1401-----+-----+----+--~
u
g
2
1201-----i---+--
.5
II)
~100~----+---~~--~+-~~
o
.s:;
U
en
'i5
80~----~--~4-~--~~~~~
.!!
~
~
60~~~~~~--~~~--~-----i-----L----~
8.
en
Use dashed line only if

computed runoff from
40r---~r_~~~----+_~w~c~a~/Ta~re~a~'~s~n~0~f~a~va~i~w~b~~~--~
O~--~2~0--~40~--~
. 6~0--~8~0----,~00----,~20--~,40
Bent Creek peak discharge in 1000 cfs
FIG. 10-28. Gage relation for the James River from Bent Creek to Scottville, Virginia,
parameter for local inflow. (U.S. Weather Bureau.)
.
wit~
u eful when dealing with streams for which di charge data are not available. It should be emphasized that any change in the channel, either
natural or artificial, may result in changes in the stage-discharge and
stage-storage relationships for the reach. An analysis in terms of stage
may be invalidated by such changes. Stage routing and gage relations
are useful in the field of flood forecasting. Here speed is of paramount
1 M. A. Kohler, A For casting Technique for Routing and Combining Flow in
Terms of Stage, Tran8. Am. GeophYB. Union, Vol. 25, Part 6, pp. 1030-1035, 1944.
PROBLEMS
2~3
importance, and stage i the de ired answer. Thus a technique which

eliminates the conver ion from tage to discharg and ba k to stage j
potentially valuable if it is ufficiently r liable.
BmUOGRAPIIY
Gilcrest, B. R.: Flood Routing, Chap. ' ,pp. 635- 709, in Hunter Rouse (ed.), "Engineering Hydraulics," Wiley, ew York, 1950.
Keulegan, G. H.: Wave Motion, Chap. XI, pp. 711- 768, in Hunter Rousc (ed .),
"Engineering Hydraulics," Wiley, New York, 1950.
Linsley, R. X ., M. A. Kohler, and J. L. H. Paulhus: "Applied Hydrology, " haps. 18
and 19, pp. 465-541, McGraw-Hill, New York, 1949.
Thomas, H. A.: The Hydraulics of Flood Movement in Rivers, Carnegie Inst. Tcclmol.
Eng. BuU., 1935.
PROBLE~
10-1. If the channel width for the stream whose stage-dis harge r latioll is shown
in Fig. 4-12 is 30 ft at a stage of 4 ft, what would b the wave celerity for a translatory
monoclinal wave of small height?
10-2. Find the ratio between wave celerity and water velocity for a semicir ular
channel when y ~ r. When y = 0. 2r. Use the CMzy formula.
10-3. A uniform reotangular channel 10 It wide (n - 0.015) on a slope of 0.0004 is
flowing at a depth of 5 ft. A sudden gate opening increases the depth to 6 ft. What
is the celerity of the resulting abrupt wave?
10-4. Given the hydrographs tabulated below, find the storage in the reach, and
plot a curve showing storage at any instant as a function of simultanoous outJow.
Ignore local inflow.
Date
Hour
--I
2
3
4
5
6
7
8
9
10
11
Midnight
Noon
Midnight
Noon
Midnight
Noon
Midnight
Noon
Midnight
Noon
Midnight
Noon
Midnight
Noon
Midnight
Noon
Midnight
oon
Midnight
Noon
Midnight
Inflow,
ors
Outflow,
oCs
40
35
37
125
340
575
40
39
37
52
130
287
722
472
740
673
456
320
245
192
144
U8
95
80
67
56
50
624
676
63S
574
394
307
235
180
142
42
114
93
77
64
55
STREAMFLOW ROUTING
10- 5. A small reservoir has an area of 300 acres at spjl!way level, and the bauks are
essentially vertical for 8 veral feet above spillway level. The spillway is 15 ft long
and bas a coefficient of 3.75. Taking the inflow bydrograph of Prob. 10-4 as the
inflow to tbe reservoir, compute the maximum pool level and maximum discharge to
be expected if the reservoir is initially at the spillway level at midnigbt on the first.
10-6. Tabulated below are the elevation-storage and eJevation-djscharge data
for a small reservoir. Taklng th inflow hydrograph of Prob. ]0-4 as the reservoir
inflow and assuming th pool elevation to be 875 at midnight on the first, find the
maximum pool elevation and peak outflow rate.
Elevation
Storage, a re-ft
Discharge, cfs
62
865
870
875
880
882
884
886
888
0
40
200
500
1000
1220
1630
2270
3150
0
0
0
0
0
100
230
394
600
10-7. Find the Musklngum K and :tfor the flood of Problem 10-4.
10-8. Taking the outflow hydrograph of Prob. 10-4 as the inftow to a reach with
K - 27 hr and x - 0.2, find the peak outflow, using the Muskingum method of
routing.
] 0-9. Write the routing equation for the case when storage is n function of al + bOo
10-10. Using th graphico.l method of ec. 10-10, find the K curve for the flood of
Prob . 10-4.
] 0-11. Using tbo outflow hydrograph of Prob. 10-4 as the inflow to a reach for which
tb lag T (. - 6 hr and K ,. 18 hr, find th peak outflow and time of p(lak by the
graphical method of Sec. 10-10.
11
FREQUENCY AND DURATION STUDIES
One of the mo t common pha es of hydrologic design is related to the

frequency with which flows of a given magnitude will b equal d or
exceeded, or "What i the average lap 'ed time betw en the occurrenc
of two floods equaling or exceeding a p cifted discharge'? " Information
concerning probable extremes which proposed structures may be required
to withstand and many other hydrologic problem can be solved by
frequency analysis, using past record of flood peaks, flood volumes,
minimum flows, etc. The selection of a design frequency must rest on
economic analysis and policy decisions. 1
Flood-frequency analysis has long been a controversial topic among
engineers, and general agreement as to methods is not yet in sight.
Criticism has been brought about largely by abuse through misunderstanding. That d ign of hydraulic structures must rest on some form
of frequency analysis i unquestionable. Those responsible mu t,
through experience or otherwise, agree on basic criteria. Frequency
analysis should then provid a reasonable basis for transposing experience.
In a report of the National Research Council2 on deficiencies of
hydrologic data it is stated," . . . No less than 60 important dams have
failed during the past 50 years as a result of overtopping by floods . . . . "
Is it then to be concluded that failure of a structure always constitutes
underdesign ? Designing for the 10-yr flood, the 100-yr flood, or in fact
any flow below the maximum possible involves u. calculated risk, and
failures here are just as c rtain as death in the life insurance business.
Efficiency of projects, such as run-of-river power plants and municipal
water supplies without appr ciabl storage, often depend on the time
distribution of streamflow; i.e., the per c nt of time flow can be expected
to exceed specified quantities. If project are to mbody appr ciable
IE. L. Grant, "Principles of Engineering Economy," 3d ed., Ronald, New York,
1950.
'''Deficiencies in Basic Hydrologic Data," p. 3, Report of the Special Advisory

Committee on Standards and Specifications for Hydrologic Data., Water Resonrces
Committee, Na.tional Resources Committee, September, 1936.
245
246
storage, the effects they would have had on pa t streamflow must be

considered in evaluating available flow. In designing storage facilities,
the engineer must determine the quantity of sto!'age required to a ure
adequate supply in low-water period. The analysi upon which such
decisions are based is discussed in the last sections of this chapter
F RE Q ENCY AN L Y I
This section is limited to a discussion of station, or point, analysis of

flood flows and rainfall inten ities. It will be noted that most of the
material of a sp cifi nature is for peak discharge, but rainfall intensities
could as well be us d in most cases. Although th importance of rainfall-frequency analysis in hydrologic de ign is well recognized, the determination of rainfall frequencies is, hydrologically speaking, "the means
to an end," the end product being the estimation of flood frequencies.
Flood frequencies in terms of stage are often required in connection
with estimating frequency of inundation of buildings and other valuable
property located in the flood plain. For reasons explained below, however, it is recommended that analysis be made in terms of discharge and
the results converted to stage by use of an applicable rating
11-1. Sel ection and compilation of d ata. Since theoretical
aspects of frequency analysis require that all data for the period of study
be comparable (all representing true, unbiased observations of the desired
element) it is extremely important that the basic data be thoroughly
crutinized. Some of the more obvious considerations are a follow :
1. Pronounced shifts in the stage-discharge relation render stage data
inconsistent for frequency studies. Although adjustment can sometimes
be made, the use of discharge data in the analysis is to be preferred. If
stage frequencies are required, the results can b transformed to stage
using the mo t recent rating.
2. Changes in the datum of a river gage directly affect the recorded
values, and, if analysis is to b made in terms of stage, adjustments must
be made.
3. Moving a river gage and/ or discharge section even a short di tance
may have a significant effect on recorded stage and/ or discharge values.
Such changes are usually fully described in data publications so that
required adjustments can be made.
4. onstruction of dam, levees, etc., can cause material changes in
the streamflow r gime. The effect of all uch constl'Uction should be
inv stigated and adjustments made as r quired.
5. Historical data, event OCCUlTing prior to the beginning of continuous
record, should be includ d in the analy is if the evidence i sufficient to
assign magnitud , order, and period. l Even though only 30 yr of com1
Hydrology Handbook, ASCE Manual 28, pp. ]02- 104, 1949.
SELECTION AND COMPILATION OF DATA
241
plete records are available, 192 1955 for example, it may b an tablished fact that the 2500-cs flood of 1 91 was the great t in e 1 50 and
has not since been exceeded. Under thes circumstances, all r quir d
information (m = 1, n = 106, and X = 2500) is known, and the flo d
can be plotted in accordance with Eq. (11-1), page 249.
The data may be as embl d in everal different way, th s lection
usually being based on personal prefer nee, availability of data, and
purpose of the study. If extreme floods are of primary concern, it is
customary to use only the annual floods, i.e., the maximum flood peak
of each calendar or water year.
uch a series ignor the s cond- and
lower-order events of each year which may be ev n greater than annual
floods of other years. This objection is met by an alternative approach
using all floods above a selected base, without regard to time of occurrence.
Such an array, known as a partial duration series (or just partial series),
is not a Lrue distribution eries since the flood event is defin d in terms of
its magnitude rather than of its occurrence. An objection is frequently
raised to the partial series on the grounds that any decision as to the
relative independence of cons cutive events, such as flood peaks, must
be arbitrary. If two con cutive events are judged to be independent,
both are listed in the series; otherwi e, only the higher is considered.
The two types of serie provide quite different distributions, particularly for the shorter return periods. 1 It can be d monstrated by theoretical analysis that the two distributions are !"elated,2 as shown in
Table 11-1. For return periods exceeding 10 yr, ther i. negligible
TABLE 11 -1. Corresponding Return Periods (Years) for Annual and
Partial Series
Partial
series
Annual
leries
0.5
1.0
1.45
2.0
1. 16
1.58
2.00
2. 54
Partial
series
Annual
series
5.0
10
5.52
10.5
50.5
100 .5
SO
100
difference between th two series. If interest is centered 011 the more

frequent events, it is perhaps advisable to base the analysis on partialduration data, judging the independence of consecutive events in the
light of the problem at hand.
1 Return period and recurrence interval are used inter hangcably in this text to s.ignify
the average number of years within which a given event will be equaled or exceeded.
The exceedance interval, defined as the average number of years between the occurrence of the event and a greater event, has received some r ognition but its usc has
been avoided in the text to minimize possible confusion.
S W. B. Langbein, Annual Floods and the Partial Duration Flood Series, Tran8.
Am. GeophY8. Union., Vol. 30, pp. 879-881, December, 1949.
FREQUENCY AND DUR.ATION STUDIES
248
A third method of summarizing data, the full series, involves the use
of all values of record, such as mean daily flows. The full series does not
provide independent events and hence is not suitable for flood-frequellcy
studies in the usual sense. Frequency relations derived from the full
series, termed duration curves, are di cussed later in the chapter.
11-2. Plotting positions and the ignificance of the N -yr event.
Frequency analysis delineates the event which can be expected to occur,
on an average, once every N yr, the N-yr event. There is no implication
that such events will occur at even reasonably constant intervals of
N yr. Rather, there is a 1 per cent chance that the 100-yr flood will
occur within any 365-day period. The theoretical distribution of the
return period is illustrated by Table 11-2. It will be seen that, over a
TABLE 11-2. Theoretical Distribution of the Return Period
Aelual return period T. exceeded various percentage. of the time
Average return
period,
T.
2
5
10
30
100
1,000
10,000
1%
5%
25 %
50%
75 %
95 %
99%
8
22
45
137
459
4,620
46,200
5
14
28
89
300
3,000
30,000
3
7
14
42
139
1,400
14,000
1
3
7
21
69
693
6932
0
1
3
8
29
288
2880
0
0
0
2
5
51
513
0
0
0
0
1
10
100
long period of years, 25 per cent of the intervals between floods equal
to or greater than the lOO-yr flood will be Ie s than 29 yr while an equal
number will be in exce s of 139 yr. In other word, for 75 per cent
assurance that the capacity of a structure will not be exceeded by a flood
within the next 29 yr, it must be designed for the 100-yr (average return
period) flood.
Mter either the partial or annual series is compiled, the items are
customarily arranged in descending order of magnitude and assigned an
order number m. There is no general agreement as to the proper return
period to be assigned to any item of the series. l The California method
a signs a. return period of n to the maximum observed flood in n yr of
I Flow in California Streams, Calif. Dept. Public Works Bull. 5, 1923.
A. Hazen, "Flood Flow," Wiley, New York, 1930.
C. S. Jarvis and others, Floods in the United tates, U.S. Geol. Survey Water-8upply
Paper 771, 1936.
R J. Gumbel, On the Plotting of Flood Discharges, Trans . Am. Geophys . U"Iion,
Vol. 24, Part 2, pp. 699-719, 1943.
2.49
PLOTTING POSITIONS AND THE SIGNIFICANCE OF THE N-YR EVENT
record, whereas Hazen a.ssigned a. return period of 2n . The plotting

formula now commonly u ed is
T r =n +1
m
(11-1)
where Tr is the return period, or recurrence interval, in year ; n, th

number of years of record; and m, the rank of the event (m = 1 for maximum and m = n for minimum eventl) . This procedure is applicable to
either the annual or partial series and is in conformance with modern
theories.
Table 11-3 is included to illustrate the reliability of return period
derived from relatively short periods of record. For example, the table
TABLE 11-3. Average Return Periods for Various Levels of Probability
Rank from
top, m
Probability
Number of years
of record,
n
0.01
0.25
0.50
0.75
0.99
1.11
1.66
2.71
4.86
13.5
1.06
1.42
2.13
3.61
9.62
1.05
1.31
1.86
3.03
7.76
1.03
1.25
1.70
2.67
6.63
2.00
4.13
7.73
14 . 9
43.8
1.48
2. 57
4.41
8.12
23.0
1.32
2.06
3.32
5.86
16.1
1.24
1.80
2.77
4.72
12.5
3.41
7.73
14 . 9
29.4
87 . 0
2.00
3.78
6.76
12.7
36.6
1.63
2.75
4.62
8.35
23.3
1.46
2.27
3.63
6.36
17.2
7.46
17 . 9
35.3
70.0
209.0
3.06
6. 20
11.4
21.8
63.4
2. 19
3.95
6. 86
12.6
35 . 8
1. 83
3.04
5.02
8. 98
24.8
200
498
996
1990
5970
17.0
37.4
71.1
138
408
7.10
14.1
25 . 6
48.6
140
4.50
8. 26
14.4
26.6
75.2
2
5
10
20
60
3
6
11
21
"
61
4
7
12
22
62
5
8
13
23
63
shows that there is a 1 per cent chance that the true average return period
of the maximum event occurring in a lO-yr record is as low as 2.71 yr
1 In the partial duration series there may be either more or fewer items than th
number of years of record, depending on the base value selected. Except when return
periods of less than one year are of interest, it is customary to limit the partial series
to the largest n events.
250
FREQUENCY AND DURATION S1UDIES
and that there is almost a 1 per cent chance that the 100O-yr flood will
occur in a particular 10-yr period of ob ervation .
11-3. Derivation of the frequency di tribution function. The
accumulative frequ ncy relation for a partial or annual series can be
graphically developed by plotting computed return periods vs. the
magnitude of the respective events (Xl, X 2 , , X .. ) and fitting a.
smooth curve. This procedure (Fig. 11-1) is quite acceptable if one is
inter st>d in only the shorter return periods, say Tr ~ n/ 5. Fitting a
curve by eye to the higher events places unjustified reliance on computed
plotting positions which are of doubtful accuracy. The desire to inc ..ease
10
~
u
80
19ft - 1948
n =. 381(5
si
.!: 6
b:
oj
e-o
.c
u
'"
i5
40
Return period,
r;,
in yeors
FIG. 11-1. Annual floods af the Clearwater River al Kamiah, Idaho.
the r liability of stimated frequency values has spurred many investigators! to dev lop methods which use the el1tir data eries as a basis for
fitting a curve. This requires the assumption of a theoretical frequency
distribution. A lengthy di cussion of the subject is well beyond the
scope of thi text. Only one of the more widely accepted approacbes is
presented here.
Gumbel 2 following the work of Fisher and Tippett,3 make use of the
distribution of extreme values. If Xl, X 2 , , X .. are the extreme
IE. J. GumboI, On the Plotting of Flood Discharges, Trans. Am. Geophys. Union,
Vol. 24, Part 2, pp. 669-719, 1943.
H. A. Foster, Theoretical Frequency Curves, Tran8. ASCE, Vol. 87, pp. 142- 173,
1924.
J. J. lad, An Asymmetric Probability Function, Trans . ASCE, Vol. 101, pp. 35104, 1936.
2 E. J. Gumbel, Statistical Theory of Extreme Va.1ues and Some Practical Applications, Nail. Bt,r. Standards (U.S.) Appl. Math. SeT. 33, February, 1954.
I R. A. Fisher and L. H. C. Tippett, Limiting Forms of the Frequency Distribution
of the Largest or Smallest Member of a Sample, Proc. Cambridge Phil. Soc., Vol. 2-1,
pp. 180-190, 1928.
DERIVATION OF THE FREQUENCY DISTRIBUTION FUNCTION
251
values observed in n sample of equal size N and if X is an unJimit d,

exponentially distributed variable, then the theory of extrem valu s
states that, as nand N approach infinity, the cumulativ probability P
that any of the n extremes will be less than X approache the expr ion
(Appendix B)
(11-2)
where e is the base of Napierian logarithm and y, term d the reduced

variate, is given by
y = a(X - X,)
(11-3)
For an infinitely large ample, it can be shown by the theory of extreme

values that the mode of the distribution X, and the dispersion param ter
a are functions of the arithmetic mean X and the standard dev iation U~,
XI = X - 0.45005u.
(11-4)
and
a=
1.28255
(11-5)
Equation (11-2) is an expression of probability of nonoccurrence.

return period can be computed from
Tr
1
1 _ P
The
(11-6)
In practice, the engineer is required to estimate return periods from

limited samples and, therefore, Eqs. (11-4) and (11-5) are not strictly
applicable. There are several possible approaches! for det rmining
values of the parameters a and XI from the annual series. That advocated by Gumbel is based on least-squares analysis of Eq. (11-3). This
equation can be represented by a straight line (X vs. y) on cartesian
coordinates (Fig. 11-5), and Gumbel's solution minimizes the squares
of the deviations measured perpendicular to the derived line of exp cted
extremes. The resulting equations are
X-
XI =
u.
fi,.
(J 1-7)
(1,.
(1,.
a = -
(11- )
(1 z
The theoretical quantities fi" and

size (Table 11-4) ,
1
(1,.
arc functions only of the sample
J. Lieblein, A New Method of Analyzing Extreme-value Data, Nail. At:toUorll

Aeronaut. Tech. Note 3053, January, 1954.
COTIITII.
252
TABLE 11 -4. Expected Means and Standard Deviations of Reduced Extremes

(After Gumbel)
y.
n
r.
-- --- - - - - - - - - --n
20
30
40
50
60
70
IT .
fT.
0.52
0.54
0.54
0.55
0.55
0.55
1.06
1.11
1. 14
1.16
1.17
1.19
80
90
100
150
200
00
0.56
0. 56
0.56
0. 56
0. 57
0.57
1. 19
1.20
1.21
1.23
1.24
1.28
ombining Eqs. (11-3), (11-7), and (11-8), we have
x = X+
IT,.
(y - fin)
(11-9)
IT"
howl has shown that most frequency functions applicable for hydrologic analy is can be resolved to the generalized form
(11-10)
wh re the "frequency factor" K takes various forms depending upon

which approach one uses. From Eq. (11-9), it will be seen that for the
Gumbel method
(11-11 )
and can be expressed by a fun ction of Tr and n. Figure 11-2 is based

on the umbel method and permits solving 2 for X-X. If (fz falls
outside the rang shown on the chart, scale values of both IT z and X - X
can be multiplied by a conv nient factor.
umulative frequency plottings on a carte ian grid di play pronounced
curvature, as may be s en from Fig. 11-1. Any curve can be adjusted
to a straight line by appropriate warping of either the X or Tr scale.
This is the procedur us d in designing probability paper. The choice
of a distribution function thus dictate the de ign of the probability pap r
to be used. Normal probability pap r, introduced by Haz n, S distorts
1 V. T.
how, A G neral Formula for Hydrologic Frequency Analysis, Tran8. Am.
Geophya. Union, Vol. 32, pp. 231- 237, April, 1951.
a L. L. Weiss, A Nomogram for Log-normal Frequenoy Analysis, Trans. Am.
Geophys. Union, Vol. 38, pp. 33- 37, Februa.ry, 1957.
I A. Hazen, Storage to Be Provided in Impounding Reservoirs for Municipal Water
Supply, Tran8. ASCE, Vol. 77, pp. 1539-1669, 1914.
253
100
II II J
90
80
70
II 1/1/
60
II
V~V~
50
dV~/
40
~';
I'~''j
30
C 20
....
.J. ~~;J, tI-
11i
I I II
~~'!tIP
.S
10
9
Ci)
a:: 8
7
0..
....
:::J
I.
1'/'1
~V '{I/I//I
~r; HIJI. ~/V
5
4
/V
/f
j, fj
fo- f0-
2
1.5 1--
r-
.,
-,/
L~
~,t
;~~ ~;2
t~
rl//J
'1jf/
WI,
[/VV
i/VV
/VV
'//V
...
:,~~
~~
I~~
i2~
r-p~\;'::< I~~O i:rr-p
j?J.
?'
pll"/
./
./
.O.efJ Ir- 2
".r-~
'Z'/VVI/ v , /
4O ~ r".,/VVv V
n.
....
r-~
~Z
v./
0.30 I-r~
/' V V _....
V
/':/ / /
_ 1.20 I- ~
~ C/
v
>r-i_f~
..... f".10
H'"
~ 1-..... f-r-
--
a~
f'/J
j,
t~
II
V'-o.,f:!
IIII
~; 5 ~
I
V~()'
I
/
~; o
II
(
I~ 1
II
~~
JJ
1 I
/
~~
'1./ I 'II. J / /
/
'III
I A
~2
/
'I l
~'/ 1'1 '7 I/VI/
It'
..l../
V
'1'1
~:
~V
"0
.g
....
v ~
~~ 8
I /-Z r{~
~VIA V
>.
/
j
II
X.~ ~~
1I1I1 v
'jv,V V /
<II
I
I
:3
Frequency factor, K
FIG. 112. Coaxial cha rt for the Gumbel method of fltting annual series of extreme values.
(Aft.r Weiss.)
TABLE 115. Frequency Analysis for the Clearwater River at Kamiah, Idaho
(Drainage area = .. 850 .q mil
(After U.S. Geological s...rvey)
Annual ftood.
Vear
Month
Day
Discharg.,
cf.
Order,
Return
period,
yr
Partial duration
series
Order,
Return
period,
yr
- - - --1911
May
June
1912
May
1913
June
April
May
191 ..
1915
1916
1917
May
June
May
April
May
June
May
June
1918
Oec.
May
1922
June
April .
May
May
June
April
May
May
1923
June
May
192..
June
May
1919
1920
1921
6
17
4
13
21
20
21
20
27
11
26
18
23
3
19
28
7
5, 9
19
29
15
30
9
17
29, 30
5
15
10
29
23
18
16
23
20
19
26
6
8- 10
26
12
13
3",600
29,"00
35,900
39,500
55,200
61,900
38,000
29,400
30,700
"5,800
76,600
"2,200
"1,500
30,700
28,200
30,000
"4,"00
36,600
56,000
36,600
63,600
69,700
56,800
70,500
37,300
52,800
35,200
52,800
30,700
52,000
"3,600
"2.900
35,200
69,700
60,600
52,100
62,400
38,800
"9,600
"3,200
"5,600
58,900
...
...
.. .
.. .
.. .
.....
.....
.....
1.35
.. .. .
3. 00
... ..
.....
.....
.....
3
27
13.0
1."4
...
...
29
...
13
...
...
38
...
...
...
16
...
...
...
...
6
...
17
...
.....
.. ...
1.03
.....
.....
.....
2 . ....
... ..
.....
.....
.....
6.50
.. ...
2.29
.... .
.....
...
...
.....
20
26
1.95
1.50
...
...
...
...
12
...
21
...
...
15
.....
.....
".88
.....
.....
3.25
.....
1.86
.....
.. ...
2.60
83
110
75
58
27
20
62
111
101
40
4
53
55
102
120
104
.. 3
69
26
70
15
10
25
8
65
28
80
29
103
33
..7
52
81
11
22
32
18
60
36
.. 9
..1
2..
0."7
0. 35
0. 52
0. 67
1. ....
1.95
0.63
0.35
0.39
0.98
9.75
0.7"
0.71
0 . 38
0. 32
0.37
0. 91
0.57
1.50
0.56
2.60
3.90
1.56
4.88
0 .60
1.39
0.49
1.35
0.38
1. 18
0.83
0.75
0.48
3.55
1.77
1.22
2.17
0.65
1.08
0.80
0.95
1.62
255
TABLE 11-5. Frequency Analysis for the Clearwater River at

Kamiah, Idaho (Continued)
Annual flood.
Year
Month
Day
Discharge,
ef.
Order,
m
--1925
1926
1927
April
May
April
May
April
May
June
Nay.
1928
May
1929
May
June
1930
1931
April
May
1932
April
May
1933
June
April
June
1934
Dec.
March
April
1935
1936
1937
1938
May
May
June
April
May
June
May
April
May
17
7
20
19
1
21
28
17
8
5
26
9
26
24
1
9
25
7
14, 16
14
14
21
13- 15
27
4
10
23
30
14
25
8
24
31
6
19
5
15
28
1
19
28
19
I
17
28
Partial duration
. e ri.1
41 ,800
44,800
59,800
35,900
35,900
32,400
46,400
64,200
68,600
43,900
29,200
65,700
72,100
52,700
28,500
35,800
31 ,000
40,800
36,500
28,500
72,100
62,200
35,100
35,800
71,200
81,400
43,600
32,300
37,800
45,900
34,300
44,000
34,400
29,900
50,600
49,800
63,200
34,300
32,900
34,300
32,200
63,400
39,400
31,500
60,800
Return
period ,
yr
Order,
m
...
...
.....
.....
54
14
33
2.79
1.18
42
23
76
...
...
...
.. .
...
.. .. .
... ..
... ..
.. ...
4 . 33
.....
.....
.....
4
18
9 . 75
2.17
...
...
...
.. .
.. ...
.....
36
28
1.08
1.39
...
... ..
.....
.. .
...
...
... ..
.... .
.... .
... ..
...
5
...
2
...
.. .
.. .
23
.. .
25
...
...
...
.. ,
11
...
...
34
'"
10
.. ,
...
...
7 . 80
19 . 5
.....
.. .. .
.. ...
1.70
... ..
1.56
... ..
... . .
.....
.. ...
3.54
.....
.... .
1.15
.....
3.90
.....
.. ...
... ..
95
37
14
12
46
113
13
5
30
117
78
100
56
72
118
6
19
82
79
7
3
48
96
63
39
85
45
84
106
34
35
17
86
93
87
97
16
59
98
21
Return
period ,
yr
0 .72
0.93
1.70
0.51
0.51
0.41
1.05
2 . 79
3 . 25
0 . 85
0 . 35
3.00
7.80
1.30
0.33
0 . 50
0.39
0.70
0 . 54
0 . 33
6.50
2 . 05
0 . 48
0 . 49
5 . 57
13.0
0.81
0.41
0.62
1.00
0.46
0 . 87
0.46
0 . 37
1. 15
1.11
2.29
0 . 45
0 . 42
0.45
0.40
2.44
0.66
0.40
1.86
256
TABLE 11-5. Frequency Analysis for the Clearwater River at

Kamiah, Idaho (Continued)
Annual floods
Yea r
Month
Day
Discharge,
eft
Order,
m
Return
period,
yr
Partial duration
series
Order,
m
Return
period,
yr
38
73
66
107
11-1
116
115
67
51
108
31
68
50
57
88
44
61
92
91
71
106
74
119
89
9
64
99
112
94
90
2
1
109
1.03
0.53
0.59
0.36
0.34
0.34
0.34
0.58
0.76
0.36
1.26
0.57
0.78
0.68
0 . 44
0 . 89
0.64
0 . 42
0.43
0.55
0 . 37
0.53
0.33
0.44
4.33
0 . 61
0.39
0 . 35
0.42
0 . 43
19.5
39.0
0 . 36
--- --1939
May
1940
May
1941
1942
May
April
1943
May
April
May
June
1944
1945
May
May
1946
April
May
1947
June
Dec.
May
1948
June
April
May
June
Period: 1911 - 1948
4
17
12
25
13
14
21
26
20
1
29
11
19
22
16
6
31
20
26
6
19
28
4
15
8
27
9
18
22
8
22
29
22
53,020
16,470
46,400
36,400
37,100
29,600
28,900
28,900
28,900
37,100
43,200
29,600
52,200
37,100
43,200
40,100
34,200
44,400
38,400
33,300
33,700
36,600
30,000
36,100
28,300
33,900
69,900
37,600
31,200
29,400
32,600
33,800
86,500
99,000
29,600
=
=
22
.. .
30
...
37
...
...
31
...
...
1.77
... ..
1.30
.. ...
1.05
.....
.....
1.26
.....
...
... ..
2.05
.....
.. .. .
.... .
35
24
1.11
1.63
19
...
...
.. ,
...
...
32
...
...
...
.. ,
...
...
...
...
...
...
1
...
.....
... ..
.... .
1.22
.. .. .
. ....
. ....
... ..
5.57
... ..
.... .
.....
.... .
.. ...
... ..
39.0
.....
mean of annual floods

standard deviation of annual floods
--
257
the frequency scale 0 that a normal di tribution plot as a straight line.!

Whipple 2 introduced log-normal paper for fitting tho di tributions in
which the logarithms of the variate are normally distributed. Examination of Eq. (11-3) will show that a plot of X vs. th r duced variat y is
linear on a carte 'ian grid, and this equation form the basi of Gwnbel
probability paper (Fig. 11-5) introduced by Pow n. a The nature of
the distribution is su h that the arithm tic averag of the annual series
has a return period which approaches 2.33 yr for large values of n, and
lin fitted by eye on t he probability paper are usually COllstru -ted so as
to pass through thi point .
..
11
.11
1.0
.E 0.9
'5 0 .8
0 .7
E 0.6
j o.~
i
0.4
iJ
2
2.~
7 8 9 10
Return period.
rr ' In years
15
20
2~
30
40
50 60
FIG. 11 -3. Clock-hour rainfall s in excess of 0.32 in. at Salt lake City, Utah.
It shou ld be emphasized that th

umbel distribution does llot provide
a satisfactory fit of partial-duration flood or rainfall data, as may be
surmised from Table 11-1. The partial rainfall series (Fig. 11-3) usually
approximates a straight line on log-log paper while the partial flood series
(Fig. 11-4) generally approximates a straight line on semi logarithmic
paper (discharge on arithmetic axis).
The learwater River at Kamiah, Idaho, has been selected for illustrative purpo es. Both annual- and partial-series data were compiled
(Table 11-5) from published records of the U. . Geological Survey.
Return periods shown in the table were computed from Eq. (11-1).
1 The mode, median, and arithmetic mean are coincident in th
symmetrical, normal distril ution and plot at a return period corresponding to 50 per cent probability.
Mor over, the line of best fit passes through the points 1 - P = 0.500 0.341 and
X = X iT
2 G. C. Whipple, The E lement of Chance in
anitation, J. Franklin lnst., Vol. 182,
pp. 37- 59, 205-227, 1916.
R. W. Powell, A Simple Method of Estimating F lood Frequency, Civil Eng.,
Vol. 13, pp. 105-107 (Discussions by W. E. Howland and E . .J. Gumbel, pp. 185 and
438, respectively), 1943.
Annual extremes are readily obtainable from the Water-supply Papers which are
issued ann ually for the several sections of the country, but the lesBer flood peaks were
not incl~ded until recent years. Some summarized data do app ar in special papers
as, for example, U.S. Geel. Survey Water-supply Papers 771 (1936) and 1080 (1949).
258
The standard deviation of the annual s ries (adj usted for sample size) is
IT ~
~(X - X)2 =
n-1
!'1:.X2 - X'1:.X
'\J
(11-12)
n-1
The partial-seri s data are plotted in Fig. 11-4, and the curve shown was
fittcd by eye. The curve shown for the annual data (hg. 11-5) was
Jeicrmined fr m Fig. 11-2.
II
r--r
10
!! 9
u
0
0
Q
.5
o_
e;;
oi
;;'"
19ff-1948
n = 38 yrs
.J:
"'
i5
/.~
4
3
~.2
//
0.3 0.4 0.5
2
3
4 5
Relurn period. T,. in years
10
20
30 40 50
FIG. 11 -4. Floods above 28,000 cfs an Cl earwater River at Kamiah, Idaho.
11-4. Selection of de 'ign frequency. The return period established by fr quency analysis as ju t do cribed indicates only the average
interval between events equal to or greater than a given magnitude, or
the probability that such an event will occur in anyone year. Tables
11-2 and 11-3 both indicate that there is a distinct po ibility that actual
return periods may be substantially less than the averag for a given
vent. Thus if it i' desir d to sel ct a design flow which i not likely to
occur during the life of the str ucture, it is ue e ary to u e a return
period greater than the stimat d u ful life. If the probability of
nonoccurrence of an event in any year is P, the probability J of the
event occurring in any n-yr period i
J=1-P"
(11-13)
which is the basis of Tables 11-2 and 11-3.

Table 11-6, also computed from Eq. (11-13), shows, for example, that
there is a 10 per cent chance that a flood with an average return period of
259
SELECTION OF DESIGN FREQUENCY

Reduced variate. y
-1
10
2
I
8
~
u
. v /.
0
0
0
sf
..
. ./
.S:
.c
u
on
is
19"- 1948
n=38yrs
./
/.
:/ .
1/
2
1.01
./
1.1
1.5
3
Return period,
r"
20
10
30
50
in years
FIG. 11-5. Flood -frequency curve for the Clearwater River at Kamiah, Idaho.
460 yr will occur within the next 50 yr. Note that it is assumed that the
true value of P is known. If P is only approximate, then an even longer
return period is necessary for a given risk.
TABLE 11 -6. Return Periods Requir ed for Specified Risk of Occurrence
within Project Life
expected life of project, yr
Per",luible risk
of failure
0.01
0. 10
0.25
0.50
0.75
0.99
10
25
100
10
4
2
1.3
1.01
910
95
35
15
8
2.7
2440
238
87
37
18
6
50
100
5260
460
175
9100
940
345
145
72
22
72
37
11
260
GENERALlZATJO
OF FREQUENCY DATA
There has always been considerable demand for generalized frequency

data, in the form of charts or maps, for estimates at locations having no
record. This expressed ne d has led to numerous studies covering a
variety of approaches as evidenced in the references cited in the following
s ctions. In addition to providing estimate of frequency data generalization can actually enhance the stability and reliability of the results.
FIG. 11-6. Fifteen-minute rainfalls to be expected, an an average, once in 10, 25, 50, and
100 yr. (After Yarnell.)
Through utilization of all available data, an areal analysis can, in effect,

increase the size of the sample beyond the length of record at a single
station. No attempt is mad to present a detailed de cription of the
various techniques for gen ralization. Most approaches fall into one
of four categories, with limited overlap, and each category is discussed
briefly.
11-5. Areal interpolation of point-rainfall frequencies. Given
sufficiently long record for a den e network of precipitation stations,
the pI' paration of isopluvial maps that show the maximum point rainfall
for selected durations and retwn period i a straightforward process.
Methods described in the previous section are applicable, and one need
only tabulate the required series for sel cted duration, fit the frequency
functions by an appropriate method, plot maps of rainfall for selected
return periods and durations, and construct the isolines. This is essen-
AREAL INTERPOLATION OF POINT-RAINFAll FREQUENCIES
261
tially the procedure followed by YarnelP in d , 1 ping a t of map of

the type shown in Fig. 11-6 and 11-7. Yarn 11' analy i , c nducted
in the early 1930' covered ouJy th shorter durati llS and was, of 1100essity, limited to the data available from the first-order" ather Bureau
stations equipped with reoording gage (about 200 stations with average
record lengths of about 30 yr).
FIG. 11-7. One-hour roinfalls to be expected, on an average, once in 10, 25, 50, and 100
yr. (After Yarnell. )
A U.S. Weather Bureau l'eport 2 present maps of the 2-yr, J-hr rainfall
and relation hips for convcrting these values to other return p riods and
durations. A detailed analysis of rainfall-frequency data from about
200 U. . Weather BW'cau offices is also available. 3 The inadeq uate
density of recording gages in regions of rugged terrain has led to studies
of the relation between rainfall frequencies and physiographio featur S.4
In an attempt to overcome deficient record length, the Miami OllI D. L. Yarnell, Rainfall Intensity-Frequency Data, U.S. Dept. Agr. Misc. Publ.
204,1935.
2 Rainfall Intensities for Local Drainage Design in Western United States, U.S.
Wealher Bur. 1'ech. Paper 28, ] 956.
3 Rainfall Intensity-Duration-Frequency Curves, U.S. Weather Bur. Tech. Paper 25,
1955.
4 R . K. Linsley, Relations between Rainfall Intensity and Topography in Northern
California, Stanford Univ. Dept. Civil Eng. Research Rept. I, H)55.
262
ervaMY District 1 used what has been termed the II station-year"

method% to derive maps of maximum exp cted 1- to 6-day point rainfalls
for return period of 15 to 100 yr (examples shown in Figs. 11-8 and 11-9).
The method assumes tbat records from everal stations in a limited area
can be combined and treated as a single eries with length equivalent to
tbe sum of the individual record lengths. For rigorous application,
the selected stations must all lie within an area of meteorological homogeneity and yet must be far enough apart that no two gages provide
FIG. 11-8. One-day rainfalls 10 be el(pected, on an average, once in 15 yr.

(Miami Conservancy District.)
FIG. 11-9. One-day rainfalls 10 be el(pecled, on an average, once in 50 yr.

(Miami Conservancy District.)
values for the same storm in the derived series. In practice, these two
criteria tend to be mutually exclusive---one must be tempered at the
expense of the othe and the homogeneity qualification renders the
metbod wholly inadequate in mountainous areas.
The reliability of station-year analysis is u ually limited by the interdependence of the lected stations. The maximum rainfall observed
by any of 10 grouped stations in 10 yr of reoord is assigned a return period
of 101 yr and the tenth highest amount, a return period of 10.1 yr. If
the 10 stations are so close as to measure essentially identical amounts
storm by storm, the 10 highest amounts would be equal and this value
should represent a return period of 11 yr. Under these circumstances
the station-year method tends to underestimate the expected rainfall.
1 Storm Rainfall in the Eastern United States, rev. ed., pp. 43- 93, 7'ech. Rept. 5,
Miami Conservan y District, Dayton, hio, 1936.
I K. C. Hafstad, Reliability of Station-year Rainfall Frequency Determinations,
Tran8. ASCE, Vol. 107, pp. 633- 683, 1912.
AREAL INTERPOLATION OF FLOW FREQUENCIES
263
Practi ally, high-intensity, hort-duration rainfall is u ually the r sult

of int ns, mall-area thunder torm , and d pend nc betw n tation
30 or more mile apart is not great. Even if ondition of homogoll ity
and independence could be sati fi d, many critics till qu tion the
assumption th8,t simultaneous ob ervation at numerous points can be
con 'idered to repre ent the ob ervation at a ingle tation over a long
period of time. The basis of this criticism becomes evid nt if one considers an observation period of 1 or 2 yr which may happen to be either
abnormally wet or dry.
The station-year method is not to be confused with averaging techniques
which are restricted only by meteorological homogeneity. Whether or
not selected stations are independent, averaging the precipitation values
(m = 1, m = 2, etc., re p tively) for several stations with equival nt
record length provides a rainfall amount which can be assigned a return
period corresp.onding to m and the record length. In other word , th
resulting averages may be treated a~ a series derived from a single tation.
This procedure tends to increase the reliability over single-station analysis
but does not provide data on long return periods, the attractive advantage
claimed for the station-year method. Averaging data for several stations
improves area sampling and provides better definition of the frequency
curve within the time range of the data. It can improve the estimated
values of rare occurrence only to the extent that an assumed frequ ncy
distribution is better fitted.
tudies directed
11-6. Areal interpolation of flow frequencies.
toward generalization of flow frequencies usually follow a two-step
procedure. All flow data over a reasonably homogeneous area are first
combined on a nondimensional, comparable basis to develop a regional
frequency r lation which can be used to estimate the return period of
any flow for those points for which discharge records are available.
Application of the regional frequency curve to un gaged points requires
the development of a relation involving physiographic factors.
In the procedure DOW employed by the U.S. Geological Survey,!
each flood at each station is divided by the mean of the annual floods
for the station. The same ba e period is used in computing mans and
dimensionless flood ratios for all stations. The base period is usually
made as long as possible by estimating missing; values from floods at
nearby stations. This procedure assures comparability of the annual
means and permits assigning order numbers to all observed peaks.
The estimated peaks are not used directly in deriving the regional
1 Floods in Youghiogheny and Kiskiminetas River Basins, Pennsylvania. and Maryland, U.S . Geol. Survey ire. 204, 1952.
Tate Dalrymple, Regional Flood Frequency, Highway Re8earch Board Research
Rept. 11.B, pp. 4-20, 1950.
264
frequ ncy curve, since thi would only give added weight to the station
used in making the e timate.
For illustration, assume that one has assembled the annual-flood series
for each of a group of stations and has estimat d any mi' ing peaks
for the select d standard period and that test have shown the records
to meet homogeneity requirements (see Dalrymple). Analysi would
then proceed as follows:
1. Assign order numbers to floods for each station and compute the
return periods from Eq. (11-1).
2.80
I
I
2.00
,C
.0
c
c
I
I
I
1.60
OJ
E 1.20
.E
"g
0.80
c
a::
0.40
' 1.01
---
~ ....
1.1
1.5
1'"
v-
I-'
10
20 30
50
100
Return period, 7i; 'in years
FIG. 11-10. Composite flood-frequency curve for selected stations in the Youghiogheny
and Kiskiminetas River Basins, Pennsylvania and Maryland. (After U.S. Geological
Survey.)
2. Plot discharge vs. return period (station by station) on extreme

probability paper and fit the curve by eye to obtain the 2.33-y1' flood.
The graphical mean annual flood, determined in thi way, tends to be
more stable than the arithmetic mean since it i little affected by the
chance inclusion or ex 'Iu ion of a major flood.
3. ompute the ratio of annual flood to mean (2. 33-y1') flood for
each year and tation and tabulate by order number. Use e timated
data in assigning order numbers, but omit entry of the corrc ponding
flood ratio.
4. Determine the median flood ratio for each order number.
5. Plot m dian flood ratio vs. return period on extr me probability
paper and fit a smooth curv of composite frequency (Fig. 11-10).
6. Corr late m an annual floods with basin characteristics (Fig. 11-11).
This relation provides a basis for estimating return periods for un gaged
areas.
To illustrate the u e of relations derived by the foregoing procedure,
as ume that the 25-yr flood i r quired for an Ullgaged site with a drainage area of 100 sq mi and within th region of applicability. Entering
265
RAINFALL RUNOFF RELATIONS AND UNIT HYDROGRAPHS
F ig. 11-11 with 100 sq mi yields an e timated mean annual flood of

4200 cf. From Fig. 11-10 it i een that th 25-yr flood is 2.16 times
the mean annual flood, or 9070 cfs. The omp) t fr qu ncy curve can
be derived by plotting di charg s for sey ral r turn p riod .
11-7. Rainfall-runoff r lation a n d unit h ydro graph i n e limating flood fr que n ey.
ne ca.nnot hope to have a long streamflow
record applica.ble to each possible project ite. Th re will, a) so, continue to be major proje 'ts for which the de ign flood is not adequately
faa
~
u
0
0
eo
60
40
v)"
20
l(
.!;;
""
0
-,..,
'r'
'"N
10
8
6
4
IY
./
f .O
0.8
'0
0.6
::0
c: 0.4
c:
0
:5
0.2
'"
:::Ii
0.1
./
f
4 6810
100
fOOO
10,000
Drainage area in square miles
FIG. 1111. Va riation of me an annual flood with drainag e are a, Youghiogh eny and
Kiskimine to s River Basins, Pe nnsylvania and Ma ryland. (After U.S. Geological Survey.)
defined by available discharge record. It is therefore necessary to use

all available data and hydrologic skill to achieve the most efficient design.
Since precipitation records are generally longer than those of streamflow and since the data are usually more readily transposabl , tb
technique discussed in this section possesses great potentiality.
Experience has shown that rainfall-runoff relations of the type shown
in Fig. 8-8 or 8-9 can be developed from only a few y ars of r cord and
that flow data for a number of ba in can often be combined for correlation purposes. In other words, a single relation may be applicable
over a considerable area. When used in conj unction with a unit hydrograph or other procedure for distributing runoff volume, the rainfallrunoff relation provides a means of extrapolating streamflow records.!
Wit h a synthetic unit hydrograph, it is even feasible to compute an
extended fl ood history for an ungagcd point. For flood-frequency
1 R. F . Kresge a nd T. J . Nordenson, Flood Frequencies Derived from lliv r Forecasting Procedures, Proc. ASCE, Vol. 81, Separate 630, February, 1955.
266
analysis, one need only compute peak flow for enough storms each year
to be certain that the maximum annual value is determined.
tudies1 have shown that the rainfall network need not be located in
the problem basin, provided that both have similar precipitation-frequency characteristics. Thus, a set of rainfall data can be applied for
flood-frequency determination within a oonsiderable area and, if data are
available for several networks of differing areal extent, one is in a position
to select the network most nearly approaching in size the study basin .
<D ,
Tecumseh
NEeR'S~'
~ '\ - -
- -
- - -
'\
I~ -
..........._
Q) '.
LEGEND
~<'::
Hor~ori
A "L.ov.nwor'h~a
Holton
R.cordfng gog. for
o~-<'
.,...",
~~
~~
/i~~
WIChlto /
BurllnQton
Newton
Mou,,' Hope
/
o
~<
C'04....
.....~..:"\.C'!:
Go_
Scol,t Mil..
tOl D '040
o~.
.~~-~
1010
Toronto
0'" :SO y.ors
Olothe
'I.
Non - recording goge

Recording gage sinc. about 1940
~-1'
Kon.os City
/I
- - -
~~
Volley Foils
- -
MISSOURI
ms'; - -- - -- - -- ..L - _P2.!..n!.._~I! (.

OELAIt'IlRE RIVER
lleOVE VIlLUY f'ALLS,KANS
- -
A~
-V~c:}~
:I
~I
~ii 1
to
~"
"
~OcS'A:)
FIG. 11-12. Precipitation networks used to compute flood frequencies of the Delaware River
at Valley Falls, Kansas (see Fig. 1113). (U.S. Weather Bureau.)
Figure 11-13 shows frequ ncy curves for the Delaware River at Valley
Falls, Kansas, as deriv d from several different networks located as
shown in Fig. 11-12. Greater consistency wa found among curves
derived from the several networks than among the curves for lO-yr
segments of the ob rved streamflow record.
11-8. Empirical formula. Before extensive collection of hydrologic data began, the practicing engineer was forced to use empirical
formulas in solution of de ign problems. Most of the formulas were
simple in form, con 'ider d only one or two of the many causal factors,
and were usually derived from limited data over a restricted region.
Although better techniques are now available, many of these formulas
are still used blindly. With the po sible exception of the o-called
"rational" formula, these generalized equations ( ec. 9-10) have become
somewhat obsolete.
1 J. L. H. Paulhus and J. F. Miller. Flood Frequencies Derived from Rainfall Data,
J. Hydraulics Div. ASCE, December. 1957.
EMPIRICAL FORMULAS
267
Many studies have been dire ted to the development of empirical

relations for timating rainfall intensity a r quir d in applying the
rational formula and other similar equations. Th r ulting r lation
80.-------r---~------r_--_r--~
Note: See Fig. 11- /2 for location of

nelworJrs A, 8, C, 0 and E
70
/923-50 and 1952 - 53, n = 30 yrs

Drainage area 922 sq mi
60
J!!
<.>
0
0
0
50
.E
CIl
E"
0
.L:
'6
40
-""
0
CIl
a.
.. -
30
10
25
50
::-
100
Return period in yeors

FIG. 1113. Flood frequencies (Gumbel method) of the Delaware River at Valley Falls,
Kansas, as observed and as derived from precipitation networks in and near the basin.
are usually of the form

(11-14)
where i is the rainfall intensity in inches per hour; TT, the return period;
t, the rainfall duration in minutes; and k, x, and b are regional constants.
268
FREQUENC Y AND DURATION STUDIES
Using data from Yarnell's charts, Bernard! derived map depicting

values of k, x, and b over eastern nited tat s. The relative con i tency
in rainfall-frequency formulas is to be expected since they need not
include elusive indice. of basin shape, slope, etc. One maya sume that
the rainfall-intensity frequency formulas are generally more reliable
than corresponding formulas for flood flows. Nevertheless, since the
physical significance of the coefficients employed is vague, there is no
reason to believe that better results can be obtained by selecting coefficients from a map than by determining the desired rainfall intensity
from a map.
RELATED
TUOIES
Preceding sections have been restricted to consideration of discharge

and point-rainfall frequencies. Many similar problems are encountered
in hydrology. There is frequent need for rainfall expectancies over
areas too large to be represented by a single station. The engineer
may be interested in the frequency of droughts as well as floods. Design
studies sometimes involve joint-frequency determination, e.g., the occurrence of heavy runoff on a tributary coincident with a major flood on the
main stream. An important phase in the de ign of power and watersupply projects is the determination of the rate or volume of flow which
will be available for use.
11-9. Fr quency of areal rainfall. Point rainfall of a given flequency is not r presentative of areas in excess of a few square miles.
Moreover, the average of point-rainfall values of a given frequency
does not give an areal av rage of corresponding frequency. An analysis 2
for seven dense networks (mostly east of the Mississippi River) indicates
the existence of a rather stable relation between storm rainfall of specified frequency and duration at a point and that for areas as large as
several hundred square miles. The resulting curves (Fig. 11-14) are
not to be confused with depth-ar a curves for a specific storm (Fig. 3-11)
which tend to be much steeper. To illustrate the use of Fig. 11-14,
assume that one wishes to obtain the 10-yr, 3-hr rainfall for a 200-sq mi
basin in a locality where the corresponding average point rainfall is
3.0 in. Entering the chart with an area of 200 sq mi and a duration of
3 hr yields a reduction factor of 80 per cent. Applying this factor to
the 3.0-in. point value gives 2.4 in. for the basin as a whole. Attention
is directed to the fact that the relation of Fig. 11-14 is based on limited
data for the astern half of the nited tates, and caution should be
exercised in its application to other climatic region .
1 M. Bernard, Formulas for Rainfall Intensities of Long Duration, Trans. ASCE
Vol. 96, pp. 592-624, 1932.
t Rainfall Intensity-Frequency Regime, Part I: Ohio Valley, Part II: Southeastern
United States, U.S. Weather Bur. Tech. Paper 29, 1957, 1958.
DROUGHT STUDIES
269
11-10. Drought tudies. A uswned p riod of time with d fici nt

rainfall i called a drought. Because of the variety of 11 ed for wa.ter,
it is not practicable to define a drought in p cific terms without dra tically limiting it applicability. A period of only a f w we ks without
precipitation may be a seriou matter to th farmer, if it i abnormal
for the sea on of the year. The type of crop und r produ tion mu t b
con. idered in defming a drought.
orne irri Tation projects ar designed
to operate successfully during periods of deficient rainfall lasting sev ral
years.
100
~ r-I'-....
--
24-hr
~ .....__ r--
\\
\~
100
--
6-hr
3-nr
f-hr
200
300
400
Area (square miles)
FIG. 11-14. Reduction of point-rainfall -frequ e ncy valu es for applicaticn t.) basins of specific
size. (U.S. Weather Bureau.)
Agriculturally speaking, an index of moisture deficiency uch as the

antecedent-precipitation index discussed in hap. 8 is equally applicable
as an index of drought everity, provided that precipitation is the Bole
source of water supply for plant growth. Normal antec dent-precipitation-index values vary widely from region to region and sea on to
eason. Drought for a particular crop in a specified locality can be
defined as a threshold index value which varies with time of year.
The hydrologist is more directly interested in drought as it affects
streamflow and groundwater. In fact, he commonly defines drought
in terms of streamflow. Extreme floods can occur several times in one
year, whereas several years of deficient runoff may be required to develop
serious drought conditions for irrigation proj cts having large volumes
of storage. Under such circumstances our records of drought are
effectively much shorter than record of floods, and much can be gain d
by tudy of long-term precipitation data.
uch studie may take either
of two courses. Annual, seasonal, or monthly precipitation-runoff relations may be developed and used to extrapolate the streamflow record,
in which case the frequency analysis would be made in terms of flow
volumes. As an alternative, the frequency of precipitation deficiencies
270
may be derived and the design precipitation converted to corresponding

streamflow.
The latter approach was us d in a study of the Missouri Basin I to
determine the relative need to design for the severe drought experienced
in the 1930's. The derived fr quency relation for the area above ioux
City, Iowa, is shown in Fig. 11-15. Based on these curves, estimated
return periods of the minimum observed 1-, 2-, 3-, and 5-yr precipitation
Return period,
fr,
in years
'8
]:::
Cl.z:.
:t:u
.e-.S
~ .5
14
~"8
[i:
... _
,8-
12
.. .r.
i~
C"c:>
Cl
10
Is
:1-
8
6_
Per cent probability (equal to or less than)

FIG. 11-1 S. Frequency of 1-, 2-, 3-, 5-, and 10-yr precipitation for the Missouri River above
Sioux City, Iowa. (U.S. Weather Bureau.)
during the period 1930 to 1940 are 560, 170, 2000, and 1100 yr, respectively. As might be expected, however, the estimated frequencies are
subject to rather larg errorsi there is roughly a 5 per cent chance that
the true average return period is as short as 30 yr for each. Deficient
streamflow for periods as short as a few days may be critical for watersupply or sewage-disposal problems. Conventional frequency analysis
of low-flow data for various durations can be u ed to obtain curves such
as those of Fig. 11-16.
11-1!. Duration curve. An accumulative frequency curve of a
continuous time series, such as mean daily di charges, di plays the
relative duration of various magnitudes and is known as a duration
curve. Flow-duration curves for the Sacandaga River are shown in
1" Adequacy of Flows in the Missouri River," Report by Missouri Basin InterAgency Committee, pp. 23-34, April, 1951.
271
DURATION CURVES
Fig. 11-17. The resulting curve i highly dependent on the ob ervation

period used in the analysi , m an daily data yielding a much st per
curve than annual data, and thi mu t be borne in mind wh n applying
100
80
60
40
."'."'""- '
Fr~
~~
"
1""+
............0
'.
"'.......
~t-,
20
Symbol
r~k".
I r,
" f ~.,J.z::._.
r\ ''~
~
r\..'
'i'_
Ii'I~! - .
-10
A~
10
8
~
Ai.
'\
JA ~o
A 0 "-
.: 4
~
CI>
co
CI>
>
<t
183 do s
1\"
~~
1
0.8
'K.
'II",..
..........
to-
1'0,,-
1~ ~ ~.l~dOYS~l' r--.
~o . " ,
" 0, .i
, I ...........
20days- f--
......
r\
0.4
7 days
i---..
"-1 day
0.2
l"- I--
0.1
1.01
i"'--
".
60 days +I
'\.
0.6
:- .
120 dOYs..t
i'.:
r\. "'-
.~.
Doys , 1 f--7 f--20

30 f--60
120
183- f--
". I
-!
~D
-\ ~'.N " '\: ,,-"\
,-
Legend
,'+,~
1.1 1.2
1.5
4. 5 6
8 10
20
30 40
Recurrence interval in years
FIG. 1116. Frequency of minimum flows for Yellow Creek near Hammondsville, Ohio
(1915-1935). (From W. P. Cross and E. E. Webber, Ohio Stream-flow Characteristic.,
Ohio Dept. Natural Resources Bull. 13, Part 2, Table 1, December, 1950.)
a duration curve. The use of daily data is recommended for most

problems.
The flow-duration curve is most frequently used for the purpose of
determining water-supply potential, particularly for run-of-river power
projects. The amount of flow available for any selected per cent of
time (assuming no added storage facilities) can be read directly from
272
the curve. In 1920 U.S. Geological urvey adopted the flows available
50 and 90 per cent of the time as the standards of flow for water-power
statistics. The higher figure is a mea ure of the prime power potential,
and the lower value is an index of the power potential with adequate
7.0
6.5
Ol/ermined from mlan daily flows in 500 cfs closs in/ervals (A)
From mean mon/hly flows (8)
Per cent of time indicated flows were equalled or exceeded
FIG. 11 -17. Flow-duration curves for the Sacandaga River near Hope, New York (19361945).
storage facilities. The chronological equence of events is completely

rna ked in a duration curve, and this deficiency greatly restricts its use.
Thu , it is not generally applicable to flood studies nor can it serve to
provide a reliable estimate of the storage required to assure any select.ed
flow.
HYDROMETEOROLOGICAL STUDIES
273
11-12. H ydl'ODl teorological t u die . For major water-control

and utilization tructure the designer need to know much mor than
can be obtairled from frequency analy es of ob erved or synth sized
streamflow data. Major dams, for example, may be designed for a
flood with a recurrence interval of several hundred years. If th dam
is to be located immediately up tream from a large city wh r its failure
could result in large loss of life or property, its design may even be
based on the probable maximum flood.
Frequency analyses of streamflow data are not satisfactory in such
cases because (1) they are of neces ity ba ed on relatively short records,
and the curves mu t be extrapolated con iderably to stimate flood
magnitudes for large recurrence intervals and (2) they do not indicat
wh ther the flood magnitudes estimated in thi fa hion are meteorologically or hydrologically possible. For these rea ons the design of
major structures is often based on hydrometeorological studies, whih
analyze the basic factors of major floods, i.e., storm rainfall and snowmelt; maximize them to their upper physical limits consistent with
current accepted meteorological and hydrologic knowledge and experience; and then reassemble them into more critical but meteorologically
and hydrologically acceptable combinations or chronological sequences
to provide the probable maximum flood.
The probable maximum precipitation is derived l by (1) taking the
results of depth-area-duration analyses ( ec. 3-9) of precipitation in
major storms that have or could have occurred in the area of interest,
(2) adjusting them for maximum moisture charge and rate of moisture
inflow, and (3) enveloping the adjusted values for all storms to obtain
the depth-area-duration curv s of probable maximum precipitation.
The use of storms outside the area of interest is called storm transposition
and involves modification for differences in factors affecting rainfall
such as elevation, latitude, and distance from moisture source. Changes
in shape and orientation of isohyetal patterns are also considered.
Storm transposition is not used in mountainous regions because it is not
possible to accurately adjust rainfall for the orographic influence.
In many area snowmelt is an important, and sometimes predominant,
factor in major floods. In such cases the probable maximum flood
requires determination of optimum snow cover, maximum melting rates,
and t he probable maximum precipitation consistent with the optimum
1 J. L. H. Paulhus and C. S. Gilman, Evaluation of Probable Maximum Precipita.tion, Trans. Am. Geophys. Union, Vol. 34, pp. 701- 708, October, 1953.
R. D. Fletcher, Hydrometeorology in the United States, in T. F. Malone (ed.),
"Compendium of Meteorology," pp. 1033- 1047, American Meteorological Society,
Boston, 1951.
A. K. Showalter and . B. Solot, Computation of Maximum POBsibl Precipitation,
Trans. Am. Geophys. Union, Vol. 23, Part 2, pp. 258- 274, 1942.
.,is....
....
.E
,~
'0
-i
'u
~
Q.
....
.s:.
.....
N
'E"
'j(
.,E
::0
o
..D
0...
cO
"7
~
HYDROMETEOROLOGICAL STUDIES
275
snowmelt conditions. Thi requir that the ea onal variation of t he

probable maximum precipitation be determined 0 that th magnitude of
possible flood , with and without snowmelt, can be e timated for variou
times of the year and compared.
TABLE 11-7. Depth-Area-Duration Relationships of Probable
Maximum Precipitation
(Percentage values ta be applied ta 200' sq mi 2<4hr valu.s af Fig. 11 18)
Area,
sq mi
Duration,
hr
Zone 1 Zone 2 Zon. 3 Zan. <4
Zon.5 Zan. 6 Zane 7 Zan. 8- 9
--- - - - --- --- --10
6
12
2<4
<48
111
123
133
1<42
117
127
141
151
102
121
134
155
112
12<4
132
141
101
121
130
1<4<4
113
123
132
143
102
120
130
140
96
108
123
137
20
6
12
24
48
103
115
126
134
108
118
131
141
98
115
126
146
105
117
124
134
95
114
12<4
138
106
116
125
137
95
113
122
133
90
103
118
132
6
12
2<4
48
92
105
116
123
97
106
118
129
92
50
107
117
133
96
108
114
125
85
10<4
115
129
97
106
116
128
86
10<4
114
12<4
83
96
111
125
100
6
12
24
48
83
97
108
114
88
98
109
119
87
100
108
124
89
100
107
118
78
96
108
123
90
98
108
122
79
96
106
118
78
91
105
119
200
6
12
24
48
74
89
100
105
80
90
100
110
81
93
100
115
82
93
100
111
70
88
100
116
82
90
100
115
73
89
100
111
72
86
100
11<4
62
77
88
94
70
79
88
98
72
74
60
500
6
12
24
<48
83
90
104
84
77
91
102
91
107
72
79
89
104
64
80
91
102
64
78
92
107
52
67
78
85
63
72
80
88
64
7<4
82
95
68
76
85
96
52
68
84
100
64
70
79
94
58
1000
6
12
2<4
48
58
7<4
86
102
73
84
96
Detailed hydrometeorological studies involve a t remendous amount

of work, and the cost may not be justified when only preliminary estimates
of the design flood are required. For thi reason generalized charts of
probable maximum precipitation and its seasonal variation have been
'06
prepared 1 for eastern and central United State by transposing and

adjusting all major storms of record to all points within their re p tive
transposition limits. Figure 11-18 show the enveloping values of
probable maximum precipitation for 200 sq mi and 24 hr. Values for
areas from 10 to 1000 sq mi and for durations from 6 to 48 hI' may be
obtained by applying the depth-area-duration per entage value of
Table 11-7. Generalized estimates have also been prepar d 2 for the
ierra slopes of the Central Valley of California.
BIBLIOG RAPHY
Bailey, S. M., and O. R. Richards: The Maximum Probable Flood and Its Relation to
Spillway Capacity, Civil Eng., Vol. 9, pp. 32-35, 1939.
Bernard, M.: The Role of Hydrometeorology in Planning the Wat r Economy of the
West, Trana. Am. Geophys. Union, Vol. 30, pp. 263- 271, April, 1949.
Bernard, M.: The Primary Role of Meteorology in Flood Flow Estimating, Trana.
ASCE, Vol. 109, pp. 311- 382, 1944.
Chow, V. T.: A General Formula for Hydrologic Frequency Analysis, Tran8. Am.
GeophYIl. Union, Vol. 32, pp. 231- 237, April, 1951.
F oster, E. E.: "Rainfall and Runoff," Macmillan, New York, 1948.
Hazen, A.: "Flood Flows, " Wiley, New York, 1930.
Jarvis, C. S., and others: Floods in the United States, U.S. Geol. Survey Water-8upply
Paper 771, 1936.
Landsb rlt, H. E.: "Advances in Geophysica," Vol. 1, Academic Press, Inc., New
York,1952.
Linsley, R. K., and J. B. Franzini: "Elements of Hydraulic Engineering," McGrawHill, New York, 1955.
McClendon, E. W.: The Role of Meteorology in Projects of the Corps of Engineers
in the Missouri River Basin, B1tU. Am. Meteorol. Soc., Vol. 31, pp. 238-243,
September, 1950.
PROBLEMS
11-1. Using the data given in Table 11-1 and the frequency curve of Fig. 11-5,
construct a partial-duration frequency curve. Plot the frequency curve of Fig . 11-4
011 the same chart for comparative purposes.
Are the two curv s of the same functional form? Are the differences b tw en the two curv significant in vi w of
possibl sampling errors?
11-2. Assuming the curve of Fig. 11-5 to ropr s nt the true average r turn period,
for what discharge would you design to provide 50 per cent assurance that failure
would not occur within the next 20 yr?
11- 3. Using the data in Table 11-5 and Fig. 11-2, compute the frequency distributions for the four 10-yr periods beginning with 1911, 1921, 1930, and 1939. Plot the
four curves so d~rived on a single sh et of extreme probability paper for comparative
purposes. What is the extrem error at the 40-yr return period, assuming the curve
of Fig. 11-5 to yield the true average return period?
1 Seasonal Variation of t he Probable Maximum Precipitation East of the 105th
Meridian for Areas from 10 to 1000 Square Miles and Durations of 6, 12, 24, and 48
Rours, U.S . Weather Bur. Hydrometeorol. Rept. 33, 1956.
t Probable Maximum Pre ipitation on Si rra lopes of tbe neDtrai Valley of Californ!!!., U.S. Weather Bur. Coop. Studies Repl. 12. 1954.
PROBLEMS
277
11- 4. Obtain annual and partial-duration ftood data for a el t,ed gaging sLation
(from Water-8upply Paper 771 or oth r source) and plot tb two s ri on xtr('me
probability pap rand milogaritbmic paper, r p cti vcly. Fit a urve to (\ h by
eye, and a.l 0 comput tb fr quen y cu rv for annual data, using Fig. 11-2. How
does the curve fitted by eye compare with tb GlImb I curve; what is th per ntagc
error at return periods of 10 and 100 yr? How do th annual and partial-durati n
return p riods compare relative to those giv n in Table 11-11
11-5. btain xcessive-precipitation data for a selccted first-ordcr Weather Bureau
station (from Bulletin W or other source) and derive inlonsity-frequen cy curves for
durations of 5, 15,30, GO, and 120 min. Compare your results for the 15- and 60-min
durations with Yarnell's maps (Figs. 11-6 and 11-7).
11-6. Equation (11-14.) involves thr' constants which can be derived by s II' ling
three points from the chart developed in Prob. 11-5. Solve for k, x, and b for each
of several combinations of three points. AII' t.he values of the constants 80 d rived
reasonably consistent? Would you conclude that Eq. (l 1-14.) is suitable in form for
th data u din Prob. 11-5?
11- 7. Compile an annual series of minimum flow for a seier ted gaging station and
plot on extreme probability pap r. Fit curves to the plott d data by eye and th n
by using Fig. 11-2. Do the data appear to follow the theory of extreme values?
11-8. Using about three or four years of mean da.ily flow data, develop th duration
curve for a selected gaging station. D velop also the duration curve of monthly
flows for the same period of record. Explain the differences in th derived curves.
(NOTE TO rNsTRUCTOR: If enoll student is assigned !l differ nt period of record, summarized data derived by the group can be used to develop a long-record curve for
comparative purposes. The magnjtude of sampling errors in a short r cord can be
demonstrated in this mann r .)
11-9. Using Fig. 11-18 and Table 11-7, construct depth-ar a-duration curves of
probable maximum precipitation for a basin centered at lat. 31 0 N., long. 910 W. and
for one cent()rcd at lat. 45 0 N., long. 69 0 W. ( up;p; at plotting on aemilogarithmic
paper, using the ordinate logarithmic scale for area and ihe abscissa linear scale for
depth, and labeling the curvc wiih duration.)
12
SEDIMENTATION
Several billion tons of soil is removed annually from the United States
by water erosion. Soil loss from agricultural land is in itself a serious
economic loss. This sediment is subsequently deposited in stream channels, lakes, reservoirs, and harbors, requiring costly remedial measures.
Erosion has been proceeding throughout geologic history, and whole
mountain systems have disappeared as a result of erosion. Man cannot
materially alter the course of natural erosion but he can avoid accelerating
it by his activities.
Because water is the primary agent of erosion and the principal vehicle
for the transport of the eroded material, sedimentation studies may be
viewed as a special field of hydrology. This chapter presents only a
general introduction to the subject whioh has become a very broad field
in itself.
12-1. The ero. ion proce s. The proce s of erosion consists of the
detachment of soil particles from their po ition in the soil mass and their
movement to a channel in whioh they
may be transported for the balanoe of
their journey. Gully erosion occurs
where a small rivulet forms during
storms. Turbulence in the flow
creates local forces capable of dislodging soil particles from the bed
and banks of the channel. As the
FIG. 12-1. Proflle of a typical gully.
gully deepens, the profile (Fig. 12-1)
is steepest at the head.
ince ero ion is most rapid in this region of
maximum slope, the gully grows headward.
Sheet erosion is the removal of a relatively uniform layer of soil from
the ground surface. No definable channels are formed , and sheet erosion
is difficult to observe except as the ground level is below old oil marks on
fence posts, tree roots are expo ed, or small pillars of soil capped by stones
278
THE EROSION PROCESS
279
remain. 1 Since overland flow is predominantly laminar, the moving

water cannot detach soil particle firmly attach d to the oil rna . However, sheet flow may pick up 100 e or du ty material from the ground and
thus aid in erosion. On very teep slopes, creep, or the gravity flow of
moist soil, may be a fa tor in shoet ero ion.
Ellison 2 has shown that the energy of falling raindrops may be a
significant factor in erosion. Raindrop vary in diameter d from about
0.02 to 0.25 in. ( ec. 3-2), and terminal velocities v vary with the diam ter
from about 12 to 25 fps. Since kinetic energy is proportional to d Sv2,
the erosive power of the drops varies
widely with size, high-intensity rains
with large drops being several thousand times more ero ive than light
drizzle. Assuming a 4-in. rain with
mean drop diameter of 0.1 in. and
terminal velocity of 23 fps, the kinetic
energy delivered to the soil is sufficient to raise the top 4 in. of soil to a
height of 6 ft.
In the absence of wind and on level
ground, splash carries particles uniformly in all directions. On sloping
FIG. 12-2. Downhill transport of soli
ground the net movement of soil by pa rticl es by splash.
splash is downhill because of the
longer trajectory in that direction (Fig. 12-2). As the particles fall to
the ground they enter the sheet of overland flow and receive a further
impetus downhill before they settle to the soil surface.
Because of the importance of raindrop impact in erosion, a vegetal
cover is one of the best protections against erosion. Generally, vegetation absorbs much of the energy of the drops. Chapman 3 has shown,
however, that drops falling from the branches of a pine tree may impart
more energy to the soil than light natural rain. Terracing, contour
plowing, and other practices which reduce urface runoff and land slope
are also important erosion-control measures ince the overland flow is
important in moving the loosened soil material to channels. Plot
measurements uggest4 that erosion E is related to phy i 'al features of the
1 C. H. Gleason, Indicators of Erosion on Watershed Lands in California, Tran8.
Am. Geophys. Union, Vol. 34, pp. 419-426, June, 1953.
2 W. D. Ellison, Studies of Raindrop Erosion, Agr. Eng., Vol. 25, pp. 131-136, 181182,1944.
'G. Chapman, Size of Raindrops and Their Striking Force at the Soil Surface in a
Red-pine Plantation, Trans. Am. Geophys . Union, Vol. 29, pp. 664.-670, October, 1948.
'G. W. Musgrave, The Quantitative Evaluation of Factors in Water Erosion-A
First Approximation, J. Soil and Water Conservation, Vol. 2, pp. 133-138, July, 19-17.
SEDIMENTATION
280
land by
E
= kS 1 .85LO.3Spl.76
(12-1)
where s is land slope in per centi L, length of slope in feeti and P, maximum annual 30-min. rainfall in inches.
Equation (12-1) represents the results of small test plot; additional
information would be required for estimates of the sediment yield of
natural watersheds. As has already been intimated, the type of vegetal
cover and land u e in the basin are factors. It has al 0 been shown that
the relative density of incised channels in the basin is significant. 1 Unless
channels which are competent to transport sediment exi t, there can be
no large movement of sediment out of the basin. Broad shallow swales
may carry water but they will transport little sediment. The physical
characteristics of the soil are also significant in determining the rate of
splash erosion. Tightly oemented soils resist erosion more readily than
100 e soils. Much research i aimed at relating mea mabIe soil characteristics with relative ero ibility. Splash erosion increases with the percentage of sand in the oil and deerea es with increasing percentages of
water-stabl ag regates. 2 A better mea ure of the rainfall characteristics
is also required. Attempts have been made to develop an index to the
erosive power of rains by exposing pans of standard sand and noting the
soil loss after each rain. Such an index of erosive power obtained on a
regular basis and combined with data on the relative ero ibility of
soils should greatly improve estimates of sediment production. 8
12-2. u spended- edim ent t r ansport.
ediment moves in streams
a suspended sediment in the flowing water and as bed load along the
channel bottom. The two processes are not wholly independent, for
material which trav Is as bed load at one section may subsequently
be ome suspended load, and vioe versa. A third proce , saltation, is
sometimes defmed a the movement of po.rticles by bouncing along the
bed. It is thus a transitional case between bed and suspended load.
Material suspended in still water tend to settle at a rate approximately
in acoordance with Stokes' law. 4 In turbulent flow this gravitational
1 Personal communication from W. C. Ackermann, Director of the Illinois State
Water urvey.
S R. Woodburn and J. Kozachyn, A Study of the Relative Erodibility of a Group of
Mississippi Gully oils, Trans. Am. Geophys. Union, Vol. 37, pp. 749-753, December,
1956.
a G. W. Eley and C. H. Lloyd, Graphic Solution of Probable Soil Loss Formula.
U.S. Soil Conservation Service Misc. Publ. 204, Januar.y, 1952.
(Stokes' law stutes that the settling velocity v, is
v, ..
where
P.
and
2(p. - p)grt
91'
ar densities of tlle particle and liquid, respectively, r is the radius of
SUSPENDED-SEDIMENT TRANSPO RT
281
settling is countera ted by upwn.rd movement in ddie. Becau e the

gravitational ettling re ult in an increa ing edim nt con entration
with depth, upward-moving ddies 'arry more sediment than downwardmoving eddies. The sy tern is aid to b in quilibrium if th gravity
movement and turbulent tran port ar in bl1lanc .
In a discu sion of the principles of edim nt movem nt Brownl sto.tes:
The movement of sediment in alluvial streams is so complex a probl m that it
may never be completely subject to rational solution. It repr s nts, in fact, the
most extreme degree of un teady, non-uniform flow, since the stream bed as w II
as the water surface may be continually changing in form. \Vith the pr nt
state of knowl >dge, an approximate understanding of th general trnnsport
mechanism can be obtained only by isolating particular details or by so simplifying the boundary conditions that only th most significant variabl s ne 'd be
considered.
In the light of this comment, di cu sion of tb Lheory of sediment

movement is held to a minimum. Analysis of suspcncled-s diment
transport usually begins with the basic equl1tion 2
e.v, =
de.
dy
(12-2)
where c, is the average concentration of sediment with settling velocity
v. at depth y and E is a mixing coefficient which is assumed to be the same

for sediment and momentum. As uming 3 that E is constant with depth
and equal to its mean value D
gDs/ 15, where D is depth and s = slope,
integration of Eq. (12-2) leads to
!!..!. =
C.a
e - 1b;dll- o)/D
(12-3)
where C. a is the concentration of sedimenL at depth a and X = v.1 gDs

for wide streams. Total suspended load M .. per unit time per unit width
.
(D
Combining the Prandtl-von Karman equation for
IS equal to Jo VC. dy.
the particle, and IJ. is the abs('lute viscosity of the Jiquid. The equatioo assumes (1)
the particles are large compared with the molecules of the liquid but that viscosity is the only source of resistan ce to their fall i (2) that the particles are rigid,
smooth, and spherical; and (3) that tbeir fall is not impeded by adjacent particlcs.
Stokes' law is g'eDcrally considered appF table to particles from 0.0002 to 0.2 mm in
diameter.
1 See Bibliography (":t:ngineering Eydraulics").
2 M. P. O'Brien, Review 01' +.he Theory of Turbulent Flow a.nd Its Relation to
Sediment Transportation, 7'rans . .lim. Geophy8. Union, Vol. 34, pp. 487-491, 1933.
a E. W. Lane and A. A. Kalinske, Engl11eering Calculations of Suspended Sediment,
TraM. Am. Geophys. Union, Vol. 22, pp. 603-tl07, 1941.
tha~
SEDIMENTATION
282
velocity distribution with Eq. (12-3) gives

(12-4)
where ~ is the ratio of the average concentration to that at the bottom

and is a f unction of X and relative roughness nj m 6 (Fig. 12-3). Equation
(12-4) applies only to particles having a common fall velocity, and
its use requires integration through the range of particle sizes in the
su pension.
1.00
0.80
0.60
"'""
0.40
~~
~ 0.20
~~
~ t:---
VI
-a'"
0.10
i/o/ties of n/o ~
> 0.08
0.06
........
0.04
0.02
0.1
0.2
0.3
...........
0.4
0.5
0.6
Values of X
--
0.7
::::
aOf =--=ao-,
0.8
=-...
~.?
~
-........
0.9
1.0
FIG. 12-3. The relation between ~ and X in Eq. (12-4) for va rious values of relative rou ghness. (After Lane and Kolinske. )
Nonequilibrium transport of sediment is far more common in nature

than are equilibrium conditions, since velocity and depth of flow usually
vary along any channel. In nonequilibrium conditions the sediment
load changes \yjth di'tancc along the strea m. The general equation is
v ac, = v. aCt
ax
ay
ac. + aEy ac, + E a2c.2 + E a2c.

ax ax ay ay '" ax
u ay2
aE,.
(12-5)
This equation has not been solved without qualifying assumptions which
render the solution of little value in natural streams.
12-3. Ded-Ioad transport. For many years analysis of b d-load
transport has been based on the c1assical equation of du Boyst
G. = T ~ (TO - To)
W
(12-6)
where G. is the rate of bed-load transport per unit width of stream, T is

an empirical coefficient depending on the size and shape of the sediment
1 P. du Boys, Le Rhone et les rivieres
5, Vol. 18, pp. 141-145, 1879.
0.
lit o.ffouillable, Ann. ponl8 et cha'U8sees, Ser.
283
BED -LOAD TRANSPORT
particle , W is the specific weight of water, TO i the hear at the str am

bed, and T c i the magnitude f h al' at. which trt1n port begins.
um 1'ou variation on the original du Boys formula have be n prop
all using the concept of a criti 'al trartiv forc to iuit.jo.te motion.
approach ignores the modern concepts of turbulen 'e and the boundary
layer as they affect entrainment of bed parti les. The uec ssful application of Eq. (12-6) lie in the proper selection of tb co ffi ient T. M t
available values are determination fr m studie with smnll flumes .
Tuble 12-1 ummarizes values giv n by traub. 2
TABLE 12-1. Factors in Eq _ (12-6) for Bed-load Movement
Particle d iamete r, mm. _'" _. _
,i
T, ft o/ lb 2-sec . _. _.... . . _ . .
0.81
0.48
0.29
0. 17
0. 10
0.06
0.01 6
0 .0 17
0.022
0.032
0. 051
0.09
Te,
Ib/lq ft .. . _. . .... . ...
Recent work on the bed-load problem has tressed attempts to apply

basic concepts of turbulent flow, including the statistical varintion of
fluid forces at a point. These more rigorou analyses lead to extremely
involved solutions suggestive of the possibility that no simple bed-load
equation exists. Kalinske 3 assumed that the rate of sediment movement G. (in lb/ sec/ lwit width) mu, t equal the product of particle volume
7rd 3/ 6, mean particle velocity Vg , average number of particles per unit of
bed area 4j/ 7rd 2, and specific weight wo, or
- 4j
Gi = 7rd3
6 Vo 7rd2
2 - 'd
Wa
= S VgJ
where j is the portion of the bed tnking shear.

by the time average of fluid velocity V,
Gi
jvdw g
2 Vg
""'
sF
Wg
(J2-7)
Rearranging Ilnd dividing
(12-8)
As uming that time variations in V follow the normal-error law, it can

be shown that. vo/v is a function of T . /TO and the intensity of tu rbulence
:r /v, where 0' is the standard deviation of the time variations in V (Fig.
12-4). From boundary theory, jj = 11 VTo/ p; hence
wqjd
G;
VTo/ p
=f (~)
(12-9)
TO
ee Linsley and Brown (1950) in Bibliography.

L. G. Straub, H. R. Doc. o. 238, 73d eong., 2d ess., p. 1135, 1935.
a A. A. Kalinske, Movement of Sediment as Bed Load in Rivers, Tran8. Am.
Geophys. Union, Vol. 28, pp. 615-620, August, 1947.
1
SEDIMENTATION
284
From experimental work by While, j = 0.35 and (I/f) = 0.25. Figure

12-5 is 11 compari 'on of Eq. (12-9) with cxp rirnentul data. Equation
(12-9) gives t he rate of transport for particles of diameter d. The total
1
1.0
I~
0.5
.......
:ri
~ ~o:...-;S.
'"
;;:
'0
is
0.1
0.05
;:
~
.:.\
"0
II>
~~G-.....
a
"O::-~'?,.
c:
'e '
~~J
0>
'0
~
<.>
~:s.
Sl.
.Q
I"'"
r:::
0.01
0.005
%!
""
1ii
0
'\.
a::
0.001
'\
0.4
08
1.2
1.6
2.0
Ratio of critical shear to bottom shear, <e/To
2.4
FIG. 12-4. The relation between (yg / v) and (T. / TO) for laminar and turbulent flow.
(After
Kalinske.)
2.4
...
.8
r--0
2.0
....
0
-....:
2~ 1.6
o<.> ..."
=E<.> 0 1.2
_c.
~
0
a::
II>
r---- h
".
SOURCE or DATA
EINSTEIN, WE ST GOOSE R. &; MTN.CR
o MEYER - PETER, ZURICH LAB.
)( u .s. W. E. S., VICKS8uRG I MISS.
LlU,IOWA H'I'DftAuLIC lAB
oeASEr, BERLIN LAB.
. OtLBERT
0
I'-...
~
0.8
,~
0.4
K~
0.001
0.01
0.1
1.0
Gil wgd 1) Toll'

FIG. 12-5. Comparison between bed-load data and Eq. (12-9).
(After Kalinske.)
tran port mu t be inte 'rated through t he mnITe of grain izes. A single

computation using the median diameter s ems to yield a reasonable
approximation.
1 C. M. White, Equilibrium of Grains on the Bed of a. Stream, Proc. Roy. Soc.
(London): A, Vol. 174, pp. 322-334, 1940.
SEDIMENT MEASUREMENT
285
Much more detail d di cussion of bed-lond movement will b found

in some of the referenc in the bibliography.
12- 4. e dim cnt mcasur 'men L.
in
th
wm nt t1'llnsport in a stream cannot be accurately 'omput d, good fi Id m aSUl'em l1t
becomes extrem ly important. Early 'usp nded- diment ob ervations
were made with open bottle or complex grab samplers whi h failed to
provide adequate data, for a number of rca ons. A good sampler must
Standard
wading rod
--
Curren!
Pint m/~k batlle
FIG. 12-6. A U.S. DH-48 d e pth-inte g rating hand sam pl e r for small stre ams.
cause minimum disturbance of streamflow, avoid errors from short-period

fluetuations in sedim nt conc ntration, and give results which can be
related to velo ity mea urements. These requirements are met in a
series of amplers 1 de igncd at the Iowa Hydraulic Laboratory under
the sponsor hip of several Federal agencie.. The samplers (Fig. ]2-6)
consi t of a streamlined shield enclosing a standard milk bottle as a
sample container. A vent permits escape of air as water enters the
bottle and controls t he inlet velocity so that it i approximately equo] to
the local stream velocity. Nozzle tips of various sizes are available to
control the rate of filling of the bottle. The large models have the bottle
fully enclosed and are fitted with tail vanes to keep the sampler headed
into the current when cable- upported.
1 A Study of Methods Used in Measurement and Analysis of Sediment Loads in
Streams, U.S. Govt. and Iowa Inst. Hydraulic Research Rep18. 3, 6, and 8.
286
SEDIMENTATION
The ampler is lowered through the stream at constant vertical speed

until the bottom is reached and i then raised to the surface at constant
spe d. The result is an integrated sample with the relative quantity
colle ted at any depth in proportion to the velocity (or discharge) at
that depth. The duration of the travel' e is determined by the time
required to n arly 1ill the sample bottle and can be computed from the
1illing-rate curves for the particular
nozzle when the stream velocity is
known. A number of traver es are
made at intervals across the section
to determine the total suspendedsediment load for the section. Thus
there is no problem of whether a
point sample is repre entative of
sediment load in the section. Point
samplers are used only where it is
impo ,ible to use the depth-integrating type because of great depths or
high velocity, or for studies of sediment distribution in streams. Becau e of the shape of the sampler,
the nozzle cannot be lowered to t.he
stream bed, and consequently a few
FIG. 12-7. Pressure- difference type of
inches of the depth near the bed is
be d-load sampler a s developed by the
not sampled. This may represent a
Netherlands government.
large error 1 in shallow streams.
The collect d samples are usually filtered and the sediment dried.
The ratio of dry weight of sediment to total weight of the sample is the
sediment concentration, usually expressed in parts per million. Other
analyses which may be performed include determination of grain-size
distribution, fall velocity, and, occasionaUy, heavy-mineral or chemical
analysis. The latter tests may be u efuI in tracing the original source
of the sediment.
There is as yet no wholly satisfactory bed-load sampler. Portable
samplers consist of a container lower d to the stream bed. The most
successful is the pre sure-difference sampler (Fig. 12-7). The expansion
section causes a pressure drop which encourage inBow at th ame rate
a the prevailing flow while at the same time the reduced exit velocity
encourages deposition of ediment in the mesh bag. Fine ediment
particles can escape through the me h, and a calibration in a flume is
desirable. The difficulty of designing a sampler which does not di turb
1 B. R. Colby, Relationship of Unmeasu.red Discha.rg to Mean Velocity, Trnns.
Am. Geophys. Union, Vol. 38, pp. 708-717, October, 1957.
ANNUAL SEDIMENT TRANSPORT
287
the flow and at the ~ame time efT ctively trap all size of bed load hould
be apparent.
One type of permanent bed-load trap for re earch purpos on small
stream con i t of a grated opening in th tream b d into which the
bed material falls. The trapped mat rial is lat r excavated or sluie d
out and mea ured. Turbulence-producing weir have al 0 b n d si n d
which throw the bed load into suspension locally '0 that it can b ampl d
with a suspended-sediment sampler. Compari on of the ample t hu
10'OOO~~II~~II~~II~all~!II!11
' ' f-:'
~ f'OOO~~II~mll!II~IIIIII~~
~",'I' 11
.....,.~..
-c
~I/)
. '.
fO
'.
',".
,..
.:tT.+ttI+t--t-+tmf-H
.. .
fOO f,OOO fO,OOO fOO,OOO
Suspended sediment discharge in tons per day
FIG. 12-8. Sediment-rating curve for the Powder River at Arvada, Wyoming.
and
(L eopold
Maddock . )
obtained with tho from a ecti n Upl'ltrCf:Lm of the weir indicates the
quantity of bed load. This method is suitable only when the bed material
is relatively finc.
12-5. Annual sediment tran port. Most sediment-sampling
methods give only pot mea urements of sediment load and must be
interpolated to determine total annual transport. This is commonly
done by use of sediment-rating curves (Fig. 12-8) in which sediment is
considered as a function of flow. The figure clearly show that the
relation leaves much to be desired. Actually, there is little reason to
expect a simple relation between stl' amflow and sediment load. A given
flow rate may re ult from melting snow, moderate rain, or inten e rain.
ome portions of
A different ediment load would re ult in each ca e.
the basin may be more prolific sediment sources than others. In this
case, areal distribution of runoff is a significant factor.
SED/MENTA nON
288
Currently, no better method of interpolating sediment measurement

Rating curves should be used with caution and applied
where po ible to small, relatively homogeneou ba ins. If u ed only
for estimating mean annual sediment production, the results should be
reasonably satisfactory. A completely sati factory approach must await
ac umulation of considerably more factual datn.. Where sediment is of
major concern, daily measurements may be advi able.
12-6. Be ervoir edimentation. The relatively quiescent waters of
a reservoir encourage the deposition of sediment transported in the
inflowing water. The rate at which the capacity of a res rvoir is reduced
by sedimentation depends on (1) the quantity of sediment inflow, (2) the
percentage of this inflow which is trapped in the reservoir, and (3) the
density of the deposited sediment.
is available.
TABLE 122. Selected Data on Rates of Sediment Production

Annual sediment production
Location
Bayview Reservoir, Ala _ .....

San Carlos Reservoir, Ariz .. .....
Morena Reservoir, Calif .....
Black Canyon Reservoir, Idaho ..
Pittsfield Reservoir, III ........ .........
Mission Lake, Kans .. .. .
High Point Reservoir, N.C ..
Tygart Reservoir, W . Va ....
Drainoge area,
sq mi
Acre-ft/sq ml
Tons/ sq mi
1.34
0 .26
2.56
0.13
3.54
2. 00
0.50
0.05
1769
389
3340
172
3090
2705
544
51
72
12,900
112
2,540
1.8
11.4
63
1,182
For re ervoir sedimentation studies, sediment inflow is usually expressed

in acre-feet or tons per year. If sediment measurements are available
near the reservoir site, the inflow may be estimated by the methods
described in Sec. 12-5. Since suspended-sediment measurements are
most common and do not include bed load, an arbitrary increase of 10
to 20 per cent is customarily included to allow for bed load. There are
relatively few sediment-sampling stations, and for mo t small proj cts
(and many large one) sediment yield of a basin i estimated on a basi
of drainage area. There are formulas for estimating sediment production (Sec. 12-1), but these are generally applicable to a specific area since
the geneml relationship are not yet thoroughly under tood. Table 12-2
presents elect d data on rate of ediment production in various parts
of the nited tates. These data are commonly obtained from surveys
of the sedim nt accumulation in xisting reservoir. L
1 L. C. Gottschalk, Measurement of
ASCE, Vol. 117, pp. 59- 71, 1952.
edimentation in
mall Reservoirs, 'Prans .
289
RESERVOIR SEDIMENTATION
100
90
.,u
Q;
80
70
Q.
..;
.,
Q.
Q.
60
8.. /
I
/I
Medion curve tor normol
1/,.j...,~--l../~II- ponded reservoirs
-+~h,..I---I--+--+-l---l
50 t---t-+-+4c._
V V ;V Y
1/ _!!/ V
30 I---+-......
I/ ;, . V
~
40
C
.,
E
,:;
.,
I/)
-~
20
10
Normol ponded reservoirs

.. Normol ponded reservoifS
wi'" sluICln9 or venting
operollofls tn effect
nveloPf CUrres lor

normol ponded reservoirs
tt-
o Deslllin9 bosins
x Semi- dry reservoirs
I-
V , V. l
II I
1 1
I
8.0~0~1--~0~.O~0~3~~O~00~7~0~.071--~0~.O~3~~0~.O~7~0~j--~0~.2~0~.3~0~.5~0~.7-L'--~2--~3--J5~7~10
Copaci ly - inflow ralia (acre - feel capa cify per acre - fool annual inflow)
L EGEND
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
Williams Reservoir
Lake Halbert (Rock Reservoir No. 1)
Lake Holbert (Rock Reservoir No. 3)
Hales Bar Reservoir (1935 and 1936)
Holes Bar Reservoir (1938)
Hale. Bar Reservoir (1937)
Keokuk Reservoir
Lake Toneycomo
Wilson Lake
lake Marinuka
lake Decatur
Bullard 's Bar Reservoir
lake Halbert (Earth Reservoir No_ 1)
lake Rockwell
Corpus Christi Reservoir (1942- 1943)
Corpus Christi Reservoir (1934- 1942)
lexington Reservoir
lloyd Shoals Reservoir
lake Michie
lake Issaqueena
Guernsey Reservoir
Arrowrock Reservair
T. and p_ Reservoir
23
Hiwassee Reservoir
24
240 Imperial Dam Reservoir (1938- 1942)
24b Imperial Dam Reservoir (1943- 1947)
Lake of the Ozorks
25
Pardee Rese rvoir
26
Possum Kingdom Lake
27
White Rock Reservoir
28
Buchanan Lake
29
Norris Reservoir
30
Senecaville Reservoir (1939- 1943)
31
H. loge Pond
32
Denison Reservoir
33
Lake Mead
34
San Carlos Reservoir
35
Conchas Reservoir
36
Fort Peck Reservoir
37
370 Elephant Butte Reservoir
All American Canal Oesllting Basin
38
Hadley Cre ek New OesUfing Basin
39
John Marlin Reservoir
40
Senecaville Reservoir (1936- 1939)
41
FIG. 12-9. Reservoir trap efficiency as a function of the capacity-inflow ratio.
(Brune.)
SEDIMENTA TlON
290
When data from reservoir sediment surveys can be compared with

measured sediment inflow it is pos ible to determine the sedimenttrapping efficiency of the reservoir. From such information Brune'
derived Fig. 12-9 which relates trn.p efficiency (sediment retained divided
by ediment inflow) and the ratio of re ervoir capacity to mean annual
water inflow. If the reservoir capacity is much les than the annual
inflow, water is retu,ined in the reservoir for a relativelY short time and
some of the suspended sediment may pass the dam without being deposited. On the other hand, if the reservoir capacity is much greater than
the annual inflow, water is retained for long periods, and practically all
the suspended sediment has a chance to settle to the bottom. Reservoirs
operated deliberately to trap debri show a marked deviation from the
curve, as do the emidry reservoir in whieh water is retained only as
needed for flood control.
The specific weight of the accumulated sediment depends on the charact r of the soil materials of which it is formed and the amouut of superimpo ed load to which it is ubj ected. Lane and Koelzer 2 found that
the ultimate dry specific weight w'" at time t can be found from
Wm
[(WI
+ K1)X + (W2 + K Z)X 2 + (wa + K S) X 3] loglo t

1
(12-10)
where the subscripts 1, 2, and 3 refer to sand, silt, and clay, respectively;
w is sp cific weight; K is a compu,ction coefficient; X is the fraction of
each soil cIa s in the sediment; and t is time in year. Values of wand K
for various re 'ervoir operating conditions are given in Table 12-3.
TABLE 12-3. Constants in Eq. (12-10) for the Speciflc Weight of Sediment
Sand
SUt
Clay
Reservoir conditian
W,
KJ
W,
K2
Wa
K.
- - - - - - - - - - - - -Sediment olways submerged

Maderate reservoir draw down
Considerable reservoir drawdawn
Reservoir normolly empty .
93
93
93
93
0
0
0
0
65
74
79
82
5.7
2.7
1.0
0.0
30
46
60
78
16.0
10.7
6.0
0.0
Equation (12-10) and Table 12-3 lead to valu of specific weight ranging
from about 46 to 93 lb/ cu ft at the end of 10 yr for ren. onablyextreme
assumption as to the ediment composition. Data from sediment
1 G. M. Brune, Trap Efficiency of Reservoirs, T,.ana. Am. Geophys. Union, Vol. 34pp. 407-418, June, 1953.
IE. W. Lane and V. A. Koelzer, Density of Sediments Deposited in Reservoirs.
St. Paul Di8t., U.S. Corps 0/ En(/ineer8 Rept. 9, 1943.
PROBLEMS
291
survey 1 show dry 'p 'Hie weight ranging from about 40 to 110 Ib/ 'u fL
with an average of about 60 Ib/ cu ft for liew s dim nt aDd 0 Ib/ eu ft for
old deposits.
BIBLlO G n APHY
Brown, C. B.: The Control of Reservoir ilting, U.S. Dept. Agr. Miac. Publ. 521, 1943.
diment Transportation, Chap. 'II, in Hunter Rouse (ed.), "EngiBrown, C. B.:
neering Hydraulics," Wiley, New York, 1950.
Ein tein, H . A.: The Bed-load Function for dim nt Transportatiou in ( p n hannel F lows, .S. Dept. Agr. Tech. Bull. 1026, eptembcr, 1950.
Linsley, R . K., M. A. Kohler, and J. L. H. PaulllUs: "AppJi d Hydrology," hap. 13,
McGraw-Bill, New York, 1949.
Trask, P. D.: "App]jed edimcntation," Wil y, New York, ]950.
U.S. Bur. Reclamation Manual, Vol. VIl, Part 9, Sedirnentt\tion, 194
DAT A
OUn CE
Sed iment data are collected by numerous agencies, but because of the variety of
measurement methods and types of data no organized publicatioll is available. For
data, contact U.S. Bureau of Reclamation, U.S. Geological urvey, U.S. Corps of
Engineers, U.S. Soil Conservation Service, and state departments of water resources.
PROBLE M S
12-1. Find the equation of a sediment rating passing through the points q - 10,
q. = 4 and q ,., 1000, q. = 8000 in Fig. 12-8.
12-2. (a) Find the equation applicable to the graph of Fig. 12-8 with sediment load
expressed in acre-fet per day. Assume a.n in-place density of 100 lb / eu ft . (b) The
drainage area of the Powder River a.t Arvada is 6050 sq mi. Write tb equation in
terms of average depth of erosion in inches. (c) What is the approximate extreme
departure (error) of the plotted poillts from the )jne in per cent?
12-3. A proposed reservoir has a capa ity of :~ OOO acre-ft and a tributary area of 50
sq mi. If the annual streamflow averages 5 in. of runoff and the s diment produotion
is 0.69 acre-ft per sq mi, what is the probable life of the reservoir before its capacity is
reduced to 500 acre-ft? Use the median curve from Fig. 12-9. Repeat tbe computations using the two envelope curves.
12-4. A reservoir bas a capacity of 50,000 acre-ft, and the annual inflow averages
78,000 acre-ft. The estimated sediment production of tbe area is 950 tonslsq mi,
and the drainage area is 1120 sq mi. Sediment samples indicate that the grain-size
distribution is sand, 24 per cent; sil t, 33 per cent; clay, 4:l per cent. (a) When wi ll 80
per cent of the reservoir capacity be filled with sediment? Use line 3 of Table 12-3
and assume minimum w'" of 40 Ibl cu ft. (b) Discuss the form of Eq. (12-10) and
its limitations.
I D. C. Bondurant,
edimentation t udies at Conchas Reservoir io New Mexico,
Tran8. ASCE, Vol. 116, pp. 1283- 1295, 1951.
13
APPLICATION OF HYDROLOGIC
TECHNIQUES
The main purpose of engineering hydrology is to derive the factun I

information on quantities and rates of flow on which the de ign of engineering projects can be based. The discussion of specific methods in
the previous chapters have neces al'ily treated individual steps in the
over~all picture. It is the purpose of this chapter to illustrate how these
techniques are combined to solve a complete problem. The usual
hydrologic problem utilizes large quantities of data and extensive computa.tions which cannot be reproduced completely. Hence, the examples
are schematic in some respects.
It should not be inferred from these examples that there is only one
possible m thod of solution for a particular type of problem. On the
contrary, almost every hydrologic problem is unique in some way. The
examples whi h follow illustrate possibl solutions adapted to the assumed
conditions of the problem. A chn.nge in these condition, e 'pecially in
the availability of data, would necessitate a chang in procedure.
13-1. Storage-reservoir design. The hydrologic problems associated with the design of storage reservoirs include determination of the
available flow, estimation of probable demand for water, and selection
of re ervoir capacity appropriate to the demand and available water.
The details of design depend to some extent on the relative size of the
re ervoir. A very small re ervoir might conceivably be overtaxed by a
drought of a few we ks' duration. In this ca e daily data are required
for the analy is, and the capacity can be selected for any desired frequency of drought by u ing the methods of ec. 11-10. At the other
extreme are res rvoirs with a capacity several times the annual inflow
(e.g., Lake Mead, Nevada-Arizona) where water is carried over in storage
for years and the critical drought period is measured in years. In uch
cases annual data may be sufficient, although for important projects
292
STORAGE-RESERVOIR DESI GN
293
detailed tati tical treatment may b employ d. 1

rvoil' op rate on an annual cycl ,with ach year" inflow enemlly ad quat for
the year' water requirement. The 'riticlli dry p ri d in uch 11 ca e
will ordinarily 0 r J or 2 yr, and monthly data are n ded for It th rough
analy j ' . The example which follow deal with thi latter case but is,
a well.
in principle, applicable to th other ca
The available flow is the a tual str amfIow at the head of the reservoir
corrected for precipitation on the re ervoir, 'vaporation 10 ,and water
I'e rved for other use .2 If ther is a record of actual flow at or near the
proposed dam site, it is often too short to include a critical p riod of dry
years. If 0, a relationship between monthly precipitation and monthly
streamflow may be used to estimate the str amflow for year when no
record is available. If the gaging station is not at the dam site, the
record must be adjusted for the inflow between the station and thi
point. This adjustment is usually based on a ratio of drainage areas,
with allowance for differences in pr cipitation.
Estimates of monthly evaporation should be prepared by the most
appropriate method for the data available ( hap. 5). PI' cipitation on
the reservoir is d termined from the best records available for the vi 'inity
of the site. The reservation of water for oth r pUl'poses is u uully fixed
by legal con iderations or plans for other projects in the ba. in.
The method for estimating the expected demand for water depends on
the projected u es. Water requirements for municipal, indu trial,
recreational, or power uses are u ually determ.in d by combining population projections with per capita requirements and arc largely independent of hydrologic considerations. The requirement for irrigation
are usually determined by combining estimates of potentially irrigl1ble
land area with estimates of probable consumptive use per unit area
(Chap. 5).
Wh n the estimates of available water and the various items of water
use and water loss are a' embled, the reservoir capacity is usually determined by an operation study. This is a simulation of the reservoir op ration which would have taken place during the critical period if specified
operating rules were followed. Such a study is illustrated in Table 13-1.
The known or estimated data on wb.ich the study is based- precipitation,
evaporation, inflow, and demand- are listed in ols. 2 to 5. A monthly
relea e for downstream u e (Col. 8) of 40 acre-ft or the natural inflow,
whichever is less, is required. The re ervoir is a sumed to be empty
on Dec. 1, 1931. For this study a reservoir capacity of 9500 acre-ft
1 H. E. Hurst, Long-term Storage Capacity of Reservoirs, Tran8. ASCE, Vol. 116,
pp. 770-808, 1951.
If the flow record is at the dam site, adjustment should be made for the difference
between reservoir evaporation and natural evapotranspiration from the reservoir area.
APPLICATION OF HYDROLOGIC TECHNIQUES
294
TABLE 131. Operation Study for a Storage Reservoir

Month
and
year
Preclpi Eva potation, ration,

ft
ft
(1)
(2)
-----
Flow,
aereft
De
Pretipi.
Evapo
mand,
totion,
ration,
ocreft
acreft
acreft
Star
Star
age
age,
change,
acre ft
acreft
acreft
Re
leale,
Area ,
acr
(3)
(4)
---
(5)
(6)
(7)
(8)
(9)
(10)
(11)
0.060
3440
100
88
40
3383
3383
165
0.079
0.111
0.199
0.294
0.411
0.482
0.500
0.450
0.385
0.252
0.137
0.074
2030
4470
96
11
11
0
0
0
0
0
0
6
100
130
140
250
440
440
440
400
370
250
130
100
72
103
13
19
5
0
0
0
0
8
15
87
15
29
62
91
125
140
138
118
96
60
32
17
40
40
40
11
11
0
0
0
0
0
0
6
1947
4170
- 133
- 322
- 560
- 580
- 578
- 518
- 466
- 302
- 147
- 30
5330
9500
9367
9045
8485
7905
7327
6809
6343
6041
5894
5864
215
315
312
308
298
282
270
256
243
235
230
229
0.083
0.102
0.210
0. 274
0.398
0.499
0.547
0.465
0.384
0.240
0 . 140
0. 079
1090
33
347
12
11
0
0
0
0
0
3
100
130
140
250
440
440
440
400
370
250
130
100
182
26
73
5
38
0
0
0
0
29
0
101
20
26
55
71
98
116
119
95
74
44
25
14
40
33
40
12
11
3
0
0
0
0
0
3
1112
- 130
185
- 316
- 500
- 556
- 559
- 495
- 444
- 265
- 155
- 13
6976
6846
7031
6715
6215
5659
5100
4605
4161
3896
3741
3728
260
257
263
253
241
224
210
198
186
179
175
174
0 . 080
0 . 098
0.214
0.266
0.392
0.480
0. 507
0.450
0.370
0.250
0.126
0.079
66
332
14
8
5
0
0
0
0
0
0
1
100
130
140
250
440
440
440
400
370
250
130
100
19
80
0
10
9
7
0
0
7
6
30
19
14
17
38
46
63
70
68
52
34
19
9
5
40
40
14
8
6
0
0
0
0
0
0
1
69
225
178
286
495
503
508
452
397
263
109
86
3659
3884
3706
3420
2925
2422
1914
1462
1065
802
693
607
173
178
175
168
153
139
128
103
81
71
65
63
0.068
0.088
0.213
0.258
1820
18
1620
3680
100
130
140
250
75
15
76
107
7
12
33
57
40
18
40
40
1748
- 127
1483
3440
2355
2228
3711
7151
137
135
175
266
1931
Dec ... 1.062
1932
0.379
0.390
0.042
0.062
0. 018
0
0
0
0
Oct. . . . 0.034
Nov . 0. 066
Dec . 0.381
1933
Jan ..... 0.745
Feb .... 0. 100
Mar . 0.280
Apr .. 0. 021
May .. 0.155
June ....
0
July ....
0
Aug ..
0
Sept .. ..
0
Oct ... 0,158
0
Nov .
Dec . 0.577
1934
Jan 0 . 110
Feb .. .. 0. 455
Mar ....
0
Apr , 0.056
May . 0. 054
June ... 0.046
July ...
0
Aug ..
0
Sept 0 .079
Oct . .. 0.084
Nov 0. 434
Dec . . 0.304
1935
Jan . 0.755
Feb 0.108
Mar . .. 0.491
Apr . 0. 486
Jan .
Feb ..
Mar . .
Apr .
May ....
Jun ..
July ...
Aug ..
S.pt
~ORAGE-RESERVOIR
DESIGN
295 .
TABLE 131. Operation Study for a Storage Reservoir (Continued )

Month
and
year
(1)
Precipi Evapo.ation, ration,

ft
ft
(2)
--_
-
(3)
--
Flow,
ocre-ft
(4)
Oemand,
acre-ft
Precipi- Evopo'otion, ration,

acre-ft acre-ft
StorReage
lease,
change,
acre-ft
acre -ft
Starage,
acre-ft
Area,
acr
(5)
(6)
(7)
(8)
(9)
(10)
(11)
- 540
- 560
-557
- 491
-439
- 284
-149
- 60
6611
6051
5494
5003
4564
4280
4131
4071
250
235
221
208
197
190
186
184
422
5007
4493
9500
194
315
1935
0
0
0
0.025
0.019
0.065
0 .037
0.288
0.390
0.494
0.515
0.447
0.360
0.244
0.140
0.068
23
8
1
0
0
0
0
8
440
440
440
400
370
250
130
100
0
0
0
5
4
13
7
53
100
120
117
96
47
26
13
23
8
1
0
0
0
0
8
Jan . 0.22S
Feb . 0.515
0.074
0.080
533
8970
100
130
43
131
14
20
40
40t
May ....
June
July .
Aug .
Sept ....
Oct .
Nov
Dec
73
1936
Plus 204 acre-ft over spillway.

Plus 3904 acre-ft over spillway.
is assumed. It i further assumed that as long as there is water in the

r servoir the demand will be met in fulL
The computations are nothing more than an application of the storage
equation. Precipitation and evaporation in acre-feet ar computed by
mUltiplying the data of Cols. 2 and 3 by the average reservoir area for
the month. This requires an estimate of the area at the end of the month,
which can u 'ually be obtained with sufficient accuracy by a trial solution
of the flow balance for the month. Area values are obtained from a curve
of storage vs. reservoir area (Fig. 13-1). The. torage change ( 01. 9) is
the sum of Cols. 4 and 6 less Cols. 5, 7, and 8. The storage at the nd
of the month (Col. 10) is the storage at the end of the preceding month
plus the storage change during the month.
1'hi study shows that the re ervoir could supply the estimated demand
and till have a re erve of 007 acre-ft at the lowe t point (Dec. 31, 1934).
If this reserve were utilized during the 3 yr when the reservoir was constantly below full capacity, the annual yield coulc! be increa ed by about
202 acre-ft. If the reservoir capacity were increased to 9700 acre-ft
an additional 200 acre-ft could have been conserved from the spill during
ince the study was based on an annual demand of
February, 1932.
3190 acre-ft, it appears that the maximum pos ible yield during this
period (with 9700 acre-ft capacity) is about 3190 + 202 + 67 = ~459
acre-ft/ yr. Practically speaking a figure of 3400 acre-ftj yr is probably
more consistent with the accuracy of the data.
296
The period analyzed in Table 13-1 repre ents the most severe dry spell
in approximately 50 yr. Hence the reservoir yield in most other years
would be larger than could be obtained in the tudy period. Occasional
deficiencies in water upply for irrigation projects are permi ible, and a.
study such as that illustrated by Table 13-1 should be extended over other
10
-I
Spillway elev.
1/
G;
~
~
o
~ 6
.5
'E""
o
Q.
'S 3
~
CI)
II>
CI)
II::
/
../
50
100
150
200
250
300
350
Reservoir surface area in acres

FIG. 13- 1. Ex.ample of relation between reservoir capacity and water-surface area.
,'ears to e tablish a reasonabl yield value. For the project under di cussion one might e timate uch a yield to be about 3400/ 0.8 = 4250 acreft/yr if a 20 per cent deficiency is tolerable in an unusually dry period.
rote that the po sibility of a drought more severe than any shown in the
record always exists. In the absence of any better data, the most
critical period of the record is commonly accepted as the basis of design.
Deficiencies resulting from the very infrequent droughts have little
influence on the conomic analy is of a project.
Operation studies involve con iderable arithmetic and may have to be
repeat d several times.
ard-opcrated digital computers are finding
SPILL WA Y DESIGN
297
more frequ nt use for such purpo e. Allowance for loss of torage
('apacity through edimentation, progre ive growth of demand, and
many other variable should be included in the analy is of major proj cts.
13-2. Spillway de ign. The flood hydr graph or p ak di charg
finally adopted as the ba i for design is call d the design flood.
orne
project may require everal design floods. For eXl1mpl ,th spill WilY
WeI basin
Dry basin
o
52
~ 200
o
l;l
i:5
(el
.s::
36
40
44
48
Time in hours
FIG. 13-2. Observed and synthesized hydrographs for the Neosho River at Council Grove.
Kansas.
de ign flood may be a far rarer event than the r servoir d ign flood
(Sec. 13-3). A design flood is based on analysi of the hydrologic
potentialities of the basin in th light of economic and other practical
considerations.
To illustrate the hydrologi a peets of spillway design, let us as ume
that a major flood-control reservoir is to be constructed on the Neosho
River just upstream of ouncil Grove, Kansas. For simplicity it is
as umed that the discharge record at ouncil Grove (250 sq mi) is
applicable to the dam site. Possible loss of life in case of a dam failure
dictates a spillway design flood well beyond that obtainable through
frequency analysis (Chap. 11).
The disch:lrge record at ouncil rove began in 1939, and the maximum observed discharge prior to July, 1951, was 69,500 cfa (October,
298
1941). Flood marks indicate that the 1903 flood reached a stage of
37.3 ft, or about 0.2 ft higher than the 1941 flood. A peak of 121,000
cfs occurred at 8:30 A.M., July 11, 1951, and just 24 hr later a econd
distinct peak of 71,100 cfs occurred. 1 The first of these two floods and
t he 3-hr unit hydrograph derived therefrom are shown in Fig. 13-2a.
The derived unit hydrograph fits the July 12 flood reasonably well, but
it is more peaked than one derived from lesser storms of record (Sec. 9-2).
The July, 1951, storm which produced the maximum observed discharge at Council Grove was not nearly as severe as the maximum
observed storm in that region. On Sept. 12, 1926, a storm centered at
Neosho Falls, 60 mi to the sout heast, deposited an average of 11.4 in.
of rain over 200 sq mi in 6 hr. The depth-area-duration data for tbis
storm and the one at Council Grove in July, 1951, are given 2 in Table
13-2. The 9-hr, 250-sq mi rain for the 1926 storm is about 11.2 in., and
TABLE 13-2. Depth-Area-Duration Data for Major Storms Centered at
Neosho Falls (September, 1926) and Council Grove, Kansas (July, 1951)
(From U.S. Corps of Engineers)
Roinfall depth, In.
Area,
sq mi
Point
10
100
200
500
Neosho Falll, Sept., 1926
Council Grove, July, 1951
6 hr
12 II,
1811r
6 II,
13 .6
13 .4
12 .2
11.4
9.5
13 .8
13.7
12.5
11.7
10 .0
14.0
13.9
12.7
11.9
10 .2
5.8
5.3
4.7
4.6
4.3
12 hr
18 h,
--
--
7.5
7 .0
6.4
6.2
5.8
8.2
7.9
7.4
7 .2
6.7
the time distribution of the storm was such that 3-hr increments of 1.5,
9.5, and 0.2 in. appear reasonable. From a runoff relation (similar to
Fig. 8-8), this rainfall would yield runoff increments of 0.4,8.5, and 0.2 in.
if the basin were initially very wet and 0.2, 6.0, and 0.0 in. if the ba in
were initially very dry. Application of the unit hydrograph derived
from the 1951 storm to these runoff incr ments results in the synthe ized
hydrograph shown in Fig. 13-2b.
From a meteorological viewpoint, there is overy reason to expect that
the 1926 storm is far short of the maximum which can occur over the
1 Kansas-Missouri Floods of July IOSI, U.S. Geol. Survey Water-8upply Paper 1139,
p.l 0,1952.
I
torm Rl1infall in the United States, U.S. Corps of E1t(Jineers (processed).
SPILLWAY DESIGN
299
problem basin (Sec. 11-12). From Fig. 11-1 the probable maximum
24-hr, 200-sq mi rainfall i about 24.5 in. Interpolation from Tabl 11-7
gives 250-sq mi values of 19.5, 22.0, and 24.0 in. for durations of 6, 12,
and 24 hr, resp tively. From the depth-duration urve defin d by
these three value it is e timatcd that the probable maximum rainfall
for the problem ba in is 17.0,19.5, and 21.0 in. for duration of 3,6, and 9
hr, respectively. Assuming that ucce sive differences of the e three
values constitute reasonable 3-hr increments and that the larg st is
most apt to occur during the second period gives a chronological equence
of 2.5, 17.0, and 1.5 in.
tonns approaching this magnitude can occur
from June through September.! From an applicable runoff relation it is
found that such a storm occurring on an initially wet basin in June or
July would generate runoff increments of 1.5, 16.0, and 1.2. in.
hould
it occur in August or September with the basin initiltlly dry, the corresponding runoff increments would be 0.2, 14.0, and ],1 in. Applying
the unit hydrograph provides the two flood hydrographs shown in
Fig. 13-2c.
The ob erved and synthe ized hydrograpbs arc indicative of flow
potentialities under natural conditions. 1 thc re ervoir capa'ity and
plan of operation are such that th water could be tored at time of
peak discharge, the hydrograph should be modified by routing ( ec.
10-5). Any anticipated flow through sluiceways should also be subtracted from the spillway flow.
Having determined the hydrologic potentialities of the problem basin,
the designer must evaluate the economic and practical aspects of several
a sumed design values. If the consequence of failure are sufficiently
severe, he may be justified in designing for the probable maximum flood
(450,000 cf in this case). If a Jes er pill way design flood is adopted, it
must be done with the full realization that some risk is being accepted.
Just what degree of risk is involved cannot be determined for
such extreme events as the Neo ho Falls storm. Even the July, 1951,
flood is well beyond the limits of reliable frequency analysis. Although
quite meaningless, straight-line extrapolation of a regional frequency
relation 2 to 1000 yr yield a flow of only 40,000 cis, a value which has
been exceeded on four occasions since the turn of the century. The
1000-yr flood may be well above the indicated 40,000 cfs and could be
as great as the 1951 flood, but the 235,000-cfs flow computed from the
Neosho Falls storm is an extremely remote event and is perhaps less
1 Seasonal Variation of the Probable Maximum Precipitation East of the 105th
Meridian for Areas from 10 to 1000 Square Miles and Durations of 6, 12, 24, and 48
Bours, U.S. Weather Bur. Hydrometeorol. Rept. 33, 1956.
I Tate Dalrymple, Flood-Frequency, Kansas-Missouri Floods of July 1951, U.S.
Geol. Survell Water-8upply Paper 1139, pp. 225-229, 1952.
300
likely to occur than other types of catastrophe facing the populace of

the area. If the co t of providing spillway capacity for the large t flood
is about the same as for Ie ser floods, the deci ion i ' simplified. In
some cases a portion of the flood flow may be as. umed to pa s over the
top of the dam with negligible damage. While called a pillway design
flood, the basic criterion is that this flow safely pass the dam by any
feasible means, emergency or otherwise.
13-3. Flood-control-re ervoir design. The primary purpose of a
flood-control reservoir is to reduce downstream flood flows to a level
which will not cause damage. For major projects the project design
flood is usually about equal to the worst flood known to hav occurred
in the general region of the project. For small projects the de ign flood
may be selected on the basis of frequency analysis. In either case,
economic and social factor control the final decision. 1
After the design flood has been selected and the permissible flow in
the reach to be protected has been determined, the safe release is equal
to the desired controlled flow less any local inflow between the dam and
the protected area. If the local inflow is small, the computation is
quite simple. In Fig. 13-3 the Neosho Falls storm, u ing wet antecedent
conditions, is taken as the project de ign flood (Fig. 13-2b). The minimum safe reservoir capacity i the difference in volume between the
design inflow and the permi sible outflow (area ABC). If the loefll
inflow is large, the flows mu t be rout d to the prot cted area, u ing
several possible plans of control until the most satisfactory plan is
determined.
The operation illustrated in Fig. 13-3 assumes that a full knowledge of
the forthcoming flood is available to the reservoir operator, for only with
a preci e knowledge of the inflow 'ould the operation hown be performed
successfully. In practice, then, an additional storage volume must be
allowed as a safety factor. It is po sible also that a e 'ond flood co uld
occur while the reservoir is filled from the first flood (Sec. 13-2) . Hence,
additional capacity must be kept in reserve against this contingency.
In determining the necessary reserve, the design flood may be a sumed to
consist of two events in series based on some case of record, or meteorological considerations may be used to determine the probable interval
between floods and the magnitude of the econd event. In an analysis
of three floods at Chattanooga, Tennes ee, Rutter2 concluded that the
TVA reservoirs were only about 50 per cent effective.
After a tentative design of the reservoir has been elected, its over-all
1 R. K. Linsley and J. B. Franzini, "Elements of Hydraulic Engine ring," pp. 336338,442- 445, 525-527, McGraw-Hill, New York, 1955.
IE. J . Rutter, Flood-control Operations of Tenness e Valley Authority Reservoirs,
Trans. ASCE. Vol. 116, pp. 671- 707, 1951.
301
FLOODCONTROL RESERVOIR DESIGN
effo t on the flood regime of Lhe basin mu t b cal ulated to e tablish the
economic benefits of the proje t. Ba i ally, thi require that the frequency curves of flood stage with and without the re ervoir be estab1i hed. If the damage re 'ulting from vat'iou 'stage i known, the averag
annual damage can be calculated by multiplying the damage r ulting
from a given stage by the probability of Lbe stage oc uffing in any year
( hap. 11). The difference in annual damag b fore and after the
250r--.---.---r--'---~--.--.--~---r~
Project design f lood

( computed hydrogroph
from Neosho Foils
storm of 1926, Fig. 13- 2)
~
u
150r---r---r--~~~~--r-~--~r--.---.--~
!2
.S:
OL_~
__- L__-L__
__
__
L-~
__- L__-L~
8
10
12
14
16
18 20
Time in hours
FI G. 133. Ideal o perating pla n for fl ood control, Neosho River at Council Grove, Ka nsas.
project represents the flood- control benefits. Presumably the frequency

curve of tag . before the projet can be d termined by conventional
frequency analysis of actual records. The frequency curve which may
be expected after project construction j usually determined by routing
several of the major floods of record through the re ervoir in accordance
with the accepted operating plan. The . ta:ge (or flows) thus computed
provide the data for plotting a second frequen r::y curve showing the
effect of the reser oir (Fig. 13-4). I
Flood-control re ervoits are commonly con. t l'ucted in conjunction
with other works such n. ch~LIlflel improvement and levees. The optimum combination of methods (in lulling zoning and flood forecasting) is
1 Note that Fig. 13-4 is for a hypothetical reservoir and does not represent conditions at Council Grove.
302
that for which the ratio of benefits to co ts i the greate t.t The methods
of reservoir design in such combination solutions are no different from
those for the isolated reservoir, but numerous trials must be made of the
possible combinations to determine the optimum plan.
13-4. Storm-drain de ign. No completely sati factory method of
storm-drain design ha bc n demonstrated as yet. Actual ob ervations
of storm flows within a city provide the be t ba is for design of additional
250
200
FreqlJency clJrve,
nollJral conditions
....u'"
150
Ob~erved flood P c : f ;;I

I
.5
E?'
"&
...
:0
0
100
If
lA f('"
/.0
Safe release_......
No attempt to
reglJlate below
",.
flow
/c
50
,I
./
1-
/~
I
I
II
I
1
I
/-r,-r
-1
"
IlomplJted peaks, :
with reglJlation ~;....- ....
11
-t.
/
/'
~L
,
FreqlJency ClJrve,
reglJlated cond/~ions
V
1.5
8 10
20
30
50
100
200
Return period in years
FIG. 13-4. Effect of
flood -control reservoir on flood frequency.
elements of the torm-drain system or for new systems in nearby locations.

Unfortunately, suoh data are generally available only for the major
cities. The method which follows is felt to be reasonably sound from a
hydrologic viewpoint but has not been clearly demonstrated on observed
data.
Becau e of the lack of runoff data from small water h ds and urban
areas, design of a storm-drain system must usually begin with an analysis
of rainfall frequency. 2 Using the nearest rainfall-intensity record, a
conventional frequency analysis of maxinlum rainfalls is made by the
1 R. K. Linsley and J. B. Franzini, "Elements of Hydraulic Engineering," Chap. 20,
I Rainfall Intensities for Local Drainage Design in the United States, U.S. lV eather
Bur. Tech. Paper 24, 1953.
STORM-DRAIN DESIGN
303
methods outlined in Chap. 11. The resulting relations are th n usually

transformed to curves showing the relation between aver ge int nsity
for various durations and fr quencies (Fig. 13-5) . Equation (11-14) i
helpful in fitting the curves, e pecially if th inten ity r ord i hort.
The mo t readily available data ar th hourly rainfall amounts publish d
by the . , Weather Bureau. The maximum clock-hour valu publi h d
10
6
"
....................
..................... ::--......
4 ...............
" , ............. t:'---
~ ~
"
.r::.
"
I'-. ~ ....
~ I'--.
~ r-.....
~
....
0
r--.: t-~
["..
lil
a.
r-.- i':--
II>
Q>
.c.
u
,!: 1.0
.!: 0.6
?-
'in
Return period,
yeors
~ t::::: ,
........ ::::
~ ~t::::
.......
"
" ............ ".......
0,6
0,4
........
............ ............
.............. ............
.......................... .............
'
"
............... .......
0.2
0.1
6 10
40
20
Duration in minutes
60 60 tOO
.....................
"
200 300
FIG. 13-5. Rainfall-intensity-d uration- frequency curves for Redwood Creek, California.
may not be the maximum GO min during the storm. On an average, the
maximum GO min is about 14 per cent greater than the maximum clock
hour. If the area under study i small, the point-rainfall values established as de cribed above are adequate, but for large areas the values
should be reduced (Fig, 11-14).
Because of the great influence of surface detention on the runoff from
small areas, t. as deftned by Eq. (9-1 ) seems to be a good definition of
the critical duration of rainfall. Combining Eqs. (9-18), (9-19), (9-21),
and (9-22), ignoring the intensity term in Eq. (9-22), and expressing the
supply rate as Ci, where C is the runoff coefficient (ratio of total runoff to
total rainfall) for the area,
41cLo
(13-1)
t. = 8oJ.il(Ci)*
where c is the retardance coefficient (Table 9-2), Lo is the distance of

overland low in feet, 8 0 is the average ground slope, and i is the rainfall
intensity in inches per hour. This equation can be solved by trial for a
combination of i and t which conform to the curves of Fig. 13-5 for the
design frequency. If Eq. (11-14) is applicable, Eq. (13-1) can be further
modified by substituting for i
tn = (41CL~l4)1 _~b/3
so)~a 3i
(13-2)
where a = CkT/', the product. of the numerator of Eq. (11-14) and the
runoff coefficient for th area, and b is the exponent of duration in the
denominator of Eq. (11-14).
Using mean values of So and Lo for the basin, and appropriate values of
c, a, and b, a value of til and the corresponding rainfall intensity i can
be determined. If equilibrium conditions were attained throughout the
entire basin simultaneously and continued until all parts of the basin
were contributing to outflow, the p ak flow would be given by the rational
formula qp = CiA [Eq. (9-26)]. However, Eq. (13-2) only provides
time for each individual portion of the area to attain equilibrium, and an
allowance for travel time and storage en route to the outlet must be
includ d. It i convenient to u e a lag-and-route procedure in which the
various subareas of the ba in are lagged to the outlet and then routed
through a hypothetical reservoir at the outlet to simulate ba in 'torage.
The determination of the proper values of lag time and . torag i the
uncertain tep in the procedure. It seems rea 'onable to a ume that the
lag time is the flow time from the subarea to the outlet as computed from
the velocity in the channels. Use of the Mu kingum method for routing
with K = lag and x = 0 appears to be a 'onservative assumption.
The procedure is illustrated by the following example. Figure 13-6
shows a portion of the drainage area of Redwood Creek, including some
of t he residential area through which it flows. The probable 30-yr flood
peaks at Stations 1, 2, and 3 are needed for the design of the channel
and channel crossings. Random sampling on a topographic map gives
average values of Lo and So above Station 1 a 200 ft and 0.1, respectively.
For a 30-yr return period, the equation of the rainfall-intensity- duration relation (Fig. 13-5) is i = 7.33/t~~, where t is the duration in minutes.
A. runoff coefficient C = 0.4 and a retardance coefficient c = 0.05 are
assumed. Hence a = 0.4 X 7.33 = 2.93 and b = 0.5. Introducing
these values in Eq. (13-2) gives
t
B
(41 O.pi
X 0.05 X 200l4) ~ = 44 .
X 2.93~
mm
STORM-DRAIN DESIGN
305
For convenience, a duration of 45 min i u ed for whieh the ra.infa.ll

ir.tensity is 1.1 in, / hr,
Using Manning's formula and eros ection and slopes of the channel,
the flow time from the extreme headwater of the area to tation 1 is
computed as 40 min, The ba e of the lagged rainfall hydrograph will
FIG, 13-6, Mop of Redwood Creek droinoge oreo, California,
therefore be 45 + 40 = 85 min, It is assumed that the lagged hydrograph will be an iso celes triangle if the travel time is greater than t or a
trapezoid if the lag is less than t, Hence, in this case the lagged hydrograph is a trapezoid with a base of 85 min and a top width of 5 min,
The total volume of flow (from 1220 acres) is
Q = 0,4 X 1.1 X
4%0
X 1220 = 402 acre-in,
I t is convenient to use this flow in second-foot-minutes so that the peak

flow is in cubic feet per second, Thus
Q = 402 X 0,5 X
712
X 1440 = 24,150 sfro
306
The height of the trapezoid is therefore computed from

24150
= 85 2+ 5 qp
,
from which qp = 537 cfs. The lagged rainfall hydrograph is tabulated
at 5-roin intervals in Col. 2 of Table 13-3. Columns 3, 4, and 5 are the
routing computations using K = 40 min. The computed hydrograph
for Station 1 is given in Col. 6. The hydrograph at Station 2 computed
in the same manner is shown in Col. 7. Columns 8 to 12 contain the
computations for routing the sum of flows at Sta tions 1 and 2 to tation 3
by the Muskingum method, using K = 6 min, x = O.
TABLE 13-3. Hydrograph Computations for Redwood Creek
,;
:0"
:0
..,..,0
...
ti
...
en
e+
:>
c/)-
0:"
N
0
0:
N
0
(7)
(8)
(9)
-(10)
ci
(3)
(4)
(5)
0
5
10
15
20
25
30
35
40
45
50
55
60
65
70
75
80
85
90
0
67
134
201
268
335
402
470
537
537
470
402
335
268
201
134
67
0
0
0
4
8
12
16
20
24
28
32
32
28
24
20
16
12
8
4
0
0
0
0
4
8
12
16
20
24
28
32
32
28
24
20
16
12
8
4
0
0
0
4
14
30
51
77
106
139
175
210
238
255
263
263
256
243
224
201
0
0
0
0
0
0
0
4
1
0
4
1
0
0
1
6
16
1 17
5
0
34
6 40 12
5
3 20
58 16 74 21 12
8 41
87 32 119 35 21 17 73
121 50 171 50 35 31 116
158 67 225 65 50 49 164
199 86 285 83 65 69 217
239 100 339 98 83 91 272
270 99 369 107 98 114 319
290 86 376 109 107 134 350
299 66 365 106 109 147 362
299 50 349 101 106 152 359
291 38 329 95 101 151 347
276 24 300 87 95 146 328
255 19 274 79 87 137 303
228 15 243 70 79 127 276
201 11 212 61 70 116 247
(6)
:0
(2)
en
"] ~
...0 :00 ..,00 0 ~Q:
0
0
-(11) -(12) -(13) -(14) -(15) -(16) -.17) -(18)
0
(1)
--
..
E .~
1= E
0
86
120
120
120
120
120
120
120
120
86
0
0
0
0
0
0
0
0
0
22
31
31
31
31
31
31
31
31
22
0
0
0
0
0
0
0
0
0
0
22
31
31
31
31
31
31
31
31
22
0
0
0
0
0
0
0
0
0
0
0 22 23
64 70
11
31
93 113
45 107 148
51 113 186
54 116 232
56 118 282
57 119 336
57 119 391
57 110 429
75 425
53
36 398
36
17 376
17
8 355
8
4 332
4
2
2 305
1 277
1
0
0 247
The urban area below Stations 1 and 2 is 240 acres. The flow time
through the street drains is computed from street lengths, slopes, and
gutter dimensions as 4 min. Travel time in the main channel from
Stations 1 and 2 to Station 3 is taken as 6 min. The lag is the sum
of the 4-roin time in the gutters and one-half the time in the channel, or
4 + 3 = 7. Thu the trapezoidal inflow graph has a base length of
52 min and a peak of 120 cfs. The routing of thi flow is shown in ols.
13 to 17. The total flow at Station 3 (Col. 18) is the sum of ols. 12
and 17.
RIVER FORECASTING
307
13-5. River foreca ting. The pr viou se tion of thi chupt r

treat problems of hydrologi' de ign. Equally important, and in many
re pects more tangible, i the role of hydrology in the op ration of wat rcontrol projects. In de ign, the hydrologi t i mo oft n l' quir d to
e timate the magnitude of extl' me events, wher as op ration i often
dependent on reliable estimate of flow for an ensuing p riod of hours,
days, or possibly longer. The time sequence of flow is an importn.nt
aspe.ct of the forecast. Each forecast should attempt to extrn.po\ate the
hydrograph as far in advance as operational planning requires, and
provision must be made for revision dUl'ing rapidly changing meterological onditions.
In addition to the role of for cast in planning water-control operations,
flood foreca ts constitute a direct means for the l' duction of flood damn.ge
and loss of life. Advance warning of an approaching flood p rmits
evacuation of people, livestock, and equipment with no los ex pt for
the cost of removal. The relatively low ratio of cost to b nefit for a
flood forecast and warning service makes it an ideal flood-protection
measure in many areas where physical means cannot be economically
justified. The soundest approach to the flood problem lies in a planned
combination of water-control structures, flood-plain zoning, adequate
forecasting, and possibly insurance.
The formulation of a river forecast requires reliable information on
current hydrologic conditions over the drainage basin, augmented by
weather reports and forecasts. The necessary data- precipitation,
river stage, water equivalent of snow pack, re ervoir storage, temperature,
depth of frost, etc.-are assembled by telephone, telegraph, and/ or radio
from an organized network of stations reporting to the forecast office on
a regular or occasional basis in accordance with specific instructions. In
this connection increasing use is being made of ro.dar for evaluating areal
rainfall.! Imminent modifications in present-day radar will greatly
increase its utility for this purpose.
The tools of the river for caster are the same as those used in design:
runoff relations, unit hydrographs, routing relations, gage relations, etc.
Because of the need to formulate the forecast in a minimum of time,
however, far more effort must be expended in the development phase to
streamline the procedures. The technique u ed in a particular case
1 S. G. Bigler and R. D. Tarble, Applications of Radar Weather Observations to
Hydrology, Final Rept. under Weather Bur. Contract Cwb-9090, Texas A & M College,
College Station, Tex., November, 1957.
H. W. Hiser, L. F. Conover,and H. V. Senn, Investigation of Rainfall Measurements
by Radar, Final Rept. under Weather Bur. Contract Cwb-9012, University of Miami,
Miami, Fla., June, 1957.
G. E. Stout and F. A. Huff, Rea arch Study on Intensity of urface Precipitation
Using Radar Instrumentation, Final Rept. under Contract DA-36-039 80-42446, Illinois
State Water Survey, Urbana, m., July, 1955.
~O8
APPLICATION Of HYDROLOGIC TECHNIQUES
45
40
35
$
30
0
0
Q 25
.5
e' 20
0
.c
u
on
15
10
5
0
./'
12
13
14
17
16
15
18
April 1956
FIG. 13-7. Storm rainfall map and predicted consequent hydrographs.
depend to a large extent on the forecasts needed and thc hydrologic

characteri tics of the basin. A forecast of crest height and time of
occurrence will usually suffice as a warning in small, headwater areas.
The rapid rate of rise and faU makes the duration above flood stage so
short that there is little value in forecasting the entire hydrograph. In
the lower reaches of large rivers where rates of rise are slow, it is important
to forecast the time when various critical stages will be reached during
the ri e and fall. Efficient operation of flood-control reservoirs requires
foreca ts of the complete hydrograpb.
TABLE 13-4. Computation of Storm Runoff for Rainfall Data Plotted in Fig.
13-7, Using the Runoff Relation of Fig. 8-8
A bove Riverton
Tobular entry
Riverton to Greendver
7 P.M., Apr. 12 7 A.M., Apr. 13 7 P.M .. Apr. 12 7 A.M., Apr. 13
Antecedent.precipltation index
Week number ............... ,
Storm duration, IIr .............
Storm rainfall, In ..
Storm runoff, in .
Previoul runoff. in ............
Runoff Increment, In
1.05
15
12
1.32
0.55
0
0.55
1.05
15
24
3.00
1.60
0.55
1.05
0.95
15
12
1.98
0.85
0
0.85
0.95
15
24
2.78
1.40
0.85
0.55
RIVER FORECASTING
309
To illu trate the manner in which forecasts of a hydrograph are fOl'mulated for a series of point over a river ba in, let us as ume that the area
mapped in Fig. 13-7 repr sent the upper renche of one of s v ral
tributary areas. The forecast are to be prepared a of 7 :00 A.M.,
TABLE 13-5. Runoff Distribution and Routing Computations
Discharge In 1000 cfs
River
station
or
Tabular entry
Apr. 12
Apr. 13
Apr. 14
Apr. 15
Apr. 16
orea
7
A.M.
,__
..IS
t:
>
ii2
7
P.M.
7
A.M.
7
P.M.
7
A.M.
.2 >~
.~
Ii
~
...
...
"
....
....
"
7
P.M.
7
A.M.
7
P.M.
-- - - - - ---- - - -0.5
0
------ ------ -- --...- -- - -- - - - -- ---- -0.5- 0
... .... - - - - ---- -- -1.0- -1.2-...- ---3.2 2.6
--- - - -- -- ---.(.7 3.8
-- -- ----... ---- - - - - ---0
0
---- - - ------ - - -0- -0. .. . .. -....- - -- - - -- -0.5- -0 .(---1.5 1.3
-- - - - - - -- ---- -2.0
- -----1.3 1.7
- - -- - ------.. -- -- - ----- 1.1
2.7
...
3.5
--_ - - - - ---- -1.2 0.9
...
- -.( .-- - - ------ -6.0
-----7
-- - - -------- - - ---2.0 1.7
-- ------ - - - - -- -8.0- -6 .-.(
----
.~
c
7
A.M.
.(.8 20.5 16.0 7.5 3.5 1.6

12-hr unit hydragraph . ...
0
-12hr storm runoff
(1.05 in.) ..
0
5.0 21.5 16.8 7.9 3.7 1.7
12hr groundwater
runoff ..
0
0.2 0 .( 0.6 0.8
Previous forecast
(7 P.M., Apr. 12) .. 2.8 5.0 12.3 11.3 7.1
5.0 ,(.2
3.6
---Total .. .... 2.8 5.0 17.3 32 . 8 2,(.1 13.3 8 . 5 6.1
2.0 9.8 6.8 3.6 1.3 0.3
12hr unit hydrograph .
0
12hr storm runoff (0.55
5 .( 3.7 2.0 0.7 0.2
1.1
In.) ........ ...
0
12hr graundwater
0 . 1 0.2 0.3 0 . .(
runoff . ..........
0
-Previaus forecast
(7 P.M., Apr. 12) . . 1.5 3.0
9.0 7.0 4.5 2.8 2.0 1.7
-Total. .. ...... 1.5 3 . 0 10.1 12.,( 8.3 5.0 3 . 0 2.3
1.,(
,
Riverton
4.9 9.3 6.9 3.8 2.4 1.7
0.285/ 2
1.6
routing
0.572/, ..
2.9 9.9 18.8 13.8 7.6 4.9
1.9
0.5
3.1
K = 10 hr 0.1430,
0.5 1.2 2.9 4.1
8.5
8.3 20 .( 28.6 21.7 13.1
3.3 3.5
X = 0.2
02
local runoff ..... 1.5 3.0 10 . 1 12.4 8.3 5.0 3.0 2.3
Total ..... 4 .8 6.5 18 .( 32.8 36.9 25.7 16.1 10 . 8
7
P.M.
_-
April 13, following a 24-hr storm with precipitation as hown. Although

it is assumed that a forecast had been made at 7 :00 P.M. on April 12, the
runoff computations for both 12-hr periods are shown in Table 13-4. In
this illustration, average runoff over each of the area has been computed
by entering Fig. 8-8 with areal average precipitation, but it is more
common to compute the runoff at each rainfall station and average these
values over the area. Unit-hydrograph and routing computations are
shown in Table 13-5, and the predicted hydrographs are plotted in Fig.
310
13-7. The computations have been based on the assumption that the
groundwater flow pel1ks I1t the cessation of storm runoff. The magnitude
of the groundwater peak j estimated subjectively. Adding th groundwater flow and direct runoff to the previous forecast provides the required
forecast for Riverton.
Local inflow from the area between Riverton and Greenriver is computed in a similar fashion. The Greenriver forecast hydrograph is
obtained by routing the Riverton hydrograph (Muskingum method,
]( = 10 hr, x = 0.2) and adding the expected flow from the local area.
In prn.ctice, stages to 7:00 A.M. of April 13 would be available, and the
forecast hycu:ograph would be smoothed into the observed data. For
flood-warning purposes the computed flows would be converted to stages
by application of stage-discharge relations.
Although simplified, the example shows how forecasts may be prepared from rainfall reports for numerous points over a river basin. The
forecasts are made in sequence proceeding downstream, point by point,
routing upstream flows and adding predicted local inflow. While this
compounding of forecasts might seem insecw'e, the dampening effect of
storage, error compensation, and the curvature of the tage-discharge
relation all combine to produc downstream stage forecasts that are
generally more accurate than the headwater forecasts from which they
are derived.
BIDLJOGRAPIlY
Davis, C. V. (ed.): "Handbook of Applied Hydrauli 's," 2d ed ., eo . 7, McGraw-Hill,
New York, 1952.
Hathaway, G. A.: Design of Drainage Facilities, Trans. AS E, Vol. 11 0, pp. 697- 848,
1945.
Jens, tifel W.: Drainage of Airport Surfaces- Some Basic Design Considerations,
Tran8. ASCE, Vol. 113, pp. 785-836, 1948.
King, R. E.: Stage Predictions for Flood Control perations, Tran8. ASCE, Vol. 117,
pp. 690-704, 1952.
Linsley, R. K., and J. B. Franzini: "EI ments of Hydraulic Engineering," Chaps. 14
to 21, McGraw-Hill, New York, 1955.
Linsley, R. K, M. A. Kohler, and J . L. H. Paulhus: "Applied Hydrology," Chaps. 21
and 22, McGraw-Hill, New York, 1949.
Riesbol, H. S.: Snow Hydrology for Multiplepurpose Reservoirs, Trans. ASCE, Vol.
119, pp. 595-627, 1954.
Turner, Robert E.: Operation of the Conowingo Hydroelectric Plant, Trans . ASCE,
Vol. 114, pp. 79- 110, 1949.
Williams, G. R.: Hydrology, Chap. IV, in Hunter Rouse (ed.) "Engineering Hydraulics," Wiley, New York, 1950.
PI{OBLEM
Instead of text problems, it is sugg 'sted that the instructor assign a typical design
problem requiring the student to refer to tho original data sources to obtain his data
and carry through an analysis following one of the cases described in tbis chapter.
APPENDIX A
GRAPHICAL CORRELATION
Hydrology is largely an empirical 'ience, and a prepondera nce of the

problems confronting the hydrologist involve corr lation analy is or the
application of a relation derived through such analysis. While the
student may be acquainted with the lea t-squares technique t of curv
fitting, it is believed unlikely that he has received instruction in the
grapbical methods now so widely us d in hydrology. For this rea on
the following brief discussion of graphical correlation has been in corporated into the text.
A-I. Two-variable correlation. If a linear relation is to be used,
the line of best fit must pass through the point deflncd by the mean
(X and Y) of the observed values of the two variabl X and Y. This
is true not only for graphical correlation but also for the lea t-squares
line. WHh one point (X, Y) on the line determined, the proper slope
can be estimated by first plotting tbe data (Fig. A-I) and then determining average coordinates for groups of points. The selected group should
comprise all points falling within specified values of th independent
variable (X) irrespective of the value of Y, the factor to be estimated.
If the total number of points be divided into two groups of approximately
equal size, then the line connecting their means will pass through the
mean of al1 points, and this is the best line that can be readily determined
graphically. The relation determined by group alJerages usually has a
'lightly steeper slope (dy / dx) than that determined by least squares.
As the d gree of corr lation increasos, the difference between the two
lines diminishes, and for perfect correlation th y nre oincident. The
relation of averages tends to minimize the absolute sum of the deviations,
while the least- quares relation minimizes the sum of the squares of the
deviations.
The group averages can be e tirnated graphically by fir t estimating
Ruccessive two-point averages (halfway between the plotted points).
M. Ezekiel, "Methods of Correlation Analysis," 2d ed., Wiley, New York, 1941.
G. W. Snedecor, " tatistical Methods," 4th ed., Iowa tate College Press. Ames,
Iowa. 1946.
1
311
312
)...
.;
;;;
.2
~
.... 3
..
-g
01 0
I,0
0-
c!
Mean location I
of points X<X !..
O~
__
____
____-L____
~_J
__
____
__
~~
__
4
3
5
Independent variable, X
FIG. A-I. Two-variable linear correlation by graphical method.
4::....
>
+c
'"c
'"c.
.i
.~.... 3-
.0
-0
"
)(
" $"
'"
"-
I
2
"
>c
Cl
"
t::. "
)(
"6
2-
I -
I
3
legend
Plotted point
" 2 - point mean
x 4 - point mean
t:. 8 - point mean
e 16- point mean
I
4
Independent variable,
I
5
FIG. A-2. Graphical determination of group means of points classifled with respect to the
independent factor X.
THREE VARIABLE CORRELATION
313
The four-point average are then halfway between the two-point av r

ages, etc. (Fig. A-2). The point should be grouped with re pe t to
values of the indep ndent variable.
nle s the corr h"tion is perfect,
a different line will result if points are grouped according to the depend nt
variable, with th) ditTerenc in'r a ing as the correla.tion decr a s.
d"
FIG. A3. Three-dimensional flgure for solution of the equation XI =- X,
+ 2.SX3
- 7.5.
If, upon examination of the group averages (two-point, four-pain L,

etc.), it is decided that the relation i curvilio o.r in form, then a curve can
be fitted to the mean point with the aid of a drafting curve. A curve
does not necessarily pass through the mean of all the data.
A-2. Three-variable correlation. Pel'haps the most logical method
of presenting a three-variable relation is by means of a three-dimensional
sketch (Fig. A-3). It will be noted that the contours of Xa (i.e., curves
connecting equal values of X a) are parallel straight lines and are equally
spaced for equal increments of X 3. This is true of any plane surface.
GRAPHICAL CORRfLAnON
31 ..
Given the values of any two of the variables, the third can be estimated
from the chart of Fig. A-3. A clo e examination of the figure will
disclose the fact that, once the thr e families of lines have been constructed on the surface ABeD r pre ented by the equation, all remaining
portions of the sketch become superfluous. In practice, the surface
ABeD is projected onto one of the coordinate planes and shown by a
family of curves on cross- ection paper.
50
40
30
,
29
r027
I!!l
027
.
029
l-
021
019
011
012
10
09
08~
00
00
,.9
.41
@)
10
20
06
oil
09
013'
'0 5
oS
03
.0
0'
04
01
1m "
00
Legend
Mean pOint,OSX2 S9
.&
,&.7
02
.1
1.14
olS
.r&
01 7
1Iru 022
.14
.24
on
.Z8
1"21
.29
21
1.'1.
r-
Y
20
,10SX2. S19
"
"
,20 sX S 29
2. I
30
XI
40
50
60
FIG. A ... Graphical determination of group means of points classified with respect to twe,
independent variables (X l and X2).
A chart similar to ABeD of Fig. A-3 can be developed from a eri s of

imultaneous ob ervations of three related variables. By plotting value
of Y vs. Xl and bbeling the points with corresponding values of X 2,
contours of X 2 can be fitted in much the ame manner as elevation contours are drawn in the preparation of a topographic map. In corr lation
analysis, however, we are attempting to derive a logical relationship
between three variables from observations subject to random error.
Moreover, the true relationship frequently involves additional variables.
For the era ons, no attempt is made to exactly fit each observation.
If the degr e of 'orrelation displayed by the plotted data is not particularly high, averages of points grouped with re pect to the independent
variables (Xl and X 2) should be determined before any attempt is made
to con truct X 2 contours (Fig. A-4). The location of the mean points
is determined graphically, and the label is the average of the X 2 values
for the group.
THREE-VARIA8LE CORRELATION
harp discontinuitie , in either spacing or urvature of th lines, should

not be embodied in a family of curve ba d on limited data unl
th y
can be logically explained. The omplexity of the d ri d curv fami]'
should be held to a minimum con i tent with available data and a prior.
Auxiliary
Xz
scale
FI G. A-S. Construction of smooth curve fami ly.
knowledge of the relation hip. With the general character of the curve
family decided upon, the method 1 of construction (Fig. A-5) is as follows:
1. Estimate the optimum position of the drafting curve for various
X 2 curves, noting in each case the location of the hole in the drafting
curve (or any other identifiable point) and the value of Y for some flxed
value of Xl.
.
2. Fit a smooth curve (A ) through the point defin d by the locations
of the hole in the drafting curve.
3. Replot the points used to define curve A by shifting them bori1
M. A. Kohler, "The Use of Crest Stage Relations in Forecasting the Rise and Fall
!If t he Flood Hydrol1;ra.t;:lh!' U.S. Weather Burea.u, 1944 (mimeo.).
316
zontally to the proper value of X 2 on the auxiliary scale (for example,

urves A and B
X 2 = 48 and Y = 68) and fit construction curve B.
together fix the position of the hole in the drafting curve for any de ired
X 2 curve.
4. Plot the values of Y (for Xl = 0) derived in step 1 at the corresponding values of X 2 on th auxiliary scale and fit const.ruction curve C.
This curve fixes the orientation of the drafting curve for any elected
X 2 curve.
5. Construct s I ct d X 2 curves as illustrated for X 2 = 10. The hole
in the drafting curve is placed on curve A at a value of Y corresponding
to X 2 = 10 on curve B. The drafting curve is then rotated so as to
fit the point Xl = 0 and Y corresponding to X 2 = 10 on curve C.
If the curve family were to consist of equally spaced straight parallel
lines, con truction curve C (also a straight line) would determine the
spacing, and the line for X 2 drawn through the point Xl, Y would determine the slope. If variable spacing and/ or convergence were to be
incorporated into a family of straight line , two con truction curves
analogous to curve C (for two selected value of Xl) would be required.
Any three-variable relation can be expres ed by a chart of the type
just di cus ed. For linear functions, the contour are equally spaced,
straight, and parallel; for curvilinear function, they may be cur\'ed,
unequally spaced, or both.
ontours which converg indicate a joint
function, i.e., a function in which the effect of one independent variable
is dependent upon the value of a second (for example, Y = X IX 2) '
A-3. Coaxial correlation of four or more variabl s. The development of a runoff relation by 'oaxial correlation is de ' 'ribed in Sec. 8-7.
The treatment of the method in this section is g neral in character, but
detailed plottings and tabular data are provided. Although minor
duplication between the two sections i intentional, they are largely
supporting and hould be tudied jointly. The coaxial method of graphical
correlation l is ba ed on the premise that, if any important factor is
omitted from a relation, th scatter of point in a plotting of observed
values of the dependent variable vs. values computed from the relation
will be at least partially explained by the omitted factor. In other word ,
if the points on such a plotting are labeled witb the corresponding value
of the new factor, a family of curves can be drawn to modify the values
computed from the original relation. It will be seen that a coaxial
relation is .in reality a series of three-variable relations arranged with
common axe to fa ilitate plotting and computing (Figs. 8-7 and A-6) .
For purposes of illustration, assum that we wi h to develop a relation
for predicting Y from Xl, X z, and Xs and that the available data have
been tabulated as hown in Table A-l. The analysis proceeds as follow:
1 M. A. Kohler, "The Use of Crest Stage Relations in Forecasting the Rise and Fall
of the Flood Hydrograph," U.S. Weather Bureau, 1944 (roimeo.).
COAXIAL CORRELATION OF FOUR OR MORE VARIABLES
317
1. For the entire series of ob ervat ions, plot Y again t X l and lab I
each point with the value of X 2, as hown in char t A of Fig. A-G. Fit
a family of curves to the plotted data ( ec. A-2) . I n this example it
appear d that traight, paraliel line would ad quately fit th data;
more complex functions hould be u ed only wh n clearly indi 'uted or
when they can be logically explain d.
TABLE A-1. Data for Coaxial-correlation IIlustrotion (figs. A-6 and 7)
Absolute error
Observation
No.
Y X, X. X,
First
opprox;
Second
approx;.
motion
motion
Absolut. .rror
Obser'
vallon
No.
88
32
74
73
86
29
35
21
30
68
57
67
87
49
65
70
15
52
80
26
37
11
23
59
57
58
21
34
47
40
2
18
26
34
43
48
57
64
8
51
25
13
31
45
16
6
6
7
24
37
77 57
58
26
53
62
82
28
27
15
10
40
30
60
66
29
31
48
3
44
61
29
30
27
24
49
51
54
X, XI X,
F,rst
approxi
mation
--- - - 1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
Second
apprax;.
motion
- - 61
51
63
30
22
59
58
41
61
77
58
19
44
84
72
76
72
39
53
34
65
19
55
52
29
48
1
3
0
1
1
4
4
1
0
4
2
2
3
0
1
2
5
0
2
2
3
3
7
1
1
0
0
1
0
1
0
0
1
3
2
4
2
2
2
3
1
1
2
1
2
1
1
1
2
1
0
0
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
Sum
Mean
44
78
50
36
54
64
61
62
65
68
55
35
59
73
32
39
74
45
43
49
87
42
47
43
45
68
27
39
16
66
33
41
11
31
40
28
53
61
10
48
54
19
21
37
52
34
57
12
24
45
35
24
51
41
47
43
62
50
44
20
42
41
36
5
50
43
46
26
72
34
18
29
53
62
64
44
41
47
56
56
39
60
36
62
43
54
31
76
49
48
27
67
40
35
54
69
.. , ... .. , ' "

... ... .. , ' "
2
2
4
2
0
4
3
1
4
0
1
3
2
4
1
1
1
5
0
2
2
1
2
1
101
2.0
1
2
2
0
0
4
3
1
1
0
1
1
3
3
3
0
1
"
0
0
2
2
1
2
71
1.4
2. Next plot observed values of Y against t hose computed from chart

A (entering with Xl and interpolating within the X 2 curves); label each
point wit h the value of X a, as shown in chart B of F ig. A-6; and again
fit a family of curves.
3. Entering with Xl, X 2, and Xa in sequence (Fig. A-6), compute the
value of Y for each observation in t he series and tabulate the corresponding error (Table A-I).
If additional variables were to be introduced, step 2 of the process
~18
10
50
Y
fiG. A-6. Development of flrst-approximation curves by coaxial method (data from
fabl e A-ll.
319
x,
SI,,, (6)
Obs.No.9
SI,,, (5)
00s.No. 8
FIG. A7. Developme nt of second-approximation curves by coaxial method (data from

Table A-l).
320
would be repeated in essence. In each case, Y as computed from the

derived chart sequence would be plotted against observed values of Y, and
the points would be labeled with values of the variable to be introduced.
ince chart A of Fig. A-6 was derived without consideration of X.,
the subsequ nt iutrodu tion of this factor may nece sitate a revi ion of
chart A, po.rticularly if appreciable correlation exists between Xa and
either Xl or X 2 In other words, the coaxial technique usually requires
two or more successive approximations to achieve best results. Although
several techniques are available,l only one method of developiug second
and succeeding approximations is discu sed here. With this method the
analysis continues as follows:
4. Reconstruct the first-approximation X 3 curves (Fig. A-6) on a
second sheet of cro -section paper, as in chart B of Fig. A-7.
5. Next, in chart A of Fig. A-7, plot values of Xl against coordinate
values derived by entering chart B with observed Y (on the vertical
scal ) and X a, labeling ach point with X 2 Fit the second-approximation X 2 urves.
6. In chart B (Fi~. A-7) , plot ob erved values of Y against those
computod from chart A in exa tly the same manner as under tep 2
above. Revise the Xa curve as indicated by the plotted points. In
this example it was found that the fir t-approximation X 3 curves were
nthely sati factory.
7. Repeat step 3 n.bove, using the econd-approximation curve of
Fig. A-7, and tabulate the corresponding error (Table A-I). If any
material change is made in any of the curve families, the average rror
of the second approximation should be appr ciably smaller than that of
the first approximation. To avoid possible bias in the final relation that
might result from inconsistent curve fitting, the algebraic sum of the
errors should be computed. If this sum is not essentially zero, one of the
curve families should be hifted as required.
Third and subsequent approximations are derived by simple repetition
of step 4 to 7. If the analy is involved five or more variables, there
would be three or more chart in the equence. In any event, the econdapproximation plotting for any chart is accomplished by entering the
chart sequence from both ends with corresponding values of all factors
(including the dependent variable). The point so determined is labeled
with tho value of the factor for which the chart is being developed, and a
revised curve family is fitted.
This method of multiple graphical correlation usually yields good
1 R. K. Linsley, M. A. Kohler, and J. L. H. Paulhus, "Applied Hydrology," pp. 652655, McGraw-Hill, New York, 1949.
M. A. Kohlor and R. K. Linsley, Predicting the RUlloff from Storm Rainfall, U.S.
Weather Bur. Research Paper 34, 1951.
321
rc ult for many problem involving joint function, such a rainfallrunoff and crest- tage relation ,wher thr e or more ind pendent fa tors
are significant. II the approximate bape and spacing of the urve families are known, the fiTst-approximation curve' for all but one variable
can be sketched without plotting tb data. The curve repr nting the
remaining variable can then be developed by plotting in the pre cribed
manner. In general, it i advi able to determine the curves for the
most important factor by plotting. The final po ition of all curve
familie can then be determined by th proces outlined. Thi approach
con titute ' the ubstitution of an" timat d fTe t" for el1ch vl1l'il1ble
in preference to the as umed " no eff ct" for tho e variables ubs q uon tly
introduced. The number of approximation required to achieve the
final solution is accordingly reduced.
APPENDIX B
PHYSICAL CONSTANTS, CONVERSION
TABLES, AND EQUIVALENTS
Conversion Table for Volume

Equivalents
Unll
Imperial
Gal
Cu In.
gal
Cu II
Cu m
Acr.-ft
SId
--Cubic Inch .. ......

U.s. gallon . .........
Imperial gallon .......
Cubic 1001 ......... . .
1
0.00(33
23 1
I
277
l. 20
7.(8
1,728
Cubic meter ..... 61,000
2604
Acr.-foot ...... ..... _ 7.53 X 10' 3.26 X 10'
Second-foot. day . ... .. 1. 49 X 10' 6 .(6 X 10'
0.00361
5.79 X 10- ' 1. 6( X 10-'
0.1l(
0.833
0.00379
0 . 00(55
I
0. 161
6.23
I
0.0283
220
35.3
I
2.71 X 10' 43,560
1230
2450
5. 38 X 10' 86,400
1. 33
3.07
3.68
2.30
8.11
I
l. 98
X
X
X
X
X
10-'
10- '
10- '
10- '
10- '
6 .70 X
1.55 X
1.86 X
1.16 X
4.09 X
0.504
I
10'"
10-'
10- '
10 -'
10 - '
Conversion Table for Discharge

Equ(vafents
Unl!
Gal/day
U.S. gallon por day . .. I

Cubic fool per day . . 7. 48
U.S. galion per minute 14(0
Imperial gallon per
minute . .... . 1728
Cu II/ day
..... 0.13(
1
193
Gpm
Imp.riol gpm
Acrelt/ day
Cis
Cv m/ .e,
6.94 X 10- ' 5.78 X 10- ' 3.07 X 10 - ' 1.55 X 10- ' ( .38 X 10- '
5. 19 X 10- ' 4.33 X 10- ' 2.30 X 10- ' 1.16 X 10- 1 3.28 X 10- '
4.42 X 10- ' 2.23X 10- ' 6 . 31 X 10 - '
0.833
I
231
l. 20
226
449
Cubic meter per cond 2.28 X 10' 3.05 X 10' 15,800
1
188
374
13,200
Acr. foot per day .... 3.26 X 10' 43,560

Cubic foot per second . 6.46 X 10' 86,400
323
5. 31 X 10 - '
I
l. 98
70 . 0
2.67 X 10- ' 7.57 X 10- '

0.504
0.0143
0.0283
I
35.3
I
PHYSICAL CONSTANTS, CONVERSION TABLES, AND EQUIVALENTS
324
Miscellaneous Conversions and Physical Constants

1 second-fool-day per square mile =
1 Inch of runoff per .quare mile =
1 miner's inch ""
1 cubic fool per second

1 pound af waler
1 horsepower
.
IOg 10 e
log , 10
0.03719 inch
26.9 second-fool-days
53.3 acre-feel
2,323,200 cubic feel
0 . 025 cubic fool per second in Arizona, California, Monlana,
and Oregon
... 0.020 cubic fool per second in Idaho, Kansas, Nebraska, New
Mexico, North and Soulh Dakola, and Ulah
- 0.026 cubic fool per second In Colorado
- 0.028 cubic fool per second in Brilish Columbia
- 0.000214 cubic mile per year
... 0.9917 acre-inch per hour
= 0.5507 inch Over 8-inch circle
~ 0.3524 Inch over 10-lnch circle
- 0.2448 Inch over 12-lnch circle
- 0.746 kilowatt
- 550 fool-pounds per second
... 2.71828
= 0.43429
... 2.30259
Properties of Water
Viscosily
Temp,
of
Speciflc
gravity
Unil
welghl,
Ib/ cu fl
Heal of
vaporlza lIan,8lu/ lb
62.416
62.423
62.408
62.366
62.300
62.217
62.118
61.998
61.719
61.386
61.006
60.586
60.135
59.843
1073
1066
1059
1054
1049
1044
1039
1033
1021
1010
999
988
Absolule,
Ib-sec/sq ft
Kinemallc,
sq fl/ sec
0.374 X 10- '

0.323
0.273
0.235
0 . 205
0.180
0.160
0.143
0.117
0 . 0979
0.0835
0.0726
0.0637
0.0593
1.93 X 10-'
1.67
1.41
1.2 1
1.06
0.929
0.828
0.741
0.610
0.513
0 . 440
0.385
0.341
0.319
Vapor pressure
Millibars
Psi
6.11
8.36
12.19
17 . 51
24.79
34.61
47.68
64.88
0 . 09
0.12
0 .1 8
0.26
0.36
0.51
0 . 70
0.95
1.69
2.89
4.74
7.51
11.52
14.70
--32
40
50
60
70
80
90
100
120
140
160
180
200
212
0 . 99987
0.99999
0.99975
0.99907
0.99802
0.99669
0.99510
0.99318
0.98870
0.98338
0.97729
0.97056
0.96333
0.95865
9n
970
.....
.....
. . ...
.....
.....
.....
325
Values of n for the Kutter and Manning Formulas [Eq. (~-6)]

Channel Condilion
Ploslic, gloss, drown lubing . . . 0 . 009

Neal cemenl, smoolh melol ....... . ........ .. . ... ........... . .. . 0.010
Planed limber, asbestos pipe .................. .. ................ 0.011
Wroughl iron, welded steel, convos. . . . . .. .. . . . . . . . . . . . . . 0.012
Ordinary concrele, ospholled casllron . . . ............... 0.013
Unplaned limber, vilrifled clay, glazed brick ......... . ...... 0.014
Cosl-lron pipe, concrele pipe. __ .. _. _ _... _. . . . . . . . . . . . . .. 0.015
Riveled steel, brick, dressed slone ..... . _ . . . . . . . . . . . . . . .. 0.016
Rubble mosonry ..... _.. ...... ... .. .. ......... 0.017
Smoolh earth ... _.... . ............. __ .. _.... 0.018
Firm grovel . ....... __ . .. . __ ... ..... ............ 0 . 020
Corrugoled melal pipe and flumes ... _............. _. . . . . . . . . . . 0 _023
Nalural channels
Clean, straighl, full olage, no pools ............................. 0.029
As above wilh weeds and slone ............................. _. 0.035
Winding, pools and shollows, cleon .... . ........... ......... 0.039
As above allow slages . . ..... . .. . .................. ... . . .... 0.047
Winding, pools and shallowl, weeds and Ilones .. ............ 0.042
As above, shallow slages, large slones ...... . .... . ....... 0 .052
Sluggish, weedy, wilh deep pools .... ..... .. ......... _... _. _.. _.0. 065
Very weedy and sluggish ......... .... . ........ . _... _. 0.112
NOTE. Values quoled above are averages of many delermlnallons. Varialions of a s much a s
20 per cenl musl be expecled, especially in Ihe nalural channels.
Variation of Relative Humidity (Per Cent) with Temperature and

Wet-bulb Depression
(Pressure = 30 In.)
Air
WeI-bulb depression, FO
temp,
OF
0
5
10
15
20
25
30
35
40
45
50
55
60
65
70
75
80
85
90
95
100
67
73
78
82
85
87
89
91
92
93
93
94
94
95
95
96
96
96
96
96
96
33
46
56
64
70
74
78
81
83
86
87
88
89
90
90
91
91
92
92
93
93
1
20
34
46
55
62
67
13
29
40
49
56
63
68
71
74
76
78
80
81
82
83
84
85
86
86
12
25
36
45
52
57
61
65
68
70
72
74
75
76
78
79
80
10
- - -- -- - - -- -- - - - -
72
75
78
80
82
83
85
86
86
87
88
89
89
89
I
16
27
37
44
49
54
58
61
64
66
68
70
71
72
73
10
22
31
38
43
48
52
55
58
61
63
65
66
68
30
16
18
20
25
12
-- - - - - - - - - --14
7
18
27
33
39
44
48
51
54
56
58
60
62
6
16
23
30
35
40
44
47
50
52
54
56
5
14
21
27
33
37
41
44
47
49
51
5
13
20
25
30
35
38
41
44
46
5
12
19
24
29
32
36
38
41
3
9
15
20
24
27
30
3
8
13
17
21
326
PHYSICAL CONSTANTS, CONVERSION TASLES, AND EQUIVALENTS
Variation of Dewpoint with Temperature and Wet-bulb Depression and of

Saturation Vapor Pressure with Temperature
(Pressure = 30 in.)
Air
temp,
OF
-0
SaturatiOfl
vapor
prenure
Mlllibars
1.29
1.66
5
10
2.1 3
15
2.74
20
3. 49
25
4.40
30
5.55
35
6.87
40
8.36
45 10.09
50 12.19
55 14.63
60 17.51
1S5 20.86
70 24 . 79
75 29 . 32
80 34 . 61
85 40.67
90 "7 . 68
95 55 . 71
100 64 . 88
In. Hg
O
Wet-bulb depr...lon, F
-- 7 - 20
10
12
-- - - - - -- -- -
0. 038 0. 049 - 1 - 9 - 24
5 - 2 - 10 - 27
0. 063
0.081 11
6
0 - 9
0.103 16
12
2 - 21
8
0 . 130 22
19 15 10 - 3 0.164 27 25 21 18
830 28 25 17
0. 203 33
0.247 38
35 33 30 25
0.298 43
41 38 36 31
0.360 48
46 44 42 37
0.432 53 51 50 48 43
0 . 517 58 57 55 53 49
0. 616 63
62 60 59 55
0 .732 69 67 65 64 61
0.866 74 72 71 69 66
1.022 79 77 76 74 72
1.201 84 82 81 80 77
1.408 89 87 86 85 82
1. 645 94
93 91 90 87
1. 916 99 98 96 95 93
14
16
18
20
25
I
I
15
7
7 - 11
7 - 14
18
7 - 14
25 18
8 - 13
32 26 18
9 - 12
38 33 27 20
45 40 35 29 21 11
51 47 42 37 31 24
57 53 49 44 39 33
63 59 55 51 47 42
68 65 62 58 54 50
7.. 71 68 64 61 57
79 76 73 70 67 63
85 82 79 76 73 70
90 87 85 82 79 76
- 8
14
26 - 11
36
15
44
28 - 7
52
39 19
59 .. 8 32
66 56 43
72
63 52
Variation of Pressure, Temperature, and Boiling Point with Elevation
(u.s. tandard
atmo.phere )
Prenure
Elevation ,
ft m.1
- 1,000
0
1,000
2,000
3,000
4,000
5,000
6,000
7,000
8,000
9,000
10,000
11,000
12,000 \
30
-- -- -
In. Hg
Millibars
Ft of
water
Air
temp,
OF
Boiling
point,
OF
31.02
29 . 92
28.86
27.82
26.81
25.84
2" . 89
23.98
23 . 09
22.22
21.38
20.58
19.79
19.03
1050 . 5
1013.2
977.3
942 . 1
907.9
875.0
842.9
812.1
781.9
752.5
724 . 0
696.9
670.2
64.....
35.12
33 . 87
32.67
31.50
30.35
29.25
28.18
27 . 15
26 . 1"
25.16
2".20
23 . 30
22.40
21.5"
62 . 6
59 . 0
55 .4
51.8
48 . 4
44.8
41.2
37.6
34.0
30.6
27.0
23 . 4
19.8
16.2
213 . 8
212.0
210 . 2
208 . 4
206.5
204.7
202.9
201.1
199 . 2
197.4
195.6
193.7
191. 9
190.1
The data of thl. table are ba.ed an average cOfIdltlon.
Map-scale Conversions
Ratio
In./mlle
Miles/ in.
Sq mi/sq in.
--1: 1,000,000
1 :500,000
1: 250,000
1: 126,720
1: 125,000
1:90,000
1: 63,360
1:62,500
1 :"5,000
1: 31,680
1 :30,000
1: 24,000
1: 12,000
1: 2,400
1: 1,200
15 . 7828
7.8914
3.9"57
2.0000
1.9728
1. "205
1. 0000
0.9864
0.7102
0.5000
0.4735
0.3788
0.1894
0.0379
0.0189
0.0634
0.1267
0.2534
0.5000
0 . 5069
0.7040
1.0000
1. 0138
1. 4080
2.0000
2.1120
2.6400
5.2800
26.4000
52 . 8000
--
249 .097
62 . 274
15.569
".000
3.892
2. 018
1.000
0. 9730
0 . 504"
0 . 2500
0. 22"2
0.1435
0.0358
0.001435
0.000358
Values of the Reduced Variate y Corresponding to Various Values of

Return Period and Probability [Eq. (11-2)]
variate
R.turn
period
Probability
T.
0.000
0.367
0.579
1. 500
2.250
2.970
3.902
4.600
5.296
6.000
1.58
2. 00
2.33
5 . 00
10.0
20.0
50.0
100
200
403
0.368
0.500
0.571
0.800
0.900
0.950
0.980
0.990
0.995
0.9975
Reduced
3'17
328
Values of kl for Various Values of k and t [Eq. (8-2)]
0.80
0.82
0.84
0.86
0.88
0.90
0.92
0.94
0.96
0.98
1
2
3
4
5
0.800
0.640
0.512
0.410
0.328
0.820
0.672
0.551
0.452
0.371
0.840
0.706
0.593
0.498
0.418
0.860
0.740
0.636
0.547
0.470
0.880
0.774
0.681
0.600
0.528
0.900
0.810
0.729
0.656
0.590
0. 920
0.846
0.779
0.716
0.659
0.940
0.884
0.831
0.781
0.734
0.960
0.922
0 . 885
0.849
0.815
0.980
0.960
0.941
0.922
0.904
6
7
8
9
10
0.262
0.210
0.168
0.134
0.107
0.304
0. 249
0.204
0.168
0.137
0.351
0.295
0. 248
0. 208
0.175
0.405
0.348
0.299
0.257
0.221
0.464
0.409
0.360
0.316
0.279
0.531
0.478
0.430
0 . 387
0. 349
0.606
0.558
0.513
0.472
0.434
0.690
0.648
0.610
0.573
0.539
0.783
0.751
0.721
0.693
0.665
0.886
0 . 868
0.851
0.834
0.817
11
12
13
14
15
0.086
0.069
0.055
0.044
0.035
0.113
0.092
0.076
0.062
0.051
0 . 147
0.123
0.104
0 . 087
0.073
0.190
0. 164
0 . 141
0.121
0.104
0.245
0.216
0.190
0 . 167
0.147
0.314
0.282
0.254
0.229
0 . 206
0.400
0.368
0.338
0.311
0.286
0.506
0.476
0.447
0.421
0.395
0.638
0.613
0. 588
0.565
0.542
0. 801
0.785
0.769
0.754
0.739
16
17
18
19
20
0.028
0.023
0.018
0.014
0.012
0.042
0.034
0.028
0.023
0.019
0.061
0.052
0. 043
0.036
0.031
0.090
0.077
0.066
0.057
0.049
0. 129
0.114
0.100
0. 088
0.078
0.185
0.167
0.150
0.135
0.122
0. 263
0. 242
0. 223
0.205
0.189
0.372
0.349
0.328
0.309
0.290
0 . 520
0.500
0. 480
0.460
0.442
0. 724
0. 709
0.695
0.681
0.668
21
22
23
24
25
0.009
0.007
0.006
0.005
0.004
0.015
0.013
0.010
0.009
0.007
0.026
0.022
0.018
0.015
0.013
0.042
0 . 036
0.031
0. 027
0.023
0.068
0.060
0.053
0.047
0.041
0.109
0.098
0.089
0.080
0 . 072
0.174
0.160
0.147
0.135
0.124
0.273
0.2S6
0.241
0.227
0.213
0.424
0.407
0.391
0.375
0.360
0.654
0. 641
0.628
0.616
0.603
26
27
28
29
30
0.003
0.002
0.002
0.002
0.001
0.006
0.005
0.004
0.003
0.003
0.011
0.009
0.008
0.006
0.005
0.020
0.017
0 . 015
0.013
0.011
0.036
0.032
0.028
0.025
0.022
0.065
0.058
0.052
0 . 047
0 . 042
0.114
0.105
0.097
0.089
0 . 082
0 . 200
0.188
0.177
0.166
0.156
0.346
0.332
0.319
0.306
0.294
0.591
0.579
0.568
0.557
0.545
40
50
60
..... ..... ..... 0.002 0.006 0.015

.. ... ..... ..... ... .. ..... O. OOS
..... ..... . .... . .... ... .. .....
0. 036
0.015
0.007
0.084
0.045
0.024
0.195
0.130
0.086
0.446
0.364
0 . 298
NAME INDEX
Collet, Mo H., 57
Colman, Eo Ao, 125
ommons,
G o, 207- 20
Conkling, Ho, J44
Conover, Lo Fo, 307
Cook, Ho Lo, 179
Corb tt, D o Mo, 64, 7
roft, Ao R, 190
rOBS, Wo Po, 5, 271
Culler, It. Co, 76, 93
Uuykendall, To R, 127
Ackermann, Wo Co, 15 , 170, 280

Allen, Co Mo, 70
Alter, Jo Co, 2 29, 49
Ammerman, Do, 43- 44, 50
Anderson, Eo R, 9 95,97- 98, 119, 182
Anderson, Lo .1., 94-95, 97- 98, 119
Andrews, Lo Ao, 27
Arnold, Co Eo, 212
Bailey, So M o, 276
Barnes, Bo 0, ]53
Baver, Lo Do, 127
Belcher, Do J o, 127
Bennet.t, R R, 143
BenLon, Go SO, 2J
Bernard, Mo, 2J, 40, 197, 212,268,276
Bigler, So G o, 307
Blackburn, Ro To, 21
Elan y, Ho Fo, 112- 113, 117, 119- 12(1
Bondurant, Do Co, 291
Bowen, I. So, 95- 96, 115
Boyer, Mo Co, 66
Bradl y, Jo No, 70
Brancato, Go No, 38
Brashears, Mo Lo, 144
Bnl.ter, Eo Fo, 5, 87
Broadfoot, Wo Mo, 127
Brooks, Co Fo, 49
Brown, Co Bo, 281, 283, 291
Browning, 00 Mo, 166
Brune, Go Mo, 289- 290
Brunt, Do, 117
Buckingham, E o, 124
Burns, Jo 10, 40
Byers, Ho R, 21, 48
Dalryrnpl , To, 203- 264, 299

Davis,
Vo, 310
Davis, Go fl., 145
D ie, H.o B o, 131
Doran, F o ,I., 57
Dorffel, Fo, 9
Dreibelbis, Fo It, 116
du Boys, P o, 282
Duckworth, Fo So, 12
Durour, Lo, 48
Dunkle, Ro Vo, 96
0
Edson, Co Go, 208

Einstein, H o Ao, 291
Eley, 00 Wo, 280
Ellison, Wo Do, 279
Erickson, Co Eo, 48
Ezekiel, Mo, 188, 311
Ferguson, 00 Eo, 57
Ferris, Jo Go, 141, 146
Fisher, R Ao, 250
Fletcher, R o Do, 21, 49, 273
Flynt, F o Lo, 212
Folsom, R o Go, 60
Foskett, Lo W., 18,236
Foster, Eo Eo, 276
Chapman, Go, 279

Chow, Vo To, 252, 276
Clark, Co 00, 240
Colby, B. R., 286
329
NAME INDEX
330
Foster, H. A., 250
Foster, N. B., 18
Fox, W. E., 103, 113, 120
Franssila, M., 96
Franzini, J. B., 173, 276, 300, 302, 310
Free, G. R., 166
Fritz, S., 103, 117
Fuller, W. E., 211
Gilcrest, B. R ., 243
Gilman, C. S., 273
Gleason, C. H., 279
Glock, W. S., 41)
Goodrich, R . D., 224
Gottschalk, L. C., 288
Grant, E. L., 245
Gregory, R. L., 212
Grov r, N. C., 87
Gumbel, E. J., 24.8, 250- 252, 257
Harstad, K C., 262
Hains, C. H., 93
Hall, F., 48
Hamilton, E. L., 27, 29
Hamon, R. W., 103
Hannaford, J., 11:)0
Hantusb, M. S., 145
Harb ok, G. E., 93-94, 96
Harrington, A. W., 87
Harrold, L. L., 116
Hathaway, G. A., 310
Haurwitz, D., 49
Hazen, A., 24 249, 252, 276
Hellman, G., 21
Hendrickson, A. H., 111, 126
Hendrix, T. M., 163
Hertzler, R. A., 150
lliatt, W. E., 4
Riser, H. W., 307
Holzman, B., 97, 115
Horner, W. W., 212
Horton, R. E., 4, 27, 36, 163, 165, 167,
1 5, 219, 238
Houghton, H. 0., 48
Howland, W. E., 257
Hoyt, W.O., 70, 167,213
Hubbert, M. K, 146
Huff, F. A., 25, 49, 307
Hurst, H . E., 293
Islitzer, N. F., 96
Izzard, C. F., 209- 211
Jacob, C. E., 138, 143, 146
Jarvis, C. 8., 211, 213, 248, 276
Jennings, A. H., 41
Jens, S. W., 310
Johnson, A. F., 117, 120
Johnson, W. M., 163
Johnstone, D.O., 5, 49
Kadel, B. C., 48
KalinBke, A. A., 281-284
Kaufman, W. J., 132
Kazmann, R. 0., 141, 145
Kendrew, W. G., 43
K ulegan, G. H., 243
Kincer, J. B., 33, 36, 50
King, H . W., 87
King, R. E., 310
Kittredge, J., 163
Knox, C. E., 113
Koelzer, V. A., 290
Kobler, M. A., 32- 33, 103, 119-120, 170171,173,187,190,232,235- 236,242,
31&--316, 320
Kozachyn, J., 280
Kramsky, M., 141
Kresge, R. F., 265
Kurtyka, J . C., 48
Landsberg, H., 21, 43, J 11, 276
Lane, E. W., 281- 2 2,290
Langbein, W. B ., 75, 7 ,93,109,119, 146,
150, 154, 213, 247
Langille, R. C., 48
Larson, C. L., 71
LI1UIitz n, C. W., 163
Laverty, F. B., 144
Leatham, P., 181
L ,C. H., 111
Leopold, L. B., 287
Lettau, H., 98
Liebl in, J., 251
Light, P., 182
Linsley, R. K., 21, 32, 158, 170- 171, 173,
186- 187,206,236,240,261,276,283,
300, 302, 310, 320
331
NAME INDEX
Lloyd, C. H., 280

Loughead, H. J., 163
Lowry, R. L., 117, 120
Ludlam, F. H., 48
McCartby, G. '1'., 228
McClendon, E. W., 21, 276
Mc rmick, R. A., 3
McDonald, C. C., 146, 150
McDonald, J . E., 33
MacDonald, T. H., 103
McDonald, W. F., 50
McEwen, G. F., 94
McGuinness, C. L., 122, 141
McMath, R E., 212
Maddock, T., 2 7
Mansfield, W. W., 109
Marciano, J. J., 94-95, 97- 98, 119
Marr, J. C., 190
Marshall, J. S., 48
Mather, J. R., 116, 1.20
Mavis, F. T., 71
Mazurak, A., 163
Mead, D. W., 4-5
Meinzer, O. E., 5, 133, 140, 146
Merriam, C. F., 33
Meyer, A. F., 5, 98
Middleton, W. E. K., 21, 48
Miller, J. F., 177,266
Mitchell, W. D., 66
Moots, E. E., 219
Morris, H. M., 71
Morris, V. B., Jr., 18
Musgrave, G. )V., 166, 279
Muskat, M., 146
Neal, O. R., 127
Neill, J . C., 48
Nelson, A. Z., 36
Nordenson, T. J., 103, 113, 120, 265
O'Brien, M. P., 60, 281
Orlob, G. T., 132
Palmer, W. McK., 48
Pa.rsball, R . L., 71
Pa.rsons, D. A., 2
Paulhus, J. L. H ., 33, 48, 177,266, 2'l3
Peck, E. L., 188

P nman, H. L., 96, 101, 112, 117- 119
Peterson, H. V., 76
Petterssen, ., 4
Poland, J. F., 145
Powell, R W., 257
PrandtJ, L., 97, 281
Reed, W. G., 36
Richards, G. R, 276
Richards, L. A., 127
Richardson, B., 94
Rider, N. E., 115
Riedel, J. '1'., 48
Riesbol, H. S., 1 I, :no
Rossby, C. G., 21
Rowe, P. B., 163
Rumbaugb, W. F., 11
Russell, M. B., 127
RuUer, E. J., 300
ack, H. S., 127

Sandberg, J. ., 12
Sanderson, E. E., 49
Schiff, L., 144
Schloemer, R. W., 48
Schmidt, W., 97
Schofield, R. K., 125
Schoklitsch, A., 219
cbulman, E., 49
Schwarz, H. E., 205 206
Seddon, J., 217, 219
Senn, H. V., 307
Shands, A. L., 38, 43- 44, 50
Shaw, C. F., 127
Shaw, N., 43
Sherman, L. K., 4, 167, 194, 213
Showalter, A. K., 273
Slade, J. J., 250
Slichter, C. S., 131
Smith, E. J ., 48
Snead, V. 0., 21
Snedecor, G. W., 311
Snyder, F. F., 184-185, 204-206
Snyder, W. M., 200
Sobel, F., 18
Solot, S., 16, 273
Spilbaus, A. F., 21, 48
Spreen, W. C., 40
NAME INDEX
332
Stevens, J. C., 57, 68
Stoltenberg, N. L., 163
Storey, H. C., 36
tout, . E ., 48- 49, 307
Straub, L. G., 283
Strau8B, F. A., 190
Suomi, V. E., 96
Suter, M., 144
Sutton, O. G., 97-98
verdrup, H. U., 97
Tannehill,!. R., 49
TarbJe, R. D., 307
Taylor, A. B., 205- 206
Taylor, E . A., 70
Taylor, G. I., 97
Theis, C. V., 136, 138, 139
Tbickstun, W. R., 18
Thiem, G., 135, 139
Thiessen, A. H., 21, 36
Thomas, H. A., 243
Thomas, H. E., 146
Thornthwaite, C. W., 97, 113, 115- 116,
118,120
Tippett, L. H. C., 250
Tolman, C. F., 146
Trask, P. D., 291
Trimble, G. R., Jr., 163
Turner, R. E., 310
Veihmeyer, F. J ., 111, 126

von Karman , T., 97, 281
Warnick,
., 4.9
Webber, E. E., 271
Wei8B, L. L., 34., 103, 252--253
WoitzIJlJl,n, ., 163
Wenzel, L. K., 137
W rner, P. W., 146
Whipple, G. C., 257
White, C. M ., 284
Wilkiuson, J . H., 219
Willett, H. C., 21
Williams, G. R., 120, 213, 310
WiUin1ns, H. M., 207- 208
Wilm, H. G., 36
Wilson, W. T., 28- 29, 34, 46, 49, 103,
231
Wisler, C. 0., 5
Wood, R. C., 18
Woodburn, R., 280
Yarnell, D. L., 260-261

Young, A. A., 100, 120
Young, K. B., 64
Zingg, A. W., 167
SUBJECT INDEX
Abrupt waves, 21 219

~cre-foot, 71, 323- 324
Adiabatic process, 11- 12
Adjustmeut of da.ta, precipitation, 3334:
streamflow, 76-78
Aeration, zone of, 122-129
Air density, 15
Air temperature, 9-13
definition of terms, 10-11
distribution of, geographic, 12- 13
vertical, 11- 12, 326
effect of, on evaporation, 91, 101-103
effect on, of cities, 12-13
of topogra.phy, 12-13
inversions, 11
lapse rates, 11- 12, 326
measur ment of, 9- 10
minimum for growing, 111
normals, 10
variation with elevation, 12- 13
Alb do, 95,117,181
Anchor ice, 69
Annual floods, 247
Annual series, 247
Antecedent precipitation index, 170-172
table of k', 328
Anticyclones, 9
Aquiclude, 129
Aquifer, 12<),. 129- 130
Aquifuge, 129
Artesian water, 123, 145
Atmosphere, general circulation, 6-9
stability of, 11- 12
U.S. standard, table, 326
Atmospheric moisture (see Humidity)
Bed load, dofinition of, 2 0

measurem nt of, 2 0- 2 7
tl'll.osport of, 2 2- 2 5
Blind drninage, 1G4
Boiling-point variation with pr ssure,
326
Bores, 219
Bowen's ratio, 95, 101
Capillary conductivity, 121)-129

Capillary moisture in soils, 123- 125
apillary potential in soils, 12 125
Channel erosion, 70
Channel precipitation, 167- 168
Chann I storag ,22 224, 227- 236
(See also Routing; torag')
ChCzy formula, 68
Ch6zy-Manning formula, 69
hinook,92
Circulation (see General circulation)
Climate, effect of general circulation on,
8
physiographic influences on, 7- 8, 1213, 18,20
Closed drainage, 164
Cloud seeding, 24Coaxial correlation methods, 316- 321
for evaporation, 102, 105, 10
for precipitation, 40- 41
for routing, 233- 234
for runoff, 173- 179
Condensation, 14
effect of, on snowmelt, 182
Conduction, heat transfer by, 95, 182
Cone of depression, 13 135
Connate water, 132-133
Consumptive use (see Evapotranspira.tion)
Control, str amfiow, 58
CODvective precipitation, 23
Conversion factors, 323-324
Backwater, 65, 122-123

Base flow, 150-160
Basin lag, 204-207, 237-241
Basin recharge, 173-175
333
334
Conversion tables, dlsoharge, 323

map 8cales, 327
volume, 323
CorioUs force, 7
Correlation methods, 311- 321
coaxial (86e Coaxial correlation methods)
Crest segment of hydrograph, 151- 152
Crest-stage relations, 241- 243
Culvert design, 71, 302- 306
Current meters, 58-61
Cycle, hydrologic, 1- 3
runoff, 167- 169
Cycles, precipitation, 43
Cyclones, 9
Cyolonic preoipitation, 23
Daroy'slaw, 130- 131
Degr e-day factor, 184-186
Degree days, 11, 183- 185
Density, of air, 15
of snow, 45
of water vapor, 15
Depletion curve, 152
Depression storage, 162, 164, 168
Depth-al'ea-duratiou relations, 37- 38, 44
maximum values, 44
probable, 274-275
use of, in design, 29 299
Design fr quency, sel otion of, 258- 259
D sign methods, Bood-control reservoir,
300- 302
spillway, 297-300
storage reservoir, 292-297
storm drainage, 302-306
Dewpoint, definition of, 16
tables, 326
Direct runoff, 151, 169
Discharge, conversion faotors, 323- 324
over dams and embankments, 7 71
effc t of ice on, 69- 70
mean daily, 73- 76
mans and extremes of, 77- 87
measurement of, 58-64
(See alao Streamflow)
Distribution graph, 197
Double-ma s ano.lysis, of precipitation
data, 33-34
of streamflow data, 76-78
Drainage, blind, 164
SUBJECT INDEX
Drainage design, 302- 306

Drought, 269- 270, 296
du Boys formula, 282-283
Dupuit formula, 135
Duration curves of f10\V, 270-272
Eddy conductivity, 280-282
E.ffiuent streams, 133, 150
Electronic rou.ting analog, 235-236
Elevation, variation with, of evapora.tion, 104-107
of precipitation, 40-41
of snowmelt, 185-186
of temperature, 12-13
of wind, 20- 21
Energy budget, 94-97, 105- 109, 181- 183
Erosion, 278- 280
ohannel,70
Evaporation, ooaxial correlation methods
for, 102, 105, 108
definition of, 13-14, 90
faotors affecting, 91-93
quality of water, 92- 93
radiation, 91, 94-97
formulas for , 97- 98, 101
geographic distribution of, 110
from la.kcs, estimates of, by energy
budget, 94-97, 105-109
by mass transfer, 97- 99, 10 109
by pans, 99- 109
by water budget, 93- 94, 108-109
measurement of, 99- 101
nomograms for, 102, 107- 108
normal annual, llO
pan coefficient, 104-108
in reservoir design, 293- 295
as runoff indcx, 170
from snow, D2
from soil, 92
suppression of, 10D
variation of, with elevation, 104-107
time, 111, 114-115, 118
Evaporation opportunity, 92
EVapotranspiration, definition of, 112ll3
estimation of, 113- 119
measurement of, 116
potential, 113, 116-118
(See also Evaporation; Transpiration)
Exceedanoe interval, 247
SUBJECT INDEX
Field capaoity, Ill, 125-126

Float measurements, 71
Flood-contl-ol reservoir design, 300- 302
Flood formula , 2tl-213, Z6G
Flood routing (see Routing)
Floods, annual, 247
oontrol of, 3(}()-302
forecasting of, 307- 310
frequency analysis of, 245-259, 263267
project-design, 300
record peaks, 82-86
spillway-design, 297-300
Flow-duration curves, 270-272
Foehn wind, 92
Forecasting, of streamflow, 307- 310
of water supply, 1 6- 190
Frazil ice, 69
Frequency, of floods, 245- 259, 263- 267,
335
Graphical oorr iation, three-variabl ,

313-316
two-variable, 311 - 313

Gravity water, 123
Groundwater, artificial r charg of,
]44-145
Cllrry-over cffeot on treamflow, 1

contribution to streamflow, 150-151
discharg of, 133-134
long-period variations of, 145- 146
movement of, 130- 132, 13 140
origin and occurrence of, 122- 123,
132-133
perched, ]23
reces ion , 152- 155
as runoff index, 170-171
safe yield of, 140-142
salt-water intrusion ill , 142-144
Growing temperatures, minim urn, 11
299-300
derived from rainfaU , 265-267

effect of flood control on, 301 - 302
regional analysis of, 263- 265
of rainfall, 247-253, 257- 263, 267- 270
use of, in design, 302- 303
theoretical distribution function s,
250- 258
Friction, effect of, on wind, 20

Front, 9
Gage datum, 52
Gage relations, 240- 243
Gages (sce Instrum nts for measuring)
General circulation, 6-9
effect on, of eartb's rotation, 7
of land and water distribution, 8
Geographic distribution, of evaporation,
110
of humidity, 18
of precipitation, 3 44
of runoff, 77, 80- 81
of snow, 47
of temperature, 12-13
of wind, 20
Geologic characteristics, influence of, on
hydrograph, 151
Ghyben-Herzberg principle, 142-143
Graphical correlation, coaxial, 316-321
(See also Coaxial correlation
methods)
Heat transfer, by conduction, 95, J 82

by convection, 95, 182
Humidity, 13- 19
geographic distribution of, 18
measurement of, 16-]8
relative, 15
specific, 16
tabl s, 325-326
Hurricanes, 9
Hydrograph, analysis of, 156- 16]
components of, 151- 152
definition of, 72-74
dim n8ionle8S, 208, 211
influence of geologic characteristics on,
151
of overland flow, 200- 211, 303- 304

recession of, 151- ] 55
sbape of, 193- 198
unit (see Unit hydJ'ograph)
IIydrologic cycle, 1- 3
Hydrology, definition of, 1
history of, 3-4
subject matter, 4-5
HydrometeorologicaJ studies, 273- 276
Hygroscopic moisture, 123
Ice, effect of, on streamflow, 69- 70

formation of, 69
types of, 69
SUBJECT INDEX
336
Images, method of, 139-140

Infiltration, 128, 164-168, 179- 181
measurement of, 167
Infiltration indices, J 80 - J 81
Influent streams, 133
Initial flow, 170- 171
Instability, atmosph ric, 11- 12
Instruments for meaJ!uring, of discharge,
58-64
of evaporation, 99-101
of evapotranspiration, 166
of humidity, 16-18
of infiltration, 167
of interoeption, 163
of permeability, 131-132
of potential evapotranspiration, 116
of precipitation, 24-30
of river stage, 52- 58
of sediment transport, 285-287
of snow, 24-30, 45-46
of soil moisture, 12 128
of temperature, 9- 10
of transpiration, 112
of wind, 19
Intensity-duration-frequency formulas,
267- 208
Interception, 162- 164, 168
measurement of, 163
Interfiow, 150, 169
Inversions, temperatul' , 11
Isochrones, 237
lsohyetal maps, 3D, 35-37
Juvenile water, 132
k' table, 328
Lag of basin, 204-207, 237- 241

Laminar flow, 209
Lapse rates, in free air, 11-12
along mountain slopes, 12
Latent hea.t, of fusion, 14, 181
of sublimation, 14-15
of vaporization, 324
Local inflow, 223- 224, 241-242, 309
Lysimeten, 116
Manning formula, 69
values of n for, 325
Map scales, table, 327
Mass transfer, 97- 99, 10 109, 182,
280- 282
Meinz r unit, 131
Melting of snow (see nowmclt)
Melting agents, 20
Meteoric water, 132
Miner's inch, 72, 324
Moisture accounting, 118- 119
Moisture equivalent of soils, 126
Monoclinal rising waves, 216- 218
Muskingllm routing, 22 229, 232,
304-306, 309
Myers rating, 79, 82-86, 211
n values, 325
Normal, definition of, 10
Normal-ratio method, 33
peTation studi s, 293- 297, 300- 301

Orographic precipitation, 23
Overland flow, definition of, 149- 150
hydrograph of, 209- 211, 303- 304
phase of runoff cycle, 162
Pan coefficient, evaporation, 104-108

PartinJ series, 247
Perched groundwat r, 123
Pereolation, definition of, 128
from streamflow, 133
Permeability, 130- 132
meaJ!urement of, 131- 132
pF of soil moisturo, 125
<.l> index, 180- 181
Phreatic surface, 122- 123
Phreatophytes, 134
Physical constants, 324
Pbytometer, 112
Piezometric Burface, 123
Plotting position, 248-250
Porosity, definition of, 129-130
Potential eva.potranspira.tion, definition
of,113
determination of, 116-118
Precipitable water, 16-17
SUBJECT INDEX
Precipitation, ajjustment of data, 33- 34

antecedent index, 170- 172
areal average of, 31- 32, 34-3
artificially induc d, 24
average annual, 39
channel, 167- 168
convective, 23
correla.tion with runoff (see Runoff correlations)
correlation methoda, coaxial, 40-41
cycles of, 43
cyclonic, 23
definition of, 23
depth-area-duration analysis of, 37- 38,
44
estimating missing data, 33
formation of, 23
forms of, 24
frequency of, 247- 253, 257- 263,
267-270
gage network, 30- 32
gag s, 2 30
geographic distribution of, 38- 44
intensity-frequcncy-duration formulas,
267-268
interception of, by vegetation, 1112- 164.,
168
maximum values of, 41- 44
monthly distribution of, 42
orographic, 23
physiographic influences on, 38-41
probable maximum, 273- 276, 297- 3.00
radar observations of, 24, 48, 307
storm analysis, 172-173
types of, 23
(See also Rainfall ; Snow; Snowf!1ll)
Precipitation-runoff relations (8 e RUDoff
correlations)
Pressure, boiling-point vnrintion with,
326
effect of, on evaporation, 104-i07
on winds, I)
mean atmospheric, 8
vapor (see Vapor pressure)
variation with elevation, 326
Probable maximum pr , cipitation,
273- 276, 297- 300
Project-design flood, 300
337
Psychrometric equation, 15
Paychrom tric tabl a, 325-326
Quality, of snow, 46
of water, off et of, on evaporation,
92-93
Radar observations of precipitation, 24,

48,307
Radiation, Cummings int gmtor, 96
efT ct of, on vap ration, 91, 94- 97
on snowmelt, 1 1- 182
Rainfall, critical duration of, 303- 304
drop size, 279
effect of, on snowmelt, 183
frequency of, 247-253, 257- 263,
267-270, 303
formula, 267
maps of, 260- 263
point-area r lation, 268- 269
gages, 24--30
maximum values, 41 - 44
v I city of drops, 279
(See also Preoipitation)
Ratings, discha.rge (Bce Stage-discharge
relations)
Rational formula, 212- 213, 304
Recurrence interva.l, 247
Reduced varia.te, 251, 327
Relative humidity, 15
tables, 325
Reservoirs, design of, 292- 297, 300302
effect of, on flood WIlN s, 220, 22 227,
300- 302
sedimentation in, 28 291
trap efficiency of, 289- 290
Return period, 247- 251, 327
selection for design, 258-259
theoretical distribution of, 248- 249
Rive~ forecasting, 307- 310
River mecha.nics, 70
River sta.ge, 52-58
Routing, in channels, 227- 236
to derive basin outflow, 237- 241,
304-306
methods for, coaxial correlation,
233- 234
graphical, 230-233
338
Routing, methods for, Muskingum, 228229, 232, 30 306, 309
scmigraphical, 229- 230
statistical, 233- 234
through reservoirs, 224-227
stage, 242
use of, in forecasting, 309
Routing devices, 233- 236
Runoff, adjustment of data, 76-78
coaxial correlation method for, 173-179
(See al80 Runoff correlations)
direct, ] 51, 169
geographic distribution of, 77, 80-81
groundwater as index, 170- 171
surface, 149- ]50
(See also Discharge; Stroamflow)
Runoff correlations, for seasonal and
annual periods, 1 6- 190
for snowmelt, 1 3- 186
for storms, 169- 181
surface (see Surface runoff)
use of, in design, 299
in forecasting, 308
Runoff cycle, 167-169
S-curv method, 200-204

Safe yield, 140-142
altation, 280
Saturation, zone of, 123, 129- 140
atu ration vapor prcssure, 14
table, 326
eddon's law, 217
Sedim nt, bed load, 2 0, 2 2- 285
rating curves, 287- 2 R
in rcservoirs, 2 8- 291
8p 'cific weight of, 290- 291
suspended load, 280-282, 285- 2 8
transport of, measurement, 285- 287
Seiches, 219- 220
Sewer design, 302- 306
Shields for wind, 28- 29
hiIting control, 58, 65, 70
lop -area determinations, 69
now, density of, 45
effect of topography on, 185-186
evaporation from, 92
geographic variations of, 46- 48
measurement of, 24-30, 45- 46
quality of, 46
water content of, 46
SUBJECT INDEX
Snow, water (.qui valent of, 45-46
water-holding capacity of, 183
Snow course, 46
Snow line, 185
Snow surveys, 45-46, 188- 190
Snowfall , gages, 24-30
geographic variation of, 46-48
mean annual, 47
(See also Precipitation)
Snowmelt, effect of condensation on, 182
estimation of, 183-186
pbysics of, 181- 183
Soil, evaporation from, 92
specific weight of, 126
Soil moisture, 123- 129
available water, 111- 112, 126
capillary, 123- 125
effect of, on infiltration, 166
field capacity, Ill, 125-126
measurement of, 126- 128
moistur equivalent" 126
movement of, 12 129
pF of, 125
as runoff index, 169- 170
wilting point, 111, 126
Soil water (see Soil moisture)
Specific gravity, of 8el\ water, 142
of wat r, 324
of water vapor, 15
Specific humidity, 16
Specific weight, of sediment, 290-291
of soil, 126
of water, 324
Specific yield, 130, 137
Spillway design, 297- 300
prings, 133- 134
Stability, atmospheric, 11- 12
Stage, river, frequency distributions of,
246
Stage-discharge relations, 64-69
effect of ice on, 69- 70
extension of, 6 69
shifting control, 58, 65, 70
simple, 64-65
variable slope, 65-68
Stage relations, 240- 243
Standard deviation, 258
Standard U.S. atmosphere, 326
Station-year method, 262- 263
SUBJECT INDEX
Stock tanks, 76, 93, 164

Stokes' law, 280- 281
Storage, bank, 157-158
channel, 222-224
depression, 162, 164, 168
determined from recession, 153, 240
equation of, 220- 221
prism, 227
wedge, 227-228
Storage constant Cor aquifer, 137
torage-reservoir design, 292-297
Storage routing (see Routing)
Storm-drainage design, 302- 306
Storm runoff, 151, 169
Storms, analysis for runoff correlation,
172-173
probable maximum, 273- 276, 297- 300
transposition of, 273
Streamflow, adjustment of data, 76- 78
carry-over effect of groundwaters on,
188
components of, 149- 151
contribution of I!;roundwater to,
150-151
control of, 58
conversion factors, 323- 324
effect of ice on, 69- 70
forecasting of, 307- 310
interpretation of records, 71- 78
means and extremes, 77- 87
measurement of, 52-64, 70- 71
normal annual, 77- 78
rating curves, 64- 69
recession of, 150- 155
routing (see Routing )
units of, 71- 72, 323- 324
variability of, 78-81
ublimation, 13- 15
latent heat of, 14--15
Subsurface flow (see Grou ndwater; lnterflow)
upply rate, 165- 166
Surface detention, 162, 193- 194
Surface retention, 164
Surface runoff, 149-150, 162
variation throughout storm, 168,
177-179
Surges, 219- 220
Suspended load, measurement of,
285-286
339
Suspended load, transport of, 280- 2 2
Syntheti unit hydrograph, 203- 20
Temp rattlre (sec Air temp ra.tur

Wa.t 'T temperatur )
Tensiometer, 126-127
Terminal velocity of drops, 279
Thermal circula.tion, 6
Thiessen method, 35-36
Time, of concentration, 212
(See also to equilibrium, below;
Rainfall, critical duration of)
to equilibrium, 193- 194, 210, 303304
of travel, 237- 241
(See also Waves, celerity of)
Time-area diagram, 237- 239
Topography, eff ct of, on climate, 8
on humidity, 1
on precipitation, 38- 41
on snow, 1 5-186
on temperature, 12-13
on wind, 20
Trade winds, 7
Transmissibility, 131
Transpiration, 111- 112
factors affecting, 111- 112
mcasurem nt of, 112
(See also Evaporation ; Evapotranspiration)
Transposition, of storms, 273
of uni t hydrogmphs, 207- 208
Trap effici noy of reservoirs, 2 9- 290
Tree rings, 43
Turbulent transport, 97- 99, 10 - 109.
182
Typhoons, 9
Unit hydrograph, o.pplication of, 209,

309
concept of, 194--198
derivation of, 19 208
distribution graph, 197
factors affccting, 195-196
by routing, 237- 238
synthetic, 203- 208
transposition of, 207- 208
use of, in design, 299
U.S. standard atmosphere, 326
SU8JECT INDEX
Vapor pressure, 13- 15

effect of, on evaporation, 91, 97- 98, 102
saturation, 14
in soils, 129
table, 326
of water, 324
Vaporization, 13-15, 90
latent heat of, 324
Vegetation, effect of, on humidity, 18
on temperature, 12
on wind, 20
interception of precipitation by,
162-164, 108
Velocity distribution, in stream channel,
61
of wind, 20
Virtual channel inflow, 238- 239
Viscosity of water, 324
W index, 1 181
Water, artesian, 123, 145
connate, 132-133
juvenile, 132
properties of, 324
vapor pressure of, 324
viscosity of, 324
Water balance, 118- 119
Water content of snow, 46
Water equivalent of snow, 45--46

Water supply, design of reservoirs,
292-297
for casting of, 186-190
Water table, definition of, 122- 123
Water temperature, effect of, on evaporation, 97- 99, 105-107
on specific weigh t, 324
Water vapor, density of, 15
Water year, 72
Wa.ves, abrupt, 218-219
celerity of, 216-219
in natural channels, 68, 219- 220
rising, monoclinal, 216- 218
Wells, hydraulics of, 13 140
Wilting point, Ill, 126
Winds, 19-21
effect of, on evaporation, 95-99,
105-107
on precipitation measurements,
20-29
measurement of, 19
varia.tion of, with elevation . 20- 21
geographic, 20
Windshields, 28-29
Zone, of aeration, 122- 129
of saturation, 123, 129- 140

Hydrology For Engineers

Uploaded by

Copyright:

Available Formats

Hydrology For Engineers

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Hydrology For Engineers

Uploaded by

Copyright:

Available Formats

What is the book about?

What is the book about?

What topics are covered in the book?

What topics are covered in the book?

HYDROLOGY FOR ENGINEERS

RA Y K. LIN LEY, JR.

Profe !lor of Hydraulic Engiue ring

~f:~ ' ' ' /

MeGRA W-HIJ...L BOOK

HYDROLOGY FOR ENGINEERS

MeGRA W-HILL CIVIL ENGINEER INO SERIES

E. DAVIS, Consulting Editor

BABDITT, DOLAND, AND CLEASBY . Wnter upply Engineering

~ I ~'r'" '" rt '

The publication in 1949 of "Applied Hydrol gy" wa well re eiv d,

Symbols and Abbreviations

The hydrologic cycle-History-Hydrology

Weather and Hydrology

General circulation- Temperatur

Interpretation of streamflow data

Evaporation and Transpiration

Characteristics of the Hydrograph

Components of runoff-Recessions- Hydrograph separation

Phenomena of runoff-Estimating the volume of storm runofiht

Eatimating snowmelt runoff- easonal- and annual-runoff relations

Unit bydrographs- Overland Bow-Flood formulas

Frequency and Duration Stu die

Frequency analysis- Generalization of frequency data- Related

Erosion- Suspended sediment- Bed load- Sediment measurement- Re ervoir sedimentation

Application of Hydrologic Technique

Appendix B. Physical Con tants, Conversion Tables, and

SYMBOLS AND ABBREVIATIONS

SYMBOLS AND ABBREVIATIONS

i. supply rate (rainfall less 'r etention)

SYMBOLS AND ABBREVIATIONS

SYMBOLS AND ABBREVIATIONS

evaporation portion of advected energy

1'~ shear at ch'artnel bed

SYM80LS AND A88REVIATIONS

Hydrology is that branch of physical geography which is concern d

cases groundwater is a source of surface streamflow.

FIG. 1 1. The hydrologic cycle.

(Courtesy of U.S. Geological Survey.)

If the discussion of the hydrologic cycle gives any impre sion of a

should be understood, in broad d tilll at least, by the hydrologist. This

Mississippi River as early as 1848, but a systematic program was not

THE 3U8JECT MATTER OF HYDROLOGY

Hydrology Handbook, ASCE Manual 28, 1949.

The hydrologic characteristics of a region are det rmined largely by

If th earth were a nonrotating phere,

2-1. Thermal cir ulation.

THE GENERAL CIRCULATION

2-2. Factor m odifying the thermal circulation. The earth

FIG. 2-2. General circulation of Northern Hemisphere.

be required to produce these changes in velocity is known as the Coriolis

WEATHER AND HYDROLOGY

FIG. 2-3. M ean-pressure pattern of the Northern Hemisphere.

lossos are distributed through rclativ ly great depths in large bodies of

WEATHER AND HYDROLOGY