## Publications with the same themes

## PDF content (text-only)

~R -
The prediction of storm rainfall in East
by
D. Fiddes, J,,A. Forsgate and A. O. Grigg
TRANSPORT and ROAD
, RESEARCH LABORATORY
Department of the Environment
TRRL LABORATORY REPORT 623
THE PREDICTION OF STORM RAINFALL IN EAST AFRICA --
x
by
D. Fiddes, B.SC., M.SC., C.Eng. M. I. C. E., DIC.,
J. A. Forsgate, B.SC., and A. O. Grigg
Any views expressed in this Report are not necessarily
those of the Department of the Environment
Environment Division
Transport Systems Department
Transport and Road Research Laboratory
Crowthorne, Berkshire
1974
CONTENTS
Abstract
1. Introduction
2. Two year, 24 hour point rainfall map for East Africa
3. Twenty-four hour storm rainfall for any return frequency
3.1 Fitting of boundaries between zones
4. Depth – Duration – Frequency Relationships
4.1 Data avaflable
4.2 Model testing
4.2.1
4.2.2
4.3 Further calibration using hourly and daily data
4.4 General depth – duration model for East Africa
4.5 Conclusions
5. Area reduction factors
5.1 Area reduction factors for East African raingauge networks
5.1.1 The Kakira network
5.1.2 The Nairobi network
5.1.3. Sambret network
5.1.4 Atumatak network
5.2 General equation for a~ed reduction factors
5.3 Comparison with published ared reduction factor
6. Conclusions
7. Acknowledgements
8. References
9. Appendix 1: Design curves and worked examples
Page
1
1
1
3
5
6
7
7
7
7
9
9
12
13
13
13
15
18
18
20
20
21
22
22
23
@ CROWN COPYRIGHT 1974
Extracts may be quoted except for commercial pu~oses,
provided the source is acknowledged
THE PREDICTION OF STORM RAINFALL IN EAST AFRICA
ABSTRACT
A simple method for predicting the characteristics of storms for the design
of drainage structures in East Africa is described. The variation of 2 year
daily point rainfall, and the 10:2 year ratio for daily rainfall, over East
Africa are given in map form. Using these, daily point rainfall for any return
frequency can be calculated. To arrive at the design storm the daily point
rainfall is adjusted using a generalised depth-duration equation and a
graphical representation of the variation of mean rainfall with area.
1. INTRODUCTION
Before the hydraulic and structural designs for a road bridge or culvert can be started, an estimate must be made
of the peak flo~v that the structure must safely pass. If flow measurements have been made for a number of
years on the river, or on a similar but adjacent river, this involves only a statistical analysis of recorded peaks
to arrive at a design flood of an a~lpropriate return frequency. In East Africa, particularly on the smaller rivers,
such data ~arel)’ exist and use must be made of the much more common rainfall measurements published by the
East African M(;teorological Department. From rainfall records a design rainstorm is constructed and routed
through a suital)le catchment model to give the design flood.
Nthough an impressive amount of rainfall data exists, it has not been published in a form that can be
readily used by highway engineers for bridge and culvert design. The purpose of this report is to extract relevant
storm data fronl the published records and, combining these with certain unpublished data, to produce a simple
method for pre ,?aring design storms for flood estimation. The method involves first estimating the 2 year, 24 hour
point rainfall from a storm rainfall map of East Africa. Three adjustments are then made:
(a) Using a generalised relationship between rainfall of any return frequency and the two year values the
24 ~our point rainfall for the design return frequency is calculated.
(b) A depth-duration rainfall equation is used to calculate the point rainfall for the appropriate time of
concentration of the catchment.
(c) An areal reduction is read off a graph to convert this to the mean rainfall depth which is the required
rair.fall input for the catchment model.
2, TWO YEAR, 24 HOUR POINT RAINFALL MAP FOR EAST AFRICA
There are abour 3,000 rainfall stations in East Africa which submit daily records to the Meteorological Department
for subsequent publication. The distribution of these is, however, far from uniform and many have been installed
ody in recent }ears.
The advice often given to engineers requiring a design storm is to select a suitable rainfall station, on, or
close to, the carchrnent and to analyse the records for this station. For much of East Africa a station on or close
to the catchmellt cannot be found or, if present, it has often been recording for such a short period of years
that it can give only unreliable estimates of flood producing rainfall. It was therefore decided to analyse dl
available publiiled records and use these to produce maps of storm rainfall from which vrdues for individud
catchments could be interpolated.
Using records in published form this would have been a mammoth task, but fortunately early in the
investigation the East African Meteorological Department transferred all their reliable daily rainfall records for
the years 1957-68 and selected stations for 1926-56 to magnetic tape and gave the Laboratory permission to
make a copy of the tapes.
From the first tape 867 stations which had at least 10 complete data years and from the combined tapes
99 stations with about 40 complete data years were selected. The first set of data was used to map the variation
of storm rainfall over East Africa and the second to establish a means of adjusting values read off the map for
alternative frequencies.
For each station selected for the first set of data the highest 24 hour fall during each calendar year was
read off. These were ranked and given ieturn frequencies using the expression:
T=— ntl
m
where T is the return frequency in years
n is the number of years of record
and m is the ranking order, m = 1 for the largest value,
m = n for the smallest.
If the rainfall depth is plotted against the assumed return frequency a non-linear relationship becomes
apparent. Many methods are available to linearise this relationship which, so long as extrapolation beyond the
period of record is not attempted, give very similar results. The most commonly accepted method is the Gumbel
method (1) which is of the form
Y=a+cloglog T