Engineering Hydrology by EM Wilson

Engineering Hydrology
E. M . WI L S 0 N, Ph.D., M.Sc., F.I.C.E., M.A.S.C.E.

Professor of Hydraulic Engineering
University of Salford
SECOND EDITION
Macmillan Education
©E. M. Wilson 1969, 1974
All rights reserved. No part of this publication

may be reproduced or transmitted, in any form
or by any means, without permission
First edition 1969
Second edition 1974
Published by
THE MACMILLAN PRESS LTD
London and Basingstoke
Associated companies in New York Dublin
Melbourne Johannesburg and Madras
SBN 333 17443 7
ISBN 978-0-333-17443-2 ISBN 978-1-349-02417-9 (eBook)
DOI 10.1007/978-1-349-02417-9
The paperback edition of this book is sold subject to the condition that it shall
not, by way of trade or otherwise, be lent, re-sold, hired out, or otherwise
circulated without the publisher's prior consent in any form of binding or
cover other than that in which it is published and without a similar condition
including this condition being imposed on the subsequent purchaser.
Contents
Preface to the Second Edition Vl
1 Introduction 1
1.1 Allied sciences 1

1.2 The hydrological cycle 2
1.3 Inventory of earth's water 3
1.4 Hydrology as applied in engineering 4
2 Meteorological Data 6
2.1 Weather and climate 6

2.2 Humidity 7
2.3 Temperature 10
2.4 Radiation 11
2.5 Wind 11
2.6 Precipitation 12
2.7 Forms of precipitation other than rain 17
2.8 The extension and interpretation of data 18
3 Evaporation and Transpiration 34
3.1 Meteorological factors 34

3.2 Transpiration 35
3.3 Methods of estimating evaporation 36
iii
iv Contents
3.4 Evaporation from land surfaces using Penman's
Eo value 44
3.5 Thornthwaite's formulae for evapo-transpira-
tion 45
3.6 Direct measurement of evaporation by pans 47
3.7 Consumptive use 48
4 Infiltration and Percolation 54

4.1 Infiltration capacity of soil 54
4.2 Factors influencingfc 54
4.3 Methods of determining infiltration capacity 58
4.4 The effect of soil moisture 61
5 Groundwater 69
5.1 The occurrence of groundwater 69
5.2 Factors of influence 70
5.3 Groundwater flow 74
5.4 The abstraction of groundwater 81
5.5 The yield of wells 82
6 Surface Runoff 91
6.1 The engineering problem 91
6.2 Flow rating curves: their determination, ad-
justment and extension 92
6.3 Duration of runoff 105
6.4 Catchment characteristics and their effects on
runoff 109
6.5 Climatic factors 115
6.6 Rainfall/runoff correlation 117
7 H ydrograph Analysis 120

7.1 Components of a natural hydrograph 120
7.2 The contribution of baseflow to stream dis-
charge 121
Contents v
7.3 Separation of baseflow and runoff 125
7.4 The unit hydrograph 129
7.5 Unit hydrographs of various durations 131
7.6 The unit hydrograph as a percentage distri-
bution 135
7.7 Derivation of the unit hydrograph 137
7.8 Unit hydrograph from complex or multi-period
storms 138
7.9 The instantaneous unit hydrograph 147
7.10 Synthetic unit hydrographs 148
8 Flood Routing 154

8.1 Introduction 154
8.2 The storage equation 154
8.3 Reservoir routing 156
8.4 Routing in a river channel 161
8.5 Graphical routing methods 168
8.6 Synthetic unitgraphs from flood-routing 171
9 Hydrological Forecasting 178

9.1 Introduction 178
9.2 Flood formulae 179
9.3 Frequency analysis 180
9.4 Synthetic data-generation 191
9.5 Rainfall data and the unit hydrograph 192
9.6 Hydro-meteorology 194
9.7 The cyclical nature of hydrological phenomena 197
Problems 199
General Bibliography 218
Answers to Problems 226
Index 228
Rijkoort's Nomogram of Penman's equation
facing page 232
Preface to the Second Edition
This book has been written for engineering students and junior
engineers who are learning about hydrology for the first time.
It is designed to introduce the reader to the elements of the subject
and to the underlying theories, to show how these are modified in
practice, and to describe the techniques which give answers in
particular cases.
Most texts of this nature are American in origin and tend to be
based exclusively on North American data and practice. This book
still relies heavily on the American development of hydrology but
also freely uses British and European data and references.
This edition has been revised and enlarged to provide more
readily available information about rainfall in the British Isles, and
in particular intensity-duration-frequency relationships. Recent
work in the United Kingdom on consumptive use and infiltration
has been incorporated and the section on stream gauging and flow
measurement expanded. The opportunity has been taken to improve
and update the references, to adopt SI units more uniformly through-
out (though still not completely) and to correct many of the errors of
the first edition. In this latter respect, I am particularly grateful to
correspondents from many parts of the world.
To improve the book when used as an undergraduate text, a
selection of problems has been given at the end of the main text, to
allow readers to use the techniques described in the text.
I must record my appreciation of the advice and help received
from many people during the book's preparation.
Manchester, 1974 E. M. WILSON

1 Introduction
The science of hydrology deals with the occurrence and movement of

water on and over the surface of the earth. It deals with the various
forms of moisture that occur, the transformation between the liquid,
solid and gaseous states in the atmosphere and in the surface layers
ofland masses. It is concerned also with the sea: the source and store
of all the water that activates life on this planet.
1.1 Allied Sciences

The engineer is normally concerned with the design and operation of
engineering works to control the use of water and in particular the
regulation of streams and rivers and the formation of storage reser-
voirs and irrigation canals. Nevertheless, he must be aware of this
application of hydrology in the wider context of the subject, since
much of the subject is derived from physics, meteorology, ocean-
ography, geography, geology, hydraulics and kindred"'sciences. He
must be aware of experience in forestry and agriculture and in botany
and biology. He must know probability theory, some statistical
methods and be able to use economic analysis.
Hydrology is basically an interpretive science. Experimental work
is restricted, by the scale of natural events, to modest researches into
particular effects. The fundamental requirement is observed and
measured data on all aspects of precipitation, runoff, percolation,
river flow, evaporation and so on. With this data and an insight into
the many bordering fields of knowledge, the skilled hydrologist can
provide the best solutions to many engineering problems that
arise.
1
2 Engineering Hydrology
1.2 The hydrological cycle
The cyclic movement of water from the sea to the atmosphere and
thence by precipitation to the earth, where it collects in streams and
runs back to the sea, is referred to as the hydrological cycle. Such a
cyclic order of events does occur but it is not so simple as that.
First, the cycle may short-circuit at several stages, e.g. the precipi-
tation may fall directly in the sea, in lakes or river courses.
Secondly, there is no uniformity in the time a cycle takes. During
droughts it may appear to have stopped altogether, during floods it
ground woler
flow
FIG. 1.1 The hydrological cycle
may seem to be continuous. Thirdly, the intensity and frequency of

the cycle depends on geography and climate, since it operates as a
result of solar radiation which varies according to latitude and
season of the year. Finally, the various parts of the cycle can
be quite complicated and man can exercise some control only on
the last part, when the rain has fallen on the earth and is making its
way back to the sea.
Although the concept of the hydrological cycle is oversimplified,
it affords a means of illustrating the most important processes that
the hydrologist must understand. The cycle is shown diagrammatic-
ally in Fig. 1.1.
Water in the sea evaporates under solar radiation and clouds of
water vapour move over land areas. Precipitation occurs as snow,
hail and rain over the land and water begins to flow back to the sea.
Introduction 3
Some of it infiltrates into the soil and moves down or percolates into
the saturated ground zone beneath the water-table, or phreatic sur-
face. The water in this zone flows slowly through aquifers to river
channels or sometimes directly to the sea. The water which infil-
trates also feeds the surface plant life and some gets drawn up into
this vegetation and transpiration takes place from leafy plant
surfaces.
The water remaining on the surface partially evaporates back to
vapour, but the bulk of it coalesces into streamlets and runs as sur-
face runoff to the river channels. The river and lake surfaces also
evaporate so still more is removed here. Finally, the remaining
water which has not infiltrated or evaporated arrives back at the sea
via the river channels. The groundwater, moving much more slowly,
either emerges into the stream channels or arrives at the coastline
and seeps into the sea, and the whole cycle starts again.
1.3 Inventory of earth's water

It is as well to have a clear idea of the scale of the events that are
being discussed. The following table lists estimates of the amounts of
water involved in the hydrological cycle and the proportion (in per-
centages) of the total water on earth involved in each part of it.
TABLE 1.1. Estimated earth's water inventory
Volume Percentage
Location 1000s ofkm3 total water
Fresh-water lakes
Rivers 12
i·25} 0·62
Soil moisture 65
Groundwater 8,250
Saline lakes and inland seas 105 0·008
Atmosphere 13 0·001
Polar ice-caps, glaciers and snow 29,200 2·1
Seas and oceans 1,320,000 97·25
TOTAL 1,360,000 100·0

or 1·36 x 101s m3
Of the 0·6% of total water which is available as fresh water about

half is below a depth of 800 m and so is not practically available on
the surface. This means that the stock of the earth's freshwater which
is obtainable one way or another for man's use is about 4 million km3
and is mainly in the ground. Spread over the earth's land surface it
would be about 30 m deep.
The four processes with which the hydrologist is mainly concerned
are precipitation, evaporation and transpiration, surface runoff or
stream flow, and groundwater flow. He needs to be able to interpret
data about these processes and to predict from his studies the most
likely quantities involved in the extreme cases of flood and drought.
He must also be able to express an opinion about the likely fre-
quency with which such events will occur, since it is on the frequency
of certain values of extreme events that much hydraulic engineering
design is based.
1.4 Hydrology as applied in engineering

To the practising engineer concerned with the planning and building
of hydraulic structures, hydrology is an indispensable tool. Suppose,
for example, that a city wishes to increase or improve its water
supply. The engineer first looks for sources of supply; having per-
haps found a clear uninhabited mountain catchment area, he must
make an estimate of its capability of supplying water. How much
rain will fall on it? How long will dry periods be and what amount
of storage will be necessary to even out the flow? How much of the
runoff will be lost as evaporation and transpiration? Would a sur-
face storage scheme be better than abstraction of the groundwater
flow from wells nearer the city?
The questions do not stop there. If a dam is to be built, what
capacity must the spillway have? What diameter should the supply
pipelines be? Would afforestation of the catchment area be bene-
ficial to the scheme or not?
To all these questions and many others which will arise the
hydrologist can supply answers. Often they will be qualified answers
and often also they will be given as probable values, with likely
deviations in certain lengths of time. This is because hydrology is
not an exact science. A contractor may be building a cofferdam in a
river and his hydrologist may tell him that, built to a certain height,
it will be overtopped only one~, on average, in 100 years. If it is a
temporary structure built for maybe 2 years' service only, the con-
tractor may decide this is a fair risk. However, it is a risk. One of
Introduction 5
the 2 years may be that in which the once in 100 years flood arrives,
and the science of hydrology cannot, as yet, predict this.
In a broader field of engineering, which is of great and increasing
importance, the development of water resources over a whole river
basin or geographical region may be under consideratio!l. In these
circumstances the role of the hydrologist is especially important.
Now his views and experience are of critical weight not only in the
engineering structures involved in water supply, but in the type and
extent of the agriculture to be practised, in the siting of industry, in
the size of population that may be supported, in the navigation of
inland shipping, in port development and in the preservation of
amenity.
Civilisation is primarily dependent on water supply. As the trend
towards larger cities and increasing industrialisation continues, so
will the role of the hydrologist increase in importance in meeting the
demands of larger populations for water for drinking, irrigation,
industry and power generation.
2 Meteorological Data
2.1 Weather and climate

The hydrology of a region depends primarily on its climate, secondly
on its topography and its geology. Climate is largely dependent on
the geographical position on the earth's surface. Climatic factors
of importance are precipitation and its mode of occurrence,
humidity, temperature and wind, all of which directly affect
evaporation and transpiration.
Topography is important in its effect on precipitation and the
occurrence of lakes, marshland and high and low rates of runoff.
Geology is important because it influences topography and because
the underlying rock of an area is the groundwater zone where the
water which has infiltrated moves slowly through aquifers to the
rivers and the sea.
The pattern of circulation in the atmosphere is complex. If the
earth were a stationary uniform sphere, then there would be a simple
circulation of atmosphere on that side of it nearest the sun. Warmed
air would rise at the equator and move north and south at a high
altitude, while cooler air moved in across the surface to replace it.
The high warm air would cool and sink as it moved away from the
equator until it returned to the surface layers when it would move
back to the equator. The side of the earth remote from the sun
would be uniformly dark and cold.
This simple pattern is upset by the earth's daily rotation, on its
own axis, which gives alternate 12 hour heating and cooling and also
produces the Coriolis force acting on airstreams moving towards or
away from, the equator. It is further upset by the tilt of the earth's
axis to the plane of its rotation around the sun, which gives rise to
6
Meteorological Data 7
seasonal differences. · Further effects are due to the different reflec-
tivity and specific heats of land and water surfaces. The result of
these circumstances on the weather is to make it generally complex
and difficult to predict in the short term. By observations of data
over a period of time, however, long term predictions may be made
on a statistical basis.
The study of hydrology necessitates the collection of data, inter
alia, on humidity, temperature, precipitation, radiation, and wind
velocity. All of these are considered in this chapter.
2.2 Humidity
Air easily absorbs moisture in the form of water vapour. The
amount absorbed depends on the temperature of the air and of the
water. The greater the temperature the more water vapour the air
can contain. The water vapour exerts a partial pressure usually
measured in either bars (1 bar= 105 N/m2 ; 1 millibar= 102 N/m 2)
or mm height of a column of mercury (Hg). (1 mm Hg = 1·36 mbar)
Suppose an evaporating surface of water is in a closed system and
enveloped in air. If a source of heat energy is available to the system,
evaporation of the water into the air will take place until a state of
equilibrium is reached when the air is saturated with vapour and
can absorb no more. The molecules of water vapour will then exert
a pressure which is known as saturation vapour pressure or e8 , for the
particular temperature of the system.
The value of es changes with temperature as indicated in Table 2.1.
These values are also plotted as a curve connecting e8 and oc in
Fig. 2.1. Referring to Fig. 2.1 consider what can happen to a mass
of atmospheric air P, whose temperature is t and whose vapour
pressure is e.
Since P lies below the saturation vapour pressure curve, it is clear
that the air mass could absorb more water vapour and that if it did
so while its temperature remained constant, then the position of P
would move vertically up dashed line 1 until the air was saturated.
The corresponding vapour pressure of P in this new position would
be ea. The increase (e8 - e) is known as the saturation deficit.
Alternatively, if no change were to take place in the humidity of
the air while it was cooled, then P would move horizontally to the
TABLE 2.1. Saturation vapour pressure e. in mm Hg (mercury) as a
function of temperature t in °C (Negative values oft refer to co~
tions over ice)
1 mmHg = 1·36 mbar
I
0·0 0·1 0·2 0·3 0·4 0·5 0·6 0·7 0·8 0·9
-10 2·15 -10
-9 2·32 2·30 2·29 2·27 2·26 2·24 2·22 2·21 2·19 2·17 -9
- 8 2·51 2·49 2·47 2·45 2·43 2·41 2·40 2·38 2·36 2·34 - 8
- 7 2·71 2·69 2·67 2·65 2·63 2·61 2·59 2·57 2·5S 2·53 -7
-6 2·93 2·91 2·89 2·86 2·84 2·82 2·80 2·77 2·75 2·73 - 6
- 5 3·16 3-14 3·11 3·09 3·06 3·04 3·01 2·99 2·97 2·95 - s
-4 3·41 3·39 3·37 3·34 3·32 3·29 3·27 3·24 3·22 3-18 - 4
- 3 3·67 3·64 3·62 3·59 3·57 3·S4 3·S2 3·49 3·46 3·44 - 3
-2 3·97 3·94 3·91 3·88 3·8S 3·82 3·79 3·76 3·73 3·70 - 2
- I 4·26 4·23 4·20 4•17 4·14 4·11 4·08 4·0S 4·03 4·00 -I
-o 4·S8 4·5S 4·52 4·49 4·46 4·43 4·40 4·36 4·33 4·29 - 0
----------------------------------------------------------------
0 4·S8 4·62 4·65 4·69 4·71 4·75 4·78 4·82 4·86 --------0
4·89
I 4·92 4·96 S·OO S·03 5·07 S·11 S·14 5·18 5·21 5·25 1
2 S·29 5·33 S·37 S·40 5·44 S·48 S·S.3 S·S7 5·60 5·64 2
3 5·68 5·72 5·76 S·80 5·84 S·89 5·93 S·97 6·01 6·06 3
4 6·10 6·14 6·18 6·23 6·27 6·31 6·36 6·40 6·4S 6·49 4
s 6·54 6·S8 6·54 6·68 6·72 6·77 6·82 6·86 6·91 6·96 5
6 7·01 7·06 7·11 7·16 7·20 1·25 7·31 7·36 7·41 7·46 6
7 7·51 7·56 7·61 7·67 7·72 7·77 7·82 7·88 7·93 7·98 7
8 8·04 8·10 8·15 8·21 8·26 8·32 8·37 8·43 8·48 8·54 8
9 8·61 8·67 8·73 8·78 8·84 8·90 8·96 9·02 9·08 9·14 9
10 9·20 9·26 9·33 9·39 9·46 9·S2 9·58 9·65 9·71 9·77 10
11 9·84 9·90 9·97 10·03 10·10 10·17 10·24 10·31 10·38 10·45 11
12 10·52 10·58 10·66 10·72 10·79 10·86 10·93 11·00 11·08 11·15 12
13 11·23 11·30 11·38 11·75 11·S3 11·60 11·68 11·76 11·83 11·91 13
14 11·98 12·06 12·14 12·22 12·96 12·38 12·46 12·54 12·62 12·70 14
IS 12·78 12·86 12·95 13·03 13·11 13·20 13·28 13·37 13·45 13·54 IS
16 13·63 13·71 13·80 13·90 13·99 14·08 14·17 14·26 14·3S 14·44 16
17 14·53 14·62 14·71 14·80 14·90 14·99 15·09 15·17 15·27 15·38 17
18 15·46 15·56 15·66 15·76 15·96 15·96 16·06 16·16 16·26 16·36 18
19 16·46 16·57 16·68 16·79 16·90 17·00 17·10 17·21 17·32 17·43 19
20 17·53 17·64 17·75 17·86 17·97 18·08 18·20 18·31 18·43 18·54 20
21 18·65 18·77 18·88 19·00 19·11 19·23 19·35 19·46 19·58 19·70 21
22 19·82 19·94 20·06 20·19 20·31 20·43 20·58 20·69 20·80 20·93 22
23 21·05 21·t9 21·32 21·45 21·58 21·71 21·84 21·97 22·10 22·23 23
24 22·27 22·SO 22·63 22·76 22·91 23·05 23·19 23·31 23·45 23·60 24
25 23·75 23·90 24·03 24·20 24·35 24·49 24·64 24·79 24·94 25·08 25
26 2S·31 25·4S 25·60 25·74 25·89 26·03 26·18 26·32 26·46 26·60 26
27 26·74 26·90 27·0S 27·21 27·37 27·53 27·69 27·85 28·00 28·16 27
28 28·32 28·49 28·66 28·83 29·00 29·17 29·34 29·51 29·68 29·85 28
29 30·03 30·20 30·38 30·56 30·74 30·92 31·10 31·28 31·46 31·64 29
30 31•82 32·00 32·19 32·38 32·57 32·76 32·95 33·14 33·33 33·52 30
0·0 0·1 0·2 0·3 0·4 0·5 0·6 0·7 0·8 0·9
left along line 2 until the saturation line was intersected again. At
this point P would be saturated, at a new temperature ta, the dew-
point. Cooling of the air beyond this point would result in condensa-
tion or mist being formed.
If water is allowed to evaporate freely into the air mass, neither
of the above two possibilities occurs. This is because the evaporation
requires heat which is withdrawn from the air itself. This heat,
called the latent heat of evaporation, hr is given by the equation
hr = 606·5 - 0·695t g cal/g (2.1)

So, as the humidity and vapour pressure rise, the temperature of the
air falls and the point P moves diagonally along line 3 until satura-
tion vapour pressure is reached at the point defined by ew and tw.
This temperature tw is called the wet bulb temperature and is the
temperature to which the original air can be cooled by evaporating
34r--,---.---.---.--.---.---.--.---.--~/r-~
321--
301------l------l------~--~--~--+---+---+---+--~/--~
C) 28 ;{ r
:I: 26
E
.E 24
q)' 22
~
~ 20
~
a.
18
~
::1 16
8.
g 14
c:: 12
.!2
E 10
.2 8
0
(f)
6
4
2
-10 -8 -6 -4 -2 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34
Temperature in °C
Flo. 2.1 Saturation vapour pressure of water in air
water into it. This is the temperature found by a wet bulb thermo-
meter.
The relative humidity is now given as
h = efes, or as a percentage, h = 100 efe8 % (2.2)
and is a measure of the air's capacity, at its existing temperature, to

absorb further moisture, It is measured by blowing air over two
thermometers, one with its bulb wrapped in wet muslin and one dry.
The air flow past the bulb has an influence on the wet bulb reading
and the two thermometers may either be whirled around on a string
or more conveniently have the air current provided by a clockwork
fan. In this latter case the instrument is called a psychrometer.
The value of e for air temperature t may be obtained from the
equation,
(ew - e) = y(t - tw) (2.3)
where tw = wet bulb temperature
t = dry bulb temperature
ew = the corresponding partial pressures for tw (from
Table 2.1)
y = psychrometer constant = 0·66 where e in mbar, t in oc
and assuming an air speed past bulbs of at least 3 m/s
(when e is measured in mm Hg, y = 0·485)
2.3 Temperature
Air temperature is recorded by thermometers housed in open
louvred boxes, about four feet above ground. Protection is necessary
from precipitation and the direct rays of the sun.
Many temperature observations are made using maximum and
minimum thermometers. These record by indices, the maximum and
minimum temperatures experienced since the instrument was last set.
The daily variation in temperature varies from a minimum around
sunrise, to a maximum from ! to 3 hours after the sun has reached
its zenith, after which there is a continual fall through the night to
sunrise again. Accordingly, max. and min. observations are best
made in the period 8 a.m.-9 a.m. after the minimum has occurred.
The mean daily temperature is the average of the maximum and
minimum and is normally within a degree of the true average as
continuously recorded.
Temperature is measured in degrees Celsius, commonly, though
erroneously, called centigrade. The Fahrenheit scale is also still used
in Britain and almost exclusively in North America.
Vertical temperature gradient. The rate of change of temperature

in the atmosphere with height is called the lapse rate. Its mean value
is 6·5°C per 1000 m height increase. This rate is subject to variation,
particularly near the surface which may become very warm by day,
giving a higher lapse rate, and cooling by night giving a lower lapse
rate. The cooling of the earth, by outward radiation, on clear nights
may be such that a temperature inversion occurs with warmer air
overlying the surface layer.
As altitude increases, barometric pressure decreases so that a unit
mass of air occupies greater volume the higher it rises. The temper-
ature change due to this decompression is about I ooc per I 000 m if
the air is dry. This is the dry-adiabatic lapse rate. If the air is moist,
then as it is lifted, expanding and cooling, its water vapour content
condenses. This releases latent heat of condensation which pre-
vents the air mass cooling as fast as dry air. The resulting saturated-
adiabatic lapse rate is therefore lower, at about 5·6°C per 1000 min
the lower altitudes.
Distribution of temperature. Generally, the nearer the equator a

place is the warmer it is. The effects of the different specific heats of
earth and water, the patterns of oceanic and atmospheric currents, the
seasons of the year, the topography, vegetation and altitude all tend
to vary this general rule, and all need consideration.
2.4 Radiation
Most meteorological recording stations are equipped with radiometers
to measure both incoming short-wave radiation from sun and sky and
nett radiation which is the algebraic sum of all incoming radiation
and the reflected short and long wave radiation from the earth's sur-
face. The nett radiation is of great importance in evaporation studies
as will be seen in Chapter 3.
2.5 Wind
Wind speed and direction are measured by anemometer and wind
vane respectively. The conventional anemometer is the cup anemo-
meter formed by a circlet of three (sometimes four) cups rotating
around a vertical axis. The speed of rotation measures the wind
speed and the total revolutions around the axis gives a measure of
wind run, the distance a particular parcel of air is moving through in
a specified time.
Because of the frictional effects of the ground or water surface
over which the wind is blowing it is important to specify in any ob-
servation of wind, the height above ground at which it was taken. An
empirical relationship between wind speed and height has been
commonly used
ufuo = (z/zo)0 "15 (2.4)
where uo = wind speed at anemometer at height z0
u = wind speed at some higher level z
In recent years there has been some effort to standardise observa-
tion heights and in Europe wind speed is usually observed 2 m above
the surface.
Fig. 2.2 shows an instrument array for making meteorological
observations at regular, short time-intervals. Instruments, which
record automatically on magnetic tape, include nett-radiation
radiometer, wet and dry bulb thermometers, wind vane, anemometer
and incident solar radiometer at the mast-top.
2.6 Precipitation
The source of almost all our rainfall is the sea. Evaporation takes
place from the oceans and water vapour is absorbed in the air
streams moving across the sea's surface. The moisture-laden air
keeps the water vapour absorbed until it cools to below dewpoint
temperature when the vapour is precipitated as rain, or if the temper-
ature is sufficiently low, as hail or snow.
The cause of the fall in temperature of an air mass may be due to
convection, the warm moist air rising and cooling to form cloud and
subsequently to precipitate rain. This is called convective precipita-
tion. This is typified by the late afternoon thunderstorms which
develop from day long heating of moist air, rising into towering
anvil-shaped clouds. Orographic precipitation results from ocean air
streams passing over land and being deflected upward by coastal
mountains, thus cooling below saturation temperature and spilling
moisture. Most orographic rain is deposited on the windward slopes.
The third general classification of rainfall is cyclonic and frontal
precipitation. When low pressure areas exist air tends to move into
them from surrounding areas and in so doing displaces low pressure
air upward, to cool and precipitate rain. Frontal rain is associated
with the boundaries of air masses where one mass is colder than the
other and so intrudes a cool wedge under it, raising the warm air to
form clouds and rain. The slope of these frontal wedges can be quite
flat and so rain areas associated with fronts may be very large.
2.6.1 Recording precipitation. Precipitation occurs mainly as rain,

but may occur as hail, sleet, snow, fog or dew. Britain has a humid
FIG. 2.2 Meteorological observation array. On lower arm 1.

nett-radiation; r. wet and dry bulb thermometers; on upper arm I.
wind direction; r. wind run: at top, solar and sky radiation.
There is a rain gauge with anti-splash screen in the middle
distance
climate and rain provides the great bulk of its moisture, but in other
parts of the world, precipitation may be almost entirely snow, or, in
arid zones, dew.
In the United Kingdom rainfall records are received and recorded
by the Meteorological Office from some 6500 rain gauges scattered
over Gt. Britain and N. Ireland, the majority giving daily values of
rainfall. In addition there are a further 260 stations also equipped
with recording rain gauges which record continuously.
Standard rain gauges in Britain are made from copper and consist of
a 5-inch diameter copper cylinder, with a chamfered upper edge, which
collects the rain and allows it to drain through a funnel into a remov-
able container of metal or glass from which the rain may be poured
into a graduated glass measuring-cylinder each day. There are
prescribed patterns for the standard gauge and for its installation and
operation.
Recording gauges (or autographic rain recorders) usually work by
having a clockwork-driven drum carrying a graph on which a pen
records either the total weight of container plus water collected, or
a series of blips made each time a small container of known
capacity spills its contents. Such gauges are more expensive and
more liable to error but may be the only kind possible for remote,
rarely visited sites. They have the great advantage that they indicate
intensity of rainfall which is a factor of importance in many problems.
For this reason some stations are equipped with both standard and
recording gauges.
The Meteorological Office has recently designed a new range of
rain gauges. [I] The new standard gauge for daily rainfall measure is a
circular catchment of 15 x 103 mm2 (5·5 in. diameter) installed at a
rim height of 300 mm above ground level. The larger and more accu-
rate new gauge has an area of75 x 103 mm 2 (12·2 in. diameter) also
set with a rim 300 mm above ground. The material used in manufac-
ture is fibreglass. New tipping-bucket mechanism has been designed
and is available with a telemetry system to provide for distant reading
by telephone interrogation. The gauge is provided with a telephone
connection and number, which can be dialled in the ordinary way.
The quantity of rain collected since the last setting of the gauge to
zero is transmitted in increments of I mm by three groups of audible
tones representing hundreds, tens and units. Interrogation can be
made as frequently as desired and hence intensities can be obtained,
by simple subtraction, with minimum delay.[2]
In recent years there has been much research into the effects of
exposure on rain gauges and it is now generally accepted that more
accurate results will be obtained from a rain gauge set with its rim at
ground level, than one with its rim some height above ground [3]. It
is necessary in a ground-level installation to make a pit to house tl;le
gauge and cover it with an anti-splash grid. Accordingly the ground-
level gauge is more expensive to install and maintain.
A typical installation is illustrated in Fig. 2.3.
Yearly records for the whole country, statistically analysed and
presented graphically, are published annually by the Meteorological
Office in a booklet entitled British Rainfall followed by the particular
year concerned. The use of rainfall data is discussed in Section 2.8.
2.6.2 Rain-gauge networks. A question which frequently arises is

that concerning the number and type of rain gauges which are neces-
sary to ensure that an accurate assessment of a catchment's rainfall is
obtained. Bleasdale [4] quotes Tables 2.2 and 2.3 as general guides
and comments as follows
The disparity between the two tables is not so great as might appear at
first sight. The first indicates station densities which are reached in
important reservoired areas and which may well be exceeded in small
experimental areas. The second indicates densities which are more
appropriate for country-wide networks. In the application of the general
TABLE 2.2. Minimum numbers of rain gauges required in

reservoired moorland areas
Rain gauges
mile2 km2 Daily Monthly Total
0·8 2 1 2 3
1·6 4 2 4 6
7·8 20 3 7 10
15·6 41 4 11 15
31·3 81 5 15 20
46·9 122 6 19 25
62·5 162 8 22 30
Source: Jot Comm. Met. Off., R. Met. Soc. "and Instn Wat. Engrs Rpt
on "The determination of the general rainfall over any area". Trans.
Instn Wat. Engrs, 42 (1937) p. 231.
FIG. 2.3 Automatic recording rain gauge set with rim at ground
level and with anti-splash screen
guidance embodied in this Table [2.3] it must be appreciated that any
large river basin will almost invariably have within it a number of sub-
basins for which the relatively dense networks would be recommended.
Moreover, the minimum densities suggested would often be substan-
tially increased in mountainous areas, and would be closely followed only
in areas of low or moderate elevation without complex topography.
TABLE 2.3. Minimum numbers of rain gauges for monthly

percentage-ofaverage rainfall estimates
Area
Number of
mile 2 km2 rain gauges
10 26 2
100 260 6
500 1300 12
1000 2600 15
2000 5200 20
3000 7800 24
Source: Bleasdale [4]
There is considerable material devoted to this question of hydro-

logical network design and the reader is referred to the further
references in the General Bibliography (p. 220).
2.7 Forms of precipitation other than rain
Snow and ice. Snow has the capacity to retain water and so acts as
a form of storage. Its density and, therefore, the quantity of water
contained, varies from as little as 0·005 for newly-fallen snow to as
much as 0·6 in old, highly compressed snow. Its density varies with
depth and so samples must be taken at various horizons in a snow
pack before the water equivalent can be computed. This is usually
done by sampling tube.
Snowfall may be measured directly by an ordinary rain gauge
fitted with a heating system, or by a simple snow stake, if there is no
drifting and density is determined simultaneously.
Snow traverses are made as field surveys along lines across catch-
ments, to determine snow thicknesses and densities at depth so that
water equivalents may be calculated for flood forecasts.
Fog. Estimates of amounts of moisture reaching the ground from
fog fQrmation have been made by installing fog collectors over
standard rain gauges. Collectors consist of wire gauge cylinders
on which moisture droplets form and run down into the rain gauge.
Comparison with standard rain gauge records at the same locality
shows differences which are a measure of fog precipitation. The
interpretation of such data requires experience and the use of
conversion factors, but can make substantial differences (of the order
of 50-100%) to precipitation in forest areas.
Dew. Dew collectors have been used in Sweden and Israel to

measure dew fall. They have been made as conical steel funnels,
plastic coated and with a projected plan area of about 1 square metre.
Dew ponds are used as a source of water in some countries. They
are simply shallow depressions in the earth lined with ceramic tiles.
Condensation. Although fog and dew are condensation effects, con-

densation also produces precipitation from humid air flows over ice
sheets and in temperate climates by condensation in the upper
layers of soil. Such precipitation does not occur in large amounts
but may be sufficient to sustain plant life.
2.8 The extension and interpretation of data
2.8.1 Definitions. The total annual amount of rain falling at a point

is the usual basic precipitation figure available. For many purposes,
however, this is not adequate and information may be required under
any or all of the following headings:
(i) Intensity. This is a measure of the quantity of rain falling in

a given time, e.g. mm per hour.
(ii) Duration. This is the period of time during which rain falls.
(iii) Frequency. This refers to the expectation that a given depth
of rainfall will fall in a given time. Such an amount may be
equalled or exceeded in a given number of days or years.
(iv) Areal extent. This concerns the area over which a point's
rainfall can be held to apply.
2.8.2 Intensity-duration relationship. The greater the intensity of

rainfall, in general, the shorter length of time it continues. A formula
expressing the connection would be of the type
. a
l=-- (2.5)
t+b
where i = intensity: mmjh, t = time: h, a and bare locality constants,
and for durations greater than two hours
. c (2.6)
l =-
rn
where c and n are locality constants.
The world's highest recorded intensities are of the order of 30 mm
(or I! in.) in a minute, 200 mm (or 8 in.) in 20 min and 26,000 mm
(or 1000 in.) in a year. Further information is given in Section 9.6.
2.8.3 Intensity-duration-frequency relationships. In 1935 E. G.

Bilham published his well-known article on these relationships in the
United Kingdom [5] containing a graph reproduced here as Fig. 2.4.
This graph used the subjective phraseology of "very rare", "remark-
able" and "noteworthy" rather than frequency of occurrence.
However, the frequencies were calculable from the formula
n = 1·25t(r + 0·1)-3·55 (2.7)
80
I I I I I
1111 I !
1..-i- 1-- ......

70
1 lloi\S
E
g
60
[..... ........
~1111
I I I I I
50
_!; ]....- I
~
v emorl<o~
'0 40
c:J
17 1-- ~lTll
0 1-- J-- I I
E 30 [....- NoleworlhY tolls
v
I
t i lI lI II II II
<t
1--
20 -1-- 1--
1-- ...... I
1-1-
[....- Falls tao numerous for discussion
10 1-1-
I I I I I I I
T TT I I
0 10 20 30 40 50 60 70 80 90 100 110 120
Time (mins)
FIG. 2.4 Bilham's rainfall classification

where n = number of occurrences in 10 years
r = depth of rain in inches
t = duration of rain in hours
The SI form of the formula is
n = 1·214 X 105t(P + 2·54)-3·55 (2.8)
where P = depth of rain in mm, and n and t have unchanged units.
In hydraulic engineering, however, it is usually the recurrence in-
terval for a given depth of rain in a given time which is wanted. This
is obtained by rearranging Eq. (2.8), so that
(1·214 X 105)NT) -3·55
P= ( - 2·54 (2.9)
600
where N =occurrence frequency, now expressed as once in N
years= 10/n
T = duration in minutes
If intensities are required, the depth P is divided by the time in hours,
i.e.
/h
'=rmm
. 60P (2.10)
and substituting the value of P from (2.9)
i = ~ ((202·3NT)-3·55- 2·54) mm/h (2.11)
Equation (2.11) is the SI form of the Bilham equation used in

reference [6].
Bilham's work was revised and extended by Holland [7] who
showed that Bilham's equations over-estimate the probabilities of
high-intensity rainfall, i.e. above about 35 mmfh. This later work is
best illustrated graphically and Fig. 2.5 shows both Bilham's formu-
lae (chain-dotted lines) and Holland's revisions (full lines). The figure
gives a return period for specific depths of rain occurring in specific
periods of time, as averaged over 14 station-decades in England.
Another way of presenting such data, this time based on the corre-
lation between annual average rainfall in Britain and one-day maxi-
mum rain-depth for various return periods is shown in Fig. 2.6.
For a specific locality it is often possible to produce curves such as
those shown in Fig. 2. 7 for Oxford, England, and Kumasi, Ghana,
-----.From a 7yr. (1956-62) assessment averaged over 20 stations in England
o o From a 35yr (1926-60) assessment averaged over a group of 10 stations
- - ·-·-·· From Bilhorns original forrrulo
~ 1ne ro1nTo11 Oepfn on eacn pow Of curves Tolls tn rne rme morkeO on rne oosneo Time curves Of rne l'llliCOTeo rrequeru y
400I '\
111 I \ -~,
300I 11
~i~ I I \ '~
~ 200I
''
1 I\\ \
~ 150
'~ ',
u II~
I . ~\ s"',;;',
100
'' ~
.Q
" BO
:E . I" to
. 60 ! 'l. ~\'I ' ........~,, ', -.
~ ' "'"'' ~
2 ~ s:
40 11 \i. \~~6~,.,, r-:::.,. ' ' !!. CD
"""'t ~' 1--.
i ~~~~\
~~\ ........._~ '~ ~...:- t---
', c-..... .......
- - -· .10
...
CD
0
-
....
20I ' ......,._ ' ....-._.., ~'----... ' 5 .2 ~
a;
:fii;
\~ <? 0. 0
'''I~~~
\\ I·"~~ \ ... ~ CCI
0 \I I I ::::::,. ' ', '
~ ~~ ~ r-.._ ......... ~"'';, ....... E c:;
10 10 ::J
B <n ~
i =I"(I)
~
........ "J . . . . .
6 - :3" ........
"0 ' .......
cD)
-~ ' ' .... ,~ 20
""~ 4
.1> ... ~ D)
0 ?; \ ' ', ' ... .... ............ ................... -

3 ~- -~ '-t, .... ' ~
1. 0~ '<so -..........::._....
~4)
~ 2 ,4)- -
'to,,..,
50
5 ... ....
z 's
I'' "';,
. -
I 100
10 -
20 -
30 40 --
50 --
60 -
70 80 90 100 110 120 130 140 150
Rain intensity mm/hr
FIG. 2.5 Rain intensity frequency graph (after D. J. Holland: "British Rainfalll961", Crown Copyright).
Example: A rainfall of 20 mmfalling in a 30 min period may be expected, anywhere in Britain, on average .....,
.....
once in 11.1 yr or 9 times a century.
respectively. The difference between coastal temperate and tropical
climates may be noted.
The data may also be presented in the form of maps of a region,
with isohyetal lines indicating total rainfall depth that may be ex-
pected in a time t, at a frequency of once in N years. A classic publi-
cation of this type by Yarnall [8] shows such maps for the United
States. Fig. 2.8 is a typical one, reproduced from Yarnall's paper and
showing the five minute rainfall that may be expected once in 50
years.
175
150
e125 5
.§
~100
·;c
~
b 1000 1500 2000 2500 3000 °

Average arniCII rainfall (mm)
FIG. 2.6 Relationship between one day maximum rainfall for
given return period and average annual rainfall in United Kingdom
(after Inst. of Hydrology)
Recently in the United Kingdom the Institute of Hydrology and

the Meteorological Office have collaborated in a flood study project.
A major part of the meteorological contribution has been the analysis
of very large quantities of daily rainfall data and data from auto-
graphic gauges. It is now possible to determine objectively the aver-
age rainfall intensity likely to be attained or exceeded at any desig-
nated point (or over any size area) in the United Kingdom in a
specified duration and with a specified return period. In addition
two sets of "storm profiles" are available-one set for the summer
and the other for the winter half-year. A profile is selected by the
engineer on the basis of the percentage of storms chosen to attain or
exceed a given degree of profile "peakedness". A designer may
apply a chosen rate of rainfall and a duration to a selected profile
with a knowledge of the return period of the composite phenomenon.
The latter information is of particular relevance to storm sewer
design.
20 808
IllI! - - Kumasi, Ghana
-
----- Oxford
II
10 ~ t--... I I
254
-1----
8 203
....... :-..... ....... ~
6
-........... r---... ....... ~ I
152
............. [""'-. I"' I:'- ~'" ,...., I
----
4 ....... 1::',-to .a 102
r-.. ~"'r-.
I
-- -- "'"<tv I
3 t-..<'crto
..... r-. 'o-~~
t----- --
i
......
r'- $ ~~'".r)
~ '
0
.JC
2
---- ~,
' j',i, .j I'~~~~

~~~~~
!:
51
!'
1--
'~ .....
j',h i',
'
-- --
I
<>
I 25
..
.E >-
0·8 '... 20 ~
"
>-
""enc:: 0·6 '' 0
'·"·
... '>,. I5 £
!!
.E
0-4 r - - ·
' '
$
...,
' ......
' ,l, "1'- 1\
~"' I0
~
!
:2c:: ! ' ~~
--
- l, 'r..; ~~
·e 03 ~
a::
a:: I
0·2
',k
' 5
I
I' ' ".....~~
0·1
I I I I
-, I I ','"
... ' '~ 3
0·08
0·06 '
- 2
I·5
i '
I ' ...
0·04 I0
003 ·-··- ·-
I
0·02 0·5
6 12 18 24 36 48 60 2 34 6810 20 24
Minutes Hours
Duration
FIG. 2.7 Rainfall frequency-intensity-duration relationships
taken at Kumasi, Ghana (courtesy of Ghana Meteorological
Service) and Oxford
2.8.4 Depth-area-time relationships. Precipitation rarely occurs

uniformly over an area. Variations in intensity and total depth of fall
occur from the centres to the peripheries of storms [9]. The form of
variation is illustrated in Fig. 2.9, which shows for a particular
FIG. 2.8 Five minute rainfall, in inches, to be expected once in

50 years in continental U.S.A. (after Yarnall)
max. depth P
E ~
E
" '-.:
........
mean depth p
..................
.........
:g .................
c:
·c; .......
. a::
................
........................
.............. .._
Area : km2 - - - -
FIG. 2.9 Depth-area curves of rainfall

storm, how the average depth of fall decreases from the maximum as
the considered area increases.
It is useful, however, to quantify this and Holland has shown [10]
that the ratio between point and areal rainfall over areas up to
10 km 2 and for storms lasting from 2 to 120 minutes is given by
~=I_ 0·3y'A (2.12)

p t*
where P = average rain depth over the area
P = point rain depth measured at the centre of the area
A =the area in km2
t* = an "inverse gamma" function of storm time obtained
from the correlation in Fig. 2.10.
/
6 /
/
v ~
/
V"
5
,.
vv
v
v
4
v
3
2
I 2 4 6 8 10 20 40 60 80 100 200 300
Time t (minutes)
FIG. 2.10 Correlation between storm t and t* (see ref [11])
Example 2.1. What is the average rainfall intensity over an area of

5 km 2 during a 60-minute storm with a frequency of once in I 0 years, in
Britain?
From Fig. 2.5, the frequency line of once in 10 years just cuts the
25 mm depth at about I hour .·. P = 25 mm. Given t = 60 min
then from Fig. 2.10, t* = 5·6 then
!_ = I - 0 "3 V 5 =I - O·I2 = 0·88

J> 5·6
P = 0·88J> = 0·88 x 25 = 22 rom in I hour
Check from Oxford's figures: From Fig. 2. 7 the 10 year frequency for
60 min duration indicates 23 rom in I hour.
.·. P = 0·88 x 23 = about 20 rom in I hour
In the assessment of total quantities of rainfall over large areas, the

incidence of particular storms and their contribution to particular
gauges is unknown, and it is necessary to convert many point values
to give an average rainfall depth over a certain area. The simplest
way of doing this is to take the arithmeti~al mean of the amounts
known for all points in the area. If the distribution of such points
over the area is uniform and the variation in the individual gauge's
amounts are not large, then this method gives reasonably good
results.
Another method, due to Thiessen [12], defines the zone of influence
of each station by drawing lines between pairs of gauges, bisecting
the lines with perpendiculars, and assuming all the area enclosed
within the boundary formed by these intersecting perpendiculars
has had rainfall of the same amount as the enclosed gauge.
A variation of this technique is to draw the perpendiculars to the
lines joining the gauges at points of median altitude, instead of at
mid length. This altitude-corrected analysis is sometimes held to be
a more logical approach but as a rule produces little difference in
result. Either method is more accurate than that of the simple arith-
metic mean but involves much labour. Thiessen polygons are illus-
trated in Fig. 2.11.
A third method is to draw isohyets, or contours of equal rainfall
depth. The areas between successive isohyets are measured and
assigned an average value of rainfall. The overall average for the
area is thus derived from weighted averages. This method is possibly
the best of the three and has the advantage that the isohyets may be
drawn to take account oflocal effects like prevailing wind and uneven
topography. A typical isohyetal map is shown in Fig. 2.12 though the
fall recorded is far from typical, this being the heaviest recorded
daily fall in the United Kingdom.
2.8.5 Supplementing rainfall records. It frequently happens when
assembling rainfall data that there are areas inadequately recorded,
particularly regarding intensities of rainfall. For example, suppose
that at two rainfall stations A and B, there is a recording gauge at A
and a non-recording gauge at B. Suppose the mass curve of rainfall
at A is as shown in the full line on the graph of Fig. 2. 13. The total
The area assumed to have the rainfall of a particular

gauging station is enclosed by the dashed I ines and
catchment boundary
FIG. 2.11 Thiessen polygons
rainfall at B is known and appears as a point on this graph. If the

physical location of B is near A and its rainfall is likely to be of
the same kind and frequency, then it is permissible to assume that
the mass curve of B will be as shown by the dashed line on the graph.
This kind of extension of data should be used with care but can be
very useful.
• • • • • ....'"'....t
•
•
• . .,
lin~
\ ..
• Rain gouges
...~...... County boundaries
Scoleo-----------t-----'lt:A-----:::;~--t-----1
1012345
I I I I I I 'mileS
FIG. 2.12 Rainfall over Dorset. July 18, 1955. (Reproduced from
British Rainfall 1955. HMSO 1957)
Another example of this is the filling in of a gap in a station's

records, when those for neighbouring stations have provided data
for the missing period. Suppose for a certain year there is no record
of the precipitation at A. In the same year the total at B has been
650mm.
Assuming that the mean annual precipitation at A and B respec-
tively is 700 mm and 600 mm then by simple proportion, assuming
the average relationship holds for the missing year also, the missing
years, precipitation at A will be ~~ x 650 = 758 mm. This result
may be checked by a similar reference to station C.
1
t
.
.s::
Q.
"0
recorded mass
curve for stn. A a_//''..-----
I
1
___ :
1
~
,.,.,./ derived mass

/-::::.--curve for sin. B
/
I
I
I
:;/
,.....:;.--
Time - -
FIG. 2.13 Derivation of rainfall data
X
c:
;;;
g
c:
c:
0
:E obrupt change indicates

.
.!2
.=:
change of circumstances
in gauge or observer
~
::>
E
::>
u
---
Cumulative mean total annual rainfall at a number of nearby stat ions
FIG. 2.14 Station check by double mass curve

2.8.6 Apparent trends in observed data. From several years' records
it may seem that annual rainfall is, say, declining. It is important
to know that this trend is independent of the gauging, and is due
only to meteorological conditions. This may be checked by plotting
a double mass curve as shown in Fig. 2.14.
A sudden divergence from the straight line correlation indicated
by the dashed line in the figures, indicates that a change has occurred
in gauging and that the meteorology of the region is probably not
the cause of the decline. Such a change might be due to the erection
of a building or fence near the gauge which changes the wind pattern
round the gauge, the planting of trees, th~ replacement of one
measuring vessel by another, even the changing of an observer who
uses different procedures.
Units of time -
01ITI!llll1fi"~
FIG. 2.15 Trends from progressive averages
2.8.7 Trends from progressive averages. Trends may be more clearly

discerned by the use of the simple statistical technique of examining
averages over longer periods, and moving the group averaged one
year at a time.
Suppose the rainfall records at a station over a number of years
are as shown in Fig. 2.15. The first five years on the record are
averaged and this average is plotted at the mid point of the group.
The next point is obtained by omitting the first year and averaging
years 2-6 inclusive, again plotting the average at the mid point of
the group. In this way the wide variations of particular years are
smoothed out and long term trends may be detected.
The same techniques may be applied to temperature, hours of
sunshine, wind speeds, cloud cover and other data.
Rainfall maps for Great Brit~in and Ireland are printed on the
following two pages; by courtesy of the Meteorological Office and
the Irish Meteorological Service respectively. Using these maps, the
long-term average annual rainfall for any locality may be deter-
mined, for use with Fig. 2.6. Note that different units are used for the
two maps.
AVERAGE ANNUAL RAINFALL
1916 - 1950
32
AVERAGE ANNUAL RAINFALL
1931 - 1960
33
3 Evaporation and Transpiration
3.1 Meteorological factors

Evaporation is important in all water resource studies. It affects the
yield of river basins, the necessary capacity of reservoirs, the size of
pumping plant, the consumptive use of water by crops and the yield
of underground supplies, to name only a few of the factors affected
by it.
Water will evaporate from land, either bare soil or soil covered
with vegetation, also from trees, impervious surfaces like roofs
and roads, open water and flowing streams. The rate of evaporation
will vary with the colour and reflective properties of the surface (the
albedo) and will be different for surfaces directly exposed to, or
shaded from, solar radiation.
In moist temperate climates the loss of water through evaporation
may typically be 600 mm per year from open water and perhaps 450
mm per year from land surfaces. In an arid climate, like that of Iraq,
the corresponding figures could be 2000 mm and 100 mm, the great
disparity in this latter case being caused by absence of precipitation
for much of the year.
Some of the more important meteorological factors affecting
evaporation are discussed below.
Solar radiation. Evaporation is the conversion of water into water

vapour. It is a process which is taking place almost without inter-
ruption during the hours of daylight and often also during the night.
Since the change of state of the molecules of water from liquid to gas
requires an energy input (known as the latent heat of vaporisation),
the process is most active under the direct radiation of the sun. It
34
Evaporation and Transpiration 35
follows that clouds which prevent the full spectrum of the sun's
radiation reaching the earth's surface will reduce the energy input
and so slow up the process of evaporation.
Wind. As the water vaporises into the atmosphere the boundary

layer between earth and air becomes saturated and it must be re-
moved and continually replaced by dryer air if evaporation is to
proceed. This movement of the air in the boundary layer depends
on wind and so the wind speed is important.
Relative humidity. The third factor affecting evaporation is the rela-

tive humidity of the air. As the air's humidity rises, its ability to
absorb more water vapour decreases and the rate of evaporation
slows. Replacement of the boundary layer of saturated air by air of
equally high humidity will not maintain the evaporation rate: this
will only occur if the incoming air is drier than that which is dis-
placed.
Temperature. As mentioned above, an energy input is necessary for

evaporation to proceed. It follows that if the ambient temperature
of the air and ground is high, evaporation will proceed more rapidly
than if they are cool, since heat energy is more readily available.
Since the capacity of air to absorb water vapour increases as its
temperature rises, so air temperature has a double effect on how much
evaporation takes place, while ground and water temperatures have
single direct effects.
3.2 Transpiration
Growing vegetation of all kinds needs water to sustain life, though
different species have very different needs. Only a small fraction of
the water needed by a plant is retained in the plant structure. Most
of it passes through the roots to the stem or trunk and is transpired
into the atmosphere through the leafy part of the plant.
In field conditions it is practically impossible to differentiate be-
tween evaporation and transpiration if the ground is covered with
vegetation. The two processes are commonly linked together and
referred to as evapo-transpiration.
The amount of moisture which a land area loses by evapo-
transpiration depends primarily on the incidence of precipitation,
secondly on the climatic factors of temperature, humidity etc. and
thirdly on the type, manner of cultivation and extent of vegetation.
The amount may be increased, for example, by large trees whose
roots penetrate deeply into the soil, bringing up and transpiring
water which would otherwise be far beyond the influence of surface
evaporation.
Transpiration proceeds almost entirely by day under the influence
of solar radiation. At night the pores or stomata of plants close up
and very little moisture leaves the plant surfaces. Evapotation, on
the other hand, continues so long as a heat input is available, and
accordingly primarily during the day. The other factor of importance
is the availability of a plentiful water supply. If water is always avail-
able in abundance for the plant to use in transpiration, more will be
used than if at times less is available than could be used. Accordingly,
a distinction must be made between potential evapo-transpiration
and what may actually take place. Most of the methods of estima-
tion necessarily assume an abundant water supply and so give the
potential figure.
3.3 Methods of estimating evaporation

3.3.1 Water budget or storage equation approach. This method con-
sists of drawing up a balance sheet of all the water entering and
leaving a particular catchment or drainage basin. If rainfall is
measured over the whole area on a regular and systematic basis then
a close approximation to the amount of water arriving from the
atmosphere may be made. Regular stream gauging of the streams
draining the area and accurately prepared flow-rating curves will
indicate the water leaving the area by surface routes. The difference
between these two can only be accounted for in three ways:
(i) by a change in the storage within the catchment, either in

surface lakes and depressions or in underground aquifers;
(ii) by a difference in the underground flow into and out of the
catchment;
(iii) by evaporation and transpiration.
The storage equation can be written generally as
E=P+I±U-O±S (3.1)
where E = evapo-transpiration
P = total precipitation
I = surface inflow (if any)
U = underground outflow
0 = surface outflow
S =change in storage (both surface and subsurface)
If the observations are made over a sufficiently long time the signifi-
cance of S, which is not cumulative, will decrease and may be ignored,
if the starting and finishing points of the study are chosen to coincide
as nearly as possible with the same seasonal conditions. The signifi-
cance of U cannot be generalised but in many cases can be assigned
second order importance because of known geological conditions
which preclude large underground flows. In such cases a good esti-
mation of evapo-transpiration becomes possible and the method
provides a means of arriving at first approximations.
3.3.2 Energy budget method. This method, like the water budget
approach, involves solving an equation which lists all the sources and
sinks of thermal energy and leaves evaporation as the only unknown.
It involves a great deal of instrumentation and is still under active
development. It cannot readily be used without much data that is
not normally available and so it is a specialist approach.
3.3.3 Empirical formulae. Many attempts have been made to

produce satisfactory formulae for the estimation of evaporation.
These are usually for evaporation from an open water surface, as
indeed, are the more general methods to follow. The reason for this
is simple. Evaporation if it is to take place, presupposes a supply of
water. Whatever the meteorological conditions may be, if there is
no water present, then there can be no evaporation. Accordingly,
estimating methods using meteorological data work on the assump-
tion that abundant water is available, i.e. a free water surface. The
results obtained therefore are not necessarily a measure of actual, but
ofpotential evaporation. Often these two are the same, as for example,
in reservoirs where a free water surface exists. When evaporation
from land surfaces is concerned, the loss of water in this way clearly
depends on availability: rainfall, water-table level, crop or vegeta-
tion, and soil type all have an influence which may be expressed by
applying an empirical factor, usually less than unity, to the free water
surface evaporation.
There are two cases that should be considered:
(i) when the temperature of the water surface is the same as the
air temperature;
(ii) when the air and water surface temperatures are different.
Case (i) rarely occurs and is empirically treated by the equation
Ea = C(e, - e)f(u) (3.2)
where Ea = open water evaporation per unit time (for air and water
temperature the same t 0 C) in mmfday
C = an empirical constant
e, = saturation vapour pressure of the air at t°C (mm
mercury)
e =actual vapour pressure in the air above (mm mercury)
u = wind speed at some standard height (m)
The following equation has been empirically obtained for this case
and is of general validity
Ea = 0·35(e, - e)(O·S + 0·54u2) (3.3)
u2 denotes wind speed in m/s at a height of 2 m: Ea is in mm/day.
Case (ii) is the normally occurring case. Again a formula should
have the form
Eo = C(e', - e)f(u) (3.4)
but now e', is the saturation vapour pressure of the boundary layer
of air between air and water, whose temperature t'8 is not the same as
either air or water and is virtually impossible to measure. Accord-
ingly empirical formulae have been developed in the form of Eq.
(3.2) which works fairly well for specific locations where the constants
have been derived, but have no general validity.
Such a formula, derived for the Ijsselmeer in Holland, and only
applicable to it and similar conditions, is
Eo = 0·345(ew - e)(l + 0·25ua) (3.5)
where Eo = evaporation of the lake in mm per day

ew = saturation vapour pressure at temperature tw of the sur-
face water of the lake in mm mercury
e = actual vapour pressure in mm mercury
ua = wind velocity in m/s at a height of 6 m above the surface.
3.3.4 Penman's Theory. The following nomenclature is used
Eo = evaporation from open water (or its equivalent in heat
energy)
ew = saturation vapour pressure of air at water surface
temperature tw
e = actual vapour pressure of air at temperature t = satura-
tion vapour pressure at dew point ta
e, = saturation vapour pressure of air at temperature t
e', = saturation vapour pressure of air at boundary layer
temperature t' s
nf n = cloudiness ratio = actual/possible hours of sunshine
RA = Angot's value of solar radiation arriving at the atmos-
phere
Rc = sun and sky radiation actually received at earth's surface
on a clear day
R1 = nett amount of radiation absorbed at surface after re-
flection
RB = radiation from the earth's surface.
In 1948 H. L. Penman presented a theory and formula [13] for the

estimation of evaporation from weather data. The theory is based
on two requirements which must be met if continuous evaporation
is to occur. These are (i) that there must be a supply of energy to
provide latent heat of vaporisation and (ii) there must be some
mechanism for removing the vapour, once produced.
The energy supply. During the hours of daylight there is a certain

measurable amount of short wave radiation arriving at the earth's
surface. The amount depends on latitude, season of the year, time
of day and degree of cloudiness. Assuming there were no clouds and
a perfectly transparent atmosphere, the total radiation to be expected
at a point has been given in tabular form by Angot, and is reproduced
in Table 3.1 as values of RA.
If Rc = short wave radiation actually received at the earth from
sun and sky and nf D = ratio of actual/possible hours of sunshine
then Penman gives
Rc = RA(0·18 + 0·55n/D)
for southern England
TABLE 3.1. Angot's values of short-wave radiation flux RA. at the outer limit of the atmosphere in gramme-calories
per cm2 per day, as a Junction of the month of the year and the latitude
Latitude 0 Jan. Feb. Mar. Apr. May June July Aug. Sept. Oct. Nov. Dec. Year
N90 0 0 55 518 903 1077 944 605 136 0 0 0 3540

80 0 3 143 518 875 1060 930 600 219 17 0 0 3660
60 86 234 424 687 866 983 892 714 494 258 113 55 4850
~ 40 358 538 663 847 930 1001 941 843 719 528 397 318 6750
20 631 795 821 914 912 947 912 887 856 740 666 599 8070
Equator 844 963 878 876 803 803 792 820 891 866 873 829 8540
20 970 1020 832 737 608 580 588 680 820 892 986 978 8070
40 998 963 686 515 358 308 333 453 648 817 994 1033 6750
60 947 802 459 240 95 50 77 187 403 648 920 1013 4850
80 981 649 181 9 0 0 0 0 113 459 917 1094 3660
890 995 656 92 0 0 0 0 0 30 447 932 1110 3540
From: Physical and Dy1111mica/ Meteorology by David Brunt, p. 112 (Cambridge University Press, 1944).
The SI unit for R.A. is joulesfm2fday. The table in gm cal/cm2fday is used so that it is compatible with Rijkoort's nomo-
gram. The conversion is
1 gm cal/cmz = 41·9 kJ/m2
and quotes Kimball
Rc = R..t(0·22 + 0•54nf D)
for Virginia, U.S.A.
and Prescott
Rc = R..t(0·25 + 0·54n/D)
for Canberra, Australia
from which it may be seen that even on days of complete cloud cover
(nf D = 0), about 20% of solar radiation reaches the earth's surface,
while on cloudless days about 75% of radiation gets through.
Part of Rc is reflected as short wave radiation; how much depends
on the reflectivity of the surface, or, the reflection coefficient r.
If Rr = the nett amount of radiation absorbed
then Rr = Rc (1 - r) = R..t(l - r)(0·18 + 0·55n/D)

In turn, some of Rr is re-radiated by the earth as long wave radiation,
particularly at night when the air is dry and the sky clear. The nett
outward flow RB may be expressed empirically as
RB = aTa4(0·41- 0·077ye)(0·20 + 0·80n/D)
where a = Lummer and Pringsheim constant = 117 ·74 x lQ-9 g cal/
cm2/day
Ta = absolute earth temperature = t°C + 273
e = actual vapour pressure of air in mm mercury
and so the nett amount of energy finally remaining at a free water
surface (r = 0·06) is given by H, where
H=Rr-RB
= Rc- rRc- RB
= Rc(l - r) - RB
= R..t(0·18- 0·5Sn/D) (l - 0·06)- RB
:. H = R..t(0·18- O·SSnfD) (1 - 0·06)- (117·4 x I0-9)
Ta4(0.·41- 0·077ye) (0·20 + 0·80n/D) (3.6)
and this heat is used up in four ways
I.e. H=Eo+K+S+C (3.7)
where Eo = heat available for evaporation from open water
K = convective heat transfer from the surface
S = increase in heat of the water mass (i.e. storage)
C = increase in heat of the environment (negative advected
heat)
Over a period of days and frequently over a single day, the storage
of heat is small compared with the other changes, and the same is
true of environmental storage, so that to a small degree of error
H=Eo+K (3.8)
Vapour removal. It has been shown that evaporation may be repre-
sented by
Eo= C(e', - e)f(u) (3.4)
but that e', cannot be evaluated if air and water are at different
temperatures. Penman now made the assumption that the transport
of vapour and the transport of heat by eddy diffusion are essentially
controlled by the same mechanism, i.e. atmospheric turbulence, the
one being governed by (e',- e), the other by (t',- t). To a close
approximation therefore
K = {J = y(t', - t) where y = psychrometer constant
Eo e',- e = 0·66 if tis oc + e in mbar
Now since
H = Eo + K = Eo(1 + {J)
H H
E - - - - ---:---
0 - 1 + {J - t'B - f
1 +y-,--
es- e
Now eliminate t's - t by substitution, since t'8 - t = e',!i- e, where

e8 = saturation vapour pressure at temperature t.
!i =slope of vapour pressure curve at t, =tan oc (see Fig. 3.1).
This is reasonable since t's is never very far from t.
H
Hence Eo = -----:----
1 + !. _e'_s_-_es (3.9)
!i e',- e
Now es must be eliminated.
Since e's - es = (e's - e) - (es - e) (3.10)
and from Eq. 3.2 Ea = Cf(u) (e, - e) } Ea e8 - e
:. - = -,- (3.11)
and Eq. 3.4 Eo = Cf(u) (e' 8 - e) Eo e' - e
where Ea = evaporation (in energy terms) for the hypothetical case
of equal temperatures of air and water,
t
::J
0
c.
~
/s
Temperature - -
FIG. 3.1 Saturation vapour pressure curve
then by substituting in Eq. (3.9) the values of Eqs. (3.10) and (3.11)
H
Eo= ---=--.,..-------=
1 + !_[(e',- e)- (es- e)]
A e's- e
H
and Eo 1 (1 -
= + !.
A
Ea)
Eo
E _ !l.H +yEa
from which (3.12)
o- A+ Y
!:!. has values obtained from the saturation vapour pressure curve,
typically as shown
t = ooc !:!. = 0·36
10 0·61
20 1·07
30 1·80
Referring to Eqs. (3.3) and (3.6) for Ea. and H respectively, it may be
seen that Eo is now computed from standard meteorological observa-
tions of mean air temperature, relative humidity, wind velocity at a
standard height, and hours of sunshine. The formula has been
checked in many parts of the world and gives very good results.
Being based on physical principles it is of general application and
gives values ijlat should serve for most project studies until supple-
mented by actual evaporation measurements. (See Section 3.6).
To overcome the computational labour involved in solving the
Penman equation, a nomogram has been designed by P. J. Rijkoort
of the Royal Meteorological Institute, Netherlands, which enables
rapid evaluations to be made. It is reproduced at the back of this
book by permission of the designer. The nomogram has been drawn
for a slightly different value of Rc from that used by Penman
Rc = RA(0·20 + 0·48nf D)
instead of Penman's RA(0·18 + 0·55nfD) but any difference will be
smaller than the probable margin of error in cloud cover estimation,
so it may be ignored.
Values of RA may be derived for any latitude from Table 3.1, p. 40.
3.4 Evaporation from land surfaces using Penman's Eo value

Penman in his original paper reported on results of experiments
conducted on turfed soil and bare soil to determine how their
evaporation rates (ET and EB) compared with open water (E0 ). He
concluded that the evaporation rate from a freshly wetted bare soil
was about 90% of that from an open water surface exposed to the
same weather.
i.e. EB/Eo = 0·90
For grassed surfaces the comparison was more erratic, and pro-
visional figures for turf with a plentiful water supply were given as
follows:
Values of ET/Eo for southern England
Nov.-Feb. 0·6
Mar.-Apr.} 0 .7
Sept.-Oct.
May-Aug. 0·8
Whole year 0·75
These figures are all less than unity because of the greater reflectivity
of vegetation compared to open water and also because the trans-
piration of plants virtually ceases at night.
3.5 Thomtbwaite's formulae for evapo-transpiration

C. W. Thornthwaite carried out many experiments in the United
States using lysimeters and studied extensively the correlation
between temperature and evapo-transpiration. From this work [14]
he devised a method enabling estimates to be made of the potential
evapo-transpiration from short, close set vegetation with an ade-
quate water supply, in the latitudes of the United States.
If tn = average monthly temperature of the consecutive months
of the year in oc (where n = I, 2, 3, ... 12)
and j = monthly 'heat index',
then
.
J=
tn
(-
)1·514 (3.13)
5
2:
and the yearly 'heat index', J is given by
2
J = j (for the 12 months) (3.14)
The potential evapo-transpiration for any month with average

temperature 1° is then given, as P Ex, by
PEx = 16 c~~y mm per month (3.15)
where
a = (675 X 1Q-9)J3 - (771 X 1Q-7)J2 + (179 X 1Q-4)J + 0·492
(3.16)
However PEx is a theoretical standard monthly value based on 30
days and 12 h sunshine per day. The actual PE for the particular
month with average temperature t 0 is given by
DT
PE=PEx-mm (3. I 7)
360
where D = number of days in the month
T = average number of hours between sunrise and sunset in
the month.
The method has been tested by Serra, who suggested that Eqs.
(3.13) and (3.16) may be simplified as follows
j = 0·09t113/2 (3.18)
a = 0·0161 + 0·5 (3.19)
This method of estimating potential evapo-transpiration is
empirical and complicated and requires the use of a nomogram for
40
30
~
r point of convergence-
IIIII I"C
26·5
27·0
PE.
135·0
139·5
25
u ~ 27·5 143·7
:.. 20
vv 28·0
28·5
147·8
151·7
v
~ 15 29·0 155·4
e8. 10 - 29·5 158·9
::J
l!f§.
v
/ 30·0 162·1
v
E
.!!
-
1-
90
80
30·5
31·0
165·2
168·0
1- 70 31·5 170·7
>- 32·0 173·1
:E 1-
~~ 32·5 175·3
~ 5 33·0 177·2
E 4 k:;;' 33·5 179·0
c::
~ 3
7 30
34·0
34·5
35·0
180·5
181·8
182·9
35·5 183·7
2 1-- 20 36·0 184·3
36·5 184·7
37·0 184·9
37·5 185·0
38·0 185·0
10 I I
10 20 30 40 60 80 100 150 200

50 70
PE, mm
Fro. 3.2 Nomogram and table for finding potential evapo-trans-

piration PEz (after C. W. Thorr•hwaite. Courtesy of "Geog.
Review"; copyright American Geographical Society, New York)
its solution. Thornthwaite published such a nomogram which is

reproduced in Fig. 3.2.
The first step is to obtain the heat index J. From Fig. 3.2 obtain
the unadjusted value of potential evapo-transpiration by drawing a
straight line from the location's J value through the point of conver-
gence at t = 26·5°C. (If t0 is greater than 26·5°C use the Table
alongside Fig. 3.2.) PE:-e for the month can then be read off, corres-
ponding to its given mean temperature. Twelve values are obtained
for each of the 12 months. These unadjusted values may then be
adjusted for day and month length by Eq. (3.17) and totalled to give
annual potential evapo-transpiration.
It has been found that the method gives reasonably good results
whatever the vegetation cover, though different types of vegetation
will affect a particular locality's true value. The formula is based on
temperature, which does not necessarily correspond to incoming
solar radiation immediately, because of the "heat inertia" ofland and
water. Transpiration, however, responds directly to solar radiation.
Accordingly, care should be exercised when using the method to
ensure that conditions do not change abruptly in a particular month,
though if figures for many consecutive months are being used, the
cumulative differences are probably negligible.
For project studies the method is a useful complement to the
Penman approach.
3.6 Direct measurement of evaporation by pans

Whenever possible, direct observations of evaporation should be
made. The instrument used for this is the evaporation pan. In
Britain the standard pan is 1·83 m (6ft) square and 610 mm (2ft)
deep filled to a depth of 550 mm (I ft 9 in.) and set in the ground so
that the rim of the pan projects 76 mm (3 in.) above the surrounding
ground. Regular observations of pan evaporation are made at some
30 points throughout the country [15].
In the U.S.A. the standard or Class A pan is circular 1·22 m (4ft)
in diameter and 254 mm (10 in.) deep, filled to a depth of 180 mm
(7 in.), set on a timber grillage with the pan bottom 150 mm (6 in.)
above ground level. Regular evaporation readings are taken at over
400 places.
A third type of pan is sometimes used in the United Kingdom, the
Peirera pan, which is circular like the Class A pan but deeper and
sunk in the ground with a 3 in. air space surrounding it.
The Class A pan has a greater daily range of temperature than the
square one, but is usually homogeneous whereas the water in the
square pan may stratify. Doubling the wind run may increase
evaporation by up to 20%.
The relatively small capacities and shallow depths of pans fin
comparison to lake and river volumes and their situation at or near
the land surface allows proportionately greater amounts of advected
heat from the atmosphere to be absorbed by the water in the pan
through the sides and bottom, than by natural open water, and by
some pans more than others. Pan evaporation is therefore usually
too high and a pan coefficient has to be applied. These coefficients
range from 0·65 to greater than unity, depending on the dimensions
and siting of the pan. Generally the standard British pan has a co-
efficient about 0·92 and the U.S. Weather Bureau Class A pan about
0·75 but there are quite wide variations. Law [16] carried out com-
parative tests over 14 years at two sites in Yorkshire and found the
ratio of evaporation from the Class A pan to that from the square
British pan ranged between 1·17 and 1·40 with an average of 1· 32.
Houk [17] gives a full account of known American values and
Olivier [18] quotes many data from African and Near Eastern
Sources.
There are difficulties in using pans for the direct measurement of
evaporation, arising from the difficulty of measuring very small
differences of elevation and the subsequent application of coefficients
to relate the measurements from a small tank to large bodies of open
water. Nevertheless, actual field measurements should form an
important part of any project studies of evaporation.
3.7 Consumptive use

Evapo-transpiration is the term used for the evaporation of moisture
from the earth's surface including lakes and streams and the vegeta-
tion that may cover the land. Consumptive use refers to the evapora-
tion and transpiration from vegetation-covered land areas only,
frequently in respect to horticulture and agriculture and associated
irrigation requirements. The terms are often used synonymously.
The consumptive use of water in an area is dependent on many
factors, including climate, the supply of soil moisture, growing
vegetation, type of soil and methods of land management. Climatic
factors include precipitation, temperature, humidity, wind and lati-
tude of the locality (which affects the length of the growing season).
Soil moisture supply depends on topography and underground
flow, as well as precipitation. Soil types and land management vary
widely over short distances. There are no formulae of general
validity but several empirical formulae may be used with local
coefficients to determine annual water use in any locality within cer-
tain broad limitations.
3.7.1 Arable crops. Consumptive use refers to water that is actually
used while evapo-transpiration formulae give potential water use.
Reference to Fig. 3.3 will show that for a particular locality, unless
the rain falls when it is needed, a large water deficit may develop in
the growing season despite quite high rainfall. In a case like this con-
sumptive use will be less than potential evapo-transpiration and a
need for irrigation in the growing period is indicated.
8r--------------------- ---,
rate af potential
evapo-transpiration in
in./ month
soil moisture
uti Iisation
precipitation
in in/month
soil moisture
accretion
JFMAMJJASOND
FIG. 3.3 Typical annual soil moisture deficiency diagram

(after Thornthwaite [19])
Actual measurements of consumptive use are very difficult and

slow to make and may be very expensive. Many such investigations,
have been made in the U.S.A., and the results may be applied in
other areas by a method developed by the Division of Irrigation and
Water Conservation, Soil Conservation Service of the U.S. Dept. of
Agriculture [20]. The existing consumptive use data is correlated
with monthly temperature, percentage of daylight hours, length of
growing period and precipitation. Coefficients are developed for
different crops and these coefficients are then used for crops in areas
where climatological data only, are available. Although other factors
are involved, the four mentioned have the principal influence and
are generally available from standard meteorological observations.
The procedure is as follows,
(1) Tabulate the mean monthly temperature for each month in
op (t).
(2) Compute the monthly proportion of daylight hours of the
year {p).
(3) Multiply t byp for each month to obtain a monthly consump-
tive use factor f
I= tp
It is assumed that, consumptive use varies directly as this
factor, if an ample water supply is available.
(4) Multiply /by k, {the monthly consumptive use coefficient) to
obtain monthly consumptive use u, in inches.
u=kf
A similar formula for the whole growing season is
U=KF='i:.kf
where
U = consumptive use of crop in inches for the period
K = empirical, seasonal, consumptive-use coefficient
F = the sum of the monthly use factors 'l:.f
Table 3.2 gives values of k which have been suggested by Blaney

for the U.S.A. The same writer, in the discussion to a subsequent
TABLE 3.2. Examples of nwnthly consumptive use coefficients (k)
for irrigated crops based on field measurements of evapo-trans-
piration and temperature
Location Crop Mar. April May June July Aug. Sept. Oct. Nou.
Arizona Alfalfa 0·74 0·84 0·91 1-10 1·30 0·90 0·75 0·1S
Cotton 0·30 0·40 0·60 0·80 0·80 0·70 0·60
Soya beans 0·35 0·60 0·90 0·80 O·SO
Guar 0·30 0·80 0·90 0·55
Grapefruit 0•55 0·65 0·65 0·70 0·70 0·75 0·70 0·70 0·6S
-California Oranaes 0·53 0·56 0·56 0·58 O·S8 0·61 0·61 0·61 0·60
Alfalfa 0·70 0·80 0·80 0·90 0·90 0·90 0·80 0·70
Lemons 0·40 0·40 0·50 0·50 0·55 0·60 0·50 0·40
Oranaes O·SO 0·50 0·55 0·60 0·60 0·60 0·60 O·SO
Beets 0·30 0·60 0·85 0·95 0·90 0·40
Tomatoes 0·4S 0·80 0·70 0·80 0·70
from Harry F. Blaney-Monthly consumptive use of water by

irrigated crops and natural vegetation. 11th General Assembly
I.A.S.H., Toronto, 1957.
100~--~~----.-----.-----r-----,
~ 80~--~4-----~~---c~~~~------1
0
-o
a;
~ 406-~~~----~------+-----~------;
Depth of irrigation water, ft.

FIG. 3.4 Water-yield relationship for crops; Cache Valley,
Utah, U.S.A. (after Harris)
paper [21] suggested that for humid areas the seasonal coefficient
K should be reduced by about 10 %.
Houk [17] points out that there is an optimum water <equirement
for every crop and that the supply of water beyond the optimum
amount reduces yields. Only general guide lines can be laid down,
however, because differences in soil texture and fertility affect the
optimum amount for the same crop in different conditions. A
typical water yield relationship for Utah, U.S.A. by Harris [22] is
shown in Fig. 3.4 and the range of variation in water requirements
for different crops is illustrated in a particular case by Table 3.3
which is reproduced from a report by Fortier [23].
TABLE 3.3 Variations in water requirements of different crops

produced in the Missouri and Arkansas River basins (after
Fortier)
Range In Range in
No. of water No. of water
Crop tests requirements Crop tests requirements
nun mm
Forage, inc. alfafa 648 590-800 Apples 4 640-79S

Barley 335 40S-SSS Beans 4 39S-490
Oats 409 410-SSO Buckwheat 3 320-395
Wheat 542 4lS-SSO Cantaloupes 10 4SS-10S
Corn 70 315-S60 Peas 168 41S-S90
Kafir corn IS 43S-480 Potatoes 3SO 420-S20
Flax SO 4SO-S6S Sugar beets 128 490-76S
Millet 14 24S-28S Sunflowers 16 36S-42S
Milo maize 27 33S-S20 Tomatoes 6 640-85S
Sorghum 26 32S-4SO Cucumbers 7 530-1140
For further information reference should be made to U.S. Dept.
of Agriculture Technical Bulletins which contain a large amount of
information on field tests for consumptive use, and to the references
[17, 18] quoted.
3.7.2 Forest. It is a matter of great difficulty to make quantitative

measurements on the consumptive use of forest and there is little
published about it.
Law [16] has carried out detailed work on the consumptive use of
a plantation of Sitka spruce near Slaidburn, Yorkshire, over 15
years. His results are summarised by the author below, as 14-year
annual averages (1956-1969) measured on a plantation lysimeter of
450 m2 area containing 74-95 trees. (21 were cut down half-way
through the period.)
Annual values
mm %
(I) Mean rainfall from 3 rain gauges outside the
plantation 1492
(2) Estimated mean rainfall over lysimeter
(based on 3 tree-top gauges) 1363 100
(3) Throughfall 923 68
(4) Interception (2) - (3) 440 32
(5) Lysimeter runoff 535 39
(6) Gross losses (2) - (5) 828 61
(7) Net losses without interception (3) - (5) or
(6) - (4), i.e. transpiration 388 29
and as measured concurrently in a neighbouring
climatological station
(8) Evaporation from 1·83 m square tank (mean
of two neighbouring tanks) 507
(9) Losses from adjacent grass-covered plot
(measured by percolation gauge) 410
(10) Calculated evaporation by Penman equation
(mean of two calculations for two sites) 511
From his experiments Law drew certain general conclusions
(a) Trees transpire roughly the same quantities as grass.
(b) A third of rainfall is intercepted by coniferous trees and
subsequently lost through evaporation in addition to transpiration.
(c) Up to 3 mm rain can be intercepted by conifers at any one
time. (There are indications that deciduous trees do not differ
much.)
(d) The evaporation of intercepted rain in coniferous forests
does not differ markedly from summer to winter because the effect of
wind is more important than ambient temperature, when adequate
evaporating surface area is available.
(e) Stemftow on Sitka spruce is about 5% ofthroughfall.
Currently (1973) in the United Kingdom extensive experiments are
under way by the Institute of Hydrology in the Plynlimon catch-
ments which should add considerably to existing knowledge on the
consumptive use of forest.
4 Infiltration and Percolation
4.1 Infiltration capacity of soU

When rain falls upon the ground it first of all wets the vegetation or
the bare soil. When the surface cover is completely wet, subsequent
rain must either penetrate the surface layers, if the surface is perme-
able, or run off the surface towards a stream channel if it is imperme-
able.
If the surface layers are porous and have minute passages available
for the passage of water droplets, the water infiltrates into the sub-
surface soil. Soil with vegetation growing on it is always permeable
to some degree. Once infiltrating water has passed through the
surface layers, it percolates downward under the influence of gravity
until it reaches the zone of saturation at the phreatic surface.
Different types of soil allow water to infiltrate at different rates.
Each soil type has a different infiltration capacity, f, measured in
mm/h or in./h. For example it can be imagined that rain falling
on a gravelly or sandy soil will rapidly infiltrate and, provided the
phreatic surface is below the ground surface, even heavy rain will not
produce surface runoff. Similarly a clayey soil will resist infiltration
and the surface will become covered with water even in light rains.
The rainfall rate, i, also obviously affects how much rain will infil-
trate and how much will run off.
4.2 Factors inftuencingfc

Nassif and Wilson [24] have recently carried out extensive studies on
infiltration using a weighable laboratory catchment 25m2 in area.
They conclude that for any soil under constant rainfall, infiltration
54
Infiltration and Percolation 55
rate decreases in accordance with an equation of the form first used
by Horton [25].
f = /c + !1-e- Kt
where f = infiltration rate at any time t (mm/h)
/c = infiltration capacity at large value oft (mm/h)
fL =/o- /c
/o = initial infiltration capacity at t =
0 (mm/h)
t = time from beginning of rainfall (min)
K =constant for a particular soil and surface (min- 1)
K is a function of surface texture: e. g., if vegetation is present K is
small, while a smoother surface texture, such as bare soil, will yield
larger values.
/o and fc are functions of both soil type and cover. For example,
a bare sandy or gravelly soil will have high values of/o and fc and a
bare clayey soil will have low values of/o and /c but both values will
increase for both soils if they are turfed.
/c is a function of slope up to a limiting value of slope (varying
between 16 and 24%) after which there is little variation.
/c is a function of initial moisture content: the dryer the soil
initially, the larger will be fc, but the variation may be comparatively
small.
fc is a function of rainfall intensity. If the intensity i increases, /c
increases. This parameter has a greater effect on fc than any other
variable.
These correlations are illustrated in Fig. 4.1 for a typical agri-
cultural soil. Table 4.1 lists some representative values of K,fo and /c
for different soil types. The parameters K and /o are relatively stable
for particular soils and do not vary noticeably with slope of catch-
ment or rain intensity; fc on the other hand, varies widely with both
and so is shown as a range of values.
Until recently it had been generally thought thatfc was a constant
for a particular soil but this does not seem to be so. The infiltration
rate appears to be largely controlled by the surface pores. Even quite
a small increase in the hydrostatic head over these pores results in
an increase in the flow through the soil surface. If the surface layer
is imagined as shown in Fig. 4.2 where the surface soil grains are
.c::
'EE
Low rain intensity

Time- Time-+ 8 16 24 32
Slope
FIG. 4.1 The variation of infiltration capacity
shown, then the governing factor is the head h over the smallest
cross section of a pore. This continues to increase with rainfall
intensity until a limiting value is reached where run-off prevents any
further increase. It seems unlikely that this limiting condition is often
reached in natural conditions.
Previous researchers [26] have found similar results but attributed
the increase in fc at higher rainfall intensities to lack of homogeneity
in their experimental catchment watershed. Others have also empha-
sised the over-riding importance of the superficial layer of a soil [27].
The infiltration rate of a soil is the sum of percolation and water
entering storage above the groundwater table. Generally the soil is
far from saturated and so storage goes on increasing for very long
periods. Accordingly, /c goes on decreasing under a steady rain
intensity for equally long periods.
TABLE 4.1. Representative values of K, fo and fc for different

soil types
fo fc K
min-1
Soil type mm/h mm/h
bare 280 6-220 1·6
standard
agricultural turfed 900 20-290 0·8
peat 325 2-20 1·8
fine sandy bare 210 2-25 2·0
clay 1·4
turfed 670 10-30
FIG. 4.2 Hydraulic head on soil-grain pores
It follows that arbitrary methods, described in the following sec-

tions, which for convenience, attribute average values of rain-loss to
catchments to determine nett rain, are gross simplifications, at least
during the early rain.
Exposed soils can be rendered almost impermeable by the com-
pacting impact of large drops coupled with the tendency to wash
very fine particles into the voids. The surface tends to become
'puddled' and the fc value drops sharply. Similarly, compaction due
to man or animals treading the surface, or to vehicular traffic, can
severely reduce infiltration capacity.
Dense vegetal cover such as grass or forest tends to promote high
values of fc· The dense root systems, all providing ingress to the
subsoil, the layer of organic debris forming a sponge-like surface,
burrowing animals and insects opening up ways into the soil, the
cover preventing compaction and the vegetation's transpiration re-
moving soil moisture, all tend to help the infiltration process.
Other effects which marginally affect the issue are frost heave,
leaching out of soluble salts, drying cracks which increase fo, and
··,,,. II=Jb,, .. ·. ,·. ·IFif. . r....1:~:....

I• mnernng ... ,
1mm ma1nta1ned
JI \ infi I I rating water

I \ \
FIG. 4.3 Infiltrometer
entrapping of interstitial air, which decreases it. Temperature has
some effect since flow in interstices is laminar and hence viscosity
has a direct effect on resistance to flow. Other things being equalfo
and /c will have higher values in the warmer seasons of the year.
4.3 Methods of determining infiltration capacity

4.3.1 Infiltrometers. An infiltrometer is a wide diameter, short tube,
or other impervious boundary surrounding an area of soil. Usually
two such rings are used concentrically as shown in Fig. 4.3. The
rings are flooded to a depth of 5 mm over the surface and continually
refilled to maintain this depth, the inflow to the central tube being
measured. The purpose of the outer tube is to eliminate to some
extent the edge effect of the surrounding dryer soil. Such tests give
useful comparative results but they do not simulate real conditions
and have been largely replaced by sprinkler tests on larger areas.
Here the sprinkler simulates rainfall, and runoff from the plot is
collected and measured as well as inflow, the difference being
assumed to have infiltrated.
While rain-simulating sprinklers are a good deal more realistic
than flooded rings, there are limitations to the reliability of results
thus obtained, which usually give higher values off than natural
conditions do. For qualitative effects, e.g. comparisons between
different conditions of vegetation, soil types etc. the methods are
simple and effective.
Consistent and repeatable results may be obtained by using
laboratory catchments with rainfall simulators, where the quantity
and thickness of soil is adequately representative of nature. Nassif
and Wilson [24] used 7 tonnes of soil in a layer 200 mm thick, and
measured all inputs and outputs and changes of storage. Such
equipment as theirs gives very good comparative and perhaps abso-
lute figures for infiltration, but it still does not simulate natural
conditions completely as there is atmospheric pressure at the bottom
of the laboratory soil layer while this is not so in nature.
4.3.2 Drainage basin rainfall-runoff analysis. Several investigators

have attempted to improve on sprinkled infiltrometers by choosing
small "homogeneous" drainage basins and carefully measuring
precipitation, evaporation and outflow as surface runoff. By elimi-
nating everything except infiltration, average f values may be ob-
tained for such basins by techniques presented by Horton [28] and
Sherman [29].
The difficulty remains of ensuring that there has been no unrecord-
ed underground outflow, nor variation in underground aquifer
storage, so that although quantitative results are obtained, the
analysis is intricate and the margin for error is wide.
4.3.3 q,-index method. In practice, it is possible to obtain infiltration

indices which enable reasonable approximations to be made of
infiltration losses. One of these is the q,.;ndex, which is defined as
the average rainfall intensity above which the volume of rainfall
equals the volume of runoff. In Fig. 4.4 a rainstorm is shown plotted
net rain =
quantity of
runoff
storm loss
time in hours -
FIG. 4.4 Infiltration loss by <1>-index
on a time base in terms of its average hourly intensity. The shaded

area above the dashed line represents measured runoff, as mm, over
the catchment area. Since the unshaded area below the line is
measured rainfall but did not appear as runoff, it represents all the
losses, including surface detention and evaporation as well as infiltra-
tion. However, infiltration is much the largest loss in many catch-
ments and although it is a rough and ready method (since it takes no
account of the variation infwith time) it is widely used as a means of
quickly assessing probable run-off from large catchments for
particular rainstorms.
Example 4.1.
Given a total rainfall of15 mm as shown· in Fig. 4.5(a) and a surface

runoff equivalent to 33 mm, establish the Cl>-index for the catchment.
Total rain = 75mm Total rain = 75mm

30
~ 25
~ 20
"'
~
c
.!!
..5
hours hours
(a) (b)
Flo. 4.5 Examples of c'!>-index computations
The ct>-index line is drawn so that the shaded area above it contri-
butes 33 mm of runoff. In this case the index is 8 mmfh.
Suppose, however, the same total rainfall had been distributed as
shown in Fig. 4.5(b). To obtain a runoff of 33 mm above the Cl>-
index line requires the line to be raised to give an index value of
9 mm/h. It may be seen therefore that one determination of the <!>-
index is of limited value and that many such determinations should be
made, and averaged, before the index is used.
4.3.4 The fav method. This method is a developed version of the

<!>-index, in that it attempts to allow for depression storage and
short rainless periods during a storm, as well as eliminating all rain
periods where the rainfall intensity is less than the infiltration capa-
city assumed.
Referring to Fig. 4.6 the approximate position of the /av line is
fixed by reference to runoff data and the raingraph. The line is then
moved vertically until the various losses are balanced and the runoff
values satisfied. The loss estimates are based on whatever data is
available and on the judgment of the analyst and are obviously
subjective but, from the analysis of many storms, values of /av may
be deduced for various conditions.
In applying a derived value of/av to a rainstorm to predict expected
runoff, the rain periods lying outside the /av period are assumed lost
and the nett rain is found directly after inserting the estimated losses
as shown. Butler [30] gives a detailed account of the method.
assumed f 0 v period
-<=.
'EE deduction for
residual
infiltration
hours
FIG. 4.6 Example of/av distribution
4.4 The effect of soil moisture

4.4.1 Initial soil-moisture conditions. The foregoing methods of
estimating losses are essentially based on records of rainfall and
runoff for particular catchments and their behaviour under rains of
varying intensity. They show average limiting values of infiltration
capacity, obtaining this inf.ormation in terms of the whole catch-
ment, rather than by sampling very small areas as the infiltrometer
does. They do not, however, enable predictions to be made with any
accuracy of the amount of rainfall that will be absorbed by the soil
and so lost to runoff in a particular case, since this depends inter alia
on the state of wetness of the soil at the start of rainfall. As well as
affecting runoff through storage capacity, the initial soil-moisture
affects the infiltration capacity and hence the runoff in the initial
stages of a storm. Some other measurement of this parameter is
therefore necessary, if forecasts of runoff are to be made from
stipulated rainfalls. $-index or /av methods will only provide average
values which may, in a particular case, be far removed from actuality,
and these methods are best used after a separate assessment of initial
losses.
There are two approaches to this problem discussed here. The
first is the Antecedent precipitation index, used in the U.S.A. and the
second, the Estimated soil moisture deficit used in Britain.
4.4.2 Antecedent precipitation index. The antecedent precipitation
index is based on the premise that soil moisture is depleted at a rate
proportional to the amount in storage in the soil. There is therefore a
logarithmic relationship
(4.1)
where Io = initial value of index (mm)
It = index value t days later
k =a recession constant with a value about 0·92 but varying
between 0·85 and 0·98
If t is unity, then any day's value is k times that of the previous day.
FIG. 4.7 Storm-runoff relationship for Monacacy River at Jug

Bridge,Md., U.S.A. (after U.S. Weather Bureau. From "Hydrology
for Engineers" by Linsley, Kohler and Paulhus. McGraw-Hill
Book Co., 1958)
If precipitation occurs it will increase the value of the index by an
amount which is indeterminate since some rain may have left the
catchment as surface runoff. The amount added to the index should
therefore strictly be the basin recharge only, but the difference in the
index by using all the precipitation is usually small.
The progressive daily reduction of the index is due to evapo-
transpiration, which alters seasonally, so Eq. (4.1) is used with a k
value which also varies seasonally. When the index is used to assess
the runoff which takes place from a particular rainstorm, this varia-
tion may be incorporated into a graphical coaxial relationship
derived from the analysis of a large number of observed rainfall and
runoff data on a particular catchment. Linsley, Kohler and Paulhus
[31] give a detailed description of how such relationships may be
derived, and Fig. 4.7 is reproduced above to illustrate the technique.
The diagram is entered at the antecedent precipitation index, and
a horizontal line is followed until the particular week-number curve
corresponding to the calendar date is met. From the intersection a
line drops vertically down to intersect with the line of appropriate
storm duration in hours, and then horizontally across to intersect
with the line of total storm precipitation. A vertical line from the
last intersection indicates the appropriate runoff.
The antecedent precipitation index is a valuable tool in obtaining
probable runoff forecasts for well documented catchments but
much work is necessary to derive relationships like Fig. 4.7.
4.4.3 Estimated soil-moisture deficit. Since evapo-transpiration is

continually removing the soil-moisture and precipitation replacing
it, continuous regular measurements of these two processes yield a
means of estimating the soil-moisture deficit (or s.m.d.) without the
need for an assumption such as Eq. (4.1).
When evaporation exceeds precipitation, vegetation draws on
accumulated soil-moisture to continue transpiration and as the
s.m.d. increases so it becomes increasingly difficult for vegetation to
transpire. Different plants and crops with different root systems
continue to transpire at their potential rates for different periods in
the same environment. In the s.m.d. estimation carried out twice a
month by the U.K. Meteorological Office [32], it is assumed that
each station where rainfall is measured and evaporation estimated is
representative of an area, 50% of which is covered by short-rooted
vegetation which can draw up to 3 inches moisture before actual
evapo-transpiration begins to fall below potential, 30% is covered
by long-rooted vegetation which can similarly draw up to 8 inches
moisture and 20% is riparian where the phreatic surface is so near
FIG. 4.8 Estimated soil-moisture deficit in inches at 0090 GMT

27th Sept. 1967 over Great Britain (Meteorological Office:
Crown copyright)
the surface that evapo-transpiration is never inhibited. When the
difference between rainfall and evapo-transpiration reaches 3 inches
the subsequent calculated evaporation figures take account of the
different rates and areas and a weighted figure for the whole catch-
ment is computed.
Estimates of s.m.d. are made from observed rainfall and estimated
evaporation at a network of 176 stations and a map showing Esti-
mated Soil Moisture Deficit over the whole country is published
within a day of the values at each station being received. Fig. 4.8 is
a typical example. Further details of the theories used in the esti-
mations, and methods used in compiling the data are available [33,
34].
Reference to Fig. 4.8 will show that the likelihood of a flood condi-
tion (for example) arising in any of the hydrometric areas (shown
in fine subdivisions of the land area) may be gauged with much
greater certainty in the knowledge of the s.m.d. there. The estimated
s.m.d. is kept up to date by daily, or weekly, additions and sub-
tractions of precipitation and evapo-transpiration, until the next issue
of the published map.
The use of estimated s.m.d. to predict the proportion of runoff
arising from particular storms does not differ from that for antece-
dent precipitation index, except in so far as coaxial relationships do
not require a week number or calendar date quadrant. Actual catch-
ment discharges are calculated in the ways described later in Sections
7.7 and 7.8, but the losses to be deducted from actual precipitation,
to give nett rain, are obtained by use of estimated s.m.d. and fav
(or <1>-index) infiltration loss data.
4.4.4 Measurement of soil moisture. An instrument for in situ

measurement of soil moisture, the Wallingford soil-moisture probe,
has been developed by the U.K. Institute of Hydrology in conjunc-
tion with the Atomic Energy Authority. The instrument is designed
for use in the field in all kinds of terrain and in any weather.
The instrument consists of a radioactive source in a 740 mm long
probe which can be lowered into an aluminium access tube perma-
nently installed in the ground, a shield and housing for the probe, a
suspension cable and meter. Fast neutrons emitted from the radio-
active source are scattered and slowed by collisions with the atomic
nuclei of soil constituents, mainly by the hydrogen of water in the
soil, thus producing "slow" neutrons. These are sensed by a slow-
neutron detector in the probe and converted into electrical pulses
which pass through the suspension cable to a meter where a visual
display indicates the rate of detection. The wetter the soil the larger
the number of collisions and hence of slow-neutrons detected. The
displayed detection rate is therefore a function of the soil moisture at
the probe depth. The moisture value indicated represents the mean
value for a somewhat indefinite "sphere of influence" within the soil
surrounding the source, with a radius which may be regarded as
being about 150-300 mm. A series of readings is taken down the
profile at intervals usually of l 00 mm or 150 mm. The weighted
Depth Indicator
Cable Clamp- --·~~~
Lock---Lr
0
0
0 0
"
0
0
...--Access Tube
0
s 0
Probe BF3 Slow

Neutron Detector
)(Neutmn Sou<CE
FIG 4.9 Wallingford soil moisture probe (Courtesy Inst. of

Hydrology)
mean of these readings gives the total moisture storage in a given
section of the proille.
If a catchment is provided with a number of permanently installed
aluminium tubes at suitable locations, then a single battery-powered
instrument can be carried from site to site to measure the in situ soil
FIG. 4.10 Wallingford soil-moisture probe

moisture at each one, thus removing much of the guess-work from
the evaluation of this parameter.
The neutron probe is normally used for making repeated measure-
ments of moisture change at the same site and depth and for such a
purpose can give a very high precision. The accuracy of absolute
moisture values on the other hand is heavily dependent upon accu-
rate calibration at each site and depth and this is generally im-
practicable.
A diagram of the instrument and a photograph of one in use are
shown in Figs. 4.9 and 4.10 respectively.
5 Groundwater
5.1 The occurrence of groundwater

Rainfall which infiltrates the soil and penetrates to the underlying
strata is called groundwater. The quantity of water which can be
accommodated under the surface depends on the porosity of the
sub-surface strata. The water bearing strata, called aquifers, may
consist of unconsolidated materials like sands, gravels and glacial
drift or consolidated material like sandstones and limestones.
Limestone is relatively impervious but is soluble in water and so
frequently has wide joints and solution passages which make the
rock, en masse, similar to a porous rock in its capacity to hold water
and act as an aquifer.
The water in the pores of an aquifer is subject to gravitational
forces and so tends to flow downwards through the pores of the
material. The resistance to this underground flow varies widely and
the permeability of the material is a measure of this resistance.
Aquifers with large pores such as coarse gravels are said to have a
high permeability, and those with very small pores such as clay,
where the pores are microscopic, have a low permeability.
As the groundwater percolates down the aquifer becomes satu-
rated. The surface of saturation is referred to as the groundwater
table or the phreatic surface. This surface may slope steeply and its
stability is dependent on supply from above. It falls during dry spells
and rises in rainy weather. The water in the aquifers is usually
moving slowly towards the nearest free water surface such as a lake
or river, or the sea. However, if there is an impermeable layer under-
lying an aquifer and this layer outcrops on the surface, then the
groundwater will appear on the surface in a seepage zone or as a
69
spring. It is equally possible for a groundwater aquifer to become
overlain by impermeable material and so be under pressure. Such
an aquifer, fed from a distance, is called a confined aquifer and the
surface to which the water would rise if it could, is called the piezo-
metric surface. Another name, used for wells drilled into such
confined aquifers is artesian wells, and the word artesian is sometimes
also applied to the aquifers. If the piezometric surface is above
I I ·$'I
~ I
I I I I I
I
I I I I I
i rn
Pe
rrne
qble
FIG. 5.1 Modes of occurrence ofgroundwater
ground level at an artesian well, it is called a flowing well, and a

fracture or flow in the impermeable overlay will, in such conditions,
result in an artesian spring. Sometimes a small area of impermeable
material may exist in a large aquifer. This may happen through
geological faulting or, for example, through a lens of clay occurring
in otherwise sandy glacial drift. A small local water table, called a
perched water table may result and this can often be a long way above
the true phreatic surface.
Some of the modes of occurrence of groundwater described above
are illustrated in Fig. 5.1.
5.2 Factors of influence

The flow of groundwater takes place in porous media. The pores
Groundwater 71
through which movement takes place may be very small indeed and
generally are between the limits 2 mm-0·02 mm. The movement is
slow by the standard of surface runoff and the flow is usually
laminar. The Reynolds number in flow of this kind is very low.
The factors of importance in the flow are
(I) the liquid-its density and viscosity

(2) the media through which the liquid moves
(3) the boundary conditions
The liquid normally is water, usually fresh but occasionally saline.

Its temperature may vary in range 0°-30°C.
E
0.
0.
,.,"'
0
.c;
u
1000 1010 1020

p kg/m 3
FIG. 5.2 Density of saline water
Density. The density of fresh water varies only very slightly with
temperature and its effect may be ignored in groundwater flow. Of
greater importance is salinity, the effect of which is shown in Fig. 5.2.
Density of Fresh Water v. Temperature
temperature oc 0 4 10 15 20
p g/1 999·868 ]()()() 999·727 999·126 998·230
(grammes per litre)
Viscosity is a measure of the shear strength of a liquid; the lower
the viscosity, the more mobile or penetrating the liquid.
Absolute viscosity, denoted by p., has units Ns/m2 ; (the CGS unit, the
poise = 10- 1 Ns/m 2). Water at 20°C has a viscosity of 1 centipoise
(0·01 poise= 10-s Ns/m2 ).
Kinematic viscosity, denoted by v, is the ratio of absolute viscosity
to density, or v = ~ and has units m2 /s (the GCS unit, the stoke =
p
10- 4 m 2/s). Kinematic viscosity occurs in many applications, e.g.
vD
Reynolds number, R = - .
v
v (water) ::::: w-l m2/day ::::: w-s m2/s ::::: I0- 2 stoke.
The kinematic viscosity of liquids is almost independent of pres-
sure and is substantially a function of temperature.
Kinematic Viscosity of Water v. Temperature
temperature oc 0 5 10 15 20
v m8/day 0·152 0·133 0·113 0·098 0·087
The media in which groundwater moves are characterised by the

properties of porosity, permeability and, to a minor extent com-
pressibility. Only the first two are considered here.
. total voids •
Porosity ts defined as n = and may range from a few
tota1 vo1ume
percent to about 90%. In a granular mass composed of perfect
uniform spheres:
in the loosest possible packing, n = 47 ·6%
in the densest possible, n =26%
Natural soils are, of course, composed of irregular particles of many

different sizes. The more regular the soil, the more porous it tends
to be, since in irregular soils the smaller particles tend to fill the
voids in the larger particle packing. It is, therefore, standard proce-
dure in any groundwater survey to analyse the soils mechanically
and plot the particle sizing in a standard way, using a logarithmic
size scale. A typical analysis is shown in Fig. 5.3. Two soils are
plotted: the more regular soil has the steeper slope and is likely to
be more porous.
When water fills the pores of a soil there is a thin layer, only a few
molecules thick, which coats the particles. This water is not free to
move and adheres to the particles even when the voids have been
drained, occupying part of the available space. This means that the
effective porosity, n6 , may be less than the true porosity n. In coarse
Groundwater 73
materials such as gravels there will be no difference between ne and n
but in fine sands the difference may be 5 %. and even more in very
fine materials. In most considerations of porosity in the flow of
groundwater it is ne, the effective porosity, which is of importance.
Permeability is a function of porosity, structure, and the geological

history of the material. By structure is meant the grain size, distribu-
tion, orientation, arrangement and shape of the particles. For
100
/
,/
80
,/
I /
"'0> I I 1/
.!2
:;; 40 I
~ /
-
"'0. v //
20
1- v
0·01 0·1 1·0 10
grain size in mm
B.s.-----"T----,-,si,..lt_ _...:..._-r----sa~n-:;d---.------
c lay 1-f,-in-e--rm-'e:.:.Od:..._,u_m-r-co-a-rs-e1---;-f,-ne--.,-m::.::ed,_.::,u::...m-r-co-a-rs~e grave 1
0·002 0·006 0·02 0·06 0·02 0·06 2·0mm
A.S.E.E. non plast1c

plcstac f.ne medium coarse fine
0·0 0·02 0·074 0·25 0·59 2·0mm
FIG. 5.3 Plotting of particle size analysis, and nomenclature
example in a sediment of predominantly flat grains, deposited in

water, the grains will tend to lie with their longest axes horizontal.
In such a soil the permeability may well be higher in a horizontal
direction than in a vertical one. Such a soil is called anisotropic.
The permeability of a particular material is defined by its permea-
bility coefficient, denoted by k; k depends on the factors listed above
which may be described as the geometry of the pore system, and is
expressed in metres per day, or feet per day.
Many attempts have been made to find a formula fork in terms of
measurable values of the properties of the material. Generally this
is very difficult and such formulae can only be used with confidence
in strictly limited applications. For example, a formula used in
connection with sand filters for water supply, and applying only to
homogeneous rounded grain media, of not too fine a size, is
k = Cd10 2
where k = permeability coefficient
in mfday
10~
d10 = the grain size in mm
gravels
where 10% of material
104
is finer and 90% is
coarser
C = a constant of value
103
~fine gravel
400-1200 (an average 10 2

value is 1000) :>,
0
"D
~ coarse sand
It must be emphasised that 'E 10 1

}-fine sand
c:
formulae such as this are of little 10°
0
value in materials of a hetero-

~ ~very f1ne sand
geneous character or outside their u
10- 1
~·::od,
precisely defined limits, k is not 8'" 5
necessarily a constant for a parti- :>, IU "'
cular soil, as chemical erosion or -
i5
0 do<
deposition may sometimes occur "'E 10-
:;;
with percolating groundwater. 0.
10-
Some values of k as they occur
naturally are indicated on the
10-5
logarithmic scale of permeability pure clays
in Fig. 5.4. 10- 6
5.3 Groundwater flow

FIG. 5.4 Range of permeability in
5.3.1 Darcy's Law. Before any natural soils
mathematical treatment of the flow
of groundwater can be attempted, it is necessary to make certain
simplifying assumptions
(I) the material is homogeneous and isotropic
(2) there is no capillary zone
(3) there is a steady state of flow
The fundamental law is Darcy's Law of 1856. This states that
the rate of flow per unit area of an aquifer is proportional to the
gradient of the potential head measured in the direction of flow. Or
v oc i
Groundwater 75
By introducing a constant of proportionality, which is k, the per-
meability coefficient
v = ki
and for a particular aquifer or part of an aquifer of area A, (area at
right angles to flow)
Q = vA = kAi
where v = velocity of the water, (measured as the time taken to pass
between 2 reference points) in mfday (or mfs or ft/s etc.)
and called the specific velocity.
i = the hydraulic gradient. This equals the potential gradient
since velocities are so small there is virtually no velocity
head. i is also written as:~. s being distance along a flow
line, and cp the potential head.
The specific velocity is not the true velocity, but is merely QfA.
The actual velocity of water in the pores is greater than the specific
velocity since the path the water follows through porous media is
always longer than a straight line between any two points.
If the average real, or pore, velocity is denoted by v then
_ discharge Q Av v
v= =-=-=-
area of water passage Ane Ane ne
:. pore velocity (average) = specific velocity-:- effective porosity
Since the velocity distribution across a pore is probably parabolic,
being highest in the centre and zero at the edges, the maximum pore
velocity v max = approx. 2 x v.
So in a typical case (say) ne = !
then v = 3v and Vmax = 6v

While these are only typical figures it is important to remember the
order of the velocities, since it is on~ Vmax that the Reynolds number
and the continuance of laminar flow depends.
5.3.2 Flow in a confined aquifer. Consider now the case of uni-

directional flow in a confined aquifer of permeability k, illustrated
in Fig. 5.5. Groundwater is flowing from left to right, the energy
required to move the water through the pores is continually using
up the available head and so the line of potential head as indicated
by piezometers introduced into the aquifer, is declining.
dc/J
From Darcy's Law l':r; =- k dx
and if q = flow in the aquifer per unit width then
q= -kH-
dc/J (5.1)
dx
FIG. 5.5 Flow in a confined aquifer
Since it is assumed that the flow is a steady state
dq =0
dx
and differentiating the equation for q, above
dq = -kHd~
dx dx2
and so
(5.2)
since both k and H have finite values.

These two equations (5.1) and (5.2) are the fundamental differential
equations for this case of a confined aquifer. By introducing suitable
Groundwater 77
boundary conditions, many problems in this case may be solved by
the equations.
Note that the Vz in the Darcy equation is the specific velocity.
5.3.3 Flow in an aquifer with phreatic surface. Consider now the

case of an aquifer with a phreatic surface resting on an impermeable
---Rdx -
equipofenflo!,..--
lmes
ground surfoc
;h=dcp
.?---....._
(ph real ic
surface
h
X
assumed impervious
· I ' ~~ri;z~~f~l) b~:~, I I > ~I I I I I ; I I I I I I I I I I I I I I I I I I
FIG. 5.6 Flow in an aquifer with phreatic surface
base. This case is illustrated in Fig. 5.6. Here the first equation from
Darcy's Law, would be
dcp where s = distance measured in the direction

Vs = - kds- of flow.
If two assumptions (due to Dupuit) are made
1. that :~ can be represented by :~ (permissible if dcp is small)

2. that all flow lines in the aquifer are horizontal and equi-
potential lines vertical (nearly true except near abstraction
. )
pomts so t h at dcp dh h
dx = dx t en t he Darcy equatiOn
.
becomes
dh
q = -kh- (5.3)
dx
and
dq d 2(h2)*
-
dx
=-tk -
dx2
. . . dq
an d smce by contmutty dx = 0, t hen
d2(h2) =0 (5.4)
dx 2
N
-- - ---
q
dx
FIG. 5.7 Flow in a phreatic aquifer with rainfall
and here Eqs. (5.3) and (5.4) are the fundamental equations for
solving problems in the case of phreatic aquifers (except where
Dupuit's assumptions are no longer reasonable).
If the aquifer is being recharged by rain falling on the ground
surface, let the rainfall beN in suitable units (as Fig. 5.7).
• The power ruIe of d"ffi · · b · d(v") 1 dv

1 erent1at10n may e wntten d~ = nv"- dx
. d(h2)- h dh
·· dx - 2 "dx
d(h2) dh . dh
.·.1 dx =h.~andsmceq = -khdx
d(h2 )
:. q = -lk dx
. dq d2(h2)
·· dx = -ik dx2
Groundwater 79
In this case dq = N . dx
dq d2(h2)
dx = -ik dx2 =N
d2(h2) 2N
dx 3 = -k (5.5)
and now Eqs. (5.3) and (5.5) are relevant.
2m
k = 12 m/doy
h 20m
JOOOm
h
FIG. 5.8 Flow between two canals
Example 5.1. Suppose there are two canals, at different levels, separ-
ated by a strip of land 1000 m wide, ofpermeability k = 12 m/day as
shown in Fig. 5.8. If one canal is 2 m higher than the other and the
depth of the aquifer is 20 m below the lower canal to an impermeable
base,find the inflow into, or abstraction from, each canal perm length.
Take annual rainfall as 1·20 m per annum and assume 60% infiltration.
Assume a reference origin as indicated on the figure. Then the

boundary conditions are simply
when x = 0, h = 20; and when x = 1000, h = 22.
N = 60% of 1·2 = 0·72 mfyear
= 0·72/365 mfday
d2(h2) 2N
FromEq. 5.5 dx 2 = -k
. . d(h2) 2N
Integratmg twtce dx =- k X + cl
N
h2 = - -k x2 + ClX + c2 (5.6)
when x = 0, h = 20
.. 400 = o + o + c2
. . c2 = 400
and when x = 1000, h = 22
0·72 X 106
.. 484 =- 365 X 12 + 1000Cl + 400
. . c1 = 0·084 + 0·164 = 0·248
Now using Eq. (5.3).
dh
q = -kh-
dx
and Eq. (5.6)
h = J(-fx2 + 0·248x + 400)

Let the expression under the square root sign = u, then h = ul
N 2x + 0·248)
dh = - 1 ( --
· -dx
·· 2ul k ·
atx = 0
1
q = -ku1 • - 1 (0·248)
2u
= -6(0·248) = -1·49 m3fday/m
at x = 1000
= _ ~ (- 2000 0·72 . )
+ 0 248
X
q 2 365 X 12
= -6( -0·328 + 0·248) = 6(0·08) = 0·48 m3fday/m

Groundwater 81
So there is discharge into both canals from the aquifer of 1·49 m3fday
to the lower, and 0·48 m3fday to the upper, for each metre length of
aquifer.
The foregoing simple cases will serve to show the way in which
the movement of groundwater can be analysed. As conditions
become more complex, the solutions become more difficult, but
standard solutions have been computed for very many groundwater
situations* and most conditions met with in nature can be analysed,
at least, approximately.
5.4 The abstraction of groundwater

The simplest and oldest way of collecting groundwater is by digging
a hole in the ground which penetrates the water table. If the quantity
which can be taken from the hole is not adequate, then the hole must
be extended either horizontally or vertically. Which method is
chosen will depend on the local geo-hydrology.
If the hole is extended horizontally it becomes an open collecting
ditch. Alternatively, it could be underground as a collecting tunnel.
These horizontal collectors must be used if the aquifer thickness is
small and if the drawdown due to abstraction has to be limited, as
for example when a layer of fresh water overlies a layer of salt water.
The vertical extension of the hole makes it a dug or drilled well,
or a borehole. This method 9an be used when the aquifer is of suffici-
ent thickness and in any case when the aquifer is more than about
6 m below ground level. Dug wells are usually 1 m or more in dia-
meter and so the shaft acts as a reservoir for a short-time, high-rate
abstraction. The large diameter is also useful when the entrance
velocity of water into the shaft has to be kept low, e.g. in fine grained
sands.
The majority of wells sunk nowadays for water supply are drilled
wells and these are commonly from 30-500 m deep. They are con-
structed by using drill bits which break the material at the bottom
of the hole into small pieces so that they may be removed with other
tools. Two principal methods are employed: percussion drilling
and rotary drilling. The percussion method involves alternately
• For a comprehensive treatment of the flow of groundwater and its recovery

the reader is referred to two companion volumes in this series, Theory of Ground-
water Flow by A. Verruijt, and Groundwater Recovery by L. Huisman.
raising and dropping the tools in the borehole; in the rotary method,
a rotating bit, cuts or abrades the hole bottom. Drilled wells can
penetrate any materials from soft clay to hard rock up to depths of a
kilometre or more.
As the well is drilled it is "cased" with steel piping to prevent wall
collapse. At the bottom of the casing a well screen is constructed.
This is the point where groundwater enters and the screen is necessary
to prevent the washing in of fine particles and consequent clogging
of the well bottom and pump. The screen should cause as small a
loss of head as possible, be structurally strong, corrosion resistant
and reasonably cheap. These requirements are to some extent con-
tradictory, since the smaller the screen openings and thus the more
effective they are in keeping out fine particles, the greater the head
loss which will be caused.
Modern well screens are usually slotted with fine slots in a plastic
material, though steel, copper, bronze, wood, vitrified clay and
glass are used. Gravel packs are customarily placed around the
screen to act as preliminary filters, and in some cases gravel packs of
reducing diameter particles, placed in concentric rings, may be used
with a simply perforated bottom pipe on the casing.
The construction of drilled wells, well screens and gravel packs
and the techniques of well development and maintenance are beyond
the scope of this book.
Once the water has entered the well it has to be pumped to the
surface. Well pumps may be classified as reciprocating, rotating
vertical shaft, jet and air lift pumps. Rotating vertical shaft may be
either surface driven or be submersible, and may be centrifugal or
rotary positive displacement.
By far the most common application is now the electric submer-
sible centrifugal borehole pump, with the electric drive motor directly
coupled to the pump stages in one long pump body which is placed
near the bottom of the well. Such pumps are manufactured in sizes
down to 100 mm diameter to supply heads up to 100 m or more if
necessary. Such a 100 mm (4 in.) diameter pump would supply about
4m3 per hour while a 250 mm (10 in.) diameter pump might supply
30 times as much. A sketch of a typical installation is shown in
Fig. 5.9.
5.5 The yield of wells
Formulae for the draw down curves of a single well may be derived
Groundwater 83
from the conditions of flow previously discussed in Section 5.3.
Only the two simplest cases are shown here, (1) steady flow to a
- - ---:::.--
- - . =· gr~ut -- --~
.·">•: ·.
:.S.ement
o "• ,_" ' . , : u.•". • : .-" :··•• •o I
~. ",~~ "' .J • o .. outer cosmg
,·_~\ ~._ ~~ ~ ,-, ~ --: i ,-'-
t ',- !. - 1 , ~ner cosmg T '

\·· ...
- .... -_ :.. --: - .:::-:: -::.-- -.:.- -- - ~

.. -- .: -- =: - _..
....• delivery main
Aquifer
well
~motor
FIG. 5.9 Sketch of bottom of a typical water supply borehole
well tapping confined groundwater, and (2) steady flow to a water

table well.
Many other factors should be considered, e.g. the influence of
partial penetration of the well in the aquifer, unsteady flow etc., but
for a full treatment of these the reader is referred to the companion
volumes in this series previously mentioned.
5.5.1 Steady confined How. Drawdown is denoted by ~ and is
measured from the undisturbed piezometric surface before pumping.
(See Fig. 5.10.) The horizontal co-ordinate is measured radially
I ground surface
..,.._...... ~ ~ ~ ~··-
.. ·.... J.p ·
. )< p;e~ometric surface
before pumping
/ ' - after pumping
-
H
flow
II, 'I 'I '/ , " ', '/
r • .. r
Flo. 5.10 Well pumping from a confined aquifer
from the well and is denoted by r, since the flow is radial to the weJJ.
The steady state discharge from the well is called Q0 •
The equations governing flow may be written: from Darcy's Law
d~
Q = vA = - k - . 21rrH
dr
and from continuity

Q = Qo = constant
combining these
Integrated between the limits r = r1, ~ = ~1, and r = rz, ~ = ~

Qo rz
~1 - ~2 = - - . ln- (5.7)
21TkH r1
Groundwater 85
Indefinitely integrated, it yields
Qo
+C
J.
'+' =- 2TTkH . In r
and if 4> = 0 when r = R then
4> = ___!k_ In !!: (5.8)

2TTkH r
Either of these two equations (5.7) and (5.8) enables the drawn-down
curve to be established if the boundary conditions are known.
/_ground surface
-·~ .. - - . - -
..-.-
H
h
.,
r -----:.,-,- 10- r .
FIG. 5.11 Well pumping from an unconfined aquifer
5.5.2 Steady unconfined flow. When the drawdown is slight com-

pared with the thickness of the aquifer, the factor kH remains nearly
constant and the formula for steady confined flow may be used.
As drawdown increases, the falling -water level reduces the area of
transmitting aquifer and the equations now become, in the notation
of Fig. 5.11, Darcy
dh
Q = 2TTr • h . k dr
Continuity
Q == Qo == constant
Combined
h dh == Qo dr
' 2TTk' r
Integrated
Qo
h2 ==- In r
1Tk
+C and if h == H at r == R
H2 - h2 == Qo In ~ (5.9)
1rk r
The value of R must satisfy the boundary conditions. Then the
drawdown at the well face (H - ho) may be deduced by introducing
the radius of the well ro since
H2- h2 == Q0 ln ~ (5.10)
1rk To
Example 5.2. A well is drilled to the impermeable base in the centre

of a circular island of 1 mile diameter in a large lake. The well com-
pletely penetrates a sandstone aquifer 50ft thick overlain by imperme-
able clay. The sandstone has a permeability of 50 ftfday. What will
be the steady discharge if the drawdown of the piezometric surface is
not to exceed 10ft at the well, which has a diameter of 1 ft?
/C)• /
piezometric surface
before pumping
---·
K= 50 It/day
, ,.,
--
5280'
FIG. 5.12 Pumping from a central well in a circular

island and confined aquifer
Groundwater 87
For a well in ~ centre of the circular island, the boundary con-
dition is c/J = 0 when r = 2640 ft.
Then from Eq. (5.8)
Qo 2640
IO = 277' X 50 X 50 In 0·5
_ 5()()()(}n- _ IS7080 _ 3
Qo - lnS280 - 8·572 - 18340 ft /day
= 0·212 ft 3 /s
5.5.3 Steady unconfined How with rainfall. When rainfall is present

the equations become
dh
Darcy Q = 27Trh • k dr
Continuity dQ = -27Tr. dr. P

integrating Q = -7Tr2P + Ct and Ct may be determined from
the condition that when r = ro ~ 0 Q = Qo
Q = -7Tr P + Qo
2
Substituting this value in the Darcy equation

Qo dr P
=
h • dh 27Tk . -; - 2k . r dr
and integrating h2 = Qo.ln r - p ,a+ C2 (5.11)

7Tk 2k
Rainfall p
J
. .
~ ~ .... ::-.~ :-: ~-~--·- ""'+-=-=
r
R '/.,.·I
Flo. 5.13 Central well in a circular island and
unconfined aquifer with rainfall
C2 being an integration constant that must satisfy the particular
boundary conditions. In the case of a central well in a circular
island of radius R when r = R, h = H and so
Qo 2k
C2 = H 2 - -In R
'Trk
+ -R2
p
substituting this value in Eq. (5.11);

Qo R P
H2 - h2 = 1rk In -;: - 2k ( R2 - r 2) {5.12)
If Qo = 0, i.e. there is no pumping, then the shape of the phreatic

surface is given by
p
H2 - h2 = - _ (R2 _ r2) (5.13)
2k
Example 5.3. A circular island 500 m radius has an effective rainfall
P = 4 mmfday. A central well is pumped to deliver a constant Q0 =
25 m3fh from an aquifer with dimensions and properties as shown in
Fig. 5.14. What is the drawdown at the well and at the water divide?
Rainfall P
~ phreatic surface with

rainfall but no pumping
phreatic surface with

both pumping
and rainfall
r= 10m
/
--~---r 0 =0·3 m
Well diameter 0·6 m ---500m---
FIG. 5.14 Circular island with central well, rainfall and an

unconfined aquifer. Solution by super position
Groundwater 89
(I) Assume no pumping. Then from Eq. (5.13)
H2 - h2 = - !_ (R2 - r2)
1 2k
0·004
I 00 - hi = - ----;ro- (250000 - r 2)
IOO - hl 2= - 25 + ( IOO
r )2 (5.14)
(2) Assume no rainfall. Then from Eq. (5.9)
2 Qo R
H2 - h2 = -ln-
TTk r
2 500
I 00 - h2 = 21 ·9 log - (5.I5)
r
Now applying the method of super position and combining Eqs.
(5.14) and (5.15)
100 _ h2 = -25 + c~Y + 21·9 log 5 ~ (5.16)
when r = 0·3 m (at well face) h = 7·39
and from Eq. (5.I4) h1 = ll·I9 and calling drawdown at the well
Sw :. Sw = 3·80 m below the phreatic surface before pumping.
If a water-divide exists, then all the output of the well is being con-
tributed by rainfall, since if the sea around the island were contribut-
ing, the hydraulic gradient would be sloping downwards and in-
wards at every point. So the area contributing is obtained from
where rx is radius of the divide
0·004
25 = TTYx 2 X --
24
rx =218m (well within the 500m radius of the land)

using this value in the combined Eq. (5.16)
h = 10·61
and from Eq. (5.14) h1 = 10·97, and calling drawdown at the divide
Ss
Ss = 0·36 m
Suppose the simple formula ofEq. (5.8) had been used, thus assum-
ing constant aquifer thickness. In this case
Qo R 600 500
Sw = 2TTkH In-; = 2TT X 20 X 10 X 2 "3 log 0·3
= 0·477 X 2·3 X 3·223

= 3·54 m (cf 3·80 m above)
500
and Ss = 0·477 X 2·3 X log 21 S
= 1·01 X 0·36
= 0·364 (cf 0·36 m above)
It will be realised that the simple formula for the confined case is
adequate in this case except in the immediate neighbourhood of the
well. It would, of course, still be necessary to compute the 'no-
pumping' phreatic surface.
6 Surface Runoff
6.1 The engineering problem

Rainfall, if it is not intercepted by vegetation or by artificial surfaces
such as roofs or pavements, falls on the earth where it may evaporate,
infiltrate or lie in depression storage. When the losses arising in these
ways are all provided for, there may remain a surplus which, obeying
the gravitation laws, flows over the surface to the nearest stream
channel. The streams coalesce into rivers and the rivers find their
way down to the sea. When the rain is particularly intense or pro-
longed, or both, the surplus runoffbecomes large and the stream and
river channels cannot accept all the water suddenly arriving. They
become filled and overflow and in so doing they do great harm to the
activities of men. The most serious effect of flooding may be the
washing away of the fertile top soil in which crops are grown, and of
which there is already a scarcity on the Earth. In urban areas there
is great damage to property, pollution of water supplies, danger to
life and often total disruption of communications. In agrarian
societies floods are feared like pestilence because they may destroy
crops, cattle and habitations, and bring famine in their wake.
The hydraulic engineer in dealing with runoff, has to try to provide
answers to many questions, of which some of the more obvious are
(i) how often will floods occur?
(ii) how large will they be and to what level will they rise?
(iii) how often will there be droughts?
(iv) how long may these droughts continue?
Questions of this kind are all variations of one, which is con-
cerned with the magnitude and duration of runoff from a particular
91
catchment with respect to time. They can be resolved only by the
determination of the frequency and duration of particular discharges
from observations over a long period of time, though if such observa-
tions are not available, estimations may still be made at various
probabilities.
A second group of questions arises in using the curves of runoff
frequency and duration once found, e.g.
(i) how can the volume of discharge be reduced?
(ii) how can the cost of flood prevention be compared with the
damage which will arise if no measures are taken?
(iii) how valuable is stored flood water in times of drought?
These questions are not directly related and each involves a quite
different and distinct approach, though the same techniques may be
used in answering more than one. In this and the following sections
ways in which some of these questions may be answered will be
sought.
6.2 Flow rating curves: their determination adjustment and extension

6.2.1 Definition. A rating curve is a graph drawn connecting the
water level elevation, or stage of a river channel at a certain cross-
section with the corresponding discharge at that section. A typical
discharge Q -
FIG. 6.1 Flow rating curve

Surface Runoff 93
rating curve is shown in Fig. 6.1. It may be seen that the curve is
drawn through a cluster of points, each of which represents the
results of a river discharge measurement. Such measurements may
be made in a number of ways of which the most important are
(I) by velocity-area methods
(2) by flow-measuring structures
(3) by dilution gauging
6.2.2 Velocity-area methods. These are conventional for medium

to large rivers and involve the use of a current meter which is a fluid
velocity measuring instrument. A small propeller rotates about a
horizontal shaft which is kept parallel to the streamlines by tail fins.
The instrument is ballasted to keep it as nearly as possible directly
below the observer. Another version of the instrument has a circlet
of small conical cups disposed horizontally about the suspension axis.
Each revolution of the propeller is recorded electrically through a
cable to the observer and the number of revolutions is counted by the
observer or automatically over a short period (say 1 or 2 minutes).
These observations are converted into water velocities from a cali-
bration curve for the instrument. By moving the meter vertically and
horizontally to a series of positions whose co-ordinates in the cross-
section are known a complete velocity map of the section may be
drawn and the discharge through the cross-section calculated. Fig.
6.2 shows a modern current meter assembled for use on its supporting
cable which is also used as a depth measure.
Observations may be made by lowering the meter from a bridge,
though if the bridge is not a single-span one, there will be divergence
and convergence of the streamlines caused by the piers which can
cause considerable errors. In many instances, however, the gauging
site, which should be in as straight and uniform a reach of the river
as is possible, will have no bridge, and if it is deep and in flood a
ca.ble to hold some stable boat must be provided, together with a
lighter measuring cable to determine horizontal position in the
cross-section.
Since the drag on a boat with at least two occupants and suspended
current meter is considerable, a securely fastened steel cable should
be used. The propinquity of suitable large trees at a particular
cross-section may often necessitate its choice for this reason.
-
•
FIG. 6.2 A modern helix current meter by Hilger and Watts Ltd
Surface Runoff 95
Alternatively, cableways are sometimes used to suspend the meter,
either from a manned cable-car or directly from the cable, the
instrument in this latter case being positioned by auxiliary cables
from the river bank.
Depths should always be measured at the time of velocity obser-
vation since a profile can change appreciably during flood discharges.
Observers should also remember such elementary rules as to observe
the stage before and after the discharge measurement, and to observe
the water slope by accurate levelling to pegs at the river level as far up
and downstream of the gauging site as is practicable, up to say
500 m in each direction.
As water velocities increase in high floods the ballasted current-
meter will be increasingly swept downstream on an inclined cable.
The position of a meter in these circumstances may be found reason-
ably accurately if the cable angle is measured. Ballast may be in-
creased but only within limits. Rods may be used to suspend the
meter but a rigid structure in the boat will then be required to handle
the rods, calling for a stable platform of the catamaran type. Rod
vibration and bending are common in deep rivers unless diameters
exceed 50 mm by which time the whole apparatus is getting very
heavy and unmanageable.
It will be appreciated that since each river is unique each will
require a careful assessment of its width, depth, likely flood veloci-
ties, cable support facilities, availability of bridges, boats, etc.
before a discharge measurement programme is started.
From many observations on many rivers it has been established
that the variation of velocity integrated over the full depth of a river
can be closely approximated by the mean of two observations made
at 0.2 and 0.8 of the depth. If time and circumstances preclude even
two observations at each horizontal intercept then one reading at
0·6 depth will approximate the average over the whole depth.
The discharge at the cross-section is best obtained by plotting each
velocity observation on a cross-section of the gauging site with an
exaggerated vertical scale. lsovels or contours of equal velocity are
then drawn and the included areas measured by planimeter. A
typical cross-section, so treated, is shown in Fig. 6.3. Alternatively
the river may be subdivided vertically into sections and the mean
velocity of each section applied to its area.
A check should always be made using the slope-area method of
Section 6.2.6(iii) and a value obtained for Manning's n. In this way a
knowledge of the n values of the river at various stages will be built
up which may prove most valuable in extending the discharge rating
curve subsequently.
10m 10m
2
m 3
4
5
6
FIG. 6.3 Isovel plotting of stream cross-section (points ringed are
those for 0·2 and 0·8 depth observations)
To ensure uniformity in the techniques of current-meter gauging

the International Organisation for Standardisation (I.S.O.) has pub-
lished various recommendations and, in the United Kingdom, BS
3680 "Measurement of liquid flow in open channels" refers.
6.2.3 Flow-measuring structures. These are designed so that stream

discharge is made to behave according to certain well-known
hydraulic laws. For example, the discharge per unit length over a
weir is a function of the head over the weir. Many specialised weirs,
such as V-notch, compound and Crump weirs, have been designed
to provide accurate discharge data by observations of water surface
level upstream of the weir. Similarly with flumes, where a stream is
channelled through a particular geometrically shaped regular channel
section for some distance before entering a length of different cross-
section, made so usually by side contractions or steps in the bed.
Trapezoidal shapes are often used and, recently, narrow vertical
sections have been employed for catchment discharge measurement
in Wales. Fig. 6.4 shows a flume of the latter type on the Plynlimon
experimental catchments of the Institute of Hydrology. Generally,
however, flow-measuring structures are confined to streams and fairly
small rivers, since for large flows and wide rivers they become
extremely expensive to build.
Surface Runoff 97
6.2.4 Dilution gauging. Dilution gauging is particularly suited to
small turbulent streams where depths and flows are inappropriate for
current-metering and flow-structures would be unnecessarily expen-
sive and/or permanent.
The method involves the injection of a chemical into the stream
and the sampling of the water some distance downstream after
complete mixing of the chemical in the water has occurred. The
chemical can either be added by constant-rate injection until the
FIG. 6.4 Narrow flow-measurement flume Plynlimon catchments
sampling downstream reveals a constant concentration level, or

administered in a single dose as quickly as possible known as gulp
injection. In this case samples over a period of time disclose the
concentration/time correlation. In both cases the concentration of
chemical in the samples is used to compute the dilution, and hence
the discharge of the stream may be obtained. Fig. 6.5 shows constant-
rate injection of sodium dichromate from a Mariotte bottle (a
constant-head device) in a mountain stream.
Analysis of the samples is by an automated colorimetric procedure
which estimates the concentration of very small amounts of the
chromium compound by comparison with a sample of the injection
solution. The equipment is expensive and specialised.
The literature on dilution gauging is not large but references [35]
and [36] will provide excellent guidance.
The methods of Sections 6.2.2, 6.2.3, and 6.2.4 used singly or in
conjunction, will establish the correlation for any stream or river
discharge with stage.
A rating curve when established in this way, enables a single
observation of stage, made each day at a set time, by an unskilled
observer, to be converted into a discharge rate and hence into a
finite quantity of water, flowing at the observation point. The diffi-
culty about rating curves is to obtain enough points at times of high
discharge to enable an accurate correlation to be obtained.
6.2.5 Rating-curve adjustment. So far rating curves have been

discussed in terms which imply that they are simply median lines
drawn through a scatter of observation points. This is not the case.
If each discharge point has been noted as being measured on a
'falling' or 'rising' stage the curve would strictly speaking become a
loop as shown by the dashed line in Fig. 6.1.
This variation, or looping effect is due to several causes.
The first of these is channel storage. As the surface elevation of a
river rises water is temporarily stored in the river channel.
Example 6.1. Suppose that the gauge shows a rise at the rate of0·2
mfh during a discharge measurement of 100m 3 /sand the channel is
such that this rate of rise may be assumed to apply to a 1000 m reach
of river between the measurement site and the reach control.*
Let the average width of channel in the reach be 100 m. Then the
rate of change of storage in the reach, dS is given by
dS = 1000 X 100 X 0·2
= 20,000 m3/h
= 5·6 m3/s.
The discharge measurement should be plotted on the rating curve as
94·4 m3fs (not as 100), since this is the discharge past the control
corresponding to the mean gauge height.
The second reason for the looping of rating curves is the variation
in surface slope that occurs as a flood wave moves along the channel.
• The control of a river reach is the section at which the profile of the river
changes. For a full treatment of river profiles and controls, and discharge
measurements, see Flow in Channels by R. H. J. Sellin, a companion volume in
this series.
Surface Runoff 99
FIG. 6.5 Dilutiongauging: dispensing sodium dichromate solution

from a Mariotte bottle
Fig. 6.6 represents a longitudinal section of a flood wave passing
along a river channel. As point a passes the gauging site, the gauge
reads h, the river cross-section is A and the slope of the water surface
s1. When the flood wave has moved on so that point b is at the gauge
site, the gauge reading h and the cross-section A are the same but the
slope is now s2. From the Manning formula
AR1S 1
Q=Av=--
n
and so two different discharges will occur in the two cases since s
changes while A, R and n remain constant.
FIG. 6.6 Flood wave slope variation
Since the rising stage is associated with the greater slope dis-
charge, measurements taken on a rising stage will plot to the right
on the rating curve of Fig. 6.1 and those on a falling stage, to the
left. Depending on the 'peakiness' of the flood wave, it often happens
that maximum discharge occurs before the maximum stage is reached,
since the influence of the steeper slope on velocity may outweigh the
slight increase in cross-sectional area.
It is generally necessary to correct the discharge measurements
taken on either side of flood waves to the theoretical steady state
condition, because the great majority of gauge readings are taken by
unskilled observers once a day without any indication of whether
the stage is rising or falling. By using a corrected, or steady state,
curve, the rising and falling stage observations will balance in the
long run, and no value judgment or second daily visit to the gauge
will be required from the observer. The correction may be made as
·follows.
Surface Runoff 101
From the Manning equation, the steady state discharge Q in a
channel of given roughness and cross-section, is given as
Q ex: ...;s (6.1)
where S = steady state slope.
When the slope is not equal to S, as is the case in conditions of
rising or falling stage, the actual discharge Qa is given by
Qa ex: v(S ± !:J.S) (6.2)
Referring now to Fig. 6.7, as may be expressed in terms of the rate
of change in stage and flood wave celerity, U.
u--
FIG. 6. 7 Change in slope of a flood wave with time
The figure represents an advancing flood wave and rising stage.

The gauge reading at the commencement of the discharge measure-
ment is h and at its conclusion, dt, later, h + dh. In this time the
wave has advanced Udt, and
!:J.S = dh = dhfdt (6.3)
Udt U
: being positive for a rising stage, negative for falling stage. Com-
bining Eqs (6.1), (6.2) and (6.3)
~= J(s + (d;fdt)/U) (6.4)
or
(6.5)
If the discharge measurement is taken at a site with two gauges, one
upstream and one downstream, in the same reach, then all the terms
in Eq. (6.5) are measured except Q and U, Q being the steady state
discharge required and U, the flood wave celerity. There are now
several ways to proceed. The first is to take an empirical value for U.
Corbett [37] suggests that in a fairly uniform channel, in flood con-
ditions, the celerity U is approximately equal to 1·3 times the mean
water velocity, or
Qa
U= 1·3-
A
where A = cross-sectional area.
From which
Q = J( +
1 A . dh/dt)
1·3QaS
(6.6)
Example 6.2. A river discharge measurement made during a flood

indicated Qa = 3160 mBfs. During the measurement, which took 2 h,
the gauge height increased from 50·40 to 50·52 m. Level readings
taken at water surface 400 m upstream and 300 m downstream of the
observation site differed by 100 mm. The river was 500 m wide with an
average depth of 4 m at the time of measurement. At what coordinate
should the measurement be plotted on the rating curve?
The cross-sectional area of the river, A = 500 m x 4m

= 2000 m 2
. Qa 3160
:. average water veloctty =A= 2000 = 1·58 m/s
Assume the flood wave celerity U = =

1·3 x 1·58 2·054 mfs
dh 0·12 m 0·1
- = - - = 1·67 x I0-5 and S = - = 1·43 x I0-4. Then
dt 7200 s 700
for a rising river, from Eq. (6.6)
- 3160 - 3160 - 3
Qcorr - J(1 + 1·67 X I0-5 ) - -yl{1·057) - 3080 m fs
2·054 X 1·43 X 10-4
and taking the mean gauge height, the corrected coordinates are
50·46 m and 3080 mBfs.
Surface Runoff 103
An alternative procedure, due to Boyer [38] is available, where
values for neither U, nor S, need be obtained. If a sufficient number
of observations are available, including measurements taken during
rising and falling stages and in steady states, then a rating curve may
be drawn as a median line through the uncorrected values. The
steady state discharge Q can now be estimated from the median
curve. Since Qa and dhfdt are measured quantities and therefore
known, Eq. (6.5) yields the term 1/US for each measurement of
discharge.
The term 1/ US is now plotted against stage and an 'average' curve
fitted to the plotted points, as shown in Fig. 6.8. From the 1/ US
~~
estimated 'normaJ' _~·
curve .&'"~corrected 'normal'
...P' curve
.."' ~·
0
;;;
~
/
o._ ljus
0
FIG. 6.8 Method of correcting discharge readings

without computing U or S
v. stage relationship, new values of 1/US are obtained and inserted

in Eq. (6.5) to yield the steady state Q. The new values of Q are
then plotted v. stage as the corrected, steady-state curve.
Another method, which uses the observed slope but avoids
evaluating U is due to Mitchell [39].
6.2.6 Extension of rating curves. As previously mentioned, the most

difficult measurements of discharge to take are those at high flood,
because of both the physical difficulties of high water velocity and
floating debris, and the rare occurrence of the condition. It fre-
quently happens, for example, that the condition for which river
structures such as dams and bridges must be designed are defined
as being "such that occur no more often than once in 100 years".
This means that the structural designers want to know the probable
discharge that will occur, on average, once every 100 years. This is
sometimes referred to as "the 100 year flood".
If discharge measurements have been taken throughout the pre-
vious 100 years this design flood may not be too difficult to find.
However, in the vast majority of cases, only stage readings will be
available and then only for limited periods. If the hydraulic engineer
can have 30 years of continuous daily stage readings and discharge
measurements to include even low flood conditions, then he is
fortunate. Almost always he will have to extend the rating curve
well beyond the last measured point, to estimate the discharge at
particular stage levels recorded. Occasionally the stage levels
reached in high flood are recorded only by lines of debris on the
banks, or grasses caught in the branches of riparian trees and scrub.
Such physical evidence is valuable.
There are a number of ways of attempting rating curve extension:
(i) By fitting an equation to the curve.
Usually an equation of the form Q = k(h - a)x is used, where
h =stage
k and x are constants derived from the observed portion of the
curve, and
a = height in m (or ft) between zero on the gauge and the eleva-
tion of zero flow.
Such a curve plots as a straight line on logarithmic paper and so
may be easily extended. At best it is a questionable procedure since
there is little theoretical justification for exponential laws operating
and at high flows there may be quite abrupt changes in cross-section
with rising stage.
(ii) Steven's method [40]. This method is based on Chezy's formula
Q = ACy(RS)
where A = cross-sectional area
C= Chezy roughness coefficient
R= hydraulic radius
S= slope of the energy line
If CyS is assumed constant and D, the mean depth substituted for R
then
Q = kAyD (6.7)
Surface Runoff 105
Known values of Ay'D and Q are plotted and often come close to a
straight line which may be extended. Field values of A v' D above
the measured rating may then be used from the extended line to plot
Q v. stage points on the rating curve.
The objection to this method is simply that Cy'S is not a constant.
However, as it takes account of the varying stream geometry it is a
more rational procedure than (i).
(iii) Slope area method. This method depends on hydraulic principles

and presupposes that it is practically possible to drive in pegs or
make other temporary elevation marks at the time of the peak flow
upstream and downstream of the discharge measuring site. These
marks may subsequently be used to establish the water slope. Cross-
sectional measurements will yield the area and hydraulic radius of
the section. Then from Manning
AR*S!
Q=-- (6.8)
n
This method is sometimes criticised because of its dependence on

the value of n. Since n for natural streams is about 0·035 an error in
n of 0·00 1 gives an error in discharge of 3 %- This objection may be
partially met by plotting n v. stage for all measured discharges, so
that the choice of n for high stages is not arbitrary, but is taken from
such a plot. If the high flood slope can be measured then this method
is probably the best one.
It should be emphasised that all methods of rating curve extension
are suspect to some degree and should only be resorted to if no flood
measurements can be obtained. The latter two methods, above, are
both subject to errors arising from alteration of cross-section due to
flood scour and subsequent low water deposition, so cross-sections
and mean depth measurements should be taken as near to the time
of discharge measurement as possible.
6.3 Duration of mnoff

While floods and droughts are important from many points of view,
they tend, as extremes, to be of comparatively short duration. For
many water resource investigations it is equally important to know
the total volumes of water that have to be dealt with over long
periods of time, e.g. in hydroelectric power generation the plant
capacity must be chosen for some discharge well below the peak
flood since otherwise much capacity would be almost permanently
idle. For such purposes the most convenient means of presenting
data are the mass curve and the flow duration curve.
With an adjusted and well measured rating curve, the daily gauge
readings may be converted directly to runoff volumes. A typical set
of such daily runoff figures is presented graphically in Fig. 6.9. Such
5400
5200
4800
4400
~ 4000
E 3600 r-
·=., 3200 l
1:' 2800
0
A f\ I ~~~
) U_
~ 2400
i5 2000
1\ ! I\ rt
\IV IV"~
1600
1200 - I~ ,. -- / f.:iroundwa ter ''~.\
.... ..,.>
.....
800 contribuiion '
400 c. --) -~--·
~ ~
0
Feb March April May June July August September
FIG. 6.9 Hydrograph of R. Euphrates at Hit. Feb.-Sept. 1957

(after Directorate of Irrigation. Iraq)
a presentation is called a hydrograph. If these volumes are plotted

against time by adding each day's volume to the previous total, a
cumulative mass curve of runoff is obtained. Such a curve is shown
in Fig. 6.1 0.
Mass curves are extremely useful in reservoir design studies since
they provide a ready means of determining storage capacity necessary
for particular average rates of runoff and drawoff. Suppose for
example that the mass curve OA of Fig. 6.11 represents the runoff
from a catchment which is to be used for base load hydroelectric
development. If the required constant drawoff is plotted on the
same diagram, as line OB, then the required storage capacity to
24
22
v--- -
/
20
"'Q 18 f.-
7
"'E 16
c: 14 >--
v
.,
E 12
.=!
-
v
0
> 10 -
0c: 8
:J
a:: 6 >--
[7
4
2 f-
0
Feb. March April May June July August September
FIG. 6.10 Cumulative mass curve of runofffor R. Euphrates

at Hit. Feb.-Sept. 1957
0 A
reservoir spillmg water -?-"'/

reservoir f~ll
again /
/
/
/
/
c/
/
/ maximum required
/
/ reservorr capacity
"'::>
E / to ensure drowoff
0 /
> Reservoir assumed /
full at start of /
0 driest per 10d /
c:
::> of record ~/
a::
c
T1me (months, years, etc.) -
FIG. 6.11 The use of mass curves in reservoir design

107
ensure this rate may be found by drawing the line CD parallel to OB
from a point C at the beginning of the dryest period recorded. The
storage capacity necessary is denoted by the maximum ordinate cd.
Normally much longer periods, as long as the record allows, are
used for reservoir design and in many instances the drawoff is not
constant nor continuous. In such cases, different techniques based
on the same principles are used [41].
52
48
44
40
Q 36
e.g. discharge is equal to or

greater than 2000 m3/s
for 17% of the time
curve tends to be
, modified in this
'-, way if storage
........ .... _
-- ---
" is introduced7
0 100
%time
FIG. 6.12 Flow duration curve for R. Euphrates, Feb.-Sept. 1957,

at Hit (derived from the data of Fig. 6.9)
Flow duration curves show the percentage of time that certain

values of discharge were equalled or exceeded. Such a curve is
shown in Fig. 6.12. Duration curves for long periods of runoff are
useful for deciding what proportion of flow should be used for
particular purposes, since the area under the curve represents
volume. Note that storage will modify the flow duration in the
manner shown by the dashed line of Fig. 6.12, reducing the extreme
flows and increasing the minimum.
The shape of the curve is important as a generally fiat curve
indicates a river with few floods which is extensively supplied from
Surface Runoff 109
groundwater, while a steep curve indicates a river with frequent
floods and dry periods, having little groundwater flow and being
supplied mainly from runoff.
The durations of particular floods or droughts, rather than the
cumulative periods of particular discharge values, are also of interest
and are usually obtainable directly from the hydrograph.
6.4 Catchment characteristics and their effects on runoff

It is now appropriate to consider how various properties of the
catchment area affect the rate and quantity of discharge from it. By
'catchment area' is meant the whole of the land and water surface
area contributing to the discharge at a particular stream or river
cross-section, from which it is clear that every point on a stream
channel has a unique catchment of its own, the size of catchment
increasing as the control point moves downstream, reaching its
maximum size when the control is at the sea coast.
There are many catchment properties which influence runoff and
each may be present to a large or small degree. The intention, in
analysing them separately, is to try to determine the effect of each
characteristic on precipitation and its subsequent drainage from the
catchment through the river channels.
(a) Catchment area. The area as defined at the beginning of this

section is usually, but not necessarily, bounded by the topographic
water-divide. Fig. 6.13 shows a hypothetical cross-section through the
topographic water-divide of a catchment. Because of the underlying
geology it is perfectly possible for areas beyond the divide to contri-
bute to the catchment. The true boundary is indeterminate, however,
because although some of the groundwater on the left of the divide
in the figure may arrive in catchment B, the surface runoff will stay
in catchment A. Here the infiltration capacity of the soil and the
intensity of the rainfall will influence the portion of the rainfall that
each catchment will collect.
If runoff is expressed, not as a total quantity for a catchment, but
as a quantity per unit area, (usually m3/s per square kilometre or
ft3fs per square mile) it is observed, other things being equal,
that peak runoff decreases as the catchment area increases. This is
due to the time taken by the water to flow through the stream
channels to the control section (the time of concentration) and also
to the lower average intensity of rainfall as storm size increases (see
Sect. 2.8.4.). Similarly, minimum runoff per unit area is increased
due to greater areal extent of the groundwater aquifers and minor
local rainfall.
(b) Slope of catchment. The more steeply the ground surface is

sloping the more rapidly will surface runoff travel, so that concen-
tration times will be shorter and flood peaks higher. Infiltration
topographical
I rain
/.
woler divide
ca tch ment catchment

A B
FIG. 6.13 Watershed defined by geology as well as topography
capacities tend to be lower as slopes get steeper, since vegetation is

less dense and soil more easily eroded, thus accentuating runoff.
Slope may be enumerated by covering a catchment contour map
with a rectilinear grid and evaluating the slope, perpendicular to
the contour lines at each grid point as shown in Fig. 6.14(a). A fre-
quency distribution of these numbers may then be plotted as in
Fig. 6.14(b). Different catchments may be compared on the same
plot, the relatively steep frequency curves indicating catchments of
fast runoff and flat curves the converse.
(c) Catchment orientation. Orientation is important with respect to

the meteorology of the area in which the catchment lies. If the
prevailing winds and lines of storm movement have a particular
seasonal pattern, as they usually have, the runoff hydrograph will
depend to some degree on the catchment's orientation within the
....
....
....
0 50 100
percentage greater than
(a) (b)
FIG. 6.14 (a) Rectilinear grid to evaluate catchment slope and orientation. (b) Frequency curves for
comparison of catchment steepness
pattern. The grid of Fig. 6.14(a) can be used for this feature also,
by the evaluation of the angle between slope direction and the N-S
meridian (say) at each grid point and the subsequent plotting of a
circular frequency diagram like that of Fig. 6.15, similar to a wind
rose.
(d) Shape of catchment. The effect of shape can best be demonstra-

ted by considering the hydrographs of discharge from three differ-
ently shaped catchments of the same area, shown in Fig. 6.16,
--- percentage of grid point

slope lines within sector
FIG. 6.15 Orientation diagram
subjected to rainfall of the same intensity. If each catchment is

divided into concentric segments, which may be assumed to have all
points within an equal distance along the stream channels from the
control point, it may be seen that the shape A will require 10 time
units (say hours) to pass before every point on the catchment is
contributing to the discharge. Similarly B will require 5 h and C
81- h. The resulting hydrographs of runoff will be similar to those
shown in Fig. 6.16, each marked with the corresponding lower-case
letter. B gives a faster stream rise than C and A, and similarly a
faster fall, because of the shorter travel times.
Surface Runoff 113
This factor of shape also affects the runoff when a rainstorm does
not cover the whole catchment at once but moves over it from one
end to the other. For example, consider catchment A to be slowly
covered by a storm moving upstream which just covers the catch-
ment after 5 h. The last segment's flood contribution will not arrive
at the control for 15 h from commencement, so that the hydrograph
Each of the catchments

has the same area
Rain is assumed to cover
the whole catchment
and last at least 10 h
A B
y/ [/
at .....
..... I..-
17vv. I ... -- J- f.-'

I ......
~ v~ I /
v /
/ ~I al
I
~/ ...... /
0 5 10 15 hours
FIG. 6.16 The effect of shape on catchment runoff
(a) of Fig. 6.16 will now have the form of curve a1 on that figure.
Alternatively if the storm were moving at the same rate downstream,
the flood contribution of time-segment 10 would arrive at the control
point only 5 h after that of segment 1, so that a rapid flood rise (a2 in
Fig. 6.16) would occur. The effect of changing the direction of storm
movement on the semicircular and fan-shaped catchments will be
less marked than this but still appreciable.
(e) Altitude of the catchment. Generally, precipitation increases
with altitude though individual catchments show wide variations
from the general rule. Fig. 6.17 shows the trend for S.W. Scotland.
More important however is the effect of reduced evaporation in
2000
!/
1500
..
1000 I
900
800
700
.. I
I.
I
600
r/
500
;;:::
.. 400
-.:;
>
.!!
. I
./
.,0' 300
.,"'> .
0
.c 200
II
0
c:
..
.2
0
.,>
i:iJ
I
100
50 I
10 20 30 40 50 60 70 80 90
Annual rainfall: inches
FIG. 6.17 Correlation between altitude and annual rainfall for
S. W. Scotland
lower temperatures and the temporary storage of precipitation as

snow and ice. This affects the distribution of the mean monthly
runoff reducing it to a minimum in winter in cold climates. There
are thus counterbalancing effects with increasing altitude and which
one predominates depends on the prevailing climate. Comparisons
of different catchments may be made by plotting frequency distribu-
tions of percentage catchment area at or above certain altitudes.
Surface Runoff 115
(f) Stream pattern. The pattern of stream development in a catch-
ment can have a marked effect on the rate of runoff. A well-drained
catchment will have comparatively short times of concentration and
hence a steeper flood-rise hydrograph than a catchment with many
surface depressions, marshy ground and minor lakes for example.
It is very difficult to convert particular drainage patterns to numbers,
but comparisons may be made in terms of stream density, i.e. length
of stream channel per unit area; average length of tributaries
or average length of overland flow, i.e. flow to the nearest stream
channel. Geology of course plays a part in the process of stream
formation, and both geological and topographical maps should
be examined in assessing the effect of different patterns.
(g) Other factors. In addition to the main factors discussed above,
the following will influence the rate and total quantity of surface
runoff and so affect the magnitude of flood peaks and hydrograph
shape.
(i) condition of stream channels: whether clean or weedy and
overgrown.
(ii) the presence of reservoirs, lakes, flood plains, swamps etc.
(see (f) above).
(iii) land use: e.g. whether arable land, grassland or forest:
artificial drainage etc.
(iv) Sub-surface conditions including initial soil-moisture, height
of the phreatic surface, depth and permeability of aquifers
and infiltration capacities.
6.5 Climatic factors

In section 6.4(d) reference has already been made to the effect of
storm movement on surface runoff. If the areal extent of a storm is
such that it does not cover the whole of a catchment, the runoff will
be less than from complete coverage.
In section 6.4(e) the effect of the form of precipitation was men-
tioned, where snowfall and freezing temperatures can effectively put
the expected runoff into storage and reduce evapo-transpiration.
The main effect of climate however is in rainfall intensity and
duration. Rainfall intensity has a direct bearing on runoff since once
the infiltration capacity is exceeded all the excess rain is available
and flows to the surface water-courses. Intensities vary greatly, the
maximum usually occurring in severe local thunderstorms. It will be
realised that for severe local maxima to be recorded at all can only
occur fortuitously, so that the highest recorded intensities will
certainly have been exceeded many times. Jennings [42] and subse-
quently Paulhus [87] list the highest recorded values, some of which
are as follows:
Depth
Duration in. mm Station Date
1 min 1·23 31 Unionville Md., U.S.A. 4 July 1956

20 min 8·10 206 Curtea de Arges, Rumania 7 July 1889
24h 73·62 1870 Cilaos, La Reunion 15-16 Mar 1952
31 days 366·14 9300 Cherrapunji, India July 1861
The highest intensities recorded in the British Isles are given below:
Depth
Duration in. mm Station Date
10 min 1·30 33 Chagford 23 Sept. 1927

about 60 min 3·63 92 Maidenhead 12 July 1901
105 min 6·09 155 Hewenden Reservoir 11 June 1956
about 14! h 11·00* 279 Martinstown, Dorset 18 July 1955
* The storm of Fig. 2.12

Since intensity represents depth/time, it cannot be considered
separately from duration. The same depth of rainfall delivered over
two different durations will obviously produce quite different runoff
rates. What can be said is that different climates will produce
different meteorological conditions leading to different types of rain
which may inherently have quite dissimilar durations. For example,
in England a thunderstorm may conceivably produce rainfall
intensities as great as 20 mm/min but it cannot be imagined that
such storms can last for more than a period of minutes, whereas the
monsoon rain in India can fall continuously for weeks at average
intensities greater than 10 mm/h, a condition never approached in
most other parts of the world. (See also Section 9.6.)
Surface Runoff 117
The influence of duration on the hydrograph of runoff may be seen
from Fig. 6.18, where a uniform-intensity storm causes the hydro-
graph of stream-rise a. Such storms may be defined as covering the
whole catchment area, over which the depth of rainfall is reasonably
constant and delivered at a constant rate. Although rare in nature,
they are used in hydrology to determine characteristics of catch-
ments. After a certain time, tc, the period of concentration, the rate
of runoff becomes constant. A hydrograph of this form is typical
only of very small catchments; e.g. paved urban areas, where such
constant runoff is quickly achieved. Natural catchments of any size
time---
FIG. 6.18 Hydrograph of short-period storm runoff derived from

two long-duration uniform-intensity storms
have periods of concentration longer, as a general rule, than periods

of uniform-intensity rain. The effect of short periods of rain can be
found by subtracting the hydrographs of two long periods, exceeding
tc, and separated in time by the short period, one from the other.
In Fig. 6.18, a and b are similar and are the result of rainfalls a' and
b' respectively. Their subtraction leaves the short-period rain c'
and its resulting hydrograph c, which is the typical shape of most
natural hydrographs.
6.6 Rainfall/runoff correladon

While there is a general cause-and-effect relationship between rainfall
and the resulting runoff, it will be clear by now that it is not a direct
one. By the time that evaporation, interception, depression storage,
infiltration and soil-moisture deficiency are taken into account and
the resulting residual rainfall at various intensities applied to catch-
ments that vary in size, slope, shape, altitude, sub-surface geology
and climate, the relationship must include extreme values that defy
rational correlation, at least in the short term.
Notwithstanding the foregoing, it may be possible to establish an
empirical relationship for a particular catchment based on annual
Rmm
Pmm
Fro. 6.19 Rainfall/runoff correlation
precipitation and runoff. To do this it is best to use a water-year

rather than a calendar year. This is a 12-month period starting and
finishing at the time of seasonal minimum flow. If precipitation and
runoff are plotted against each other as catchment depths, a correla-
tion like that shown in Fig. 6.19 may be obtainable. In temperate
and tropic, humid climates such a straightline relationship generally
is found, where if P is annual precipitation, annual runoff R is
expressed as
p
R=--X (6.7)
y
Surface Runoff 119
so that the annual rainfall can be used to obtain a first approximation
to the annual runoff.
Variations from the straight line may be due to conditions in the
preceding year which gave markedly higher or lower groundwater
levels, or to variations in the seasonal distribution of rainfall. The
method can be used also for wet months in humid climates when the
ground is saturated, but beyond such narrow limits it is not valid.
Although the application of a relationship like Eq. (6.7) is restric-
ted, it can nevertheless be a useful method for estimating total annual
runoff of completely ungauged catchments, if they are in similar
climates and of similar size and character.
Further refinements are possible taking into account the particular
period of the year, the antecedent precipitation index (see Section
4.4.2) and the storm duration as well as depth, so that relationships
may be derived for particular storms on a particular catchment.
Coaxial graphs may be produced which take the various variables
into account. A relationship of this form is shown in Fig. 4. 7.
7 Hydrograph Analysis
7.1 Components of a natural bydrograpb

The various contributing components of a natural hydrograph are
shown in Fig. 7.1. To begin with there is base-flow only, i.e. the
groundwater contribution from the aquifers bordering the river
which go on discharging more and more slowly with time. The
hydrograph of base-flow is near to an exponential curve and the
quantity at any time may be represented very nearly by
Qe = Qoe-a.t
where Qo. = discharge at start of period
Qe = discharge at end of time t
oc = coefficient of aquifer
e = base of natural logarithms
As soon as rainfall begins there is an initial period of interception
and infiltration before any measurable runoff reaches the stream
channels and during the period of rain these losses continue in a
reduced form as discussed previously, so that the rain graph has to
be adjusted to show nett, or effective rain. When the initial losses
are met, surface runoff begins and continues to a peak value which
occurs at a time tp, measured from the centre of gravity of the rain-
graph of nett rain. Thereafter it declines along the recession limb
until it completely disappears. Meantime the infiltration and perco-
lation which has been continuing during the gross rain period results
in an elevated groundwater table which therefore contributes more
at the end of the storm flow than at the beginning, but thereafter is
again declining along its depletion curve.
Surface runoff is, for convenience, assumed to contain two other
120
Hydrograph Analysis 121
components; channel precipitation and inteiflow. Channel precipi-
tation is that portion of the total catchment precipitation that falls
directly on the stream, river and lake surfaces. It is usually small but
if large lakes are present in the catchment it may be quite important
and then requires separate treatment. Intc;:rflow refers to water
net or effective rain

C.G.
lass curve
.tf:V
recession curve
"
.groundwater ' ,
depletion curve '
extrapolated ...--"
----
Groundwater contribution
to streamflow
Time h (or days)
FIG. 7.1 Component parts of a natural hydrograph
travelling horizontally through the upper horizons of the soil, per-

haps in artificial tile drain systems or above hard-pans or impermeable
layers immediately below the surface. Such flow can vary from
nothing to appreciable fractions of total runoff.
Since the groundwater contribution to flood flow is quite different
in character from surface runoff it should be analysed separately
and one of the first requirements in hydrograph analysis therefore
is to separ~te these two.
7.2 The contribution of baseftow to stream discharge

Since baseflow represents the discharge of aquifers, changes occur
slowly and there is a lag between cause and effect which may easily
extend to periods of days or weeks. This will depend on the trans-
missibility of the aquifers bordering the stream and the climate.
Some of the infinite number of natural conditions are considered
below.
A broad distinction should be made between influent and effluent
streams. An influent stream is one where the baseflow is negative,
i.e. the stream feeds the groundwater instead of receiving from it,
e.g. irrigation channels operate as influent streams and many
time -->-
FIG. 7.2 Hydrograph of an ephemeral stream
natural rivers which cross desert areas do so. The negative contribu-
tion is taking place at the expense of contributing aquifers on other
parts of the stream, since there can be no baseflow from a wholly
influent stream. Such a stream, e.g. a Middle Eastern wadi, will dry
up completely in rainless periods and is called ephemeral, and has
a hydrograph of the form of Fig. 7.2.
An effluent stream on the other hand is fed by the groundwater
and acts as a drain for bordering aquifers. The great majority of
streams in Britain and Europe are in this category.
Intermittent streams are those which act as both influent and
effluent streams according to season, tending to dry up in the dry
season.
Perennial streams are greatly in the majority, with a low dry-season
flow fed by baseflow and are mainly effluent streams, though many
perennial rivers crossing different geological formations of varying
permeability and subject to different climates are both influent
and effluent at different parts of their courses. A good example of
this is the river Euphrates in Iraq. Fig. 6.9 shows a part-annual
hydrograph of the Euphrates and the slow seasonal variation of the
baseflow may be observed. This baseflow is derived principally
from the headwaters of the catchment in northern Iraq, Turkey and
Syria. At Hit, where the hydrograph was observed, the river for
much of the year is influent.
Bank storage describes the portion of runoff in a rising flood which
is absorbed by the permeable boundaries of a water course above the
FIG. 7.3 Influent stream
normal phreatic surface. It is illustrated in Figs. 7.3 and 7.4. In the

latter figure the direction of the arrows showing influx of ground-
water to the stream will be reversed during the flood period while
Bonk storage
Flood level
normal out~f=lo..c.w_~
of ground water
FIG. 7.4 Effluent stream
the surface level of the stream is above the phreatic surface. As a

result the hydrograph of a particular flood might well have a baseflow
contribution as indicated in Fig. 7.5. Such a separation as is shown
there is virtually impossible to make quantitatively but it is quali-
tatively correct.
In many natural rivers, depending naturally on bank permeability
and the slope of the phreatic surface, the variation in baseflow will
.
..
surface
~ runoff
,
.r.
.,u
.... increase in baseflow
after fload
,,
..............
/,. ....
11 baseflow
'\ I
time-
""""v.--- ornegative baseflow
bank storage
Fxo. 7.5 Negative basejlow
1
extrapolated
depletion curve
-·~-1 .
···· .....~~---
basel low
time-
Flo. 7.6 Basejlow separation
124
be much less than indicated in Fig. 7.5 and will only cause a slight
dip from the extrapolation of the depletion curve, followed by a
gradual rise to a higher-than-initial value as indicated in Fig. 7.6.
7.3 Separation of baseflow and runoff

It has been shown in the preceding paragraphs that the dividing line
between runoff and baseflow is indeterminate and can vary widely.
Since, to analyse its precise position would require a detailed
knowledge of the geohydrology of the catchment, including the areal
extent and transmissibility of the aquifers, it is generally more
practical to use a consistent separation technique. Which of the
following is used depends on the data available.
If a continuous discharge record of the stream, over a period of a
few years is available, the hydrograph should be plotted in the
manner of Fig. 7.7(a). This is examined for portions which include
recession curves running into baseflow contribution only, after
runoff has ceased, at as many different stages as possible. These
sections are abstracted from the continuous hydrograph and plotted
again to a log Q vertical scale and linear time scale, as shown in
Fig. 7.7(b). Starting with the lowest recession flow in the record, a
curve is now constructed such as is tangential to the lower portions
(i.e. the true depletion curves) of the log Q abstracted plots. This is
most easily done by moving tracing paper over the plots, with the
abscissae coincident, until each log Q plot in successive increasing
magnitude fits into the growing curve and extends it fractionally
upward. The tangential curve thus established to the highest stage
possible is then converted back to linear vertical scale and is called
the master depletion curve for the particular gauging station. It
may now be applied to the hydrograph of a particular storm period
in the manner indicated in Fig. 7.8 whereby the depletion curves are
fitted together at their lower ends and the point of divergence
marked as N. N represents the point at which surface runoff has
effectively ceased and a straight line is drawn to it from the point of
sudden rise. This line, shown dashed in Fig. 7.8, represents the base
line of the hydrograph of surface runoff which can then be analysed.
While the procedure outlined above is probably the best available,
it does depend on previously observed data which is not always
available. An alternative procedure is to establish the point of
greatest curvature on the recession limb of the hydrograph. This is
C)
(I)
.,~
.s::
0
..
'0
time-
(a)
.....
N
0'\
Recession curves ore plotted so that their

lower ends (where the flow is only boseflow)
--.__~!.flo 0 " " " " " liM

Moster depletion curve
replotted Iinearly
C) C)
2"' (I)
plot of I he ----::::!; q h . .,~
~~~~~e
master depletion curve 1 i .s::
0
o5
lime units
00
- ~
time units
FIG. 7.7 Derivation of a master depletion curve. (a) Normal hydrocroph with recession curves selected.
(b) Log plot of recession curves. (c) Linear plot of master depletion curve
.
CD
2'
..<:
u
"'
'6
time-
Flo. 7.8 Procedure to separate base flow
I I I I
<::l'f.l·3~
v,
120 l'"uk- f:hange af slope-
<::!'- .,
110
......... 2 1·2
100
I ~-.....
~1·1
90 I I
1·0
......... ~ 1
6h intervals
II
I \
..
.!!!
E
80
70 I I
\ This vertical bisect s
.,
eo 60 I I N' \ I the 6 h time interval
between Qand Q+& f or
0
..<:
u 50 I surface runoff /I\ I the point at change of slope
·c"' I \ll
40
- - --
-- v - -
30 ~
.... ['....
20 ~---
base flow
-r--
10
2 3 4 5 6 7
Days
FIG. 7.9 Alternative method of separating basejlow

127
perhaps most easily done by computing the ratio between Q at any
time and say 2 h (or any convenient interval) later. An example will
illustrate the method. Fig. 7.9 is the observed hydrograph of a
river over a period of several days. It is desired to separate surface
runoff and baseflow. Starting at point A and using a 6 h separation
for successive ratios, the computations are as shown in Table 7.1.
TABLE 7 .I. Computation to assist in finding N
Ratio
Day Hour Q Q+6
m3fs m3js Q
Q+&
1200 60·1 47·5 1·27

5 1800 47·5 39·0 1·22
2400 39·0 33·2 1·18
0600 33·2 28·6 1·16

1200 28·6 25·2 1·13
6
1800 25·2 22·7 1·11
2400 22·7 20·9 1·09
0600 20·9 19·7 1·06

1200 19·7 18·9 1·04
7
1800 18·9 18·2 1·04
2400 18·2
It may be seen from the inset graph of the ratio-time interval on

Fig. 7.9 that two separate slopes are apparent, the upper being associ-
ated with runoff and the other with groundwater depletion. At their
intersection, the critical ratio may be determined and the first point
beyond the region of intersection on the groundwater side gives a
conservative position for N. The subsequent hydrograph analysis
is not very sensitive to the precise position of N and either 0300 or
0600 h on Day 6 would be satisfactory. A straight line is now
drawn to N from the point where the hydrograph started to rise, as
before. The total amount of runoff may now be obtained by measur-
ing the area under the curve and above the straight line.
The position of N is important also in synthesising hydrographs,
as will be seen in Sect. 7.10, since it partly defines the base length of
the hydrograph. The baselength is made up of the part before the
peak, which depends on the length of the period of rain and tc the
period of concentration, and the recession limb after the peak which
depends primarily on the character of the catchment. From obser-
vations on many natural catchments the position of N may be
established empirically, from the Table 7.2 given below, in days after
the peak of the flood.
TABLE 7.2. Catchment area as a guide to N
Catchment Time from

Area peak toN
km2 days
250 2
1250 3
5000 4
12500 5
25000 6
7.4 The unit hydrograph

Having derived the hydrograph of surface runoff by the methods
discussed in preceding sections, the problem now arises of how it
can be correlated with the rainfall which caused it. Clearly the
quantity and intensity of the rain both have a direct effect on the
hydrograph but it has not yet been made clear how, and to what
extent, each of these affects it. The method of doing this is a part-
empirical, part-theoretical technique which uses the concept of the
unit hydrograph (also called the unitgraph), first described by
Sherman [43].
It should be emphasised that the correlation sought is between the
nett or effective rain (i.e. the rain remaining as runoff after all losses
by evaporation, interception and infiltration have been allowed for)
and the surface runoff(i.e. the hydrograph of runoff minus basefiow).
The method involves three principles which are as follows:
1. With uniform-intensity nett rainfall on a particular catchment,
different intensities of rain of the same duration produce
runoff for the same period of time, although of different
quantities. This is an empirical rule which is approximately
true and is illustrated in Fig. 7.1 0.
2. With uniform-intensity nett rain on a particular catchment,
different intensities of rain of the same duration produce
hydrographs of runoff, the ordinates of which, at any given
time, are in the same proportion to each other as the rainfall
intensities. That is to say, that n times as much rain in a given
time will give a hydrograph with ordinates n times as large.
In Fig. 7.10 the ordinates at time t1 are np and p respectively
for rainfall intensities of ni and i.
.s::
eE
;.:. ni
~ ; nell rain only shown
r::
~ .
.E
hydrograph of ni mm/h for I hr•
hydrograph of I mm/h for I hr
T • base length, the same

in both cases
F1o. 7.10 Proportional principle of the unitgraph
3. The principle of superposition applies to hydrographs resulting

from contiguous and/or isolated periods of uniform-intensity
nett rain. This is iilustrated in Fig. 7.11 where it may be seen
that the total hydrograph of runoff due to the three separate
storms is the sum of three separate hydrographs.
Having established these principles the concept of unit rain is now
introduced. A unit of rain may be any specified amount, measured
as depth on the catchment, usually 1 em or 1 in. but not exclusively
so. The unit rain then must all appear as runoff in the unit hydro-
graph. The area under the curve of the. hydrograph has the dimen-
sions of instantaneous discharge multiplied by time, or
£3
T x T = L3 = volume of runoff
so that although unit rain is spoken of as 1 em over the whole of the
catchment area the resulting runoff is given in cubic metres, and the
quantities involved are identical. If the unit graph for a particular
catchment, and a particular duration of rain is known, then from
principle 2, the runoff from any other rain, of the same duration may
be predicted.
This is a first step towards the complete correlation sought, but
if the rainfall should be of different duration from that of the unit-
graph then the unitgraph must be altered before it can be used.
;=-
.
"iii
c:
]
nett rain only
resulting runoff hydrograph

obtained by summing the
ordinates of the three
compopent hydragraphs
time
Flo. 7.11 Principle of superposition applied to unitgraphs
7.5 Unit bydrograpbs of various durations

7.5.1 Changing a short duration unitgrapb to a longer duration unit-
graph. The simplest way to produce a unitgraph for a longer dura-
tion of rain is illustrated in Fig. 7.12.
Suppose a 2 h unitgraph is given and a 4 h unitgraph is wanted.
This may be obtained by assuming a further 2 h period of nett rain
immediately following the first, which will give rise to an identical
unitgraph, but shifted to the right in time by 2 h. If the two 2 h
unitgraphs are now added graphically, the total hydrograph ob-
tained represents the runoff from 4 h of rain at an intensity of
t em/h. (This must be so because the 2 h unitgraph contains 1 em
rain.) This total hydrograph is therefore the result of rain at twice
the intensity required and so the 4 h unitgraph is derived by dividing
its ordinates by 2. This is shown as the dashed line on Fig. 7.12. It
will be observed that it has a longer time base by 2 h than the 2 h
2 4 6 8 time in h
4 h unit hydrograph
lime-
Fro. 7.12 Changing a short period unitgraph to a long period one

(if the long is an even multiple of the short)
unitgraph; this is reasonable since the rain has fallen at a lower
intensity for a longer time.
1.5.1 Changing a long duration unitgraph to a shorter duration unit-

graph. To derive a short-rain period unitgraph from that for a
longer period it is necessary to use an S-curve technique. An S-
curve is simply the total hydrograph resulting from a series of
continuous uniform-intensity storms delivering 1 em in t 1 h on the
catchment, i.e. it is the hydrograph of runoff of continuous rainfall
at an intensity of 1/tl. Such a hydrograph has the form of Fig. 7.13,
the discharge of the catchment becoming constant after tc, the time
of concentration, when every part of the catchment is contributing
and conditions are in a steady state. Thus each S-curve is unique
for a particular unitgraph duration, in a particular drainage
basin.
If a second S-curve is drawn one unit period to the right of
the first, then clearly the difference between the two S-curves ex-
pressed graphically equals the runoff of one t1 h unitgraph.
If the unitgraph for a shorter period storm of t 2 h is required, it
may be obtained by drawing the S-curve again, but shifted only 12 h
along the time axis. The graphical difference between ordinates of
the two S-curves now represents the runoff of 12 h rain at an intensity
of 1/t1 cmfh. The ordinates of this S-curve difference graph must
therefore be multiplied by t1 ftz so that the rain intensity represented
is l/t2 cmfh, which is the intensity required for the 12 unitgraph. The
procedure is illustrated in Fig. 7.13.
If the time base of the unitgraph is T h, then steady state runoff
must occur at Th and so only T/II unitgraphs are necessary to
develop constant outflow and so produce an S-curve. The constant
2·78A (km 2 )
a.= ttlhl
0 time-
FIG. 7.13 Transposing unitgraphs by S-curves
outflow, Qe, may easily be obtained since I em on the catchment is

being supplied and removed every 11 h.
Q, == 2·78A
11
where A is catchment area in km2
11 is duration in h
and Qe is in m3 /s.
It will be apparent that the method may be used for altering the
unit period either way, longer or shorter and that if changing from
shorter to longer duration, that 12 need not be a direct multiple of 11.
Although the method has been described graphically, in practice its
application is usually made in tabular form and Example 7.1 illus-
trates it.
Example 1.1. Given the 4 h unit hydrograph listed in column (2)
derive the 3 h unit hydrograph. The catchment area is 300 km2.
2·78 X 300
The S-curve equilibrium flow Qe =
4
= 208 m 3 fs.
It will be noted that Qe = 208 m3fs, as calculated, agrees very
well with the tabulated S-curve terminal value 207. This is an indi-
cation that the 4 h period of the unit graph is correctly assessed.
Very often with an uneven rainfall distribution, an attempt has to be
made to reduce the nett rain to a uniform-intensity rain of particular
duration. The S-curve can in this way serve as a check on the chosen
TABLE 7.3. S-curve method

(1) (2) (3) (4) (S) (6) (7)
time 4h S-curve 4
S-curve lagged col co/6 X 3
h unitgraph additions co/s S-curve (4)-(S) = 3h
(2)+(3) unitgraph
0 0 0 0 0
1 6 6 6 8
2 36 36 36 48
3 66 66 0 66 88
4 91 0 91 6 85 113
5 106 6 112 36 76 101
6 93 36 129 66 63 84
7 79 66 145 91 54 72
8 68 91 159 112 47 63
9 58 112 170 129 41 55
10 49 129 178 145 33 44
11 41 145 186 159 27 36
12 34 159 193 170 23 31
13 27 170 197 178 19 25
14 23 178 201 186 15 20
15 17 186 203 193 10 13·5·
16 13 193 206 197 9 12•
17 9 197 206 201 5 6·5·
18 6 201 207 203 4 5·5·
19 3 203 206 206 0 o•
20 1·5 206 207 206 1 1·5·
21 0 206 206 207 -1
• Slight adjustment is required to the tail of the 3 h unitgraph. This is most
easily done by eye (see Fig. 7.14). All figures except col. 1 in m 3 /s
value. If the S-curve terminal value had fluctuated wildly and not
steadied to a minor variation it would have indicated an incorrect
rainfall-time for the unit graph.
Note also that it was not necessary in Table 7.3 to set out T/t1
columns of the 4 h unitgraphs, and add them laterally. The S-curve
additions are the S-curve ordinates shifted in time by 4 h. Since the
first 4 h of unitgraph and S-curve are the same, the S-curve additions
and S-curve columns are filled in, in alternate steps. The effect is the
120
4
110 1\
100
90
Ij "''\
I
I • \
~~h unitgraph derived from S- curve
if 4 h unitgraph given
II
80
., 70
'\\
o;;.. \~\
'..
..
E
60
~
0 50
J/ '\\
.<:::
.,
u ~\
0 40 I
{• '.t.\
30 'l\.,.,
20 !I '\.
••--.::::.,
f-9
t;
10 -~,
0
I
I r 'I 'f.:-_!:,_
0 2 4 6 8 10 12 14 16 18 20 22
Hours
FIG. 7.14 Unitgraph derived by S-curve method
same as setting out rows of unitgraph ordinates successively stag-

gered 4 h, since the S-curve additions represent the sum of all
previous unitgraph ordinates.
7.6 The unit hydrograph as a percentage distribution

The distribution graph, first used by Bernard [44], represents the
unitgraph in the form of percentages of total flow occurring in par-
ticular unit periods. Since the discharge represented by a unitgraph
is directly proportional to nett rain, the percentages in unit
times will remain constant whatever the nett rain. This is a useful
means of applying the unitgraph method in some cases.
In Fig. 7.15 a unit hydrograph is shown, together with the derived
distribution graph which represents it. The areas under the curve
and under the step line are the same and so, in deriving unitgraphs
from distribution percentages a smooth line must be drawn through
the steps to give equal areas.
The distribution graph is therefore less precise than the hydro-
graph but is much better suited for iterative processes of derivation,
as will be seen in Section 7.8.
hours
(a)
unit periods
(b)
FIG. 7.15. (a) Unit hydrograph. (b) Derived distribution graph

7.7 Derivation of the unit hydrograph
The unitgraph for a particular catchment may be derived from the
natural hydrograph resulting from any storm which covers the
catchment and is of reasonably uniform intensity. If the catchment is
very large, i.e. greater than (say)5000 km2 , it may never be covered by
a uniform-intensity storm, since these are limited in size by meteoro-
logical conditions. In such a case the catchment should be divided
coorcl. position of average peak
p31----"""-- ( ~ +lz+l3 P1+Pz+P3 )

PII---~I:n 3 • 3
Pz 1-----f'-.h~...
average unitgroph
sketched to fit and
to contain unit runoff
time
FIG. 7.16 Average unit hydrographfrom a number
up into tributary catchments and the unit graphs for each of these
determined separately.
The first step is to separate the baseflow from surface runoff
(Sect. 7.3) and plot the runoff and the rain graph on the same time
base. The quantity of nett storm rain must then be estimated and its
intensity and duration established. A check is now made on the
quantity of nett rain on the catchment and the amount of runoff
under the hydrograph. These should be the same and one or the
other may require adjustment.
The unitgraph may now be obtained by dividing the runoff
hy<jrograph ordinates by the nett rain in em. The adjusted ordinates
represent the unit graph for the particular· duration established.
It is always advisable to determine several unit graphs, using
separate and distinct isolated uniform intensity storms, if available.
Natural events like rainstorms and runoff, are affected by a multi-
plicity :>f factors and no two are precisely the same. Frequently the
best natural data will be for different rain durations and the resulting
unit graphs will require to be altered to the same duration (Sect. 7.5).
Once a number of such hydrographs has been obtained for the same
duration, an 'average', or typical unitgraph may be determined as
shown in Fig. 7.16. The ordinates are not averaged since this would
produce an untypical peak. The peak values of the separate unit-
graphs are averaged, as are the values of the time from the
beginning of runoff to the peak. These values are assigned to the
average unitgraph which is then sketched in to a median form on
both rising and falling limbs, so that the total area under the curve is
equal to 1 em runoff.
7.8 Unit hydrographs from complex or multi-period storms

While the approach outlined in Section 7. 7 is simple and direct, it
presupposes that the records contain a number of isolated, uniform-
intensity storms and the corresponding natural hydrographs. Fre-
quently this is not the case and methods are required for deriving
unitgraphs from more complex storms, involving varying intensities
of rain with runoff hydrographs consisting of several superimposed
separate storm hydrographs.
To derive unit graphs from such records is more laborious than
for simple storms but a variety of methods is available, two of
which are discussed below.
The first, described by Linsley, Kohler and Paulhus [45] requires
the writing and successive solving of a series of equations for each
ordinate of the complex hydrograph, baseftow being assumed
previously separated. The process may be illustrated by reference to
Fig. 7.17.
The first rain period, of duration and intei\sity t and ii respectively,
gives rise to runoff illustrated by the hypothetical hydrograph
bounded by the lower dashed line. Each ordinate of this hydrograph
is til times the unitgraph ordinate U1, U2 ..• Un. Similarly, the second
and third rains of intensity i2 and is respectively produce additional
runoff whose ordinates in each case are ti2 and tia, multiples of the
th unitgraph shifted appropriately in time. If the complex hydro-
graph is now defined by ordinates at suitable equal intervals (con-
veniently but not essentially fixed as a whole multiple of t h) then
the first ordinate of the unitgraph, U1, may be obtained from Q1 =
tiJU1 where Q1 IS the observed runoff, hence U1 may be found.
net rain only
Recorded hydrogroph
cv
~
0
.t::.
0
"'
Ci
'' ~
' ·~
.....
~ ' ' ....
., ~
.,
~ .....
..... ~
~
~
FIG. 7.17 Derivation of unit hydrographfrom a

multi-period storm hydrograph
For the second ordinate, Q2 = thU2 + ti2U1 in which equation U2

is the only unknown.
The third ordinate is similarly obtained from Qa = th Ua +
ti2U2 + tiaU1 where Ua is now the only unknown. Proceeding in
this way, the t h unitgraph ordinates may be successively determined.
In the above illustration all the rain periods, although of different
intensities, were assumed of the same duration t h. This is a condi-
tion for the use of the method since otherwise other variables U{,
U~ etc. (the ordinates of the t' h unitgraph) would be introduced.
Although the method &ppears simple, since each U ordinate
depends on the preceding ones, and on the assumptions about
intensity and duration of rainfall and deduction of an assumed
baseftow, errors accumulate, and several trials and restarts may be
necessary to find a reasonable unitgraph.
The second method is due to Collins [46] and is amongst the
simplest of various iterative methods proposed. To illustrate its use
a unit hydrograph will be derived from the data of rainfall and
natural discharge for the catchment of the River Rother at Wood-
house Mill in Yorkshire.
Collin's method for the determination of a unit hydrograph from a
multi-period storm. This method requires the initial selection of a
set of coefficients, or distribution percentages (see Sect. 7.6) of the
unitgraph. This distribution graph is then applied to the various
rain periods, excepting the largest one, and the resulting discharge
subtracted from the actual discharge to obtain a set of 'residuals'.
These residuals should represent the discharge of the unitgraph
applied to the largest rain. If the correspondence is poor, the initial
coefficients are altered and another trial is made. By a series of con-
verging approximations, the residual graph is made to correspond
with the assumed distribution graph.
The procedure is set out below step by step and referred to the
particular case of the River Rother at Woodhouse Mill, for which a
unit hydrograph is derived from the storm of May 14/15, 1967.
Phase A is standard procedure; Phase B is the Collins approach.
Phase A: Assembling and preparing the data:

1. Assemble all rainfall data available for the catchment under
consideration and the storm period, including daily observa-
tions, recording rain gauge records and the synoptic weather
maps of the region, if available.
2. Derive a mean mass curve of rainfall for the catchment, for
the period of rain producing the hydrograph under study.
Make provisional separation of the rain into separate uniform
periods.
The catchment of the River Rother at Woodhouse Mill, York-
shire is shown in Fig. 7.18. The continuous rainfall record (mass
curve) at Sutton-in-Ashfield }or the period of the storm is shown
in Fig. 7.19. The dashed line superimposed on the mass curve
represents the idealised intensities used and plotted in the
appropriate time period of Fig. 7.20. The total rainfall depth is
measured daily at Chesterfield in the centre of the catchment.
Catchment area =
136 miles2
012·345 SUTTON-IN-ASHFIELD
0
·Statute miles
FIG. 7.18 Catchment of the River Rother at Woodhouse Mill
w ....---
1·4 I J I
.,..
.<:
·=
u
1·2
Sultan rain recorder -----/;•
trace 1
v
·= 1·0
.<:
a.
., j/
'0 /I
c 0·8 rainfall rates assumed
·~
.,
.~ 0·6
~ in the analysis
c-;
/
v
E 0·4
"
u
./
0·2
vf""
~
1200 1400 1600 1800 2000 2200 2400 0200 0400 0600 0800
- - - - - 14th May 15th M a y - -
FIG. 7.19 Rainstorm of 14th-15th May, 1967, at Sutton-in-Ashfield

141
The total recorded at Chesteifield on 15th May at 9.00 a.m. was
1·42 in., as against 1·43 in. at Sutton. The precipitation was
frontal, widespread and comparatively uniform and so the
recorded rainfall at Sutton has been assumed catchment-wide.
0·40
·2 0·05 /~lilt rain =0·65in
·I ~ •i index 0·05 in/h
0
2200
t
2000 r~
1800
I ~/ Observed hydrograph
r/
IGO0
~
/Derived distribution graph
~ 1400
~
.5 ·1200
~ 1000
_g
<> 800 1'
r)j Runoff volume=206·5xl0~- ~
--
.!!!
060 0 = 0·65in on 136 mL2
~
400
-·
2oc ~
_l
- -- I
f . - - - ~----
I
Bose flow
I I I l I
0 4 8 12 16 20 24 28 32 36 40 44 48 52 56 60 64 68 72 76 80
Hours
Flo. 7.20 Hydrograph of storm runoff and raingraph for River

Rother at Woodhouse Mill, Yorkshire, 14th-17th May, 1967
3. Plot the hydrograph of total flow and separate baseftow

(Sect. 7.3.).
In the hydrographfor the Rother at Woodhouse Mill (Fig. 7.20)
the position of the basejlow separation line was very indeterminate
and so the end of surface runoff (the point N) was made about
2 days after the peak.
4. Decide upon a unit period. As a general rule this should be
not greater than ! of the time from start of runoff to peak.
Consider the provisional rain graph (from 2 p. 140).
The choice of 4 h in this analysis was suitable for both rain and
runoff.
5. Consider the soil-moisture deficit of the catchment and use
cD index or /av method for estimating storm loss for each rain
period. If antecedent precipitation indices have been kept for
the catchment, use these. Make an estimate of initial and subse-
quent loss rates. Compare the nett rain derived in this way
with the surface runoff expressed as depth on the catchment.
If the two do not agree, one or the other must be adjusted.
The Rother catchment was wet, but not saturated, from light rains
the day previous to the storm. It seemed reasonable to assume
the first period's rain lost entirely in making good the remaining
soil-moisture deficit. The index was subsequently chosen to
balance the nett rain with surface runoff.
Phase B: Using the data to derive the unitgraph:

6. Tabulate the relevant data in columns 1-9 of a table similar to
Table 7.4 and provide columns for the number of unit periods
in the unit hydrograph under 'Distribution coefficients'.
The number of columns in Table 1.4 under' Distribution coeffici-
ents' is 16. This is the number of unit periods from the beginning
of the last rain period to the end of surface runoff.
7. Assume distribution coefficients of the unitgraph (representing
percentage of total runoff per unit period) and arrange in
appropriate columns.
8. Determine the discharge, which, flowing constantly for one
unit period, would just equal I in. nett rain on the catchment.
This figure is found in this case to be:
136 X 27·9 X 1()6 = 21960 ft3js
12 X 4 X 3600
9. The first nett rain is multiplied by this discharge and the product
is distributed in percentages across the distribution coefficient
columns by being multiplied in turn by each percentage
coefficient. The various numbers are entered in the columns
diagonally.
In this case 0·05 X 21960 = 1098, so in the first column 0·05 X
1098 = 55 and so on. Note that 55 was entered opposite the
corresponding rain and not on the top row which is ignored.
10. The procedure of 9, above, is repeated for all nett rains except
the largest for which a dash is entered throughout.
It is purely fortuitous that the largest rain is at the end in this case.
TABLE 7.4. Derivation of unit hydrograph from a multi-period storm
(I) (2) (3) (4) (S) (6) (7) (8) (9) (10) (11) (12) (13)
Nett Aver- Base- Nett Distribution Coefficients '7. Resi~
Day Time Period R~in Losses rain
no. JD. age flow Q
I: duals %
in. 5·0 10·014·015·013·511·0 8·5 6·2 4·0 3·3 2·6 2·1 1·7 1·4 1·0 0·7
0·16 0·20 0 180 145 35
14 1200
May I 0·25 0·20 0·05 290 160 130 55 55 75 (0·8)
2 0·40 0·20 0·20 580 175 405 220 110 330 75 (0·8)
---2400
3 0·60 0·20 0·40 1380 185 1195 - 440 154 594 601 6·9
4 1950 200 1750 165 780 970 11·1
15 5 2030 215 1815 148 808 1007 11·5
- -615 660
May 6 2030 225 1805 - 594 121 715 1090 12·5
.j>. 7 1995 240 1755 - 484 93 577 1178 !3·4
.j>.
- ---2400
8 1820 255 1565 - 374 68 442 1U3 12·8
9 1450 270 1180 - 273 44 317 863 9·8
10 1095 280 815 36 212 603 6·9
16 11 840 300 540 29 174 366 4·2
- -176 145
May 12 720 315 405 24 138 267 3·1
13 640 116 199 2·3
- 114 97 19
325 315 -
14 570 340 230 - 75 15 90 140 1·6
--2400
15 520 350 170 - 62 11 73 97 1·1
17 16 480 365 liS - 44 8 52 63 0·7
May 17 440 380 60 - 31 31 29 0·3
18 420 395 25 - 0 25 0·3
2nd trial coefficients 6·4 10·912·613·5 13·412·5 9·4 6·7 4·1 3·2 2·4 1·8 1·3 1·0 0·6 0·4 100·4
3rd trial 7·2 11·5 12·3 12·812·912·3 9·5 6·7 4·2 3·2 2·5 1·8 1·3 1·0 0·5 0·3
Accepted coefficients 7·0 11·612·512·912·712·09·66·8 4·3 3·2 2·5 1•8 1·3 1·0 0·5 0·3
11. The various discharges are now summed horizontally and
entered in the I: column.
12. The I: column discharge totals are now subtracted from
column 9 and the remainders entered in the 'Residuals' column.
These residuals are then converted into percentages of the unit
distribution graph by dividing by the discharge of 8 (p. 143)
multiplied by the largest rain, and subsequently multiplying by
100. The sum of the percentage column should be 100. The
percentages which cannot have been influenced by the largest
rainfall are bracketed and redistributed over the other coeffici-
ents so that the total, 100 %, remains constant. These per-
centages represent the distribution that would be necessary for
the largest rain to make good the nett Q of column 9. If they
are the same as the assumed distribution coefficients then the
unit distribution graph has been determined.
The residual1090 ofperiod 6 for example is converted thus
1~~l
0·40 960 X 100 = 12·5%
The percentages total comes to 100·4 due to rounding off. The
bracketed figures have not been redistributed since the next trial
requires fairly substantial changes in any case.
13. If the differences between the trial coefficients and the adjusted
(after redistribution) coefficients are large, then a new trial
set must be adopted and steps 9-12 repeated, until the differ-
ences are sufficiently small to be ignored (say < 1 %). A weigh-
ted average of the previous trial and resulting adjusted coeffici-
ents should be used, as follows:
If P = the sum of residuals
Q =the sum of discharges of all the periods during which
the largest rainfall would have been contributing
cl = the trial coefficient
C2 = the calculated and adjusted coefficient
Cs = the proposed new trial coefficient
C _ QCt +PC2
then s- Q+P
Alternatively, the weighting is often simply done in proportion

to the total rain depth, the trial coefficients having the weighting
of rain depths actually used, and the computed coefficients that
of the largest rainfall.
14. It is always wise to plot the distribution graph before deciding on
acceptable coefficients. It may be found that small adjustments
can help to give a smooth curve for the adopted unitgraph.
280 0
~~-
--
260 0
240 0 ~
~-
Distribution graph from the
2200
2000
I /accepted coefficients of Table 7·4
I
1800 V Derived unit hydrograph
"l
....."' 1600 r ,_
".:::
.S 140 c I
I
I
"'~ 120 I
~
.<:: I
(.)
1000 I
Ci"' I
I
800
I
~,
:i
I
600
400
20
0 A ~
~ -......;..-
0 4 8 12 16 20 24 28 32 36 40 44 48 52 56 60 64 68
Hours
FIG. 7.21 4-h unit hydrograph for River Rother at Woodhouse
Mill derived from the multi-period storm of Fig. 7.19
In the illustrated case three trials were made with calculations

and a .final adjustment (without recalculation) made after plotting
the distribution graph and hence deriving the unit graph shown
in Fig. 7.21.
The method is particularly useful when the largest rainfall is
very great compared with the others since rapid convergence
of the coefficients then takes place. This was not the case in
the example illustrated. Too much reliance should not be
placed in unit graphs derived in this way, until they have been
used in practice and/or derived from a series of different
storms, since the loss rates chosen have a critical influence on
the resulting unitgraphs.
7.9 The instantaneous unit hydrograph
An extension of unitgraph theory is the concept of the instantaneous
unit hydrograph or IUH. The IUH is the hydrograph of runoff
from the instantaneous application of unit effective rain on a catch-
ment.
Referring to Fig. 7.13 of Section 7.5, the S-curve was seen to be a
simple method of deriving a unitgraph of period T h from the unit
In the limit (S,-5/-r) --d S,

and T--- dt
,_
FIG. 7.22 The instantaneous unit hydrograph as the S-curve derivative
graph of any other period t, by drawing two t-h S-curves, T-h apart.
This is expressed in the equation
U(T, t) = Tt (St - Se-T) (7.1)
where U(T,t) represents the ordinates of the T-h unitgraph derived

from those of the t-h unitgraph. Now as T, progressively diminish-
ing to zero, ~dt, the right hand side of Eq. (7.1) ~the S-curve
derivation, as may be seen graphically in Fig. 7.22. In equation
form this is
U(O t) = d(St) (7.2)
' dt
. the ord"mate of t he IUH at any time
1.e. . . g1ven
t 1s . by dt
dSt at time
. t.
The IUH is a unique demonstration of a particular catchment's
response to rain, independent of duration, just as the unitgraph is its
response to rain of a particular duration. Since it is not time-
dependent the IUH is thus a graphical expression of the integration
of all the catchment parameters of length, shape, slope condition etc.
which control such a response.
The conversion of an IUH to a unitgraph of finite period is simple.
The ordinate of an n-h unitgraph at time t is the average ordinate of
the IUH for n h before t. From Fig. 7.23, it may be seen that the
Instantaneous unitgroph
points
FIG. 7.23 The n-h unitgraph derived from the IUH

/UH is divided into n h time intervals, and the averages of the
ordinates at the beginning and end of each interval are plotted at
the end of the intervals to provide the n-h unitgraph.
The IUH may be used to derive unitgraphs by flood-routing as
explained in Section 8.6.
7.10 Synthetic unit hydrographs

In preceding sections it has always been assumed that some records
have been available for the derivation of the unitgraph, but there are
many catchments for which there are no runoff records at all and
for which unitgraphs may be required. In these circumstances
hydrographs may be synthesised on the basis of past experience in
other areas and applied as first approximations to the unrecorded
catchment. Such devices are called synthetic unitgraphs.
The best known approach is due to Snyder [47] who selected the
three parameters of hydrograph base width, peak discharge and basin
lag as being sufficient to define the unit hydrograph. These are shown
in Fig. 7.24.
Snyder considered the catchment characteristics likely to affect
unit hydrograph shape as being catchment area, shape of basin,
topography, channel slopes, stream density and channel storage.
He eliminated all these parameters except the first two by including
net storm rain
1. ~ :
.,
basin lag (h)
synthetic unilgraph
T = runoff duration (days)
lime
Flo. 7.24 Synthetic unitgraph parameters
them in a coefficient Ce. He dealt with the size and shape of catch-
ment by measuring the length of the main stream channel and he
proposed that
tp = Ce(Lca£)0 · 3 (7.3)
where = basin lag in h, i.e. the time between mass centre of unit
tp
rain of tr h duration and runoff peak flow.
Lea = distance from gauging station to centroid of catchment
area, measured along the main stream channel to the
nearest point, in miles.
L = distance from station to catchment boundary measured
along the main stream channel, in miles.
Ct = a coefficient depending on units and drainage basin
characteristics and varying between 1·8 and 2·2 for the
Appalachian Highlands catchments studied.
The equation for peak flow (per square mile of area) of the t,
unitgraph was given by
640
qp= Cp.- (7.4)
tp
where qp is in cubic feet per second.
Cp = a coefficient depending on units and basin characteristics
and varying between 0·56-0·69 for the Appalachian
catchments and generally approaching its largest value
as Ce approaches its lowest and vice versa,
and therefore the peak flow for the unitgraph
640A
Qp= Cp.--
tp
where A = catchment area in square miles.
The duration of surface runoff, or unit hydrograph baselength,
T was given by Snyder by the empirical expression
T= 3 + 3 G~) (7.5)
where Tis in days and tp in hours This expression gives a minimum

baselength of 3 days for even small areas, a period much in excess
of delay attributable to channel storage.
Snyder comments on this as being due to the "subsurface storm
flow" which has been defined by Hursh [48] as "that portion of the
storm flow which infiltrates into the surface soil but moves away from
the area through the upper soil-horizons at a rate much in excess
of normal groundwater seepage". This is what is referred to in
recent literature as interjlow and for most practical purposes it is
regarded as surface runoff.
t,, the unit rain period, was assumed to equal tpf5·5 in the study,
since it was necessary to choose a single standard in all the catch-
ment for the derivation of the formulae. This particular value was
selected to make the unit of time equal to a minimum value below
which further decrease would have little or no effect on basin lag or
unitgraph peak discharge. If the actual length of the storm is not
equal to tr, but is IR, the Eq. (7.4) becomes
640
qpR = Cp. tp + (tR- tr)/4 (7.6)
where qpR =peak discharge (per square mile) of the IR unitgraph

which allows for the generally observed reduction in unitgraph peak
flows with longer periods of rain. Snyder proposed subsequently [49]
an expression to allow for some variation in basin lag with variation
in effective rainfall duration
lpR = lp + (tR - lr)/4 (7.7)
where tpR =basin lag for a storm of duration IR.
Linsley subsequently presented data [50] based on a study of
Californian catchments and suggested modifications of Snyder's
formulae and gave values of the various coefficients, as follows:
Basin lag: a new factor tpo was introduced-the basin lag of an
instantaneous storm-and used to derive lpR which has the same
meaning as before.
tpo = Ce(Lca£) 0 •3 with Ct (average) = 0·5
fpR = lpo + (Cs - 0·5)tR (7.8)
where Cs (average)= 0·85.
Unitgraph peak flow (per square mile): qpR = Cp. 640ftpR (1.9)
where Cp varies 0·35-0·50, and QpR = qpRA
Time base of unitgraph: T = 3 + (3tpRf24) days
The degree of divergence of the coefficients is an indication of how
important it is to attempt to obtain some actual data on the lag in an
ungauged catchment, and thus enable the value of Ct to be used with
reasonable assurance.
A further advance in the subject was made by Taylor and Schwarz
(51] in a study based on 20 drainage basins from 20-1600 square
miles in area, in north and central Atlantic states of the U.S.A.
They used the parameters L, Lea as before and also introduced the
slope of the main watercourse by defining S8t as the slope of a uni-
form channel having the same length as the longest watercourse and
an equal time of travel. The equations derived are as follows:
Basin lag
lpR = C'em'ta (7.11)
where tpR = lag in h from centroid of nett rain to hydrograph peak
t R = time in h from beginning to end of nett rain
e = 2·7183
m' = rate of change of lag with storm duration
C' = lag of instantaneous unit hydrograph
m' and C' are derived from the following equations
m' = 0·212/(LLca,)0·3& (7.12)
C' = 0·6/v(S,t) (7.13)
where L and Lea have the same definitions as previously and
•.. + Sat = [(1/St + 1~! + 1/S!)T (7.14)

where n = Manning's coefficient of roughness for the natural
watercourse
S1, S2 etc. = the slopes of individual sections, of equal length, into
which the main watercourse may be conveniently divided.
Peak discharge (per square mile) of unitgraph

qpR = C"em"tR (7.15)
where
C" = 382(LLca)-o·3& (7.16)
m" = 0·121S8 to·14 2 - 0·05 (7.17)
Base width of unitgraph

T = 5(tpR + tR/2) h (7.18)
The authors give a nomogram in their paper for the solution of
equations and make a number of observations about the use of their
method. These include the suggestions that major tributaries should
be treated separately, and that the equations should be limited to the
results of moderate and major storms of uniform distribution over
geographical areas similar to those from which they were derived.
The equations are given here as they form a useful addition to the
literature on synthesis of unitgraphs.
Unit hydrographs may also be synthesised by the methods of
streamflow routing and Section 8.6 describes such a technique. It
has been placed in Chapter 8 because a knowledge of routing pro-
cedure is essential to its understanding.
The coefficient n in either form of Manning's equation (British units

or SI) has the same numerical value. Some typical values are listed
below.
Type of channel n
Rough-paved trapezoidal channel or canal without 0·021
obstructions or sharp curves
Natural stream channel, smoothly flowing in very clean 0·030
condition
Standard natural stream or river in stable condition 0·035
Rivers with shallows and meanders and noticeable 0·045-Q·OSO
aquatic growth. Streams with gravel banks and
meanders
Slow flowing meandering rivers with pools and slight 0·060-Q·lOO
rapids and overgrown banks
8 Flood Routing
8.1 Introduction
Civilisation has always developed along rivers, whose presence
guaranteed access to and from the sea coast, irrigation for crops,
water supplies for urban communities and latterly power develop-
ment and industrial water supply. The many advantages have always
been counterbalanced by the dangers of floods and in the past levees
or flood banks were built along many major rivers to prevent inunda-
tion in the flood season. In more recent times storage reservoirs have
been built as the principles of dam construction became better under-
stood and other measures like relief channels, storage basins and
channel improvements are continually under construction in many
parts of the world. It is important for such works that estimates can
be made of how the measures proposed will affect the behaviour of
flood waves in rivers so that economic solutions may be found in
particular cases. Flood routing is the description applied to this
process. It is a procedure through which the variation of discharge
with time at a point on a stream channel, may be determined by con-
sideration of similar data for a point upstream. In other words it is a
process which shows how a flood wave may be reduced in magnitude
and lengthened in time (attenuated) by the use of storage in the reach
between the two points.
8.2 The storage equation

Since the methods of flood routing depend on a knowledge of storage
in the reach, a way of evaluating this must be found. There are two
154
Flood Routing 155
ways of doing this. One is to make a detailed topographical and
hydrographical survey of the river reach and the riparian land and
so determine the storage capacity of the channel at different levels.
The other is to use the records of past levels of flood waves at the
two points and hence deduce the reach's storage capacity. It is
assumed that such storage capacity will not change substantially in
time and so may be used to route the passage of larger and more
critical, predicted floods. As much data as possible is required for
the second method, which is the one generally used, including flow
wedge storage
A 8
inflow prism storage outflow
stream bed
reach
FIG. 8.1 Storage in a river reach
records at the beginning and end of the reach and on any tributary
streams joining it, and rainfall records over any areas contributing
direct runoff to it.
Storage in the reach of a river is divided into two parts, prism and
wedge storage, which are illustrated in Fig. 8.1. This is simply
because during floods the slope of the surface is not uniform.
(See Section 6.2).
If the continuity of flow through the reach shown in Fig. 8.1 is
now considered, it is clear that what enters the reach at point A
must emerge at point B, or temporarily move into storage.
dS
I= D + dt
where I = inflow to the reach
D = discharge from the reach
dt = rate of ch ange m
dS . reach storage wtt
"h respect to time.
.
This equation is approximated to, for a time interval t, by

h+ht_DI+D2t=S2- Sl
2 2 (8.1)
where subscripts l and 2 denote values at the beginning and end of
the time.t, respectively. The timet is called the routing period and it
must be chosen sufficiently short, so that the assumption implicit in
Eq. (8.1), i.e. that the inflow and outflow hydrographs consist of a
series of straight lines, does not depart too far from actuality. In
particular, if t is too long it is possible to miss the peak of the inflow
curve, so the period should be kept shorter than the travel time of
the flood wave crest through the reach. On the other hand, the
shorter the routing period the greater the amount of computation to
be done.
8.3 Reservoir routing

If Eq. (8.1) is now arranged so that all known terms are on one side,
the expression becomes
!(h + l2)t + (S1 - !D1t) = (S2 + !D2t) (8.2)
The routing process consists of inserting the known values to obtain
S2 + ! D2t and then deducing the corresponding value of D2 from the
relationship connecting storage and discharge. This method was
first developed by L. G. Puis of the U.S. Army Corps of Engineers.
The simplest case is that of a reservoir receiving inflow at one end
and discharging through a spillway at the other. In such a reservoir
it is assumed there is no wedge storage and that the discharge is a
function of the surface elevation, provided that the spillway arrange-
ments are either free-overflow or gated with fixed gate openings.
Reservoirs with sluices may be treated as simple reservoirs also if
the sluices are opened to defined openings at specified surface-water
levels, so that an elevation-discharge curve may be drawn. The
other required data are the elevation-storage curve of the reservoir
and the inflow hydrograph.
Example 8.1. An impounding reservoir enclosed by a dam has a

surface area which varies with elevation as shown by the relationship of
Fig. 8.2(a). The dam is equipped with 2 circular gated discharge ports,
each of2·1 m diameter, whose centres are at elevation 54·0, and a free
overflow spillway 72·5 m long with crest level at elevation 66·0. The
discharge gates are open and surface water level is at elevation 63·5, at
time t = 0. The flood hydrograph of column 3, Table 8.2, is forecast.
What will the maximum reservoir level be and when will it occur?
Flood Routing 157
1. Assume the ports have a coefficient of discharge, Ca = 0·8, then
Q = 2(CaA'\1'(2gH)) and at time 0, Q = D (column 5 of Table
8.2) = 2(0·8 x 5·7 x y'186·5) = 125 m3/s. (Note g = 9·81
mfs 2 .) Insert this value on first line of column 5.
2. Compute the elevation-storage curve of Fig. 8.2(b). Remember
that live storage starts at 52·65, the invert level of the discharge
ports, and amounts to 5·5 x 106m3 by 54·0. The storage be-
tween 54·0 and 56·0 = mean area of reservoir between these
levels x 2m = 8·33 x 106 m3. Successive increments, com-
puted in this way, are plotted cumulatively with 52·65 as datum.
3. Compute the elevation-D tabulation of Table 8.1 below.
TABLE 8.1. Elevation-discharge table
Head Head
Elevation of over Discharge from over Spillway Total
water surface 54·0 gated ports 66·0 = H Ht discharge discharge
m m m 3 /s m m 3 /s m 3 /s
58·0 4·0 81·0 81
60·0 6·0 99·5 100
62·0 8·0 Il4·8 II5
64.0 10·0 128·0 128
66·0 12·0 140·7 0 0 0 141
66·1 12·1 141·2 0·1 0·032 5·1 148
66·2 12-2 141·8 0·2 0·089 14·2 156
66.3 12·3 142·6 0·3 0·164 26 169
66·4 12·4 143·0 0·4 0·252 40 183
66·5 12·5 143·7 0·5 0·353 56 200
66·7 12·7 144·8 0·7 0·58 93 238
66·9 12·9 146·0 0·9 0·85 136 282
67·0 13·0 146·4 1·0 1·0 160 306
67·5 13·5 149·3 1·5 1·84 294 443
68·0 14·0 152·0 2·0 2-83 453 605
Assume Q spillway= CLHf and use C = 2·2 mijs.
4. From Fig. 8.2(b) and Table 8.1, the D-Storage curve of Fig. 8.3
may now be drawn, i.e. the central curve. The abscissa of Fig.
8.3 is graduated in "storage units". Each storage unit =
routing period x 1 m3js. Since the forecast hydrograph of
column 3, Table 8.2, is given at 6 hour intervals it is convenient
to make this the routing period. Then each storage unit =
6 x 3600 x 1 = 21·6 x 103 m3 = t m3fs day. The use of
v
/
72
/
70
v v
.5 6866 / v
~ / '/
.!!! 64
.,u 62 / v
v
0
1-- -
-; 60 (a} w.l. v. area
/
L{b}w.l. v. storage -
en 58
56 v /
54 I
4·0 4·2 4-4 4·6 48 50 5·2 0 20 40 60 80 100
Reservoir surface area: km 2 Live storage: m3 x 106
Flo. 8.2 Reservoir characteristics
these units is necessary to keep the dimensions of columns 4, 6

and 7 in Table 8.2 compatible.
The S ± ! Dt curves are now added on either side of the
storage curve. Since the abscissa is in storage units, then t = 1
and the curves may be plotted without calculation, e.g. at
D = 200, !Dt = 100 and so two points may be set off 100
storage units on either side of the G curve, and similarly for
other points.
5. The routing calculation can now be started in Table 8.2. The
figures in heavy type are known.
"i
700
600
0 Q"
500 -1" C/) ' ·j
enI -!'!
V)
400 ,....__. g)
Ui300 r-ID _g "'c: ~I
z u
"''
.5 200 r--~~ 1/
, ,
Cl 150 "' I '/
"'E!' 130 ~ ~
0
£
u
¢P
c5"' 110
~
90
/
/"
70
50
500 1000 1500 2000 2500 3000 3500 4000
Storage units (I unit =21·6 x 106 m3=J m3/s day}
Flo. 8.3 Reservoir routing storage curves

Flood Routing 159
TABLE 8.2 Reservoir routing computation
1 2 3 4 5 6 7 8
S-tDt S+tDt
Time Routing Inflow h+ht D im3js tm3/s Surface
h period m3/s 2 m3fs day day level m
0 1 50 62 125 2095 2157 63·4
6 2 75 127 122 2035 2162 63·0
12 3 180 265 122 2040 2305 63·0
18 4 350 400 127 2178 2578 63·8
24 5 450 485 136 2442 2927 65·1
30 6 520 512 200 2727 3239 66·5
36 7 505 475 425 2814 3289 67.4
42 8 445 402 460 2829 3231 67·5
48 9 360 325 416 2815 3140 67·35
54 10 290 270 347 2793 3063 67·15
60 11 250 230 288 2775 3005 66·95
66 12 210 192 242 2763 2955 66·7
72 13 175 157 208 2747 2904 66·55
78 14 140 125 190 2714 2839 66·45
84 15 110 97 165 2674 2771 66·25
90 16 85 75 144 2627 2702 66·05
96 17 65 60 140 2562 2622 66·0
102 18 55 52 138 2484 2536 65·3
108 19 50 47 134 2402 2449 64·7
114 20 45 42 132 2317 2359 64·3
120 21 40 39 129 2230 2269 64.0
126 38 127 63·7
To begin with, compute column 4 by averaging successive pairs

of inflow values. Now find from the S - 1;Dt curve of Fig. 8.3,
the value of this parameter corresponding to D = 125 m3js.
The value is 2095 and this figure is inserted in the first space of
column 6.
6. The figure in column 4 is added to column 6 and the total put in
column 7 (e.g. 2095 + 62 = 2157). The left side of Eq. (8.2)
has now been evaluated. Find the column 7 value on the
S + 1;Dt durve and read off the corresponding value of D,
entering it in column 5 (e.g. from the S + 1;Dt curve find
D = 122 corresponding to 2157).
7. Use this new value of D to find S - ! Dt again as in step 5. Note
that this is directly obtainable without using the curve by sub-
tracting the value of D from the column 7 value in the line
above (e.g. 2157 - 122 = 2035). The column 4 figure is then
added to the new column 6 to get a new column 7 figure
(e.g. 127 + 2035 = 2162).
n
Reservoir
I
600
I I I I I
j_ ~Inflow hydrograph
500 I
400 I
I I \~ / Discharge hydrograph
1/ I '\
;)I.JV
~
v
-----
200
I ~~
100 ...........
/
0 12 24 36 48 60 72 84 96 108 120 132
Hours
I 2 3 4 5 6 7 8 9 10 II 12 13 14 15 16 17 18 19 20 21 22
6 hour routing periods
FIG. 8.4. Inflow and discharge hydrographs for the reservoir

of Example 8.1
Complete the table and plot the outflow hydrograph (Fig. 8.4).
The peak outflow should fall on the recession limb of the inflow
graph.
The time difference between the peaks of the inflow and dis-
charge hydrographs is termed reservoir lag and the reduction
in peak flows together with the spreading out of the recession
curve is referred to as attenuation.
8. The column 8 values of surface water level are derived from
the values of discharge and levels in Table 8.1. They are most
conveniently found by plotting a graph and reading off the
levels corresponding to the values of column 5.
The maximum water level in the case of this example is
67·5 m occurring about hour 40.
Flood Routing 161
8.4 Routing in a river channel
The solution of the storage equation in this case is more complicated
than for the simple reservoir, since wedge storage is involved. Stor-
age is no longer only a function of discharge as was the case in
Example 8. I. McCarthy [52], in what has become known as the
lime-
FIG. 8.5
Muskingum method, proposed that storage should be expressed as

a function of both inflow and discharge in the form
S = K[xl + (1 - x)D] (8.3)
where x = dimensionless constant for a certain river reach

K = storage constant with dimension of time, that must be
found from observed hydrographs of I and D at both
stations.
The two constants may be found as follows. Let Fig. 8.5 represent
the simultaneous inflow I and outflow D, of a river reach. While
I> D, water is entering storage in the reach and when D > I, water
is leaving it. A difference diagram can now be drawn showing this
(Fig. 8.6) and subsequently, a mass curve of storage. (Fig. 8.7).
Now assume a value of x, say x = 0·1 and compute the value of
the expression (O·ll + 0·9 D) for various times and plot these against
corresponding S values taken from Fig. 8.7. The resulting plot,
known as a storage loop is shown in Fig. 8.8(a); clearly there is no
linear relationship. Take further values of x (say 0·2, 0·3 etc.) until
a linear relationship is established, as in Fig. 8.8(c) when the particu-
lar value of x may be adopted. K is now obtained by measuring the
slope of the line.
water enterinQ storoQe

., +
;;;-
E
Flo. 8.6
Care about units is required. It is often helpful to work in some-

what unusual units, both to save computation and to keep numbers
small. For example, storage Sis conveniently expressed in ms/s day:
such a unit is that quantity obtained from 1 mSfs flowing for 1 day =
86·4 x 10s ms. If Sis expressed in mSfs day and the ordinate of
Fig. 8.8 is in mSfs then K is in days.
-time
Fxo. 8.7
The following excerpt from Carter and Godfrey [53] concisely sums
up the choice of values for x and K:
The factor x is chosen so that the indicated storage volume is the
same whether the stage is rising or falling. For spillway discharges from
Flood Routing 163
a reservoir, x may be shown to be zero, because the reservoir stage, and
hence the storage, are uniquely defined by the outflow; hence, the rate
of inflow has a negligible influence on the storage in the reservoir at any
time. For uniformly progressive flow, x equals 0·50, and both the inflow
and the outflow are equal in weight. In this wave no change in shape
occurs and the peak discharge remains unaffected. Thus, the value of x
will range from 0 to 0· 50 with a value of 0·25 as average for river reaches.
The factor K has the dimension of time and is the slope of the storage-
weighted discharge relation, which in most flood problems approaches a
straight line. Analysis of many flood waves indicates that the time re-
quired for the centre of mass of the flooq wave to pass from the upstream
end of the reach to the downstream end is equal to the factor K. The
time between peaks only approximates the factor K. Ordinarily, the value
of K can be determined with much greater ease and certainty than
that of x.
,..."")
/ I
/ I
/ /
I /
I /
I /
1/
.:: 1/
+
..... /(
"' /I m
/ I
/ I Cbl (c) K= -'f- days
--------{a)
--------{b)
----------~c)
S: storage (m 3 /s doys)
FIG. 8.8 River routing storage loops
Having obtained values of K and x, the outflow D from the reach

may be obtained, since by combining and simplifying the two
equations
(8.1)
and
S2 - S1 = K(x(h - h) + (1 - x) (D2 - D1)] (8.4)
(the latter being Eq. {8.3) for a discrete time interval) the equation,
D! = Col2 + C1lt + C2D1 (8.5)
is obtained, where
Co=-
Kx- O·St
, Ct = Kx + O·St '
K- Kx+ O·St K- Kx +O·St
K-Kx- O·St
c2 = (8.6)
K- Kx + O·St
where t = routing period, which should be taken as between i and l
of the flood wave travel time through the reach (obtained
from the inflow hydrograph)
0 6 12 18 24 30 36 42 48 54 60 66 72 78 84 90 96 102 108 114 120 126

hours
FIG. 8.9 Inflow and discharge hydrographs for a river reach
A worked example illustrating the application of the method is

set out below.
Example 8.2. Routing in a stream channel by the Muskingum method.

Part I
Given the inflow and outflow hydrographs of Fig. 8.9, derive the
constants x and K for the reach. ·
Flood Routing 165
The data is set out in tabular form in Table 8.3. In columns 1 and
2, the given hydrographs are listed at a routing period interval,
taken as 6 h. The storage units are taken here as Ct m3 /s day) since
the routing period is t day. The columns 4, 5 and 6 are simply
tabular statements of the processes illustrated in Figs. 8.6 and 8.7.
A value of xis then chosen, in the first instance 0·2, and the value
inside the square brackets of Eq. (8.3) is then evaluated in columns
7, 8 and 9. Columns 6 and 9 are now plotted in Fig. 8.10 and produce
the loop on the left side of the figure. (Note, incidentally, that this
3 00
280
2 I l J
240
/1 jJJ J
1/ I
.,220
r-
.g'200 iff I
v
1240-! m3 /s days
I1/
.. I
;;-
~
=60 m3/s days
E I60
-i~ I40
8.1 20
e x=0·2 !J x=0·25/ x=O·~~
Vj
.21 00
VI
80
lf If__ I
A' fl-4o m•t;-1 [//V
60
40
20
v
o~~~~~romoo
/ .".K=~=I·5days
v
o~~~~~rorooo
o 10 20 30 40 50 60 ro ao
xi+( 1-x) D: m3 /s
FIG. 8.10 Storage loops for the reach of Example 8.2
figure has ordinate and abscissa reversed compared with Fig. 8.8.
This is quite unimportant and is done only for convenience of fit.)
A second value of x = 0·25 is now tried (columns 10-12 refer)
and the resulting plot is the central one of Fig. 8.10. A third value,
x = 0·3, is also tabulated and plotted on the right of the figure. By
inspection, the central value of x = 0·25 approximates a straight
line most nearly, so this is chosen as the x value.
K is determined by measuring the slope of the median line as
shown on the figure, and is found to be 1·5 days. This confirms the
TABLE 8.3. StorQKe loops calculatiom
1 2 3 4 s 6 7 8 9 10 11 12 13 14 IS
I D Mean Cumulative
Inflow Outflow 1-D storage storQKe x=0·2 X= 0·25 x=0·3
!Hour m3js m3js m 3/s (i m 3/sday) (i m 3/s day) 0·21 0·8D Total 0·251 0·1SD Total 0·31 0·1D Total
0 22 22 0 0 0 4 17 21 s 16 21 7 15 22
6 23 21 2 1 I s 17 22 6 16 22 7 15 22
12 35 21 14 8 9 7 17 24 9 16 25 10 15 25
18 71 26 45 29 38 14 21 35 18 19 37 21 18 39
24 103 34 69 57 95 20 27 47 26 25 51 31 24 ss
30 111 44 67 68 163 22 35 57 28 33 61 33 31 64
36 109 ss 54 60 223 22 44 66 27 41 68 33 38 68
...... 42 100 66 34 44 267 20 53 73 25 49 74 30 46 76
48 86 75 11 22 289 17 60 77 21 56 77 26 52 78
"'"' S4 71 82 -11 0 289 14 66 80 18 61 79 21 57 78
60 59 85 -26 - 18 271 12 68 80 15 64 79 18 59 77
66 47 84 -37 -31 240 9 67 76 12 63 75 14 59 73
72 39 80 -41 -39 201 8 64 72 10 60 70 11 56 67
78 32 73 -41 -41 160 6 58 64 8 ss 63 10 51 61
84 28 64 -36 -38 122 6 51 57 7 48 ss 8 45 53
90 24 54 -30 -33 89 5 43 48 6 40 46 7 38 45
96 22 44 -22 -26 63 4 35 39 s 33 38 7 31 38
102 21 36 - 15 -18 45 4 29 33 5 27 32 6 25 31
108 20 30 -10 -12 33 4 24 28 s 22 27 6 21 27
114 19 25 - 6 - 8 25 4 20 24 s 19 24 6 17 23
120 19 22 - 3 - 4 21 4 18 22 s 16 21 6 IS 21
126 18 19 - I - 2 19 4 15 19 4 14 18 s 13 18
--- ----- ----
Flood Routing 167
approximate peak to peak time of 33 h from Fig. 8.9. For this
reach, therefore, use
x = 0·25 and K = 1·5 days
Part 2
Using the x and K values derived from the hydrographs, route the flood
of column 2, Table 8.4, through the reach and derive the outflow hydro-
graph.
TABLE 8.4. Derivation of discharge
1 2 3 4 5 6
I - 0·2/s 0·41I 0·8 D1 Da
Hours m3/s m3js m3/s m3/s m3/s
0 31 31·0*
6 so -10·0 12·4 24·8 27·2
12 86 - 17·2 20·0 21·8 24·6
18 123 - 24·6 34·4 19·7 29·5
24 145 -29·0 49·2 23·6 53·8
30 150 - 30·0 58·0 43·0 71·0
36 144 - 28·8 60·0 56·8 88·0
42 128 - 25·6 57·6 70·4 102·4
48 113 -22·6 51·2 81·9 110·5
54 95 - 19·0 45·2 88·4 114·6
60 79 - 15·8 38·0 91·7 113·9
66 65 -13·0 31·6 90·8 109·4
72 55 - 11·0 26·0 87·5 102·5
78 46 - 9·2 22·0 82·0 94·8
84 40 - 8·0 18·4 75·8 86·2
90 35 -7·0 16·0 69·0 78·0
96 31 - 6·2 14·0 62·4 70·2
102 27 - 5·4 12·4 56·2 63·2
108 25 - 5·0 10·8 50·6 56·4
114 24 -4·8 10·0 45·1 50·3
120 23 -4·6 9·6 40·2 45·2
!26 22 - 4·4 9·2 36·2 41·0
• assumed value
First, compute Co, C1 and C2 from Eq. (8.6). Use a routing period
t = 6 h = t day as before.
(1·5 X 0·25) - (0·5 X 0·25) 0·25
Co= - = - - = -0·2
1·5 - (1·5 X 0·25) +
(0·5 X 0·25) 1·25
Similarly calculated, C1 = 0·4 and C2 == 0·8, check that -0·2 +
0·4 + 0·8 = 1·0 from Eq. (8.5)
~ = -0·2/2 + 0·4/t + 0·8D1
h, /2 etc. are known from the hypothetical flood hydrograph, but
D 1 is unknown. Assume a value for D 1 = /1 = 31 m3fs. This
will be very nearly correct since the river is at a low level and will be
near to a steady state. Then the first equation to be solved is
D2 = -0·2(50) + 0·4(31) + 0·8(31)

= -10·0 + 12·4 + 24·8 = 27·2
This value of D2 becomes the D1 for the next calculation and the
values are tabulated as in Table 8.4.
The outflow hydrograph is plotted as a dashed line to a little way
beyond the peak in Fig. 8.9.
8.5 Graphical routing methods

If Eq. (8.3) is written with x = 0, then
S=KD (8.7)
. d'fti . . dS dD
and smce, 1 erenttatmg, dt = K dt
and
dS
I - D = dt (from Sect. 8.2)
1- D dD
J(=di (8.8)
This equation has been used [54] to provide a simple graphical

method of routing, since ~~ represents the slope of the outflow
hydrograph and I - D and K are measurable quantities in m3/s
and days. In Fig. 8.11, which is a plotted inflow hydrograph /,
with discrete values It, h Ia etc. marked at intervals of time, the
storage constant K is plotted horizontally from the position of each
Flood Routing 169
I value and a line drawn from the end of the Kline to the previous
discharge value D. Since this line represents dD' the lower part
dt
may be used to denote the actual outflow hydrograph. Naturally,
the smaller the time interval the more accurate the method will be,
but there is no need to have the intervals equal.
K may be varied, if its variation is known, and reference to Fig.
8.10 will suggest K tr.ay well vary and could be obtained in a rela-
tionship with outflow from such a storage loop, giving a K v. D
curve as illustrated in Fig. 8.13(b).
The method can also be used in reverse, so that K at any time
may be obtained from simultaneous hydrographs of I and D.
t--
FIG. 8.11 Graphical routing method
The foregoing description is all qualified by the initial statement

that x = 0, so it applies only to simple reservoir action. It may be
extended, however, to include positive x values since the effect of
increasing x, with K constant, is to move the outflow graph bodily
to the right so that the peak value no longer falls on the recession
limb of the inflow graph, and also to increase the magnitude of the
outflow peak.
If a succession of historic floods is analysed, the lag that is
caused by x having a positive value may be determined. The lag
due to this cause, T:r:, is measured from the peak of the outflow graph
to the same discharge on the recession limb of the inflow graph, as
illustrated in Fig. 8.12 and a plot may be made connecting T:r: with
corresponding I (Fig. 8.13(b)).
,_
FIG. 8.12 Lag due to the constant x > 0
,,, ~Inflow lagged in accordance

/ \ with T; 'l I
curve
\
\
\ .
\ .~Outflow denved from
·"\ ., logged inflow using
\ ' · variable K
\ '
''
"" .
' ' ', " "·

.
', "
.
...........
.......... ,
(a) (b)
,_ T, and K
FIG. 8.13 Graphical routing with variable lag and K
Now the inflow graph of a reach with x > 0 is lagged as illus-

trated in Fig. 8.13, the amount of lag at each horizon being deter-
mined from the T x:-1 curve, to give the dashed inflow graph, which
Flood Routing 171
is then routed by the graphical method as though it were a simple
reservoir inflow. It is often convenient to plot the curve of variation
of K v. D alongside the T:rl curve so that both variations may be
taken account of in the same plot. Fuller descriptions of this and
similar routing techniques are available [31, 55, 56].
8.6 Synthetic unitgraphs from flood-routing

The principles of flood-routing may now be used to derive unit
hydrographs for a catchment where almost no rainfall or runoff
short rainburst
storage
water
leaving
storage
I ----
r----"'"----- - - -
base flow
time -------
FIG. 8.14 Hydrographfrom short rain approximates /UH
records exist. The method is not entirely synthetic since at least one
observation of a runoff hydrograph must be made.
Consider a catchment as a series of sub areas, each of which,
under a sudden burst of rain, contributes inflow into the system of
drainage channels, which have storage. The instantaneous unit
hydrograph is therefore in two parts, the first representing the inflow
of the rain, and the second the gradual withdrawal of the catchment
storage. The dividing line between these two parts may be conveni-
ently taken at the point of inflexion on the recession limb as shown
in Fig. 8.14.
The assumption is now made that the catchment discharge Q
and the storage S are directly proportional, so that
S = KQ (8.9)
(i.e. Eq. (8.3) with x = 0, and Q used instead of D)
dS
and I- Q = dt (8.10)
where I represents the inflow resulting from the instantaneous rain.

Since:~= K~; by differentiating Eq. (8.9), then
KdQ =I- Q
dt
and using the condition Q = 0 when t = 0, the equation may be
solved to
(8.11)
Since the inflow ceases at the inflexion point at time T(say), then the
outflow at time t in terms of discharge QT at Tis given as
Qe = QTe-<t-Tl/K (8.12)
The determination of the storage coefficient K must be made
from an observed hydrograph on the catchment, as illustrated in
Fig. 8.15, by taking two values unit time apart at the point of inflex-
ion. The hydrograph should be of an isolated period of rain. It is
not necessary that the magnitude of the rain be measured but it is
necessary that it should be reasonably short, say of 1 or 2 hr only.
Then, Q1 = QT and from Eq. (8.12) Q2 = QTe-<t-TliK
rt=T+l
the shaded area A= Jt=T QTe-<t-T)fK
= [ -KQTe-<t-T)/K J:
= [KQT - QTe-lfK]
Flood Routing 173
The second observation that must be made from the observed
hydrograph is the catchment lag (TL), i.e. the maximum travel time
through the catchment. This may be taken as the time from the mass
centre of the causative rain (hence the requirement that it should be
short-so that no large error is introduced here) to the inflexion
point on the recession limb.
The storage of the catchment is now thought of as a hypothetical
reservoir, situated at the point of outflow; the inflow is expressed
inflexion point
l
Q
basel low
r---------------------1----
Flo. 8.15 Determination of K
as the time-area graph of the catchment, where each sub-area is

delineated so that all rain falling on it instantaneously has the same
time of travel to the outflow point, as illustrated in Fig. 8.16. The
time-area graph (/) now has instantaneous unit rain applied to it
and is routed through the reservoir, in the manner of Section 8.3,
and the outflow (Q) derived. This outflow represents the IUH for
the catchment and may be converted if required to the n-h unit graph.
The method is basically due to Clark [57], though K is as derived
by O'Kelly [58]. It is open to criticism in several respects and more
advanced techniques [59, 60, 61, 62] are now available, but is has the
advantage of comparative simplicity. Its derivation is not dependent
on an observed hydrograph of unit intensity.
Another advantage is that instead of deriving the IUH (and hence
the n-h unitgraph) design rain may be applied directly to the
time-area graph, with areal variation and in any desired quantity.
This produces an instantaneous design-storm hydrograph which
can then be directly converted to a design-storm hydrograph of
required intensity by averaging ordinates as discussed before.
=Sh
Q~ K =7·5h
catchment areo = 250km2
FIG. 8.16 Catchment with isochrones
A worked example of the method is given below.
Example 8.3. Given the catchment area of Fig. 8.16, of area 250 km2,
and the information derived from a short rain hydrograph that TL =
8 h and K = 7.5 h, derive the 2 h unit hydrograph
1. Divide the catchment area into eight hourly divisions by

isochrones, or lines of equal travel time. It will be assumed that
all surface runoff falling in one of these divisions will arrive
during a 1 h period at the gauging point.
2. Measure by planimeter the area of each of the hourly areas.
The areas of the figure are :
Hour 1 2 3 4 5 6 7 8
Area in km2 10 23 39 43 42 40 35 18
Flood Routing 175
3. Draw the distribution graph of the runoff using the sub-areas
as ordinates and 1 h intervals as abscissa. The result is the
Fig. 8.17-the time-area graph drawn in full lines.
4. This time-area graph is now treated as the inflow I due to unit
nett rain of 1 em on the catchment of a hypothetical reservoir,
50 r-
.--1--- . Dotted I i nes show the

40- - distortion of graph to
account for non- uniform
oreal distribution of rain.
301-
-
201-· ....
101---
hours
FIG. 8.17 Sub-area distribution or time-area graph
situated at the outlet, with storage equal to that of the catch-

ment. Then
(from Eq. (8.1))
and
S1 = KQ1 and S2 = KQ2 (from Eq. (8.9))
From these equations
where
0·5t 0·5t K- 0·5t
mo - m1 = m2=---
- K + 0·5t
-::----c~ -----o-
K + 0·5t K+ 0·5t
and since a distribution graph is being used and It = /2, then
Q2 = m I I+ m2 Q1 where m I = K +t 0 .51
VIUH
60
/ ........
~',
/ 2 h1 unitgropn
/
50 I / •
I / " '\.
40
v
1 I
"'· r"
'\
."'"',
I /
I
'
"~~
30
I I
I
V._i
20 I
I
10
.I,~,/
.,?/
0
2 4 6 8 10 12 14 16
hours
FIG. 8.18 Derived IUH and 2 h unitgraph
and in this case

I I
m1 = =-=0·I25
7·5 + 0·5 8
and
7·5- 0·5 7
m2 = = - = 0·875
7·5+0·5 8
Q2 = O·I25/ + 0·875Ql
5. Tabulate the data and compute Q2 as in Table 8.5. Q2 is the
required synthetic instantaneous unit hydrograph. Compute
the conversion constant for column 3.
IQ6 X IQ-2
1 em rain on I km2 in 1 h = '"" 2·78 m3/s
3600
Flood Routing 177
TABLE 8.5. IUH by routing
2 4 6
Tlm~-area 0·1251 = 2·78 02=cot.3+ 2h

Time x 0·125 x col. 2 0·875 X col. 5 col. 4 unitgraph
diagram =IUH
h km2 m 3/s m3js m3js m3js
0 0 0 0 0 0
I 10 3·5 0 3·5
2 23 8·0 3·1 11·1 5·5
3 39 13·5 9·7 23·2
4 43 14·9 20·3 35·2 23·1
s 42 14·6 30·8 45·4
6 40 13·9 39·6 53·5 44·3
7 35 12·1 46·8 58·9
8 18 6·2 51·4 57·6 55·5
9 0 0 50·5 50·5
10 0 0 44·1 44·1 50·8
II 0 0 39·6 39·6
12 0 0 34·6 34·6 39·3
13 0 0 30·2 30·2
14 0 0 26·4 26·4 30·5
IS 0 0 etc. etc. etc.
6. Plot the IUH and 2 h unitgraphs of columns 5 and 6 as Fig.

8.18.
To illustrate the ease of the method in accommodating areal
variation in rainfall, suppose that a hypothetical rainfall is specified
of20 mm on sub-areas 1 and 2, 7·5 mm on 3, 4, 5 and 6, and 15 mm
on areas 7 and 8, all falling in 1 h. The technique used then simply
converts the time area graph by these proportions, as shown by the
dotted graph lines of Fig. 8.17 before the routing operation derives
the IUH as before and then converts it to an n h unitgraph by
averaging each pair of ordinates at n h spacing. In this latter case a
degree of licence is being employed using the terms /UH and unit-
graph since the rainfall is not uniform and catchment-wide as re-
quired by their definition.
9 Hydrological Forecasting
9.1 Introduction
In the previous chapters the various physical processes involved in
the hydrologic cycle have been enumerated and examined in detail.
Methods of evaluating each process have been suggested and often
explained, and techniques discussed which may be used to provide
quantitative answers to many questions.
The remaining problem which must now be tackled is how to use
this knowledge to predict from existing data, however meagre it
may be, what will happen in future. This is a fundamental problem
of all engineering design, since the engineer designs and constructs
work to provide for future needs, whether he be a structural engineer
designing an office block, an electrical engineer designing power
systems to meet future electrical demand, or a hydraulic engineer
designing reservoirs to meet future demand for water.
There is one major difference in these three cases. The struc-
tural designer is working with homogeneous materials whose
behaviour is known within narrow limits. His buildings will be
used by people, whose spacing, dimensions, weight and behaviour,
en masse, can be predicted quite accurately. He has to cope with
natural events only in the form of wind loads, which usually form a
small proportion of the total load, and earthquakes. For both
these eventualities there are codes of practice and recommendations
available to him.
The electrical system designer has to extrapolate the rising demand
curve of recent years, and to examine the trends of industry and
personal habits, to decide how much capacity should be available
in future years. While this is a complex and continuous task, it is
178
Hydrological Forecasting 179
almost completely immune from natural events other than disasters
for which he cannot be expected to provide.
The hydraulic engineer on the other hand is dealing, in reservoir
design, almost exclusively with natural events: in the incidence of
precipitation, evaporation and so on. These events are usually
random in nature and may have any of all the non-negative values. It
is true that if the rainfall at a place is measured daily for a
period of time, a knowledge about what is a probable daily rainfall
will be built up, but it will not, however long it goes on, lead to any
limiting possible value of daily rainfall, other than intuitively.
The hydrologist is frequently asked what the maximum possible
discharge of a particular river will be. There is simply no such value.
The only answer that can be given is that from the data available,
and making various assumptions, it would appear that a certain
value will not be exceeded on average more than once in a specific
number of years. On such estimates all hydrologic design must be
performed, and this chapter deals with methods whereby some of the
uncertainties may be removed or narrowed in range.
9.2 Flood formulae

The particular random variable of river flood discharge has been of
interest to engineers and hydrologists from the earliest days of
hydrology and many formulae have been proposed to define lhe
"maximum flood" that could occur for a particular catchment. The
formulae are empirical by nature, derived from observed floods on
particular catchments and usually of the form Q = CAn
where Q = flood discharge in m3fs (or ft 3/s)
A =catchment area in km2 (or mile2)
n = an index usually between 0·5 and 1·25
C = a coefficient depending on climate, catchment and
units.
An early example of such a formula due to Dickens was developed
in India
Q = 825a0·75
with Q in ft3fs and a in square miles; but since the formula takes
no account of soil moisture, rainfall, slope, altitude etc., it is clearly
of very little value in general application. This is true of all such
formulae although they are frequently used to obtain a quick first
estimate of the order of "maximum flood" that may be expected.
For such purposes Morgan [63] proposed the formula for a cata-
strophic flood in Scotland and Wales of
Q = 3000M0 · 5
where Q is in ft3js and M is catchment area in square miles, and
added the sophistication of a recurrence period T (in years) by
quoting
design flood = catastrophic flood X (T/SOO)l
for cases where the adoption of the catastrophic flood was not
justified by danger to human life or the safety of a dam. A similar
formula of the same type, due to Fuller, has been widely used in the
U.S.A.
Qav = CAO·B
where A is catchment area in square miles
C is a coefficient often taken as 75
Qav is average value of annual flood discharge in ft 3/s
The value of Qav is then substituted in the formula
Qm = Qav(l + O·Slog T)
where T is a return period in years and Q711 is the 'most probable'
annual maximum flood. Such calculations, while simple to make,
are of limited value, in that they are 'envelope' expressions derived
to cover all recorded occurrences with indeterminate safety margins
and as such they take no account of the physical processes involved
in runoff, and are frequently very conservative.
9.3 Frequency analysis

9.3.1 Series of events. The next approach is to use the methods of
statistics to extend the available data and hence predict the likely fre-
quency of occurrence of natural events. Given adequate records,
statistical methods will show that floods of certain magnitudes may,
on average, be expected annually, every ten years, every 100 years
and so on. It is important to realise that these extensions are only as
valid as the data used. It may be queried whether any method of
extrapolation to 100 years is worth a great deal when it is based on
(say) 30 years of record. Still more does this apply to the '1000 yr
flood' and similar estimates.
Another point for emphasis is the non-cyclical nature of random
events.* The 100 yr flood, (i.e. the flood which will occur on average,
once in 100 years) may occur next year, or not for 200 years or may
be exceeded several times in the next 100 years. The accuracy of
estimation ofthe value of the (say) 100 yr flood depends on how long
the record is and, for flood flows, one is fortunate to have records
longer than 30 years. Notwithstanding these warnings, frequency
analysis can be of great value in the interpretation and assessment
of events such as flood flows and the risks of their occurrence in
specific time periods.
It is particularly important to define what is meant by an event.
For example if a river has been gauged every day for 10 years, there
t annual water- year peaks
* secondary peaks greater than on annual peak
water years
-:--------:-~-;------:
calendar y e o r s
FIG. 9.1 Annual and partial duration series events
will be about 3650 observations. These are not independent random

events since the flow on any one day is dependent to some extent
on that of the day before, and so the observations do not comprise
an independent series. The array of these observations is termed a
full series.
Suppose from the 10 year record we extract in each year the
maximum event. These would constitute an independent series since
it is highly unlikely that the maximum flow of one year is affected
by that of a previous year. Even so, care is necessary, as may be seen
from Fig. 9.1, where water years, measured between seasons of
minimum flow, are marked, as well as calendar years. One calendar
year might contain two water-year peaks, so it is necessary to specify
that water years should be used in defining events. A selection like
• See also Section 9.7.
this is called an annual series. Such a series is open to the objection
that some of the peaks are smaller events than secondary peaks
(marked with an asterisk in Fig. 9.1) of other years. The objection
may be overcome by listing a partial duration series, in which strict
time segregation is no longer a condition and all peaks above some
arbitrary value (say the lowest annual peak) are included provided
that, in the judgment of the compiler, they are independent events,
uninfluenced by preceding peak flows. Partial duration series events
therefore allow the objection of subjective judgment and are not,
strictly speaking, independent and random.
Which series is used depends on the purpose of the analysis. For
information about fairly frequent events, e.g. the size of a flood that
might be expected during the construction period of a large dam (4
years say}, then a partial series may be best, while for the design
flood for the dam's spillway which should not be exceeded in the
dam's lifetime (say 100 yr) the true distribution series, or annual
series, will be preferable. Actually, with very large floods there is a
very small difference in recurrence interval between the two. Full
series events, although not independent, are most valuable in design
where quantity rather than peak values are required.
9.3.2 Probability of the N-yr event. The term recurrence interval

(also called the return period), denoted by Tr, is the time which, on
average, elapses between two events which equal or exceed a par-
ticular level. Putting it another way, the N-yr event, the event which
is expected to be equalled or exceeded, on average, every N years, has
a recurrence interval, Tr of N years.
As mentioned previously there is no implication that the N-yr
event occurs cyclically. It does, however, have a probability of
occurrence in any particular period under consideration.
Let the probability P(X ~ x) represent the probability that x will
not be equalled or exceeded in a certain period of time.
Then P(X ~ x)n will represent the probability that x will not be
equalled or exceeded in n such periods.
For an independent series and from the multiple probability rule
P(X ~ X)n = [P(X ~ x)]n

= [1 - P(X;;?: x)]n
P(X;;?: x)n = 1 - [I - P(X;;?: x)]n
Now 1
Tr = -P-(X---~-x-)
P(X ~ x)n = 1 - [1 - :J"
So, for example, the probability of X ~ x, where x is the value of a
flood with a return period of 20 years, occurring in a particular 3 year
period is
P(X ~ 20 yr flood)a = I - [I - 1\]3
=I - [0.95]3
= I - 0.857
= 0·143 or 14·3%
Table 9.1 shows the probability of the N-yr flood occurring in a
particular period.
TABLE 9.1 Percentage probability of theN-year flood occurring
in a particular period
Number of N = Average return period Tr: years

years in
period 5 10 20 50 100 200 500 1000
1 20 10 5 2 1 0·5 0·2 0·1
2 33 19 10 4 2 1 0·4 0·2
3 45 27 14 6 3 1·5 0·6 0·3
5 63 41 22 10 5 2·5 1 0·5
10 87 65 40 18 9 5 2 1
20 98 88 64 33 17 10 4 2
30 99·8 96 78 45 24 14 6 3
60 99·8 95 70 43 26 11 6
100 99·4 87 60 39 18 9
200 98·2 74 63 33 18
500 99·6 92 63 39
1000 99·3 96 63
Where no figure is inserted the % probability > 99·9.
For example, it may be seen from the table there is a 1% chance

of the 200 yr flood occurring in the next 2 years and an 8% chance
that it will not occur for the next 500 years.
If the probability P(X ~ x)n is defined by a policy ruling, the
value of n, the design period, may be found from
P(X ~ x)n = 1- (1- ~;r

1-P(X~x)n = -1)"
(1-T,1)" = (T.T,
log (1 - P(X ~ x)n) = n log (Tr~ 1)
log 1 - P(X ~ x)n)
n = ----!::.....,.-~----=-..,..,......:.=
log (T,-1)
T,
Example 9.1. How long may a cofferdam remain in a river, with an
even chance ofnot being overtopped, if it is designed to be secure against
a 10 year flood?
Here, the policy ruling is that there should be an even chance, so
P(X ~ x)n = 0·50 and T, = 10 then
n = log (1 - 0·5) = log 0·5 = I·699 = 0·301 = 6.5 r
log 19o log 0·9 I·954 0·046 y
9.3.3 Probability plotting. Having listed a series of events they may

each then be accorded a ranking, m, starting with m = 1 for the
highest value, m = 2 for the next highest and so on in descending
order. The recurrence interval T, (in years) of each event is now
computed from
(9.1)
where m = event ranking, and n = no. of events.

(Note: when plotting partial duration series, it is customary to
limit the number of events to the n highest, where n = number of
years of observation.)
Various other formulae are used in place ofEq. (9.1)which is most
commonly used: for example, the California formula [64]
n
Tr=-
m
and Hazen's formula [65]
T. _ 2n
r-2m-I
Table 9.2 shows a listing of the annual maximum daily mean flows
of the River Thames at Teddington Weir, for the years 1882-1967.
This is a true annual series of random events with return periods
computed.
TABLE 9.2. Maximum mean daily flows for water years 1882-1967,
for R. Thames at Teddington
Water- Percelll Water- Perulft

year Returlf proba- year Returlf proba-
eJU~IIfr Om Rlllfk period blllty eltdlllr Om Ralfk period blllty
30 Sept m•/s m Tryr p 30 Sept. m 3 /s m Tryr p
1882 1925 522 7 12·3 8·1

1883 292 46 1·87 53·5 1926 370 25 3·44 29·1
1884 231 65 1·32 75·5 1927 375 23 3·74 26·8
1885 230 67 1·28 78·0 1928 526 6 14·3 7·0
1886 244 59 1·46 68·6 1929 235 62 1·39 72·0
1887 284 48 1·79 SS·8 1930 552 4 21·5 4·6
1888 208 73 1-18 84·9 1931 228 69 1·25 80·3
1889 237 61 1·41 71·0 1932 274 49 1·75 57·0
1890 205 74 1·16 86·0 1933 478 9 9·SS 10·5
1891 171 81 1·06 93·2 1934 95 86 1·01 99·0
1892 339 32 2·69 37·2 1935 227 71 1·21 112·5
1893 300 42 2·04 48·9 1936 478 10 8·6 11·6
1894 173 79 1·09 92·0 1937 438 14 6·15 16·3
1895 789 1 86 1·16 1938 247 58 1·48 67·5
1896 202 76 1·13 88·4 1939 369 26 3·30 30·2
1897 351 29 2·96 33·8 1940 410 IS 5·74 17·4
1898 171 80 1·07 93·0 1941 384 19 4·52 22·1
1899 262 51 1·69 59·3 1942 298 44 1·95 51·2
1900 533 5 17·2 5·8 1943 457 11 7·8 12·8
1901 200 77 1·12 89·5 1944 115 83 1·04 96·5
1902 162 82 1·05 9S·S 1945 261 52 1·65 60·5
1903 386 17 5·06 19·8 1946 257 53 1·62 61·6
1904 516 8 10·8 9·3 1947 714 2 43 2·3
1905 229 68 1·26 79·0 1948 227 70 1·23 81·5
1906 249 57 1·51 66·3 1949 299 43 2·00 50·0
1907 220 72 1-19 83·7 1950 324 35 2·46 40·7
1908 376 21 4·1 24·4 1951 385 1i 4·78 20·9
1909 204 75 1·15 87·1 1952 377 20 4·3 23·2
1910 231 66 1·30 76·7 1953 263 so 1·72 58·1
1911 395 16 5·38 18·6 1954 231 64 1·34 74·5
1912 367 28 3·07 32·6 1955 453 13 6·61 15·1
1913 2SS 55 1·57 64·0 1956 316 38 2·26 44·2
1914 256 54 1·59 62·8 1957 314 39 2·20 45·4
1915 585 3 28·6 3·5 1958 317 37 2·32 43·0
1916 373 24 3·58 27·9 1959 375 22 3·91 25·6
1917 327 34 2·53 39·6 1960 308 41 2·10 47·7
1918 351 30 2·86 34·9 1961 456 12 7·16 14·0
1919 334 33 2·60 38·4 1962 344 31 2·78 36·0
1920 251 56 1·54 65·1 1963 286 47 1·83 54·6
1921 240 60 1·43 69·8 1964 369 27 3·18 31·4
1922 198 78 1·10 90·6 1965 113 85 1·02 97·6
1923 231 63 1·37 73·2 1966 324 36 2·39 41·9
1924 298 45 1·91 52·4 1967 313 40 2·15 46·5
If= 85 0 11v = 319·5 m 3 /s standard deviation a = 124·56 Tr =(If .. 1)/m P% = 100/Tr
Having obtained the Tr values, the question now arises, can the
listed values of discharge, Q and Tr be used to extrapolate the data
and infer the return periods of extreme floods? This may be attemp-
ted in various ways, which are discussed below.
(a) Q v. Tr may be plotted on simple plane co-ordinates. An
example of this is illustrated in Fig. 9.2 where the River
Thames data of Table 9.2 has been plotted. The extrapolation
of the curve to high Q and Tr values depends almost entirely
on the few highest existing points.
(b) Q linear and Tr logarithmically. The same data has now been
plotted in Fig. 9.3 and a straight line fitted to it. Since the
--
judgment required in this case is to fitting a line through all
the points rather than extrapolating from a few, it is simpler
800
-
l..--
~
v--
700 r-
/
~
600
/
~ 500
./
I
E
·=~400
/
0
J::.
~
0 300
200
100
0 10 20 30 40 50 60 70 80 90 100
Return period T, in years
FIG. 9.2 Annual maximum daily flow of River Thames at

Teddington, 1882-1967
to do than the first case. However, unless the return period

follows a logarithmic law, it does not necessarily provide
more accurate extrapolation.
(c) Another approach is to assume that the events have a normal
distribution, so that on normal probability paper (due to
Hazen [66] they will plot along a straight line. This plot is
shown in Fig. 9.4. Clearly the points do not lie along a straight
line, so a shallow curve has been fitted to them. It should be
noted that the Tr scale has now become probability P. P is
1000
900 _____..
800
~
;:. 700
E
·= 600
___... v
~ 500 ~
0
.....:~
~ 400
;.;:.--""
....
0 300 /
200 ... ~
100
I 2 3 4 5 6 7 8910 20 30 40 50 70 90 200
Return period 7;, in years 60 80100

Teddington, 1882-1967 (semi log)
1000
900
800
./
..,~ 700
E
,!:"
600 /
e. 500 /"
0
..c;;
" 400
/"
..·.
"'
i5 ,.,-"
,;:
,.,.
--
300
,• ·;?"" ~
200
_:..;.-
100
- ~
99·8 99 98 95 90 80 70 60 50 40 30 20 10 5 2 I 0·5
Per cent probability

Teddington, 1882-1967 (normal probability)
187
the probability of the corresponding value of Q being equalled
or exceeded in any one year.
P=_!_
Tr
(d) A variation of this approach is to assume that the logarithm
of the variate (Q) is normally distributed and this leads to
the use of a logarithmic-normal distribution or log-normal
paper (first used by Whipple [67].) The same data as before
are presented in this way in Fig. 9.5 and again the abscissa
represents probability P.
1000
900
800 I-'
700
600 /
./·
...,.."' 500
/
E 400
.:
"' 300
V-'1'"
v v ... ..
12'
0
..c. ;;.;~
l;l 200
Ci f/
/
1_.......
100
98 95 90 80 70 60 5040 30 20 10 5 2 I 0·5
Per cent probability

Teddington, 1882-1967 (log normal)
(e) Other investigators have proposed methods assuming other

theoretical frequency distributions. Gumbel [68, 69, 71] used
extreme-value theory to show that in a series of extreme
values X1, X2 ... Xn where the samples are of equal size and
X is an exponentially .distributed variable, (e.g. the maximum
discharge observed in a year's gauge readings), then the cumu-
lative probability P' that any of the n values will be less than a
particular value X (of return period T) approaches the value
P' = e-e-71
where e =natural logarithm base
and
i.e. P' is the probability of non-occurrence of an event X in

Tyears.
1
or T=--
1-P'
(Note: This argument refers to Gumbel's method. The reader
1
should not confuse this with the normal usage of T, = p
where P =probability of occurrence.)
The event X, of return period Tyears, is now defined as Qp,
then Qp = Qav + a(0·78y - 0·45) (9.2)
where Qav = average of all values of 'annual flood' Qm
and a = standard deviation of the series
where n = number of years of record = number of Qm

values
and I: Q!. = sum of the squares of n values of Qm
Table 9.3 gives values of y as a function ofT.

TABLE 9.3
T y T y T y
1·01 - 1·53 5·00 1·50 50 3·90

1·58 0·00 10·00 2·25 100 4·60
2·00 0·37 20·00 2·97 200 5·30
Powell [70] suggested that if plotting paper is prepared in which

the horizontal lines are spaced linearly and the vertical lines' spacing
is made proportional toy, then from Eq. (9.2) Qp and Twill plot
as straight lines. This is the basis of Gumbel-Powell probability
paper, used to plot the River Thames data in Fig. 9.6. The return
period T has been computed, as before, as T = n + 1. The straight
m
line on this figure has been drawn between the two points Qav and
Q2oo. Qav, from Eq. (9.2) occurs when 0·78y = 0·45 or y = 0·577,
which corresponds toT= 2·33 yr. Eq. (9.2) holds for large values of
n, say n > 50, when Qav at 2·33 yr is included on the line through the
points. The other point Q2oo, represents the '200-yr flood' and is
found by inserting the appropriate values in Eq. {9.2).
Q2oo = Qav + 124·56(0·78 X 5·30 - 0·45) = 778 m3/s
The correspondence between the plotted data and Gumbel's theoreti-
cal line is demonstrated. Gumbel paper should not be used for
900
v
800
vv
700
~
"'E 500
600
·/
v
·= v.
e. 400
o., "
v /'
0
.s::
l;l "';;...
i5 300
... ·· /
200
f:."y
100
I~
1·01 1·05 1·1 1·2 1·4 2 2·5 3 4 5 6 78910 20 30 4050 100 200
1·3 1·5 7; :Recurrence interval, in years

Teddington, 1882-1967 (Gumbel distribution)
partial series which usually plot better on semi-log paper, as used

in Fig. 9.3.
From the plots presented in Figs. 9.2-9.6, it may seem there is
little to choose between the particular plotting papers available.
This is very often the case and investigators should use whichever
distribution makes their particular job of fitting and extrapolation
simplest and the line apparently of best fit.
The foregoing is a necessarily brief resume of the methods of
plotting flood events, in current use. For the underlying theories the
reader should refer to original papers and more comprehensive
treatments available [72, 73].
9.4 Synthetic data-generation

One of the perennial problems of the hydrologist is insufficient data,
whether it be rainfall or, more usually, discharge observations. If he
wishes to predict flood flows of relatively large magnitude and hence
large return periods, he may find that he has perhaps a decade or two
of daily observations which represent discharges. Using an annual
series such a record might yield 20 or 30 points from which it is
clearly dubious to predict rare events of the order of the 100 year
flood. By using only one or two of these measurements from each
year's record, an enormous amount of information about the dis-
charge characteristics of catchment is being left untouched. Is there
not contained within this mass of routine observation a guide, not
only to the catchment's response to rain, but to the incidence of the
rain itself? By studying the fundamental nature of the variations in
discharge, even at low flows, might it be possible to reproduce them,
acting in their random and apparently uncoordinated ways and so
automatically reproduce series exhibiting the variations that the
natural data show?
The advent of powerful computers has provided answers to these
questions by analysing and reproducing synthetic data in large
quantities. While the quality of the synthetically generated data is
mainly dependent on the original natural data, the method makes
use of all the information available rather than the very small
proportion of it in the form of extreme values.
In essence, the method of artificially generating time-series relies
on using historical records as a sample of a total population, while
conventional methods consider the records to be the total popula-
tion. It follows that designs will be based on estimates of what
might have happened instead of on what has happened.
Any time series of observed values may contain a trend compo-
nent, a periodic component and a stochastic component. The first
two components are deterministic in nature, i.e. they are not inde-
pendent of the time at which a series starts nor of the length of the
series, while the stochastic component is stationary, i.e. the statistics
of the sample do not differ from the statistics of the population
(except as a result of sampling variability) and are time-independent.
If the trend and periodic components are removed from the series,
a stationary stochastic component is left. This component will
contain a random element and may or may not contain a correlation
element. Series correlation describes how each term in a series
is affected by what has gone before, e.g. a wet summer may lead to
higher autumn flows than average. Accordingly, the random element
and correlation structure of the stochastic component must be
isolated and quantified.
By now the time series has been taken apart and its various parts
examined. Each of the parts is now reproduced by mathematical
simulation using randomly occurring numbers, Markov series,
serial correlation coefficients etc., including the re-introduction of
the periodicity and trend components. The 'model' thus created can
now be used to generate synthetic data in whatever quantities are
desired, and the series produced used to estimate particular N-year
events as though the data had been observed.
9.5 Rainfall data and the unit hydrograph

In using unit hydrographs for hydrologic forecasting it is only
necessary to decide what effective rain to apply, since the unitgraph
can be adjusted to suit the chosen storm length (Sect. 7.5). This,
however, is no simple task and is one of the least well documented
subjects in hydrological literature. The reason for this is mainly
that the operation of applying rain to unit graphs is carried out for
many different reasons. The design and construction of different
projects may involve the estimation of many different floods. For
example, the dam designer wants to make his spillways sufficiently
large so that the safety of the dam is not threatened in its lifetime.
The dam builder wants to know what risk he runs of various floods
occurring in the 3 or 4 years he may be building in a river channel,
so that his cofferdams and tunnels may be economically sized. The
flood protection engineer wishes to be sure his levees will not be
overtopped more often than at some design frequency which he
regards as economic, or perhaps not at all if the safety of human
life is concerned. The estuarial or river port manager wants to know
the depth of water he can rely on in navigable channels and so on.
In any or all of these cases it may be desirable to choose a design
rain of particular frequency and magnitude and apply it to a catch-
ment unitgraph as a means of flow prediction. This is particularly
so because in many parts of the world there are reasonably long-term
rainfall records though very sparse or short-term runoff records.
The rainfall records must therefore be examined and frequency
analyses made of the incidence of particular 24 h rainfall depths.
Rainfall observations are comparatively rarely made more often
than this, although clearly very intense storms of rain frequently last
for shorter periods. Much depends on catchment size. If it is very
large, then 24 h frequencies may be adequate for choosing design
rain. Most often, however, shorter periods of rain at greater inten-
sity will represent the greatest flood danger and efforts must be made
to find shorter interval data, or to start measuring it by autographic
recorder as soon as project investigations commence. Extrapolations
may be made by recording the frequencies with which certain rainfall
depths occur in 96 h, 72 h, 48 h, and 24 h periods. Much data is
already available, from U.S. sources [74, 75, 76, 77, 78] which
though directed obviously to U.S. catchments, is applicable with
care elsewhere. In Britain, reference should be made to the Flood
Study Report expected to be published in 1974 [79]. The Meteoro-
logical Office contribution to this report allows estimates of expected
rainfall amounts to be made for any place in the United Kingdom
with a given return period for durations from 15 seconds to 30 days.
It is also possible to obtain reduction factors to be applied to point
rainfall to calculate areal rainfall for the same return period, for a
rainfall duration from 2 minutes to 8 days and for areas between
I and 10,000 km2.
Reference may also be made to Bleasdale [80], to British Rainfall
1939 et seq. and others [81].
The process of using the unit-hydrograph therefore involves the
following steps:
(a) An examination of all relevant data (including frequency
analyses), in particular, any autographic rain recordings,
and a subsequent selection of a total precipitation that could,
in the judgment of the individual concerned, occur in a
reasonably short time period, over the catchment concerned.
(Note: where the catchment is large and it seems unlikely
that uniform precipitation could or would occur, it should be
split into a series of sub-catchments and each of these should
be assigned a precipitation.)
(b) From the total precipitation, thus chosen, all losses must be
subtracted. These will include interception, infiltration
(having regard to the soil-moisture deficiency) and evapo-
transpiration. (See Section 4.4, and Chapter 3.)
(c) The·resultant nett, or effective rainfall is then applied to a unit-
graph adjusted to the correct time-base, and the resulting
storm hydrograph ordinates obtained. If more than one
hydrograph is concerned the temporal spacing of the rain
must be decided, usually, but not necessarily, to obtain the
maximum possible combination of ordinates.
(d) Base-flow, assessed separately, is now added to give the total
storm runoff.
The choice of the s.m.d. on which a design storm may fall is par-
ticularly significant. It may not always be reasonable to presuppose
a saturated catchment, and a hypothetical infiltration capacity
curve, a ~-index or coaxial correlation graph (e.g. Fig. 4.7) may be
used with this in mind.
9.6 Hydro-meteorology
Over the last 25 years there has been an increasing use of the science
of meteorology, in combination with hydrology, by engineers con-
cerned with water resources and flood control. The development of
this trend is due to the need to use large values of rainfall to apply to
the unit hydrograph, which is a powerful tool in the estimation of
extreme values of flood flows used in the design of water control
structures. Immediately, the problem is met of how large is 'large'?
The need to define the upper limits of possible precipitation more
precisely than can be done by statistical methods (based on what are
comparatively short records) led to the investigation of just how
much water is stored in a storm.
Hydro-meteorology, like many other hydrological techniques,
has been developed mainly in the U.S.A. and it seems reasonable,
therefore, to discuss it in the light of procedures used there. The
U.S. Corps of Engineers prepare flood estimates of three classes
under the headings:
(a) Statistical analyses of stream flow records. These are made
generally on a regional basis along the lines discussed in Sect. 9.2.
They are used mainly in assessing mean annual benefit over the long-
term, as a result of implementing particular projects.
(b) Standard project flood (SPF) estimates. These estimates are of

the floods likely to occur from the most severe combination of
meteorological and hydrological conditions which are reasonably
characteristic of the geographical area being considered, but exclud-
ing extremely rare combinations. The concept is a good one but
inevitably subjective. The difference between "extremely rare" and
"rare" and decisions about what is "reasonable" obviously call for
judgment and have caused discussion (without resolution) before [82].
(c) Maximum-probable-flood (MPF) estimates. These estimates

differ only from the SPF in that they are reckoned to include the
extremely rare event, or what has been termed in much British
literature "the catastrophic flood", and are usually confined to spill-
way design for high dams.
The SPF and MPF estimates are examples of the continuing
modern tendency to rely on the science of meteorology for the
prediction of very large floods. This meteorological approach is
concerned with those physical conditions of the atmosphere in
particular climates, which represent the upper limits of rainfall pro-
duction. The variation of these conditions in magnitude, season,
temporal distribution and frequency all require consideration.
Probable maximum precipitation (PMP). The hydrometeorologist

starts from the premise that if the greatest amounts of rain that can
fall over a given catchment in a given time can be estimated, then the
application of this rainfall with appropriate reduction for losses to
the unit hydrographs of the catchment will provide an estimate of
the probable maximum flood. This is likely to be a better estimate
of the MPF than the extrapolation of flow records, to 20 or 30 times
their period of observation, by fitting statistical distributions.
The essential requirement for precipitation is a supply of moist air.
The air's water content is measured by its temperature and dew-
point, which are standard meteorological observations and are
therefore often available in catchments where few flow records exist.
The amount of water in the atmosphere above a catchment can be
established by the assumption of a saturated adiabatic lapse rate
(see Sect. 2.3) or by actual weather balloon observations. The
amount is usually between I 0 mm and 80 mm in depth. This water is
precipitated by cooling, which is accomplished almost always by
vertical movement, the air mass expanding adiabatically, thus
precipitating moisture which in turn liberates latent heat, which
accelerates the vertical movement. The process is self-stimulated and
in extreme cases is responsible for the "cloud burst" of the severe
thunderstorm. The vertical lifting may also be due to orographic or
frontal effects as discussed in Section 2.6.
The inflow of moist air at the base of an ascending storm column
may be measured by wind-speed observations around storm peri-
pheries. These observations yield data about the amount of water
which may be carried into a storm-column and hence precipitated.
The data may also be inferred from synoptic weather charts if these
are based on a sufficiently dense network of observer stations.
From these observations of air-moisture, (through dew-point and
temperature) and air-inflow (through wind speed and barometric
pressure) a storm moisture inflow index may be found. This method
is described in the literature for various areas of the world [78, 83,
84, 85]. Maxima for separate seasons or months may be subse-
quently derived by taking maximum recorded seasonal or monthly
values of air-moisture and air-inflow and assuming they occur
simultaneously. These may be compared with historic storms
which have occurred in the catchment and meteorologically-similar
neighbouring regions. In this way an array of data about the
historic storms may be built up and compared with the hypothetical
maxima.
The PMP may be derived from the hypothetical maxima by taking
the peak of the envelope curve covering them. It is argued sometimes
that the elements used to define the maxima should themselves be
subject to frequency analysis before the adoption of a PMP. This,
however, presupposes th,at the physical processes of rain producing
storms are random events which can occur in vastly different magni-
tudes. This may not be true to the same extent as it is of precipita-
tion recorded at a rain gauge or the flood in a river. Also, in taking
the known maximum values of the determining factors and combining
them in space over the catchment, the concept of return period is
invalidated. The approach is deterministic rather than statistical.
The subject is under active development in many countries and
more precise techniques will no doubt be developed presently.
General application of data. Many large storms have been analysed,
particularly in the United States, and maximum values of rainfall
depth for various durations and areas have been published [75, 76,
86]. Such data are usually presented as sets of curves, each represent-
ing a rainfall depth plotted on rectangular coordinates of storm
duration in hours and storm area in square miles. These are a useful
guide to limiting values but must be used with judgment for particu-
lar catchments, since topography and elevation as well as climate
may modify the results appreciably for other regions.
In the light of this section it is interesting to look again at the
world maximum point rainfalls quoted in Section 6.5. Paulhus [87]
suggests that if rainfall is plotted against duration, both scales
logarithmic, the world's greatest observed point rainfalls lie on or
just under a straight line whose equation is
R = 16·6D0·475
Where R is rainfall in inches and D is duration in hours. Bleasdale
[88] has suggested that the simpler relationship
R = 15·5D*
is a reasonably good fit and is easier to calculate, and points out that
British maxima also lie very close to a straight line on the same
plotting, with values close to one quarter of the world's values.
It is probable, however, that the simple relationship quoted
represents a smoothing of several overlapping complex relationships
and naturally it would be unwise to use the straight line values in
areas or on catchments whose records show precipitation falling
well below those where the maxima have occurred.
9.7 The cyclical nature of hydrological phenomena

In all the foregoing sections of this book it has been tacitly assumed
that the processes described and studied are based on non-changing
physical conditions. For example in the analysis of frequencies it is
assumed that events which occurred in the last 50-100 years can be
used to predict the probability of similar events occurring in future.
From time to time this assumption is challenged but has rarely been
disproved.
This is at least partly due to misconceptions about what periodicity
implies, when associated with hydrological events. The implication
of a periodicity in hydrological phenomena is that the likelihood of
certain values of random events appears greater at certain times than
at others. In other words there is a cyclical change in probabilities,
rather than events. This does not rule out the possibility of maximum
events (for example) occurring at times when their probabilities are
least.
Brooks and Carruthers [89] describe the periodicity of annual
rainfall in England which seems to have a period of 51·7 years.
They go on to show that the probability of a wet year was almost
twice that of a dry yearnear the maximum of the 51· 7 year cycle, whilst
near the minimum it was less than half. This was true despite both
very wet and dry years occurring at corresponding times of mini-
mum probability.
It seems futile to employ statistical methods to drive probabilities
of occurrence of certain events without recognising that certain
processes which are part-causative agents for these events, may be
subject to cyclic probability themselves, thereby altering the derived
probability to some degree. Although the opportunity for using
such information may be rare, its presence should be investigated
for any short-term probability analysis where the time-span is of
the same order as the derivative data.
Cochrane, in an analysis of the hydrology of Lake Nyasa and its
catchment [90] has demonstrated a correlation between the rate of
change of sunspots and the "free water" on Lake Nyasa. (The term
"free water" refers to the residual runoff from the catchment and
storage on the lake, after losses have been deducted from rainfall.)
In another paper [91] he quotes many references to support the case
for cyclic behaviour in hydrologic phenomena, concerning work in
many parts of the world. A dispassionate examination of the evi-
dence leads the present writer to the conclusion that periodicity in
hydrological phenomena exist<>, however imperfectly as yet, we
understand it.
Problems
Some of the following problems have appeared in the University of
Salford's B.Sc. (Civil Eng.) examination papers, and are published by
permission of the University.
Answers follow on p. 226.
Chapter 2-Problems
2.1 An air mass is at a temperature of 28°C with relative humidity
of 70%. Determine
(a) saturation vapour pressure
(b) saturation deficit
(c) actual vapour pressure in mbar and mm Hg
(d) dewpoint
(e) wet bulb temperature
2.2 Discuss the relationships between depth, duration and area of

rainfall for particular storms.
2.3 The following are annual rainfall figures for four stations in
Derbyshire. The average values for Cubley and Biggin School have
not been established.
Average (in.) 1959 1960
Wirksworth 35·5 26·8 48·6

Cubley 19·5 42·4
Rodsley 31-3 21·6 42·1
Biggin School 33·1 54·2
(a) Assume departures from normal are the same for all
stations. Forecast the Rodsley "annual average" from that at
Wirksworth and the two years of record. Compare the result with
the established value.
199
(b) Forecast annual averages for Cubley and Biggin School
using both Wirksworth and Rodsley data.
(c) Comment on the assumption in part (a). Is it reasonable?
2.4 One of four monthly-read rain gauges on a catchment area

develops a fault in a month when the other three gauges record re-
spectively 37, 43 and 51 mm. If the average annual precipitation
amounts of these three gauges are 726, 752 and 840 mm respectively
and of the broken gauge 694 mm, estimate the missing monthly
precipitation at the latter.
2.5 Compute the average annual rainfall, in inches depth, on the

catchment area shown in Fig. 1 by
45 8
376
-Catchment area
boundary
29 2
(i) arithmetic means

(ii) Theissen method
(iii) plotting isohyets
Comment on the applicability of each method.
2.6 Discuss the setting of rain gauges on the ground and comment
on the effects of wind and rain falling non-vertically on the catch.
2. 7 Annual precipitation at rain gauge X and the average annual

precipitation at 20 surrounding rain gauges are listed in the following
table
Problems 201
(a) Examine the consistency of station X data.
(b) When did a change in regime occur? Discuss possible
causes.
(c) Adjust the data and determine what difference this makes to
the 36-year annual average precipitation at station X.
Annual precipitation Annual precipitation

mm mm
Year Year
20 station 20 station
Gauge X average Gauge X average
1972 188 264 1954 223 360

1971 185 228 1953 173 234
1970 310 386 1952 282 333
1969 295 297 1951 218 236
1968 208 284 1950 246 251
1967 287 350 1949 284 284
1966 183 236 1948 493 361
1965 304 371 1947 320 282
1964 228 234 1946 274 252
1963 216 290 1945 322 274
1962 224 282 1944 437 302
1961 203 246 1943 389 350
1960 284 264 1942 305 228
1959 295 332 1941 320 312
1958 206 231 1940 328 284
1957 269 234 1939 308 315
1956 241 231 1938 302 280
1955 284 312 1937 414 343
2.8 The data for the mean of the 20 stations in Question 2.7 should
be plotted as a time series. Then plot 5-year moving averages and
accumulated annual departures from the 36-year mean. Is there
evidence of cyclicity or particular trends?
2.9 At a given site, a long-term wind-speed record is available for

measurements at heights of 10 and I 5 m above the ground. For
certain calculations of evaporation, the speed at 2 m is required, so
it is desired to extend the long-term record to the 2m level. For one
set of data, the speeds at 10m and 15m were 9.14 and 9.66 m/s
respectively.
(a) what is the value of the exponent relating the two speeds
and elevations?
(b) what speed would you predict for the 2 m level?
2.10 A rainfall gauge registers a fall of9 mm in 10 minutes.

(a) how frequently would you expect such a fall at a particular
place in Britain?
(b) what total volume of rain would be expected to fall on
3 km 2 surrounding the gauge?
2.11 What is the maximum one-day rainfall expected in Britain for

a 50-year period at location X (average annual rainfalllOOO mm) and
a 30-year period at location Y (average annual rainfall1750 mm)?
2.12 What is the average rainfall over an area of 8 km2 during a

storm lasting 30 minutes with a frequency of once in 20 years in (a)
Oxford, (b) Kumasi. Does your answer for (b) require qualification?
Chapter 3-Problems
3.1 Determine the evaporation from a free water surface using the
Penman equation nomogram for the following cases
Locality Month h nfD U2

Amsterdam (52°N) July 0.5 0.5 1.2 m/s
Seattle (47°N) Jan 0.8 0.3 1.5 mfs
3.2 Use the nomogram for the solution of Penman's equation to

predict the daily potential evapo-transpiration from a field crop at
latitude 40°N in April, under the following conditions
mean air temp. = 20°C mean h = 70%
sky cover = 60% cloud mean U2 = 2.5 mfs
ratio of potential evapo-transpiration to
potential evaporation = 0·7
3.3 Compute the potential evapo-transpiration according to

Thornthwaite for two locations A and B where the local climate
yields the following data
Problems 203
% %
daylight daylight
hours of hours of
A B year at A A B year at A
Jan. -5 -2 6 July 19 16 11
Feb. 0 2 7 Aug. 17 14 10
Mar. 5 3 7t Sept. 13 10 St
Apr. 9 7 St Oct. 9 8 7t
May 13 10 10 Nov. 5 3 7
June 17 15 11 Dec. 0 0 6
(a) at A for April (mean temp 10°C) and Nov. (mean temp =
30C)
(b) at B for June (mean temp 20°C) and Oct. (mean temp=
goq
At A the average number of hours between sunrise and sunset is
13 for April and 9 for Nov. At B the figures are 14 for June and 10
for October. Use the Serra simplification for A and the nomogram
for B.
3.4 Water has maximum density at 4°C; above and below this
temperature its density is less. Consider a deep lake in a region where
the air temperature falls below 4°C in winter.
(a) Describe what will happen in the lake in spring and autumn
(b) What will be the effect of what happens on (i) the time lag
between air and water temperatures? (ii) the evaporation rate in the
various seasons ?
(c) Will there be a difference if the winter temperature does not
drop below 4°C and if so explain why.
3.5 Calculate for location A (question 3.3) the consumptive use of

water of a crop of tomatoes during a growing season from June to
October if the relevant consumptive use coefficient is 10% less than
that for California.
3.6 Discuss the advantages and disadvantages of evaporation pans

placed above the ground surface (e.g. the U.S. Class A pan) com-
pared to those sunk in the earth.
3. 7 Draw up a water-budget for 100 units of rainfall falling on to a
coniferous forest in a temperate coastal climate. Describe the pro-
cesses involved and indicate the proportions of the rain that becomes
involved in each.
3.8 Describe fully Penman's evaporation theory for open water

surfaces. Show how each parameter used affects the evaporation and
discuss how the theory differs from other evaporation formulae.
3.9 A large reservoir is located in latitude 40°30'N. Compute

monthly and annual lake evaporation for the reservoir from given
data using nomogram of Penman's theory. If the Class A pan
evaporation at the reservoir for the year is 1143 mm, compute the
pan coefficient. Assume the precipitation on the lake is as given and
that the runoff represents unavoidable spillage of this precipitation
during floods, what is the nett annual anticipated loss from the
reservoir per square kilometre of surface in cubic metres per day?
What would the change in evaporation be for the month of July if
the reservoir was at 40°8?
Average
Mean air Dew wind Cloud in Precipita-
temp. point speed tenths lion Runoff
oc oc m/s coverage mm mm
Oct. 14·4 7·8 0·8 5·9 51

Nov. 8·3 1·7 1·3 7·2 99 23
Dec. 3·9 2·2 1·7 9·5 102 43
Jan. 2·2 1·9 2·1 8·7 117 58
Feb. 2·2 1·4 2·2 6·3 91 20
Mar. 4·4 H 1·3 5·1 69 12
Apr. 8·9 3·3 H 3·4 51
May 15 10 0·9 2·6 28
June 20 15·6 0·8 0·2 3
July 23·9 16·7 0·75 0·1 0
Aug. 22·8 17·8 0·7 0·0 0
Sept. 17·8 12·8 0·75 1·5 20 8
Chapter 4--Problems
4.1 Discuss the influence of slope of catchment and rainfall intensity

on infiltration rates under constant rainfall.
Problems 205
4.2 Discuss the influence of forestation and agriculture on ground-
water. Present arguments for and against
(a) livestock rearing
(b) crop growing
(c) forestation
on the catchment of a public water-supply reservoir.
4.3 The table below gives the hourly rainfall of three storms that
gave rise to runoff equivalent of 14, 23 and 18·5 mm respectively.
Determine the lP index for the catchment.
Storm 1 Storm 2 Storm 3

Hour mm mm mm
1 2 4 3
2 6 9 8
3 7 15 11
4 10 12 4
5 5 5 12
6 4 3
7 4
8 2
4.4 Why is the method of subtracting infiltration rates from rainfall

intensities to compute hydrographs of runoff not applicable to large
natural river basins?
4.5 The Antecedent Precipitation Index for a station was 53 mm on

I October; 55 mm rain fell on 5 October, 30 mm on 7 October and
25 mm on 8 October.
Compute the API
(a) for 12 October, if k = 0·85
(b) for same date assuming no rain fell.
4.6 Use the co-axial relationship of Fig. 4. 7 to determine how the

runoff in this river changes seasonally. Assume that during week
number 1 a storm of 5 inches rain lasting 72 hours occurs. Compare
what happens with the effects of the same storm in week number 25,
if the API in each case is 1·5 inches. Suggest which seasons of the
year the weeks are in and explain why there should be a difference in
runoff.
Chapter 6-Problems
6.1 A river gauging gives Q = 4010 m3 fs. The gauging took 3
hours during which the gauge fell 0·15 m. The slope of the river
surface at the gauging site at the time was 80 mm in 500 m, and the
cross-section approximated a shallow rectangle 200 m wide by 11 m
deep. What adjusted value of discharge would you use? What value
of n in Manning's formula results?
6.2 The following discharge observations have been made on a

river. Using Boyer's method adjust the figure for slope variation to
produce a steady flow discharge rating curve for the river.
Gauge height Measured discharge Rise+ or Fall-

ft ft3js x 1000 ft/h
10·4 50
12·2 65
13·9 77
14·3 80
22·3 150
27·3 180 -0.32
28·1 228 +0·80
30·8 256 +0·525
32·6 225 -0.36
35·2 251 -0.355
38·9 338 +0·345
40·3 316 -0·22
40·8 352 +0·18
41·5 333 -0.235
42·2 362
6.3 Explain how observations of river discharge at particular gauge

heights may be corrected, so that they fall on a smooth curve, and
why this is desirable.
A river discharge was measured at Q = 2640 m3 /s. During the
I 00 min of the measurement the gauge height rose from 50·40 to
50·52 m. Level readings upstream and downstream differed by
100 mm in 700 m. The flood wave celerity was 2·2 mfs. Give the
corrected rating curve co-ordinates.
6.4 An unregulated stream provides the following volumes over an

80-day period at a possible reservoir site.
Problems 207
(a) Plot the data in the form of a mass diagram.
(b) Determine average, maximum and minimum flow rates.
(c) What reservoir capacity would be needed to ensure main-
tenance of average flow for these 80 days if the reservoir is full to
start with?
(d) How much water would be wasted in spillage in this case?
Runoff volume Runoff volume Runoff volume

Day m3 x 106 Day m3 x 106 Day m3 x 10&
0 0 32 0·8 64 2·0
2 2·0 34 0·7 66 2·3
4 3·2 36 0·7 68 3·2
6 2·3 38 0·5 70 3·4
8 2·1 40 0·4 72 3·5
10 1·8 42 0·7 74 3·7
12 2·2 44 0·8 76 2·8
14 0·9 46 0·4 78 2·4
16 0·5 48 0·3 80 2·0
18 0·3 50 0·2
20 0·7 52 0·2
22 0·7 54 0·4
24 0·6 56 0·6
26 1·2 58 1·2
28 0·7 60 1·4
30 0·8 62 1·8
6.5 The average domestic per capita demand for water in an ex-
panding community is 0·20 m3fday. Industrial demand is 30% of
total domestic requirements. The town has I 00,000 inhabitants now
and is expected to double its population in future.
Water is supplied from a river system with existing storage capacity
of 107 m3 and whose mean daily discharges for each month of the
year are as follows (thousands of m3)
Jan. Feb. Mar. Apr. May June July Aug. Sep. Oct. Nov. Dec.
290 250 388 150 64·5 50 64·5 117 283 388 317 385
Compensation water of 1·5 m3fs is to be provided constantly.
Find, to a first approximation and for an average year, the addi-
tional storage capacity which will have to be provided if the popula-
tion doubles in size. Also determine the quantity of water spilled to
waste in such a year and compare it with the wastage now. Assume
the existing storage is half full on I January.
6.6 A community of 60,000 people is increasing in size at a rate of
10% per annum. Average demand per head (for all purposes) is
currently 0·20 m3jday and rising at a rate of 5% per annum. The
existing water supply has a safe yield of 0·5 m3fs. A river is to be
used as an additional source of supply. Its mean daily discharges for
each month of the water-year are listed below in thousands of m3.
Allowing for compensation water of 3 m3/s from October-March
inclusive and 5 m3Js from April-September inclusive, determine to a
first approximation the storage capacity required on the river to
ensure the community's water supply 20 years from now, assuming
present trends continue, and that the reservoir would be full at the
end of November
April 220 July 670 October 670 January 300
May 250 August 865 November 530 February 280
June 370 September 1630 December 270 March 280
6. 7 List eight characteristics of drainage basins affecting their

discharge hydrographs.
Chapter 7-Problems
7.1 A catchment area is undergoing a prolonged rainless period.
The discharge of the stream draining it is 100 m3/s after 10 days
without rain, and 50 m3/s after 40 days without rain. Derive the
equation of the depletion curve and estimate the discharge after 120
days without rain.
7.2 Describe how to derive a master depletion curve for a river.

What would you use it for and why?
7.3 The recession limb of a hydrograph, listed below, is to be

divided into runoff and baseflow. Carry out this separation
(a) by finding the point of discontinuity on the recession limb
(b) by finding the depletion curve equation and extrapolating
back in time.
Comment on your results.
Problems 209
Time Flow Time Flow
h m3js h m 3/s
15 41·1 33 10·0
18 35·8 36 8·3
21 25·0 39 7·0
24 19·2 42 5·8
27 15·1 45 4·9
30 12·2 48 4·1
7.4 The hydrograph tabulated below was observed for a river

draining a 40 square miles catchment, following a storm lasting
3 h.
Hour ft 3 /s Hour ft 3 /s Hour ft 3 /s

0 450 24 3500 48 1070
3 5500 27 3000 51 950
6 9000 30 2600 54 840
9 7500 33 2210 57 750
12 6500 36 1890 60 660
15 5600 39 1620 63 590
18 4800 42 1400 66 540
21 4100 45 1220
Separate base-flow from runoff and calculate total runoff volume.

What was the nett rainfall in inches per hour? Comment on the
severity and likely frequency of such a storm in the United Kingdom.
7.5 Write down the three major principles of unit hydrograph

theory iiiustrating their application with sketches.
Given below are three unit hydrographs derived from separate
storms on a small catchment, all of which are believed to have
resulted from 3 h rains. Derive the average unit hydrograph and
confirm its validity if the drainage area is 5·25 square miles.
Hours Storm 1 Storm 2 Storm 3
0 0 0 0
1 165 37 25
2 547 187 87
3 750 537 260
4 585 697 505
5 465 608 660
6 352 457 600
7 262 330 427
8 195 255 322
9 143 195 248
10 97 135 183
11 60 90 135
12 33 52 90
13 15 30 53
14 7 12 24
15 0 0 0
All values in cubic feet per second.
7.6 The 4-hour unit hydrograph for a 550 km 2 catchment is given

below.
A uniform-intensity storm of 4 hours' duration with an intensity
of 6 mm/h is followed after a 2 hour break by a further uniform-
intensity storm of 2 hours' duration and an intensity of II mm/h.
The rain loss is estimated at 1 mm/h on both storms. Base flow was
estimated to be 10 m3fs at the beginning of the first storm and 40 m3Js
at the end of the runoff period of the second storm.
Compute the likely peak discharge and its time of occurrence.
Q Q
Hours m3/s Hours m3fs
0 0 12 62
1 11 13 51
2 71 14 40
3 124 15 31
4 170 16 27
5 198 17 17
6 172 18 11
7 147 19 5
8 127 20 3
9 107 21 0
10 90
11 76
Problems 211
7.7 The 4-hour unit hydrograph for a river-gauging station draining
a catchment area of 554 km 2 , is given below. Make any checks
possible on the validity of the unit graph. Find the probable peak
discharge in the river, at the station from a storm covering the
catchment and consisting of two consecutive 3-hour periods of nett
rain of intensities 12 and 6 mmfh respectively. Assume baseflow
rises linearly during the period of runoff from 30 to 70 m3js.
Time Unit hydrograph Time Unit hydrograph

h m3fs h m3fs
0 0 12 62
1 11 13 51
2 60 14 39
3 120 15 31
4 170 16 23
5 198 17 16
6 184 18 11
7 153 19 6
8 127 20 3
9 107 21 0
10 91
11 76
7.8 A drought is ended over a catchment area of 100 km 2 by uni-

form rain of 36 mm falling for 6 hours. The relevant hydrograph of
the river draining the area is given below, the rain period having been
between hours 3 and 9. Use this data to predict the maximum dis-
charge that might be expected following a 50 mm fall in 3 hours on
the catchment. Qualify the forecast appropriately.
Discharge Discharge
Hours m3fs Hours m3/s
0 3 24 25
3 3 27 21
6 10 30 17
9 25 33 13-5
12 39 36 10·5
15 43 39 8
18 37 42 5·5
21 30·5 45 4
48 3·9
7.9 Using the data and catchment of Question 7.7 find the probable
peak discharge in the river, at the station, from a storm covering the
catchment and consisting of three consecutive 2-hour periods of rain
producing 7, 14 and 12 mm runoff respectively. Assume base flow
rises from 10 m3jsec to 20 m3jsec during the total period of runoff.
Chapter 8-Problems
8.1 A catchment can be divided into ten sub-areas by isochrones in
the manner shown in the table below, the catchment lag TL being
10 hours.
Hour 1 2 3 4 5 6 7 8 9 10
Area in km2 14 30 84 107 121 95 70 55 35 20
A single flood recording is available from which the storage coefficient
K is found as 8 hours. Derive the 2-hour unit hydrograph for the
catchment.
8.2 Tabulated below is the inflow I to a river reach where the storage
constants are K = 10 h and x = 0. Find graphically the outflow
peak in time and magnitude.
What would be the effect of making x > 0?
Time Inflow I Time Inflow I

h m3fs h m3/s
0 28·3 40 90·6
5 26·9 45 70·8
10 24·1 50 53·8
15 62·3 55 42·5
20 133·1 60 34·0
25 172·7 65 28·3
30 152·9 70 24·1
35 121·8
Assume outflow at hour 11 is 28·3 m3 /s and starting to rise.
8.3 A storm over the catchment shown in Fig. 2 generates simul-

taneously at A and B the hydrograph listed below. Use the Mus-
kingum streamflow-routing technique to determine the combined
maximum discharge at C. The travel time for the mass centre of the
flood between A and Cis 9 hours and the factor x = 0·33. Any local
inflow is neglected.
Problems 213
FIG. 2
Hours Q m3/s Hours Qm3/s

0 10 24 91
3 35 27 69
6 96 30 54
9 163 33 41
12 204 36 33
15 210 39 27
18 190 42 24
21 129
8.4 Define the instantaneous unit hydrograph of a catchment area,

and describe how it may be used to derive the n-h unit graph.
A catchment area is 400 km 2 in total and is made up of the sub-
areas bounded by the isochrones tabulated below. From a short-
storm hydrograph it is known that TL = 9 h and the storage
coefficient K = 5·5 h.
Derive the 3 h unitgraph.
Sub-area
bounded by
isochrone Area
h km2
1 15
2 30
3 50
4 75
5 80
6 60
7 45
8 25
9 20
2.14 Engineering Hydrology
8.5 Listed below is the storm inflow hydrograph for a full reservoir
that has an uncontrolled spillway for releasing flood waters. De-
termine the outflow hydrograph for the 48-hour period after the
start of the storm. Assume outflow is I m3fs at time 0. The inflow
hydrograph, and the storage and outflow characteristics of the
reservoir and spillway are tabulated below.
INFLOW HYDROGRAPH
3-h 3-h
intervals m3/s intervals m3/s
0 1·5 12 54
1 156 13 45
2 255 14 40
3 212 15 34
4 184 16 28
5 158 17 23
6 136 18 17
7 116 19 11
8 99 20 8·5
9 85 21 5·5
10 74 22 3·0
11 62
RESERVOIR CHARACTERISTICS
Height above Height above

spillway crest Storage Outflow spillway crest Storage Outflow
m m3 x 106 m3/s m m3 X 106 m3/s
0·2 0·30 1·21 3·0 6·80 70·15
0·4 0·62 3·42 3·2 7·38 77·28
0·6 0·96 6·27 3·4 7·98 84·64
0·8 1-35 9·66 3·6 8·60 92·21
1·0 1·70 13·50 3·8 9·25 100·00
1·2 2·10 17·75 4·0 9·90 108·00
1·4 2·57 22·36 4·2 10·50 116·20
1·6 3·00 27·32 4·4 11·21 124·60
1·8 3·52 32·60 4·6 11·90 133·19
2·0 4·05 38·18 4·8 12·62 141·97
2·2 4·57 44·05 5·0 13-35 150·93
2·4 5·10 50·19 5·2 14·10 160·08
2·6 5·68 56·60 5·4 14·88 169·40
2·8 6·22 63·25
Problems 215
Chapter 9-Problems
9.1 A contractor plans to build a cofferdam in a river subject to
annual flooding. Hydrological records over 30 years indicate a
maximum flood flow of 7800 ms/s and a minimum of 2000 m3fs.
The observed annual maxima plot as a straight line on semi-loga-
rithmic paper where return period is plotted logarithmically.
The cofferdam will be in the river during four consecutive flood
seasons and it is decided to build it sufficiently high to protect
against the 20-year flood.
Evaluate (without plotting) the 20-year flood and determine the
probability of its occurrence during the cofferdam's life.
9.2 The annual precipitation data for Edinburgh is given below for
the years 1948-1963 inclusive. From this data
(i) estimate the maximum annual rainfall that might be ex-
pected in a 20-year period and a 50-year period;
(ii) define the likelihood of the 20-year maximum being equalled
or exceeded in the 9 years since 1963.
Year Precipitation (in.) Year Precipitation (in.)
1948 36·37 1956 28·17

1949 28·01 1957 25·68
1950 28·88 1958 29·51
1951 30·98 1959 18·04
1952 24·41 1960 24·38
1953 23·64 1961 25·29
1954 35·15 1962 25·84
1955 18·08 1963 30·24
9.3 Discuss the methods in common use for plotting the frequency
of flood discharge in rivers: List the separate steps to be taken to
predict, for a particular river, the flood discharge with a probability
of occurrence of 0·005 in any year. Assume 50 years of stage obser-
vations and a number of well-recorded discharge measurements with
simultaneous slope observations.
9.4 The following table lists in order of magnitude the largest

recorded discharge of a river with a drainage area of 12,560 km2.
Date rriljs Date rril/s
1948 29 May 2804 1918 10 June 1495
1948 22 May 2450 1929 24 May 1492
1933 10 June 2305 1943 29 May 1478
1928 26 May 2042 1922 26 May 1476
1932 14 May 2042 1919 23 May 1473
1933 4June 2016 1936 10 April 1433
1917 17 June 1997 1936 5 May 1410
1947 8 May 1980 1923 26 May 1405
1917 30 May 1974 1927 28 April 1314
1921 20May 1974 1939 4 May 1314
1927 8 June 1943 1934 25 April 1300
1928 9 May 1861 1945 6 May 1257
1927 17 May 1818 1935 24 May 1246
1917 15 May 1801 1920 18 May 1235
1938 19 April 1796 1914 18 May 1195
1936 15 May 1790 1931 7 May 1155
1922 6June 1767 1911 13 June 1119
1932 21 May 1762 1940 12 May 1051
1912 20 May 1753 1942 26 May 1051
1938 28 May 1722 1946 6 May 1037
1922 19 May 1716 1926 19 April 1017
1925 20 May 1694 1937 19 May 971
1924 13 May 1668 1944 16 May 969
1917 9 June 1609 1930 25 April 878
1916 19 June 1586 1941 13 May 818
1912 21 May 1563 1915 19 May 799
1918 5 May 1495
The mean of annual floods is 1502 m3fs and the standard deviation
of the annual series is 467 mBfs.
Compute return periods and probabilities for both partial and
annual series. Plot the partial series data on semi-log plotting paper
and the annual series on log-normal and Gumbel probability paper.
Estimate from each the discharge for a flood with a probability of
once in 200 years.
9.5 The annual rainfall in inches for Woodhead Reservoir for the
period of record 1921-1960 is listed below.
The mean and standard deviation for the period are 49·69 in. and
7.08 in. respectively.
Arrange the data in ranking order. Compute return periods and
probability. Plot the data on probability paper.
Problems 217
Year Rainfall Year Rainfall
1921 44·48 1941 46·79

1922 56·25 1942 43·38
1923 65·57 1943 45·87
1924 45·72 1944 59·00
1925 44·78 1945 42·74
1926 48·02 1946 55·39
1927 54·48 1947 41·04
1928 51·53 1948 45·03
1929 48·11 1949 44·11
1930 59·03 1950 52·15
1931 60·59 1951 55·43
1932 47·20 1952 48·87
1933 38·16 1953 43·26
1934 45·38 1954 63·13
1935 54·62 1955 36·46
1936 53.03 1956 56· 57
1937 40·98 1957 51·79
1938 50·88 1958 51·22
1939 49·95 1959 39·58
1940 46·27 1960 60·66
(a) What are the 50-year and 100-year annual rainfalls? How
do these compare with predictions made using Gumbel's theory?
What qualification needs to be made if using the latter results?
(b) What is the probability that the 20-year rainfall will be
exceeded in a 10-, a 20- and a 40-year period?
(c) A certain waterworks plant is to be designed for a useful
life of 50 years. It can tolerate an occasional rainfall year of 70 in.
What is the probability that this amount may occur during the
project's life?
General Bibliography
Chapter 2
1. MAIDENS, A. L.: New Meteorological Office rain-gauges, The Me-
teorological Magazine, Vol. 94, No. 1114, p. 142, May 1965.
2. GooDISON, C. E. and BIRD, L. G.: Telephone interrogation of rain-
gauges, Ibid., p. 144.
3. GREEN, M. J.: Effects of exposure on the catch of rain gauges,
T.P. 67 Wat. Res. Assoc., July 1969.
4. BLEASDALE, A. : Rain gauge networks development and design with
special reference to the United Kingdom, lASH Symposium on
Design of Hydrological Networks, Quebec, 1965.
5. BILHAM, E. G.: The Classification of Heavy Falls of Rain in Short
Periods, H.M.S.O., London, 1962 (republished).
6. A guide for engineers to the design of storm-sewer systems. Road
Res. Lab., Road Note 35, H.M.S.O., 1963.
7. HOLLAND, D. J.: Rain intensity frequency relationships in Britain,
British Rainfall 1961, H.M.S.O., 1967.
8. YARNALL, D. L.: Rainfall intensity-frequency data, U.S. Dept. Agric.,
Misc. pub. 204, Washington D.C., 1935.
9. LINSLEY, R. K. and KoHLER, M. A.: Variations in storm rainfall over
small areas, Trans. Am. Geophys. Union, Vol. 32, p. 245, April1951.
10. HoLLAND, D. J.: The Cardington rainfall experiment, The Meteoro-
logical Magazine, Vol. 96, No. 1140, pp. 193-202, July 1967.
11. YouNG, C. P.: Estimated rainfall for drainage calculations, LR
595, Road Res. Lab., H.M.S.O., 1973.
12. THmssEN, A. H.: Precipitation for large areas, Monthly Weather
Review, Vol. 39, pp. 1082, July 1911.
Further references
Standards for methods and records of hydrologic measurements, Flood
Control Series no. 6, United Nations 1954.
218
LANGBEIN, W. B.: Hydrologic data networks and methods of extrapolat-
ing or extending available hydrologic data, Flood Control Series No. 15,
United Nations 1960.
Guide to Hydrometeorological Practices, U.N. World Met. Org., No. 168,
T.P. 82, Geneva 1965.
Chapter 3
13. PENMAN, H. L.: Natural evaporation from open water, bare soil and
grass, Proc. Roy. Soc. (London), A vol. 193, p. 120, April 1948.
14. THORNTHWAITE, C. W.: An approach towards a rational classification
of climate, Geographical Review, Vol. 38, p. 55, (1948).
I 5. British Rainfall 1939 (and subsequent years), H.M.S.O., London.
16. LAw, F.: The aims of the catchment studies at Stocks Reservoir,
Slaidburn, Yorkshire (unpublished comm. to Pennines Hydro-
logical Group, Inst. Civ. Eng., September 1970).
17. HouK, I. E.: Irrigation Engineering, Vol. 1, Wiley, New York, 1951.
18. OLIVIER, H.: Irrigation and climate, Arnold, London 1961.
19. THORNTHWAITE, C. W.: The moisture factor in climate, Trans. Am.
Soc. Civ. Eng., Vol. 27, No. 1, p. 41, Feb. 1946.
20. BLANEY, H. F. and CRIDDLE, W. D.: Determining water requirements
in irrigated areas from climatological and irrigation data, Div. Irr.
and Wat. Conserv., S.C.S. U.S. Dept. Agric., SCS-TP-96 Washing-
ton D.C., 1950.
21. BLANEY, H. F.: Definitions, methods and research data, A symposium
on the consumptive use of water, Trans. Am. Soc. Civ. Eng., Vol.
117, p. 949, (1952).
22. HARRis, F. S.: The duty of water in Cache Valley, Utah, Utah Agr.
Exp. Sta. Bull., 173, 1920.
23. FoRTIER, SAMUEL: Irrigation requirements of the arid and semi-arid
lands of the Missouri and Arkansas River basins, U.S. Dept. Agr.
Tech. Bull. 26, 1928.
Further references
HoRSFALL, R. A.: Planning irrigation projects, J. Inst. Engrs. Aust., 22,
No. 6, June 1950.
WHITE, W. N.: A method of estimating ground water supplies based on
discharge by plants and evaporation from soil. Results of investigations
in Escalante Valley, Utah, U.S. Geological Survey Water Supply, Paper
659-A, 1932.
HILL, R. A.: Operation and maintenance of irrigation systems, Paper
2980, Trans. Am. Soc. Civ. Eng., 117, p. 77, 1952.
CRIDDLE, W. D.: Consumptive use of water and irrigation requirements,
J. Soil Wat. Conserv., 1953.
CRIDDLE, W. D.: Methods of computing consumptive use of water, Paper
1507, Proc. Amer. Soc. Civ. Eng., 84, Jan. 1958.
ROHWER, CARL: Evaporation from different types of pans, Trans. Am. Soc.
Civ. Eng., Vol. 99, p. 673, 1934.
HicKOX, G. H.: Evaporation from a fr_ee water surface, Trans. Amer. Soc.
Civ. Eng., Vol. 111, Paper 2266, 1946, and discussion by C. Rohwer.
LoWRY, R. and JOHNSON, A. R.: Consumptive use of water for agriculture,
Trans. Amer. Soc. Civ. Engrs., Vol. 107, paper 2158, 1942, and discussion
by Rule, R. E., Foster, E. E., Blaney, H. F. and Davenport, R. W.
FoRTIER, S. and YoUNG, A. A.: Various articles in Bull. U.S. Dep. Agric.
Nos. 1340 (1925), 185 (1930), 200 (1930), 379 (1933).
PENMAN, H. L.: Estimating evaporation, Trans. Amer. Geophys. Union,
Vol. 31, p. 43, Feb. 1956.
Chapter 4
24. NASSIF, S. and WILSON, E. M.: The influence of slope and rain
intensity on runoff and infiltration (to be published).
25. HoRTON, R. E.: The role of infiltration in the hydrologic cycle,
Trans. Am. Geophys. Union, Vol. 14, pp. 443-460, 1933.
26. BOUCHARDEAU, A. and RODIER, J.: Nouvelle methode de determina-
tion de Ia capacite d'absorption en terrains permeables, La Houille
Blanche, No. A, pp. 531-526, July/Aug. 1960.
27. SoR, K. and BERTRAND, A. R.: Effects of rainfall energy on the
permeability of soils, Proc. Am. Soc. Soil Sci., Vol26, No.3, 1962.
28. HORTON, R. E.: Determination of infiltration capacity for large
drainage basins, Trans. Am. Geophys. Union, Vol. 18, p. 371, 1937.
29. SHERMAN, L. K.: Comparison ofF-curves derived by the methods of
Sharp and Holtan and of Sherman and Mayer, Trans. Am. Geophys.
Union, Vol. 24 (2), p. 465, 1943.
30. BUTLER, S. S.: Engineering Hydrology, Prentice-Hall, Englewood
Cliffs, 1957.
31. LINSLEY, R. K., KOHLER, M.A. and PAULHUS, J. L. H.: Hydrology
for Engineers, (p. 162), McGraw-Hill, New York, 1958.
32. Estimated Soil Moisture deficit over Gt. Britain: Explanatory Notes
Meteorological Office, Bracknell (issued twice monthly).
33. PENMAN, H. L.: The dependence of transpiration on weather and soil
conditions, Journal of Soil Science, Vol. 1, p. 74, 1949.
34. GRINDLEY, J.: Estimation of soil moisture deficits, Meteorological
Magazine, Vol. 96, p. 97, 1967.
Further references
BELL, J. P.: Neutron probe practice, Institute of Hydrology Report No.
19, Wallingford, U.K.
HoRTON, R. E.: Analyses of runoff-plot experiments with varying
infiltration capacity, Trans. Amer. Geophys. Union, Pt. IV, p. 693, 1939.
WILM, H. G.: Methods for the measurements of infiltration, Trans. A mer.
GeCJphys. Union, Pt. III, p. 678, 1941.
Chapter 5
WENZEL, L. K.: Methods for determining permeability of water bearing
materials, U.S. Geol. Surv. Water Supply, Paper 887, 1942.
KIRKHAM, DoN: Measurement of the hydraulic conductivity of soil in
place, Symposium on Permeability of Soils, Am. Soc. Testing Materials,
special Tech. Publ. 163, p. 80, 1955.
CHILDS, E. C. and COLLIS-GEORGE, N.: The permeability of porous
materials, Proc. Roy. Soc., A201: p. 392, 1950.
ARONOVICI, V. S.: The mechanical analysis as an index of subsoil perme-
ability, Proc. Am. Soc. Soil Sci., Vol. 11, p. 137, 1947.
ToDD, DAVID K.: Ground Water Hydrology, John Wiley, New York,
1959.
HUISMAN, L.: Groundwater Recovery, Macmillan, London, 1972.
VERRUIJT, A.: Theory of Groundwater Flow, Macmillan, London, 1970.
CEDERGREEN, H. R.: Seepage, -Drainage and Flow Nets, John Wirey, New
York, 1967.
Chapter 6
35. HoswooD, P. H. and BRIDLE, M. K.: A feasibility study and de-
velopment programme for continuous dilution gauging, Institute
of Hydrology, Report No. 6, Wallingford, U.K.
36. ISO/R. 55, 1966, Liquid flow measurement in open channels;
dilution methods for measurement of steady flow, Part 1, constant
rate injection.
37. CoRBETT, DoN M. and others, Stream-Gaging Procedure, Water
Supply Paper 888, U.S. Geol. Survey, Washington D.C., 1943.
38. BoYER, M. C.: Determining Discharge at Gaging Stations affected
by variable slope, Civil Eng., Vol. 9, p. 556, 1939.
39. MITCHELL, W. D.: Stage-Fall-Discharge Relations for Steady flow
in prismatic channels, U.S. Geol. Survey, Water Supply Paper 1164,
Washington D.C., 1954.
40. STEVENS, J. C.: A method of estimating stream discharge from a
limited number of gagings, Eng. News, July 18, 1907.
41. KoELZER, V. A.: Reservoir Hydraulics, Sect. 3, Handbook of
Applied Hydraulics, ed. by Davis and Sorenson, 3rd Edition,
McGraw-Hill, New York, 1969.
42. JENNINGS, A. H.: World's greatest observed point rainfalls, Monthly
Weather Review, Vol. 78, p. 4, Jan. 1950.
Further references
B.S. 3680, Part 3, 1964; Part 4, 1965.
Logarithmic plotting of stage-discharge observations, Tech. Note 3.
Water Resources Board, Reading, 1966.
HoRTON, R. E.: Erosional development of streams and their drainage
basins, Bull. Geo/. Soc. Am., Vol. 56, p. 275, March 1945.
STRAHLER, Statistical analysis of geomorphic research, Journ. of Geol.,
Vol. 62, No. 1, 1964.
AcKERS, P. and HARRISON, A. J. M.: Critical depth flumes for flow
measurement in open channels, Hyd. Res. Paper No. 5, London,
H.M.S.O., 1963.
PARSHALL, R. L.: Measuring water in irrigation channels with Parshall
flumes and small weirs, U.S. Dept. Agric. Circ. 843, 1950.
AcKERS, P.: Flow measurement by weirs and flumes, Int. Conf on Mod.
Dev. in Flow Measurement, Harwell, 1971, Paper No. 3.
WHITE, W. R.: Flat-vee weirs in alluvial channels, Proc. A.S.C.E., 91
HY3, pp. 395-408, March 1971.
WHITE, W. R.: The performance of two dimensional and flat-V triangu-
lar profile weirs, Proc. I.C.E., Supplement (ii), pp. 21-48, 1971.
BuRGESS, J. S. and WHITE, W. R.: Triangular profile (Crump) weir:
two dimensional study of discharge characteristics, Rpt. No. INT 52,
H.R.S. Wallingford, 1952.
HARRISON, A. J. M. and OwEN, M. W.: A new type of structure for
flow measurement in steep streams, Proc. I.C.E., 36, pp. 273-296,
1967.
SMITH, C. D.: Open channel water measurement with the broad-crested
weir, Int. Comm. on Irrig. and Drainage Bull., 1958, pp. 46-51.
Chapter 7
43. SHERMAN, L. K.: Stream flow from rainfall by the unitgraph method,
Eng. News Record, Vol. 108, p. 501, 1932.
44. BERNARD, M.: An approach to determinate stream flow, Trans. Am.
Soc. Civ. Eng., Vol. 100, p. 347, 1935.
45. LINSLEY, R. K., KOHLER, M. A. and PAULHUS, J. L. H., Applied
Hydrology, pp. 448-49, McGraw-Hill, New York, 1949.
46. COLLINS, W. T.: Runoff distribution graphs from precipitation occur-
ring in more than one time unit, Civil Engineering, Vol. 9, No. 9,
p. 559, Sept. 1939.
47. SNYDER, F. F.: Synthetic unitgraphs, Trans. Am. Geophys. Union,
19th Ann. meeting 1938, Pt. 2, p. 447.
48. HURSH, C. R.: Discussion on Report of the committee on absorption
and transpiration, Trans. Am. Geophys. Union, 17th Ann. meeting,
1936, p. 296.
49. SNYDER, F. F.: Discussion on ref. 47.
50. LINSLEY, R. K.: Application of the synthetic unitgraph in the western
mountain States, Trans. Am. Geophys. Union, 24th Ann. Meeting,
1943, Pt. 2, p. 580.
51. TAYLOR, A. B. and ScHWARZ, H. E.: Unit hydrograph lag and peak
flow related to basin characteristics, Trans. Am. Geophys. Union,
Vol. 33, p. 235, 1952.
Further references
BARNES, B. S.: Consistency in unitgraphs, Proc. Am. Soc. Civ. Eng., 85,
HY8, p. 39, Aug. 1959.
MoRRIS, w. V : Conversion of storm rainfall to runoff, Proc. Symposium
No. 1, Spillway Design Floods, N.R.C., Ottawa, p. 172, 1961.
MoRGAN, P. E. and JoHNsoN, S. M.: Analysis of synthetic unitgraph
methods, Proc. Amer. Soc. Civ. Eng., 88 HY5, p. 199, Sept. 1962.
BUIL, J. A.: Unitgraphs for non uniform rainfall distribution, Proc. Am.
Soc. Civ. Eng., 94, HY1, p. 235, Jan. 1968.
Chapter 8
52. McCARTHY, G. T.: The unit hydrograph and flood routing, un-
published paper presented at the Conference of the North Atlantic
Division, Corps of Engineers, U.S. Army, New London, Conn.
June 24, 1938. Printed by U.S. Engr. Office, Providence R.I.
53. CARTER, R. W. and GoDFREY, R. G.: Storage and Flood Routing,
U.S. Geol. Survey Water-Supply, Paper 1543-B, p. 93, (1960).
54. WILSON, W. T.: A graphical flood routing method, Trans. Am.
Geophys. Union, Vol. 21, part 3, p. 893, 1941.
55. KoHLER, M. A.: Mechanical analogs aid graphical flood routing,
J. Hydraulics Div., ASCE 84, April 1958.
56. LAWLER, E. A.: Flood routing, Sec. 25-11, Handbook of Applied
Hydrology, ed. VenTe Chow, McGraw-Hill, New York, 1964.
57. CLARK, C. 0.: Storage and the unit hydrograph, Trans. A mer. Soc.
Civ. Engr., Vol. 110, p. 1419, (1945).
58. O'KELLY, J. J.: The employment of unit hydrographs to determine
the flows of Irish arterial drainage channels, Proc. lnstn. Civ.
Engrs., Pt. III, Vol. 4, p. 365, (1955).
59. NASH, J. E.: Determining runoff from rainfall, Proc. Instn. Civ.
Engrs., Vol. 10, p. 163, (1958).
60. NASH, J. E.: Systematic determination of unit hydrograph para-
meters, Journ. Geophys. Res., Vol. 64, p. 111, (1959).
61. NAsH, J. E.: A unit hydrograph study, with particular reference to
British catchments, Proc. lnstn. Civ. Engrs., Vol. 17, p. 249, (1960).
62. VEN TE CHow, Handbook of Applied Hydrology, Sect. 14, McGraw
Hill,\ New York, (1964).
Chapter 9
63. MORGAN, H. D.: Estimation of design floods in Scotland and Wales,
Paper No. 3, Symposium on River Flood Hydrology, lnstn. Civ.
Engrs., London, 1966.
64. Flow in California streams, Calif. Dept. Public Works, Bull. 5, 1923.
65. HAZEN, A.: Flood Flow, Wiley, New York, 1930.
66. HAZEN, A.: Storage to be provided in impounding reservoirs for
municipal water supply, Trans. A.S.C.E., Vol. 77, p. 1539, 1914.
67. WHIPPLE, G. C.: The element of chance in sanitation, J. Franklin Inst.,
Vol. 182, p. 37, et seq., 1916.
68. GuMBEL, E. J.: On the plotting of flood discharges, Trans. Am.
Geophys. Union, Vol. 24, Pt. 2, p. 699, 1943.
69. GUMBEL, E. J.: Statistical theory of extreme values and some practical
applications, Nat/. Bur. Standards (U.S.) Appl. Math. Ser., 33,
Feb. 1954.
70. PowELL, R. W.: A simple method of estimating flood frequency,
Civil Eng., Vol. 13, p. 105, 1943.
71. Ibid., discussion by E. J. Gumbel, p. 438.
72. YEN TE CHOW and YEVJEVICH, V. M.: Statistical and Probability
Applied Hydrology, ed. Yen Te Chow, McGraw-Hill, New York,
1964.
73. DALRYMPLE, TATE: Flood Frequency Analysis, U.S. Geol. Water
Supply, Paper 1543-A, (1960).
74. PAULHUS, J. L. H. and GILMAN, C. S.: Evaluation of probable maxi-
mum precipitation, Trans. Am. Geophys. Union, 34, p. 701, 1953.
75. Generalised estimates of probable maximum precipitation over the
U.S. east of the 105th meridian, Hydrometeorological Report No.
23, U.S. Weather Bureau, Washington, 1947.
76. Generalised estimates of probable maximum precipitation of the
United States west of the 105th meridian for areas to 400 square
miles and durations to 24 hours. Tech. Paper 38, U.S. Weather
Bureau, Washington, 1960.
77. Manual for depth duration area analysis of storm precipitation, U.S.
Weather Bureau Co-operative Studies Tech. Paper, No. 1, Washing-
ton, 1946.
78. HERSHFIELD, D. M.: Estimating the probable maximum precipitation,
Proc. Am. Soc. Civ. Eng., 87, p. 99, September 1961.
79. Flood Study Report: Institute of Hydrology, 1974 (unpublished at
time of going to press).
80. BLEASDALE, A.: The distribution of exceptionally heavy daily falls of
rain in the United Kingdom, Journ. Inst. Wat. Eng., Vol. 17, p. 45,
Feb. 1963.
81. WIESNER, C. J.: Analysis of Australian storms for depth, duration,
area data, Rain Seminar, Commonwealth Bureau of Meteorology,
Melbourne, 1960.
82. WoLF, P. 0.: Comparison of methods of flood estimation, Symposium
on River Flood Hydrology, Instn. Civ. Engrs., London, 1966 and
discussion by T. O'Donnell.
83. PETERSON, K. R.: A precipitable water nomogram, Bull. A mer. Met.
Soc., 42, p. 199, 1961.
84. SowT, S.: Computation of depth of precipitable water in a column
of air, Mon. Weath. Rev., 67, p. 100, 1939.
85. BINNIE, G. M. and MANSELL-MoULLIN, M.: The estimated probable
maximum storm and flood on the Jhelum River-a tributary of the
Indus, Paper No. 9 Symposium on River Flood Hydrology, Instn.
of Civ. Engrs., London, 1966.
86. Handbook of Meteorology, ed. by Berry, Bollay and Beers, p. 1024,
McGraw-Hill, New York, 1949.
87. PAULHUS, J. L. H.: Indian ocean and Taiwan rainfalls set new
records, Mon. Weather Rev., 93, p. 331, May 1965.
88. BLEASDALE, A.: Private communication to the author, Met. Office,
Bracknell, May 1968.
89. BROOKS, C. E. P. and CARRUTHERS, N.: Handbook of Statistical
methods in meteorology, H.M.S.O., p. 330, London, 1953.
90. CocHRANE, N. J.: Lake Nyasa and the River Shire, Proc. I.C.E.,
Vol. 8, p. 363, 1957.
91. CocHRANE, N. J. : Possible non-random aspects of the availability of
water for crops, Paper No. 3, Conf. Civ. Eng. Problems Overseas
Inst. C.E., London, June 1964.
Further references
LANGBEIN, W. B.: Annual floods and the partial duration flood series,
Trans. Am. Geophys. Union, Vol. 30, p. 879, Dec. 1949.
WIESNER, C. J.: Hydrometeorology and river flood estimation, Proc.
Instn. Civ. Engrs., Vol. 27, p. 153, 1964.
Symposium on Hydrology of Spillway Design by the Task Force on Spill-
way Design Floods of the Committee on Hydrology, Proc. Am. Soc.
Civ. Eng., Vol. 90, HY3, May 1964.
ALEXANDER, G. N.: Some aspects of time series in hydrology, J. Instn.
Engrs. Australia, Vol. 26, pp. 188-198, 1954.
ANDERSON, R. L. : Distribution of the serial correlation coefficient,
Ann. Math. Stat., Vol. 13, pp. 1-13, 1941.
BEARD, L. R.: Simulation of daily streamflow. Proc. Internal. Hydro/.
Symposium, Fort Collins, Colorado, 6-8 Sept. 1967, Vol. 1, pp. 624-
632.
FIERING, M. B.: Streamflow Synthesis, Macmillan, London, 1967.
HANNAN, E. J.: Time Series Analysis, Methuen, London, 1960.
KISIEL, C. C. : Time series analysis of hydrologic data, in Advances in
Hydroscience, Vol. 5, pp. 1-119, ed. Yen Te Chow, Academic Press,
New York, 1969.
MATALAS, N. C.: Time series analysis, Water Resources Res., Vol. 3,
pp. 817-29, 1967.
MATALAs, N. C.: Mathematical assessment of synthetic hydrology,
Water Resources Res., Vol. 3, pp. 937-45, 1967.
MoRAN, P. A. P.: An Introduction to Probability Theory, Clarendon
Press, Oxford, 1968.
O'DONNELL, T.: Computer evaluation of catchment behaviour and
parameters significant in flood hydrology, Symp. on River Flood
Hydrology, Inst. Civ. Eng., 1965.
QVIMPO, R. G.: Stochastic analysis of daily river flows, Proc. Am. Soc.
Civ. Engrs, J. Hydraul. Div., Vol. 94, HY1, pp. 43-57, 1968.
YEVJEVICH, V. M. and JENG, R. I.: Properties of non-homogeneous
hydrologic series, Colorado State University, Hydrology Paper
No. 32, 1969.
Answers to Problems
Chapter 2
2.1 (a) 28·32 mm Hg; (b) 8·50 mm Hg; (c) 19·82 mm Hg or 26·96 mb;
(d) 22·0 oc; (e) 23·7 oc
2.3 (a) 29·99 in; (b) Cubley 29·78 in; (c) Biggin 42·0 in
2.4 39 mm
2.5 29·8 in; 31·2 in; 27·0 in
2.7 (b) about 1951; (c) assuming earlier period correct, increases it from
279 to 330 mm/yr
2.9 (a) 0·136; (b) 7·34 m/s
2.10 (a) once in 4 years; (b) 21·5 x 103 m 3
2.11 at X, 75 mm; at Y, 92 mm
2.12 (a) 19 mm Oxford; (b) 59 mm Kumasi (if equation 2.12 derived
for United Kingdom applies in Ghana)
Chapter 3
3.1 Amsterdam 3·8 mm/day; Seattle 0·3 mmjday
3.2 2·5 mm/day
3.3 (a) April 5·24 em; Nov 1·01 em; (b) Jun 11·67 em; Oct 3·70 em
3.5 17·6 in
3.9 Annual Eo= 950 mm; pan coefficient 0·83; surface loss= 1·32 x
103 m3jkm 2 /day; change = 142 mm less
Chapter 4
4.3 4·1 mm
4.5 (a) 53 mm; (b) 9 mm
Chapter 6
6.1 4085 m3js; n = 0·032
6.3 2560 m 3 /s at 50·46 m
226
Answers to Problems 227
6.4 (b) 0·705, 1·85 and 0·1 million m3jday; (c) 20·5 x 106 m3; (d)
5·0 x 10 6 m 3
6.5 Additional storage 3·65 x 10 6 m 3; future spillage = 11·9 x 10 6
m 3; present spillage= 20·8 x 106m3
6.6 45 x 106 ma
Chapter 7
7.1 8 m 3 /s
7.3 Point N at hour 33
7.4 Point N at 36 h; 1·61 in/h; extremely severe; frequency probably
< once in 100 yr
1.5 peak of u.h. at 710 ft 3/s and 4·2 h
7.6 about 686m 3/sat 9 h
7.7 923 m 3/s at 6 h
7.8 76 m 3/s assuming same <I> index and baseflow of 4 m 3/s
7.9 614 m 3/s at 7 h
Chapter 8
8.1 At 2 h intervals: 0, 6·95, 42·1, 95·6, 126·6, 125·9, 106·3, etc.
8.2 146m3/sat 32 h
8.3 353 m 3/s
8.4 At 3 h intervals: 0, 19·8, 72·7, 100·2, 74·7, etc.
Chapter 9
9.1 7 060 m3js; probability 18·6%
9.2 (i) 37·0 in; 39·3 in; (ii) 37%
9.4 About 3 220 ma;s
9.5 (a) Gumbel P 50 = 68·04 in, P10o = 71·91 in; (b) 0·471; 0·642;
0·871; (c) 0·395
Index
ADIABATIC lapse rate, dry 11 altitude 114

saturated 11, 195 area 109
albedo 34 climate 115
anemometer 11, 13 orientation 110
Angots' radiation flux 39, 40 shape 112
anisotropic soil 73 slope 110
annual maximum mean daily flow stream density 115
of R. Thames at Teddington stream pattern 115
185 catchment lag 171, 173
annual series 182 celerity of flood waves 101, 102
antecedent precipitation index 61 channel precipitation 121
aquifers 69 channel storage 98
confined 70 Chezy's formula 104
flow in 76 cloudiness ratio 39, 41
transmissibility of 122 condensation 18
unconfined or phreatic 77 consumptive use 48
artesian wells 70 arable crops requirements 49,
attenuation of flood waves 160 51
coefficients 50
BANK storage 123 forest 52
baseflow 120, 121 control (of a river reach) 98
separation from runoff 125, Coriolis force 6
127 cumulative mass curve 107
basin lag 149 current meter 93, 94
basin recharge 63 cyclical nature of natural phenom-
boreholes 81 ena 197
pumps for 82
boundary conditions 77 D' ARcY's law 74
density of fresh water 71
CALIFORNIA formula (for Tr) 184 of saline water 71
casing (to wells) 82 depletion curve 120
catastrophic floods 180 depression storage 91, 118
catchment characteristics 109 dew 18
228
Index 229
dew point 8 full series 181
dew ponds 18
dilution gauging 97 GRAPHICAL routing method 168
discharge measurements 93-103 gravel packs 82
distribution graph 135, 136 groundwater 69
double mass curve 29 abstraction of 81
drawdown (of wells) 84-90 flow of 74
drilled wells 81 flow in a confined aquifer 75
Dupuit's assumptions 77 an unconfined aquifer 77
with rainfall 78
EFFECTIVE rain 129 influencing factors 70
effluent streams 123 occurrence of 70
elevation-discharge curves 92 potential gradient of 75
energy budget 37 potential head 75
ephemeral streams 122 velocity in soil pores 75
estimated soil-moisture deficit 61, Gumbel frequency analysis 188
63
evaporation 34 HAZEN's formula (for Tr) 184
empirical formula 37 humidity 7
potential 37 relative 9, 35
from open water 39 hydraulic gradient 75
from pans 47 hydrograph (see also unit hydro-
from soil and turf 44 graph) 106
evapo-transpiration 35 analysis 120
potential 36 baseflow separation 125, 127
events, of random nature 181 components of 120
ranking of 184 from differently shaped catch-
extreme value theory 188 ments 113
instantaneous unit 147, 171
FLOOD routing 154 unit 129
formulae 179 hydrological cycle 2
waves 100 forecasting 178
celerity 101, 102 hydro-meteorology 194
flow duration curve 108
modification by storage 108 INDEPENDENT series 181
flow measuring structures 96 infiltration 54
flow of ground water 74 capacity 55
laminar 71 constants 56
adjustment 98-103 drainage basin analysis 58
fog 18 indices 59, 60
frequency analysis 180 influencing factors 54
frequency plots 184 methods of determining capacity
Gumbel distribution 188 57, 58
log-normal 188 infiltrometer 57, 58
normal probability 186 inflexion point 171
plane coordinates 185 influent stream 123
semi-logarithmic 186 instantaneous unit hydrograph
fronts 12 147, 171
interception 118 perched water table 70
interflow 121, 150 percussion drilling 81
intermittent stream 122 perennial stream 122
inventory of earth's water 3 period of concentration 110, 117
isochrones 174 permeability 73
isohyets 26 coefficient 73
isohyetal maps 26 phreatic surface 69
isovels 95 piezometric surface 70, 76
plotting papers 186, 187, 188
K, STORAGE constant 161, 162 porosity 72
variation of 169 effective 72
precipitation 12
LAPSE rate 10 channel 121
dry adiabatic 11 climatic factors affecting 115
saturated adiabatic 11, 195 convective 12
latent heat of evaporation 8 cyclonic 12
log-normal frequency plot 188 frontal 12
orographic 12
MANNING formula 100 pressure, partial 7
values of n 153 saturation vapour 8
mass curve 106 prism storage 155
in reservoir design 107 probable maximum precipitation
of channel storage 161, 162 (PMP) 195
master depletion curve 125 probability 182, 188
maximum and minimum ther- of N-yr flood in a period 183
mometers 10 plotting 184
maximum probable flood (MPF) psychrometer 9
195 psychrometer constant 10
maximum rainfall depths 116,
197 RADIATION 11
mean daily temperature 10 radiometers 11, 13
monsoon 116 rainfall 11
Muskingum routing method 164 applied to unitgraphs 192
areal extent 18
NEGATIVE baseflow 124 Hilham's classification 19
nett rain 129 in British Rainfall (pub.) 15
normal distribution 188 depth-area 25
normal probability 186 depth-area-time relationship
N-year event 182 23
frequency 18, 21
OuTFLOW lag (in channel routing) intensity-duration 18, 116
169, 170 intensity-duration-frequency
19, 23
PAN, evaporation 47 intensity of 14, 18, 116
coefficients 48 maps of Great Britain and Ire-
partial duration series 182 land 32, 33
partial pressure 7 runoff analysis 58
Penman's theory 39 runoff correlation 117
Index 231
supplementary records 27 standard deviation of 189
trends from progressive averages slope area method (rating curves)
30 105
trends in data 30 snow and ice 17
rain gauges anti-splash screen 16 soil moisture 61
standard 14 antecedent precipitation index
recording 13, 14, 16 61, 62
raingauge networks 15 measurement of 65
rain graph 120 Wallingford soil moisture
random events, nature of 181 probe 65
rating curves 92 soil moisture deficiency (s.m.d.)
adjustment 98-103 61, 63
(Boyer's method) 103 solar radiation 34
extension of 103 snow traverses 17
(Steven's method) 104 specific velocity 75
(slope area method) 105 springs 70
reach (of a river) 154 artesian 70
recession limb 120 stage 92, 96
recurrence interval 182 standard deviation 189
reflection coefficient 41 standard project flood (SPF) 195
relative humidity 9 stomata 36
reservoir lag 160 storage constant, K 161, 163,169
reservoir routing storage curves variation of 169
158 storage equation (evaporation) 36
computation 157, 159 storage equation (routing) 155
return period 182 storage loops 161
Rijkoorts' nomogram for evapora- calculation of 163, 165
tion 44 and facing p. 232 storm duration 21
rotary drilling 81 storm moisture inflow index 196
routing 154 streams 122, 123
period 156 superposition, method of 130
in reservoirs 156 surface runoff 91
in river channels 161 synthetic data generation 191
runoff 91 periodic component 191
correlation with rainfall 118 stochastic component 191
duration 105 synthetic unit hydrographs 148
from flood routing 171
S-CURVE 132
saline water density 71 TEMPERATURE 10
saturation deficit 7 inversion 10
saturation vapour pressure 8 Theissen polygons 26, 27
screens (for wells) 82 Thornthwaite's formula 45
semi-log frequency plot 187 nomogram 46
series of events 180 thunderstorms 116
annual 181 rainfall intensity in 116
full 181 time-area graph 175
independent 181 time of concentration 110
partial duration 182 transpiration 35
UNCONFINED aquifer 77, 79, 85, WADI 122
87, 88 water budget equation 36
uniform intensity storm 117 water divide 109
unit hydrograph or unitgraph 129 water year 181
as a percentage distribution 136 wedge storage 155
average, from a number 137 wells 81, 82
derivation of 137 artesian 70
from complex storms 138 drawdown of 84--90
(Collins) 140 drilled and dug 81
from S-curves 132 flowing 70
instantaneous 147 pumping from 84--90
of various durations 131 confined flow 84
principles of 129 unconfined flow 85, 87
steps in using 192 pumps for 82
synthetic 148, 171 screens for 82
yield of 82
wet bulb temperature 9
VAPOUR removal 39, 42 wind 11
vertical temperature gradient 10 wind run 11
viscosity, absolute 71
kinematic 72 YIELD of wells 82
Additional material from Engineering Hydrology,
ISBN 978-0-333-17443-2,
is available at http://extras.springer.com

Engineering Hydrology by EM Wilson

Uploaded by

Copyright:

Available Formats

Engineering Hydrology by EM Wilson

Uploaded by

Document Information

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

Engineering Hydrology by EM Wilson

Uploaded by

Copyright:

Available Formats

Engineering Hydrology

E. M . WI L S 0 N, Ph.D., M.Sc., F.I.C.E., M.A.S.C.E.

All rights reserved. No part of this publication

Preface to the Second Edition Vl

1.1 Allied sciences 1

2.1 Weather and climate 6

3 Evaporation and Transpiration 34

3.1 Meteorological factors 34

4 Infiltration and Percolation 54

7 H ydrograph Analysis 120

8 Flood Routing 154

9 Hydrological Forecasting 178

Manchester, 1974 E. M. WILSON

The science of hydrology deals with the occurrence and movement of

1.1 Allied Sciences

FIG. 1.1 The hydrological cycle

may seem to be continuous. Thirdly, the intensity and frequency of

1.3 Inventory of earth's water

TABLE 1.1. Estimated earth's water inventory

TOTAL 1,360,000 100·0

Of the 0·6% of total water which is available as fresh water about

1.4 Hydrology as applied in engineering

2.1 Weather and climate

hr = 606·5 - 0·695t g cal/g (2.1)

Flo. 2.1 Saturation vapour pressure of water in air

h = efes, or as a percentage, h = 100 efe8 % (2.2)

and is a measure of the air's capacity, at its existing temperature, to

Vertical temperature gradient. The rate of change of temperature

Distribution of temperature. Generally, the nearer the equator a

2.6.1 Recording precipitation. Precipitation occurs mainly as rain,

FIG. 2.2 Meteorological observation array. On lower arm 1.

2.6.2 Rain-gauge networks. A question which frequently arises is

TABLE 2.2. Minimum numbers of rain gauges required in

mile2 km2 Daily Monthly Total

TABLE 2.3. Minimum numbers of rain gauges for monthly

Source: Bleasdale [4]

There is considerable material devoted to this question of hydro-

2.7 Forms of precipitation other than rain

Dew. Dew collectors have been used in Sweden and Israel to

Condensation. Although fog and dew are condensation effects, con-

2.8 The extension and interpretation of data

2.8.1 Definitions. The total annual amount of rain falling at a point

(i) Intensity. This is a measure of the quantity of rain falling in

2.8.2 Intensity-duration relationship. The greater the intensity of

2.8.3 Intensity-duration-frequency relationships. In 1935 E. G.

1..-i- 1-- ......

FIG. 2.4 Bilham's rainfall classification

and substituting the value of P from (2.9)

i = ~ ((202·3NT)-3·55- 2·54) mm/h (2.11)

Equation (2.11) is the SI form of the Bilham equation used in

0 ?; \ ' ', ' ... .... ............ ................... -

b 1000 1500 2000 2500 3000 °

Recently in the United Kingdom the Institute of Hydrology and

' j',i, .j I'~~~~

2.8.4 Depth-area-time relationships. Precipitation rarely occurs

FIG. 2.8 Five minute rainfall, in inches, to be expected once in

FIG. 2.9 Depth-area curves of rainfall

~=I_ 0·3y'A (2.12)

FIG. 2.10 Correlation between storm t and t* (see ref [11])

Example 2.1. What is the average rainfall intensity over an area of