Lecture Notes ECE 2402: Hydrology I by Dr. J.K.

Mwangi
PRECIPITATION
Precipitation is a general term for all forms of moisture emanating from the clouds and
reaching the earths surface. Usual forms are rainfall, snowfall, frost, hail and dew. Of all these,
the first two contribute significant amounts of water. Others such as fog and dew are only of
interest to meteorologists. Precipitation is important to hydrologists and forms the basis of
many hydrological studies. It becomes an element of hydrology once it reaches the ground and
is considered an input to the hydrologic system. Variations in rainfall distribution cause floods
and droughts. It is expressed as a depth of liquid water on a horizontal surface in a day/month
/year. Precipitation is the sum total of liquid precipitation i.e. rainfall and liquid equivalent of
frozen snow, hail, sleet and freezing rain. Our focus will be on rainfall as it is the major
contributor of moisture in tropical countries like Kenya and the predominant form of
precipitation causing stream flow, especially flood flow in a majority of rivers. In this context,
rainfall will be used synonimously with precipitation.
Rainfall can be classified into:
Light intensity trace to 2.5 mm / hr.
Moderate intensity 2.5 mm/hr. to 7.5 mm / hr
Heavy >7.5 mm / hr.
Formation and occurrence of precipitation: This requires lifting of an air mass in the
atmosphere, so that it cools as it gains height through condensation process. For precipitation to
occur, the following essential requirements must be met:
1) There must be a mechanism to cool the air sufficiently to cause condensation and
growth of droplets.
2) There must be a condensation nuclei around which the clouds will form. These are
normally present in the atmosphere in large numbers.
3) There must be a large scale cooling mechanism for significant precipitation to occur
and which is achieved by lifting of air mass.
These conditions can occur in a relatively short period and may be observed simultaneously.
Water droplets in clouds are formed by nucleation of vapor on aerosols, and then go through
many condensation-evaporation cycles as they circulate in the cloud, until they aggregate into
large enough drops that fall through the cloud base.
Precipitation types
Convective precipitation: Is the precipitation that occurs when the air near the ground is
heated by the earths warm surface. This warm air rises, cools and creates precipitation. During
this period, increasing quantities of water vapor are drawn upwards and the warm moisture
laden air becomes unstable resulting in pronounced vertical currents. Dynamic cooling then
takes place, causing condensation and precipitation. Convective precipitation may be in the
form of light showers or storms of extremely high intensity that change rapidly and is typical of
the tropical regions for example the lake basin.
Orographic precipitation
Results from mechanical lifting of moist horizontal air currents over natural barriers such as
mountains. Precipitation resulting from this type of lifting is often of low intensity unless it is
associated with cyclonic or convective action. Further, since mountains are fixed in direction,
1

Lecture Notes ECE 2402: Hydrology I by Dr. J.K. Mwangi

the resulting precipitation falls on the same general location leading to high rainfall on the
windward side only. The leeward side of the mountain receives less rainfall. It is the most
common form of precipitation in Kenya.

a. Frontal precipitation:
Cyclonic storms and their resulting precipitation are associated with planes (frontal
surfaces) between air masses of different temperatures and moisture contents. Fronts may
be of two types cold and warm depending on weather there is an active or passive ascent of
warm air mass over the cold air mass. Warm front precipitation is normally light to
moderate and is formed in warm air moving gradually upward over a wedge of cold air.
Cold front precipitation on the other hand is generally of showery nature formed in warm
air which is formed upward by an advancing wedge of cold air.
b. Non- frontal
Precipitation occurs when a moving cold air mass meets a stationary moist air mass and the
warm air being lighter, gets lifted up over cold air. At higher altitude this warm air mass
cools down resulting in precipitation. Precipitation is often classified according to the
factors responsible for the lifting of the air mass although under natural conditions various
types of cooling, causing precipitation are often interrelated making it difficult to classify
precipitation as of one particular type or another.
.

Lecture Notes ECE 2402: Hydrology I by Dr. J.K. Mwangi

Rainfall:
Is the total liquid product of precipitation or condensation from the atmosphere as received and
measured in a rain gauge (mm). Rainfall intensity is the amount of rain occurring in unit
interval of time expressed in mm/hr.
Measurement of rainfall:
Is measured on the basis of the vertical depth of water accumulated on a level surface during a
given interval of time; if all the rainfall remained where it fell (no losses occurred) units mm.
Rainfall measurement is necessary:
Agriculture what to plant, where and when to plant, when to harvest etc.
Horticulture - how and when to irrigate and by how much.
Civil Engineering- design of runoff control structures; storm drains, bridges etc.
Scientists - Data for catchment modelling.
Rainfall data is basic to water resources planning and management and can be used to extend
runoff in both time and space. Proper rain gauge network in a watershed is required to collect
adequate data. Measurement is made at the point of fall (point rainfall) using rain gauges.
Non-recording gauges:
Standard Rain Gauge (SRG) used for daily measurement of rainfall. Rainfall caught in the
gauge is measured manually. It consists of a collector to intecept the sample of rainfall and a
receiver consisting of a bottle to collect and store rainfall until it is measured. The collector
extends above ground surface while the receiver is fixed partially below ground level with a
capacity to hold extremes of rainfall likely at a location in 24 hrs. They give only the total
depth of rainfall (not intensity or duration) during different times of the day. In very remote
areas storage gauges (totalizers) may be used to accumulate rainfall caught for a period e.g.
week / month / season before measurement.They are useful in areas of low data demand and
poor accessibility.
Recording (automatic) gauges: These give record of rainfall (integrated gauge).The
mechanical arrangement in it records rainfall for extended periods in the form of a pen trace or
clock driven chart. The plot of cummulative rainfall is called mass curve of rainfall and
indicates the amount of rainfall, its duration and intensity for the given period thus obviating
need for an observer. Examples are tipping bucket, weighing bucket, float type and radio
reporting rain gauges. They are more useful than non-recording gauges and very suitable for
inaccessible hilly areas where it is not possible to visit the station daily. Their installation is
based on consideration of economic feasibility and need for short duration rainfall intensities
normally required in hydraulic design. At least 10% of all rain gauges should be of the
recording type in order to determine rainfall intensities. They should also be installed with
standard raingauges 3m away which to serve as standard for checking and adjusting records of
automatic gauges. They are expensive and mechanical failure may affect rainfall measurement.
Weighing bucket type:
Is used for recording rainfall or snow and consists of a receiver bucket supported by a spring
balance (see below). Catch from the funnel empties into a bucket mounted on a weighing scale.
The weight of the rain falling into the bucket depresses the entire mechanism which movement
is transmitted to a recording pen through levers and links which record the increased weight of
3

Lecture Notes ECE 2402: Hydrology I by Dr. J.K. Mwangi

the bucket and its contents on a chart held by a clock-driven drum and which gives a trace of
accumulated amounts of ranfall with time. The record is in the form of a conventional rainfall
mass curve.The drum rotation sets the time scale whereas the vertical motion of the pen records
the cumulative rainfall. Slope of the mass curve at any point gives intensity of rainfall.

Weighing type rain gauge

Merit: Simple mechanism and can be used to digitise output signal for use in telemetry.
De-merit: Rainfall below the capacity of the bucket may not be recorded.
Tipping Bucket type gauge:
Consists of two small buckets placed below a funnel fitted in a 30cm dia. receiver. The buckets
are balanced in an unstable equilibrium about a horizontal axis such that at any one time only
one bucket remains below the funnel. One bucket is always higher than the other. As the
rainfall is collected by the receiver, it passes through the funnel to the higher bucket. After a
small amount of rainfall (usually 0.25 mm) is received by the higher bucket, it becomes
unstable and tips to empty itself into a measuring tube placed below it. At the same time, the
other bucket comes to the higher position and starts colleting rainfall. The tipping of the
bucket actuates an electric circuit, which causes the pen to make a mark on the chart wrapped
around a drum rotated by a clock-driven mechanism. Therefore each mark on the chart
corresponds to a rainfall of 0.25 mm. By counting the number of marks and noting the time, the
intensity and amount of the rainfall can be determined (see figure below).

Lecture Notes ECE 2402: Hydrology I by Dr. J.K. Mwangi

Tipping bucket type rain gauge

.
Float type rain gauge:
In this type of gauge, precipitation falling onto the receiver passes through the funnel and filter
and is collected in a float chamber. The filter prevents dust, debris, etc, from entering the float
chamber. The float chamber contains a float with a vertical stem fixed over it. A pen is
mounted on top of the vertical stem As the rain is collected in the float chamber, the float
rises and the pen moves on a chart fixed on a drum rotated by a clock-driven mechanism. The
pen marks a trace of the cumulative depth of rainfall on the chart. The clock mechanism rotates
the drum once in 24hrs. Rewinding is needed once every week when the chart fitted round the
drum is replaced. Rainfall record is in the form of the conventional mass curve, from which
intensity and duration of rainfall can be determined (see figure below).

Float type rain gauge

5

Lecture Notes ECE 2402: Hydrology I by Dr. J.K. Mwangi

Automatic radio reporting rain gauge:
This signals rainfall data over radio at regular intervals or on command. Information sent
icludes total amount of rainfall, intensity and wether it is raining at the time of broadcast or not.
Areal rainfall integrators: e.g. radars measure the intensity of a storm and total amount of
rainfall over an area. This technique is suitable for large inaccessible catchments and gives
more accurate results compared with other methods.
Challenges of rainfall measurement:
All rain gauges must extend above the surface of the earth in order to avoid splash from
the surroundings, flooding and accidents. They thus create eddy currents which may
affect the amount of catch.
Few sites are sufficiently sheltered from winds to be able to reduce these effects or are
sufficiently clear of obstructions.
Rainfall measurement is not subject to check by repetition or duplication.
Sample measured is small representing only a small part of rainfall occurring over an
area.
It is difficult to locate sites free from wind effects and clear of obstructions especially in
urban areas
Installation of a SRG: As a single instrument or as part of a hydrometric station.
Amount of rainfall collected depends on location and exposure.
Requirements of a rain gauge site:
Open, away from obstruction. Distance from the nearest obstruction, at least four times
the height of the obstructing object.
Representative of the area for which rainfall measurement is required.
Level ground to ensure gauge mouth is horizontal. Not on slope, terrace, wall etc.
In hilly areas where it is difficult to find a level ground, site should be shielded from
strong winds to avoid formation of eddies.
Secure (normally located in secure public places).
Site that is easily accessible at all times and weather conditions.
Common errors in rainfall measurement:
Initial catch (about 0.25mm) used to moisten the funnel and inside surfaces.
Errors due to natural causes (e.g. evaporation, turbulence etc.).
Gauge errors (e.g. leaks, inclined gauge mouth etc.)
Observer errors (improper reading, spillage etc.)
Proper training is required to ensure uniformity of procedure to facilitate comparison.
Design and exposure of rain gauges has a profound effect on the amount of catch. Both factors
must therefore be considered when obtaining a representative sample by minimizing errors due
to the gauge itself and presence of obstructions near the gauge mouth. Rainfall depth is
measured, by pouring it in a measuring cylinder, calibrated to indicate water depth. Where
rainfall is measured may times in a day the total is summed to obtain total rainfall for the day.

Lecture Notes ECE 2402: Hydrology I by Dr. J.K. Mwangi

Example
30% of RG collectors area is covered during a storm event by a plant leaf. If the total amount
of rain recorded was 25mm using a dip rod in a cylinder whose area is one-tenth the area of the
collector, compute the correct point rainfall in mm.
Solution
SRG collector diameter = 127mm (standard)
Area of collector = (d2/4 = 3.14*1272) /4 = 12667.7mm2
Effective area = 70% = 8867.4 mm2 (The rest is obstructed)
Correct rainfall estimate is directly proportional to collectors area
Collectors area of 8867.4mm2 gives 25mm,
Collectors area of 12667.7 mm2 assuming no obstruction will give 35.71mm

Plotter pen

Graph paper rolled around

a rotating Drum

Components of a recording gauge type gauge

Analysis and interpretation of rainfall data:
Precipitation process is random in nature hence it is difficult to predict with certainty rainfall
for any given period in future. Rainfall magnitudes are therefore estimated with some
probability attached to them. Analysis of rainfall data collected over long periods helps
hydrologists to make reasonable probabilistic estimates of rainfall for use in designs. Methods
of analysis used and aspects analyzed depend on the purpose of data.
Frequency of point rainfall:
In many engineering applications such as those concerned with floods, probability of
occurrence of a particular extreme flood is important. Such information is normally obtained
from frequency analysis of rainfall data. Rainfall at a place is a random hydrologic process and
rainfall data when arranged in a chronological order constitutes a time series (annual/monthly).

Lecture Notes ECE 2402: Hydrology I by Dr. J.K. Mwangi

Components of a Standard Rain Gauge

Recurence interval is the average number of years within which a given rainfall event can be
equalled or exceeded. Frequency of occurrence of amaximum and minimum rainfall is
determined to predict future rainfall trends for the design of river protection and drainage
works. Methods of determining reccurrence intervals include:
N
( N 1)
Hazen williams method: T
, Weibulls method: T
and Gumbels
m 0.5
m
N
method T
where in all cases T= recurrence interval, N = number of years of
m C 1
record, m = serial number of a particular event arranged in descending order so that the
maximum rainfall is at the top to find out maximum rainfall value and in ascending order to
determine minimum rainfall value. C is Gumbels correction normally given in tables.
Gumbles method has the best theoretical basis and is commonly used.
The purpose of frequency analysis is to determine the relationship between the magnitude of an
event and its probability of exceedence. The probability P of an event being equalled or
exceeeded is given by:
m
P=
called plotting position.
N 1
Recurrence interval T= 1/P.
8

Lecture Notes ECE 2402: Hydrology I by Dr. J.K. Mwangi

Example:
Table 1 shows the recorded annual 24hr. maximum rainfall. Estimate the maximum rainfall
with return periods of 13 and 50 years.
Table 1.
Year
Rainfall (cm)
61
7.8

62
9.0

50
13.0

51
12.0
63
10.2

52
7.6
64
8.5

53
14.3
65
7.5

54
16.0

55
9.6

66
6.0

67
8.4

56
8.0

57
12.5
68
10.8

58
11.2
69
10.6

59
8.9
70
8.3

60
8.9
71
9.5

Solution
Arrange the data in a descending order and calculate probability and reccurrence
intervals
Plot the graph of rainfall magnitude and return period on a semi-log paper and draw a
smooth curve through the points. This curve (line) can be extended to obtain data
outside the original data (13years = 14.55cm and 50years = 18.00cm)
Average rainfall depth over a basin:
Rainfall recorded by a gauge represents rainfall at that particular point. Where there is only one
gauge station the recorded rainfall is taken to apply to the entire basin. Where more than one
gauge station exists, average rainfall is obtained through computational methods. If 30 or more
years of record are available, normal (monthly or annual) rainfall and standard deviation can
be computed which shows rainfall variability. If rainfall is less than normal rainfall, then the
year is called a dry year and if more a wet year. Annual rainfall data can be presented in
chronological charts, bar charts or ordinate graphs. Normal rainfalls values are revised every
year by deleting the oldest 10 years data from record and adding the most recent 10 years data.
Mean rainfall
This is the average (representative) rainfall at a place. It is determined by averaging the total
rainfall of several consecutive years. Since annual rainfall at a station varies over the years, a
record number of years are required to get a correct estimate. Similarly, mean monthly rainfall
at a place is determined by averaging the monthly total rainfall for several consecutive years.
Example:
A rainfall station recorded the following annual values from 1970 to 1988:
520,615,420,270,305,380,705,600,350,550,560,400,520,435,395,290,430,1020 and 900. Find
the mean, standard deviation and plot the data in form of a chronological and bar chart. What is
the significance of standard deviation? (Mean 508.68mm, std 198.55mm).
In many hydrological studies, average rainfall depth over an area resulting from a storm is often
required. For accurate determination of average rainfall over an area, a large number of rain
gauges are normally required. Design of a proper rain gauge network is required in order to
collect a representative sample of rainfall over an area.

Lecture Notes ECE 2402: Hydrology I by Dr. J.K. Mwangi

A typical bar chart for three rainfall stations (showing monthly values)

Reliability of data measured at one station in representing average rainfall depth over a large
area is a function of:

Distance from the gauge to the center of the area being represented.
Size of the area under consideration
Topography of the area
Nature of the rainfall e.g. local storm characteristics.

In using rainfall data for runoff estimation in large basins, it is required to estimate average
depth of rainfall for different rainfall durations. Methods of estimating areal rainfall from point
rainfall include:
Arithmetic mean method:
In flat areas where rain gauges are uniformly distributed, rainfall of individual stations do not
show much variation from the mean rainfall for the area. In such case, average depth of rainfall
(Pav) over the area is taken as the arithmetic mean of rainfall depths of all stations and is
obtained by dividing the sum of the depths of rainfall recorded at all rain gauge stations by the
number of stations. Thus

Pav

P1 P2 .......... .. Pn
N

It is the simplest method and is based on the assumption that all rainfall values are of equal
importance. It gives good results in flat areas, if gauges are uniformly distributed and if rainfall
values at the stations do not vary widely from the mean. It is quick and can be easily adapted to
computer application. However, it gives inaccurate results and hence is rarely used in practice.

10

Lecture Notes ECE 2402: Hydrology I by Dr. J.K. Mwangi

Thiessen Polygon method:
In this method, rainfall recorded at each rain gauge station is given a weight depending on the
area it represents. The method assumes a linear variation of rainfall between stations and that
the gauge at any station can be used for the proportion of drainage area nearest to it. It is
suitable where rain gauges are few compared to basin size and is more accurate than the
arithmetic mean method which gives equal weightage to all stations. The method involves the
following procedure:
1. The positions of the rain gauge stations are marked on the plan of the catchment area
over which average rainfall depth is required.
2. Adjacent rain gauge stations are then joined by straight lines; so that the entire area is
divided into a series of triangles (Fig 6). Rain gauge stations outside the catchment but
in its neighbourhood are also considered.
3. Perpendicular bisectors are then drawn on the connecting lines to form polygons around
the stations with each polygon containing only one rain gauge station. The entire area of
a particular polygon is nearer to the rain gauge station contained therein than to any
other rain gauge station. Thus each polygon represents the area of the influence of that
rain gauge station.
4. For determination of the average depth of rainfall, the boundary of the catchment is
taken as the outer limit of the Thiessen polygons. The areas of the polygons are
determined either with a planimeter or with an overlay grid.
Average percipitation of the catchment is then given by:

Pav =

P1 A1 P2 A2 Pn An
A1 A2 A3 An

where A1,A2,.........An are the areas of the Thiessen polygons representing the stations while
1,2,....... n and P1, P2,......Pn are the corresponding precipitations. A is the total catchment area.
The procedure ignores orographic influences and hence is not suitable for mountainous regions.
The method is fixed for a given gauge configuration and polygons must be reconstructed a
fresh if any gauges are relocated / added which is a major limitation. The method makes use of
rainfall stations located a short distance from the basin boundary although their influence
A
diminishes as the distance from the boundary increases. The ratio 1 is known as Weightage
A
Factor or Thiessen constant and once determined, the computation of average rainfall for
different storms becomes easy until the number or station configuration changes. No
adjustment can be made for variations due to altitudes and other factors which is a limitation of
the method.

11

Lecture Notes ECE 2402: Hydrology I by Dr. J.K. Mwangi

Figure 6: Thiessen polygon method

Isohyetal method: Also a weighted average method is the most accurate method and takes into
consideration orographic effect of rainfall. Weighting is done by averaging rainfall between
two consecutive isohyets and multiplying by the area enclosed between them.The weightings
are then summed up for the whole basin and divided by the total basin area to give average
rainfall depth (fig 7). For the area between the last isohyet and catchment boundary, a suitable
value of average depth is taken, depending on the distance of the outside isohyet and actual
basin boundary. To ensure accuracy, isohyetal interval should be kept small. If P1, P2, . . .Pn are
the values of isohyets and A1, A2, . . .An-1 are the inter-isohyetal areas respectively, then
average precipitation Pav over the catchment is given by:

P P3
P Pn
P P2
A1 1
... . An n1

A2 2
2
2
2

Pav
A1 A2 ... An1
The method takes into account actual rainfall distribution patterns but its accuracy depends on
the skill of the analyst in plotting the isohyets. The method is slow, laborious and cannot be
easily adopted to computer processing. The choice of which method to use in a particular
situation depends on:
Network of rain gauge stations in the basin
Basin size and its topography
Accuracy required
It is the most accurate of all as all relevant data is used and properly interpreted.The method
can be used to make adjustments for variations in station altitudes due to orographic influences.

12

Lecture Notes ECE 2402: Hydrology I by Dr. J.K. Mwangi

Procedure:
Locate all rainfall stations on a base map and record rainfall amounts.
Draw isohyets (lines of equal rainfall) by proportioning the distances between adjacent
gauges according to differences in rainfall.
Calculate the mean precipitation for the area corresponding to each isohyet.
Calculate the fraction of catchment area under each isohyet, multiply by the mean
precipitation for that area and sum to get average for the catchment.
An isohyet is a line joining points of equal rainfall.

Figure 7: Isohyetal method

Figure 7: Isohyetal method

Isohyets
(Upper)
38
37
36
35
34
Total

Isohyets
(lower)

A
Mean rainfall on area

B
A*B
Area between Weighted
isohyets (ha)
mean rainfall
37
37.5
130
4875
36
36.5
150
5475
35
35.5
300
10650
34
34.5
450
15525
33
33.5
200
6700
1230
43,225
Mean rainfall 43225/1230 = 35.14 mm

Example
If a dam is built at a catchment outfall and a statutory minimum discharge of 0.1m3/s is
maintained throughout the year in the river downstream, assess the available water supply per
year in m3. (Assume the drainage basin is water tight and evaporation loss is 400mm / year).

13

Lecture Notes ECE 2402: Hydrology I by Dr. J.K. Mwangi

Rainfall measurements within the basin as shown in Table 1.
Table 1:
Rain gauge No.
1
2
3
4
Rainfall (mm)
2052 1915
1969
1723
Thiessen Polygon area (km2)
7.8
8.3
10.2
11.5

5
1640
5.4

6
1510
6.8

Solution
Area of basin = summation of all areas = 50km2
Assessment is by subtracting outputs from inputs:
Ai hi = 1798mm
Input (rainfall): Areal mean depth =
Ai
Volume per year = Area x mean depth = 89.9 x106 m3
Outputs: Minimum downstream release per year = 0.1x60x60x24x365 = 3.1536x106 m3
Evaporation loss per year
= 0.4x50x106
= 20x106 m3
Available water supply = input losses = 89.9 x106 m3- 3.1536x106 m3- 20x106 m3
= 66.7464x106m3
Losses from precipitation include:
(1) interception
(2) depression storage
(3) evaporation
(4) infiltration
(5) transpiration
Depression storage and interception are called initial losses.
Network of rain gauge stations:
The most common method of measuring rainfall is through a network of rain gauge stations.
Rain gauges stations should be adequate in order to give a good representation of rainfall
intensity and duration of an area because gauge catch area is small compared to areal extent of
storms.To get a representative picture of a storm over a basin, gauges should be as many as
possible and evenly distributed within the area. However, their number is often restricted by
topography, economic considerations, accessibility etc. The optimum density depends on
intended use of data. In order to get reasonably accurate rainfall information, the World
Meteorological Organization (WMO) has recommended the following rain gauge densities.

In flat regions of temperate, Mediterranean and Tropical zones

Ideal 1 station for 600 900km2
Acceptable -1 station for 900 3000km2
In mountainous regions of temperate, Mediterranean and Tropical zones
Ideal 1 station for 100 250km2. Acceptable 1 station for 25 1000km2
In arid and polar zones: 1 station for 1,500 10,000km2 depending on the feasibility.
The network density obtained using these recommendations is such that all gauges have almost
equal Thiessen weights. Ideally 10% of all rain gauge stations should be equipped with
automatic gauges for measurement of rainfall intensities.

14

Lecture Notes ECE 2402: Hydrology I by Dr. J.K. Mwangi

Rain Gauge Network
The ratio of catchment area to the total number of rain gauges (network density) indicates the
average area served by each rain gauge. A network should give a representative picture of the
areal distribution of rainfall. Network density varies from country to country and may vary
from region to region within the same country (compare Tana and Uaso Nyiro basins).
The optimal number of rain gauge stations required in order to achieve an assigned percentage
error in the estimation of mean annual rainfall is obtained statistically as:
2

C
N v Where N = optimal number of rain gauges required.

= degree of error allowable in the estimation of mean annual rainfall (varies)

Cv = Coefficient of variation of rainfall values at the existing m stations (%)
If there are m stations in the catchment each recording rainfall values P1, P2, P3.............. Pm in a
known time, the coefficient of variation Cv is calculated as

Cv

100 * m1

where m1 is the standard deviation, P is the mean annual rainfall while

P
is normally taken as equal to 10%
=

If N>m, then additional stations are required and these should be placed such that together with
existing gauges they are evenly distributed over the catchment.
Example:
A catchment with 6 rain gauges has annual rainfall recorded as given in table 2:
Table 2:
Station
Rainfall (cm)

A
82.6

B
102.90

C
180.3

D
110.3

E
98.8

F
136.7

Calculate the optimum number of rain gauge stations required in the catchment if only 10%
error is allowed in the estimation of mean annual rainfall.
Solution
N = 9, therefore 3 additional gauges are required.
Data preparation
Rainfall data should be interpreted correctly to avoid erroneous conclusions. This may not
always be possible due to changes in gauge site, gauge surroundings e.g. construction of new
buildings, growth of trees or changes in rainfall measurement techniques. Rainfall data should
therefore be checked for continuity and consistency before use.
Consistency of data relates to the type, measurement technique, sampling interval and manner
of processing. Inconsistency is a change in the amount of systematic error associated with data
15

Lecture Notes ECE 2402: Hydrology I by Dr. J.K. Mwangi

recording and occurs if the conditions relevant to the recording of a raingauge station have
undergone significant changes during the period of record.
Common causes of inconsistency in rainfall record include:
(i) Shifting of a rain gauge station to a new location
(ii) Neighbourhood of the station undergoing a marked change
(iii) Change in ecosystem due to forest fires, land slides etc.
(iv) Occurrence of observational error from a certain data.
(v) Use of different instruments and observation techniques
Check for consistency of a record is done using the double-mass technique.
Continuity: Rainfall records suffer breaks because of non-observation of rainfall by the reader
or equipment failure. However, the missing record can be approximated with reference to the
stations close to and evenly spaced around the station with missing records. In these
calculations, normal rainfall is used as a standard of comparison.
Presentation of rainfall data:
Rainfall data can presented in the form of a mass curve or hyetograph. A hyetograph is a bar
diagram of average rainfall intensity plotted against time. It is obtained from mass curve of
rainfall. To plot a hyetograph, a convenient time interval is chosen and the corresponding
reading of accumulated rainfall noted from the rainfall mass curve. It gives a representation of
storm characteristics when developing design storms for predicting extreme floods.Time
interval chosen depends on purpose; with small durations used in urban drainage problems and
long durations in flood flow computations in large catchments.
Hyetograph of a storm

Total depth of rainfall = 10.6 mm

Duration of rainfall 46 hours

Intensity, cm/hr

0.5
0.4
0.3
0.2
0.1
0
08

8 16

16 24

24 32

32 40

40 48

Time, hours

Figure 8: A storm Hyetograph

Area under the hytegraph represents the total rainfall received in the period.
A rainfall mass curve is a plot of accumulated rainfall depth at a station against time in a
chronological order. Recording gauges give mass curves directly. In non-recording gauges it is
plotted from knowing the beginning and end of a storm and distributing the total rainfall depth
over the various periods according to mass rainfall curves of nearby recording gauges. Mass
curves are useful for determining storm magnitudes and durations. Slopes of the mass curve
give intensities at various times (fig. 9).
16

Lecture Notes ECE 2402: Hydrology I by Dr. J.K. Mwangi

Figure 9: Rainfall mass curve showing determination of intensity.

Interpolation of rainfall records of missing data:
Frequently rainfall records may be missing for certain periods. To fill the missing data a
number of mathematical approaches are used which require at least three concurrent records of
nearby stations and mean annual rainfall for all the four stations. The methods include:
(1) Arithmetic average method- if the normal annual rainfall at each of the index stations is
within 10% of that for the station with missing record, then a simple arithmetic average of
precipitation at the three stations provides the estimated amount as:

PX

1 n
Pi
n i 1

where

PX is the missing precipitation value for station X

P1, P2 Pn are precipitation values at adjacent stations for the same period and n is the number
of nearby stations.
(2) Normal ratio methodWhen there is a short break in precipitation record of a rain rain
gauge station, it can be estimated from observed data of three adjoining index stations A, B and
C, evenly distributed around the station with missing data (X).
The following two cases can arise and are dealt with separately.
(a) When the mean annual rainfall at each of the index stations A,B and C is within 10% of the
mean annual rainfall of station X, a simple average of the rainfall values of the index
stations is taken, thus:
1
PX PA PB PC
3
17

Lecture Notes ECE 2402: Hydrology I by Dr. J.K. Mwangi

(b) When the mean annual rainfall at each of the index stations differs from that of
station X by more than 10%, the normal ratio method is used. Thus

PX

NX
3

PA
P
P
B C where N is the mean annual rainfall and P is the rainfall.

N A N B NC

When there are M index stations:

PX

NX
3

PA
PB
PM

NM
N A NB

A minimum of three adjoining stations evenly distributed around the station is required.
Example 1:
Compute the storm rainfall at station A from the following data:
Station
Daily rainfall (mm)
Normal annual rainfall (mm)

A
?
1150

1
125
1250

2
145
1450

3
169
1300

Solution:
First compute maximum normal annual rainfall departure from the value at A. (14501150)/1150=26% which is more than 10%. Normal ratio method is applicable. Substitute
NA=1150, P1=125, N1=1250, P2=145, N2=1450, P3=169 and N3=1300 and solve for PA.
Example 2:
Rain gauge X malfunctioned for part of a month during which a storm occurred with rainfall
values of 84, 70 and 96mm at three nearby stations A, B and C. The normal annual rainfalls at
stations X, A, B and C are respectively 770, 882,736 and 944 mm. Estimate the missing storm
rainfall at station X.
1
3

770
770
770
1
882 * 84 736 * 70 944 * 96 = 3 73.333 73.234 78.305 = 75mm

(3) Comparison method- If the rainfall record of a rain gauge station (X) is missing for a
relatively long period e.g. a month or a year, it can be estimated by comparing the mean annual
rainfall of the station X with that of an adjoining station A.

Where PX and PA are the precipitations of stations X and A for the missing period and
NX and NA are the mean annual rainfalls of the stations X and A.
Isohyetal map method
Isohyets are lines of equal rainfall depth. The method is suitable for the estimation of missing
data of a station X due to a particular storm. An isohyetal map is prepared from data of the
various rain gauge stations and the precipitation of station X estimated from the two isohyets
between which the station lies. In Figure 10, precipitation of station X is estimated as 5.3 cm.
18

Lecture Notes ECE 2402: Hydrology I by Dr. J.K. Mwangi

Figure 10: Isohyetal method of estimating missing station rainfall

Testing and adjusting precipitation records:
Double mass curve analysis tests consistency and accuracy of record at a station suspected to
contain discrepancies. This is by comparing its accumulated annual rainfall with concurrent
accumulated values of mean rainfall for a group of neighbouring stations. A change due to
meteorological causes would not cause a change in slope as all base stations would be
similarly affected. Records that deviate from the acceptable trend on a double mass curve are
corrected by multiplying them with a correction factor (ratio of the slopes of the two lines).
This ensures that the data is reasonably homogeneous throughout its length and is related to a
known station with reliable data which enables estimation of data for missing periods and
extrapolation of data beyond the period of record.

Figure 10: An example of inconsistent rainfall record

From the calculated slopes S0 and Sc from the plotted graph, we can write
19

Lecture Notes ECE 2402: Hydrology I by Dr. J.K. Mwangi

S
Pc P0 C Where
S0
Pc and P0 are the corrected and original rainfalls at the suspected station at any time while Sc
and S0 are the corrected and original slopes of the double mass-curve.
Example:
The annual rainfall at station X and the average annual rainfall values at 25 surrounding base
stations are as given below in cm.
Year
1941
1942
1943
1944
1945
1946
1947
1948
1949
1950
1951
1952
1953
1954
1955
1956
1957
1958
1959

Rainfall at
station X
163
119
121
129
126
120
153
172
127
108
126
190
112
97
86
111
68
88
112

Average rainfall
of base stations
135
111
124
111
123
90
138
119
108
107
111
142
112
99
93
131
92
142
123

Year
1960
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
1972
1973
1974
1975
1976

Rainfall at
station X
95
106
81
116
112
80
88
85
90
120
72
113
82
116
112
73
74

Average rainfall
of base stations
142
92
91
131
104
97
111
114
92
146
93
138
112
117
152
90
104

Check the consistency of data at station X and when regime change occurred.
Compute the mean annual rainfall at station X at its present location for the first 36 yrs
without adjustment and then with data adjusted for change in regime.
Compute the adjusted annual rainfall values at station X for the affected period.
Solution:
Year Annual rainfall
at station X
1976
74
1975
73
1974
122
1973
116
1972
82
1971
113
1970
72

Cumulative rainfall
at station X
74
147
269
385
467
580
652
20

Average annual rainfall

for base stations
104
90
152
117
112
138
93

Cumulative rainfall
of base stations
104
194
346
463
575
713
806

Lecture Notes ECE 2402: Hydrology I by Dr. J.K. Mwangi

1969
1968
1967
1966
1965
1964
1963
1962
1961
1960
1959
1958
1957
1956
1955
1954
1953
1952
1951
1950
1949
1948
1947
1946
1945
1944
1943
1942
1941

120
90
85
88
80
112
116
81
106
95
112
88
68
111
86
97
112
190
126
108
127
172
153
120
126
129
121
119
163

772
862
947
1035
1115
1227
1343
1424
1530
1625
1737
1825
1893
2004
2090
2187
2299
2489
2615
2723
2850
3022
3175
3295
3421
3550
3671
3790
3953

146
92
114
111
97
104
131
91
92
142
123
142
92
131
93
99
112
142
111
107
108
119
138
90
123
111
124
111
135

952
1044
1158
1269
1366
1470
1601
1692
1784
1926
2049
2191
2283
2414
2507
2606
2718
2860
2971
3078
3186
3305
3443
3533
3656
3767
3891
4002
4137

The slope of the curve is not uniform so the record at station X is not consistent
The break in slope occurs in 1953. Data prior to 1953 therefore requires adjustment.
Total rainfall for 36 years at station X before adjustment = 3953. Mean = 109.8cm
From the curve cumulative rainfall at station x for 36 years after correction = 3480cm.
New mean = 96.7cm
Slope of adjusted mass curve (1953-1941)=0.85. That of unadjusted mass curve = 1.17
Correction factor for adjustment = 0.85/1.17 = 0.7265. This is used to give adjusted
rainfall values at station X for the period 1941- 1945 as an example.
Year
1941
1942
1943
1944
1945
1946

Observed rainfall at station X (P)

163
119
121
129
126
120
21

Adjusted rainfall at station X (p)

118.4
86.5
87.9
93.7
91.5
87.2

Lecture Notes ECE 2402: Hydrology I by Dr. J.K. Mwangi

1947
1948
1949
1950
1951
1952
1953

153
172
127
108
126
190
112

111.2
125.0
--------------------

Figure 11: An example of a double mass curve application

The corrected values can also be read from the extended mass curve.
Exercise:
Check the consistency of precipitation data for station E by the double mass curve technique.
Year
1926
1927
1928
1929
1930
1931
1932
1933
1934
1935
1936
1937

A
99.4
98.9
105.3
103.5
78.9
138.9
120.3
99.9
113.5
112.2
81.6
114.7

B
114.3
96.3
120.7
86.5
112.8
133.2
100.2
79.3
104.2
120.3
98.7
110.3

C
76.7
102.5
101.1
81.2
90.5
155.6
119.9
81.9
90.3
76.8
88.5
97.9
22

D
93.4
77.2
105.0
99.8
90.8
91.5
96.5
67.2
81.0
103.90
78.3
110.4

E
82.1
70.2
83.8
74.0
59.4
96.0
115.6
75.9
117.0
95.2
107.1
99.8

Lecture Notes ECE 2402: Hydrology I by Dr. J.K. Mwangi

1938
1939
1940
1941
1942

115.1
124.4
118.1
91.7
114.7

97.4
101.0
124.2
79.8
95.4

108.2
124.6
119.7
80.4
131.0

126.6
102.7
97.7
86.1
118.3

126.7
117.1
266.3
86.0
119.0

Intensity Duration Analysis

Intense storms normally last for very short durations and as the storm duration increases the
maximum average intensity of the storm also decreases. Observed maximum intensities plotted
against respective durations gives an intensity duration graph.

Figure 12: A typical Intensity-Duration-Frequency curve

Extreme rainfall events often result in water quality degradation, property damage and loss of
life due to flooding. Statistics of historic rainfall events (intensity,duration and return period)
are used to design storm water management facilities, erosion and flood control structures,
flood protection structures and many other CE structures. If data from a recording gauge is
available for a period of 30-50 years, the frequency of occurrence of given maximum
intensities can be determined and IDF curves developed. An IDF curve gives the probability of
a given rainfall intensity and duration expected to occur at a particular location.
Although standards have been developed for designing infrastructures based on IDF curves,
revision and updating of IDF curves is often necessary in face of climate change. This is
because changes in extreme rainfall events require revision of design standards for CE
structures and/or reconstruction and/or upgrade of existing ones. For example a dam designed
to control a 100-year flood event will provide a significantly lower level of protection if
intensity and duration of the 100-year flood event increases.
Computation of maximum depth or intensity of rainfall for a chosen duration: Maximum
rainfall depth / intensity recorded in a given time interval in a storm is found by computing a
series of running totals of rainfall depths for that time interval starting at various points in the
storm then selecting the maximum value from this series. Computation of maximum rainfall
depth and intensity performed in this way gives an index of how severe a storm is, compared to
other storms recorded at the same location and provide useful data for design.
23

Lecture Notes ECE 2402: Hydrology I by Dr. J.K. Mwangi

Example:
The table below shows rainfall record for a particular storm at station Y.
Calculate the maximum rainfall intensities in 10 and 20 minute time intervals in mm/hr.
Time (min)
Rainfall (mm)
Solution:
Time
0
5
10
15
20
25
30
35
40

0
-

5
3.6

10
10

Rainfall Cumulative (0 min)

0.0
3.6
3.6
10.0
13.6
12.4
26.0
11.0
37.0
10.6
47.6
8.0
55.6
6.0
61.6
1.6
63.2

15
12.4

20
11.0

25
10.6

Running Totals
10 min
20 min

13.6
22.4
23.4
21.6
18.6
14.0
7.6

37.0
44.0
42.0
35.6
2.2

30
8.0

35
6.0

40
1.6

Maximum

140.4
132mm/hr

In the 10 minutes column, maximum depth is 23.4mm which gives 2.34 mm/min and an
intensity of (2.34*60) = 140.4 mm/hr. In the 20 minute column, maximum depth is 44mm
which gives an intensity of 132 mm/hr.
In most design applications e.g. culverts, irrigation schemes etc the maximum depth of rainfall
likely to occur over a given area, its duration and frequency is often required. A storm of a
given duration in a particular area rarely produces uniform rainfall depth over the entire area. It
has a centre where the rainfall is maximum and larger than the average depth of rainfall for the
whole area. The difference between these two values increases with increase in area and
decreases with increase in duration.
Depth-Area-Duration (DAD) Curves:
In designing water resources structures, one has to know the areal spread of rainfall within
watershed. However, it is often required to know the amount of high rainfall that may be
expected over the catchment. It is usually observed that a storm event starts with a heavy
downpour which gradually reduces as time passes. Rainfall depth is therefore not proportional
to the time duration of rainfall observation. Similarly, rainfall over a small area may be more or
less uniform.
However, if the area is large, then due to the variation of rain falling in different parts, the
average rainfall would be less than that recorded over a small portion below the high rain fall
occurring within the area. Due to this, a Depth-Area-Duration (DAD) analysis is carried out
based on records of several storms on an area and, the maximum areal precipitation for
different durations corresponding to different areal extents.
In many hydrological studies involving estimation of severe floods, it is often necessary to have
information on the maximum amount of rainfall of various durations occurring over various
areas during standard passage of time, for example the largest depth over 1000km2 in 24 hours.
24

Lecture Notes ECE 2402: Hydrology I by Dr. J.K. Mwangi

Such values when determined for each transposable storm, provide data for estimation of the
probable maximum precipitation (PMP) for the basin.
DAD analysis is the development of relationships between maximum depth- area- duration for
a region. A DepthArea-Duration curve expresses graphically the relationship between
progressively decreasing average depth of rainfall over a progressively increasingly area from
the eye of the storm outward to its edge for a given duration of rainfall. It is an important
aspect of hydrometeoro;ogical study. Depth-Area relationships for various durations, (fig.12)
are derived through DAD analysis in which isohyetal maps are prepared for each duration from
tabulation of maximum n-hour rainfalls recorded in a densely gauged area. The area contained
within each isohyet on the maps is then determined and a graph of average precipitation depth
versus area plotted for each duration.
DAD analysis is devised to determine the greatest precipitation amounts for various area sizes
and durations over different regions and for certain seasons. The resulting relationships are
used to determine hypothetical storm events for the design of hydraulic structures.

Figure 12: Depth Area Duration curves

DAD relationships indicates the areal distribution characteristic of a storm of given duration.

Rainfall intensity
Is the amount of rainfall for a given rainfall event recorded at a station divided by the time of
record, counted from the beginning of the rainfall event.
25

Lecture Notes ECE 2402: Hydrology I by Dr. J.K. Mwangi

Return period
Is the time interval after which a storm of given magnitude is likely to recur. It is obtained by
analyzing past rainfall events recorded at a station. Frequency of a rainfall event (storm event)
is the inverse of return period and is often multiplied by 100 and expressed as a percentage. The
frequency a rainfall or storm (expressed as percentage) of a given magnitude means the number
of times the given event may be expected to be equaled or exceeded in 100 years.
Intensity-Duration-Frequency analysis:
The analysis of continuous rainfall events, usually lasting less than a day, requires evaluation of
rainfall intensities. Assessment of such values may be made from records of several part storms
over the area and presented graphically (fig.13). IDF analysis gives the frequency of occurrence
of a given maximum intensity. From IDF curves maximum intensity of rainfall for any duration
and return period can be obtained. It is often necessary to know the rainfall intensities of
different durations and return periods, in many design problems e.g. runoff disposal, erosion
control, highway construction, culvert design etc. The curve that shows the inter-dependency
between i (cm/hr), D (hour) and T (year) is known as the IDF curve.
In addition, One of the steps in urban drainage design is determination of the rainfall event
/events to be used in the Rational Formula which is derived from IDF curves already drawn for
the site concerned. A graphical relationship of rainfall intensity that can be expected to occur in
a given period of time (Duration) on average once every so many year (Frequency) is called
an IDF curve (Figure 13). Rainfall intensity of the desired frequency is computed from IDF
curves for the duration equal to the time of concentration of the watershed. The frequency
chosen should reflect the economics of flood damage reduction (1 to 10 years for residential
areas and longer periods for urban areas.

Figure 13: Intensity Duration - Frequency- curves

Application of IDF curves the in rational formula
The Rational formula Q p CiA where Qp is peak discharge (m3/s), C is runoff coefficient
(dimensionless) reflecting soil type, topography, surface roughness, vegetation and land use, i

26

Lecture Notes ECE 2402: Hydrology I by Dr. J.K. Mwangi

is rainfall intensity (mm/hr) and A is catchment area (km2). C values are obtained from
published data. The equation takes different forms depending on units used.
Basic assumptions in the use of the formula include:
Rainfall intensity is constant for a time interval equal to the time of concentration (tc)
which is taken as the time for water to flow from the most remote part of the catchment
to the outlet.
Maximum rate of runoff equal to a certain rainfall intensity occurs after elapse of t c,
when the entire area is contributing at the outlet and continues at a constant rate until
there is a change in rainfall intensity.
Runoff coefficient C is constant during the entire storm.
Contributing catchment area does not change during the storm.
The rational formula is most applicable in urban and sub-urban areas.

Stepwise procedure of determining peak flow rates using the rational formula:
Estimate time of concentration tc of the drainage basin by an appropriate formula e.g.
0.77 0.385
Kirpich equation. t c 0.00032 L S
Where L= maximum length of travel of
water (m), S = slope (H/L) where H = is the difference in elevation between the
remotest point on the basin and the outlet (m) and L= basin length (m).
Note that tc is taken to be equal to storm duration.
Estimate runoff coefficient C from standard tables e.g. that of American Society of
Engineers or any other recommended source.
Select a return period T (from curve) and find storm intensity i corresponding to tc on
the selected curve. Methods of computing i from the equation have been developed
for some regions in the world.
Compute Qp, now that C, I, and A are known.
Maximum intensity varies inversely with duration and is generally of the form

t a b

Where a, b, c are coefficients obtained through regression analysis.

t is duration in minutes while i may be in mm/hr or cm/hr. The regression constants are
different for different locations.
Probable Maximum Precipitation (PMP): is the estimated limiting value of precipitation that
can occur at a given location in a given duration. Consequently it is defined as the
analytically estimated greatest depth of precipitation for a given duration that is physically
possible and reasonably characteristic of a particular geographical region at a certain time of
the year. It is the amount of rainfall over a region which cannot be exceeded. It is obtained by
studying all storms that have occurred in the region and maximizing them for the most critical
atmospheric conditions. PMP varies over the Earths surface according to local climatic factors
and would be expected to be much higher in hot humid equatorial regions than in the colder
regions of mid-latitudes where the atmosphere is not able to hold as much moisture. From an
operational point of view, PMP can be defined as that rainfall over a region which would
produce a flood flow with virtually no chance of being exceeded. The PMP concept is not
entirely reliable because it cannot be accurately estimated and its probability of occurrence is
27

Lecture Notes ECE 2402: Hydrology I by Dr. J.K. Mwangi

unknown. However, in operational terms, PMP has been found useful and its use will continue
because of public concern for safety of projects such as large dams. If PMP for a given basin
can be estimated, it can be used to provide an estimate of the probable maximum flood (PMF)
after adjustments for infiltration losses etc.
Methods of estimating PMP include:
Application of storm models- ideal in large areas with insufficient historical data and
rugged topography which complicates storm phenomena.
Maximizing of actual storms: For small areas, if records of actual storms are available,
they may be maximized to obtain PMP values.
PMP x Where k is frequency factor empirically determined and which
depends upon the statistical distribution of the series, number of years of record and the
return period. Usually it is in the neighborhood of 15, x is the mean of all maximum
daily data for a number of consecutive years of record (say 30yrs) and is standard
deviation of sample. The value of k ranges from 5 to 30. The PMP value so estimated
for a location is then adjusted through appropriate DAD relationships to make it
applicable over a given area.
Generalized PMP charts: PMP estimates may be displayed as isohyetal maps that depict
regional variation of PMP for some specified duration, basin size, and annual / seasonal
variation. These maps are known as generalized PMP charts.
Standard Project Storm (SPS)
Is the storm which is reasonably capable of occurring over the basin under consideration, and is
generally the heaviest rainstorm, which has ever occurred in the region of the basin during the
period of rainfall records. It is not maximized for the most critical atmospheric conditions but
may be transposed from an adjacent region to the catchment under consideration.
A common use of rainfall data is in the assessment of probabilities or return periods of given
rainfall at a given location. Such data can then be used in assessing flood discharges of given
return period through modelling or some empirical system and can thus be applied in schemes
of flood alleviation or forecasting and for the design of bridges and culverts.

28