JABATAN PENGAIRAN DAN SALIRAN (JPS)

KEMENTERIAN SUMBER ASLI DAN ALAM SEKITAR (NRE)

MALAYSIA

NOTA KURSUS TAHUN 2006

KM 7, JALAN AMPANG
68000 AMPANG, KUALA LlJMFW

URBAN STOR WATER

MANUAL FOR MALAYSIA
(MANUAL SALIRAN MESRA ALAM MALAYSIA)

VOLUME 4
DESIGN FUNDAMENTALS

DEPARTMENT OF IRRIGATION AND DRAINAGE

MALAYSIA

i3.2
Rainfall is, obviously, the driving f ~ r c e behind ail
stonwater studies and designs. An understanding of
rainfali processes and the significance of the rainfall design
data is a necessary pre-requisite for preparing satisfactory
drainage and stormwater management projects.
13.1.1 Rainfall Patterns in Malaysia
An overview of the climate of Malaysia, with general
rainfall characteristics is given in Chapter 1.
The frequency and intensity of rainfall in Malaysia is much
higher than in most countries, especially those with
temperate climates. Rainfall design methods, which have
been developed in other countries, may not always be
suitable for application in Malaysia.
The design
calculations for these methods have been adjusted in this
Manual to suit Malaysian conditions.

13.1.2 Application
This Chapter supersedes Hydrologic Procedure HP1-1982
for urban stormwater drainage only. The Chapter does not
deal with non-urban situations, such as dams or river
engineering, for which the HP1 or other special hydrologic
procedures should continue to apply.
The material in this Chapter draws upon that in HP1-lW2,
and its presentation has been revised to be more directly
applicable to urban drainage problems. No additicnal
analyses were performed. It is envisaged that both HP1
and this Chapter will be revised in the future, using
additional data that is becoming available.

13.1.3 Climate Change

Thcre is potential for global climate changes to occur due
to the "Greenhouse Effect". A further discussion on
climate change due to the Greenhouse Effect is given in
Chapter 46, Lowland, Tidal and Small Island Drainage.
Some authors have suggested that climate change due to
the Greenhouse Effect will cause an increase in storm
rainfall intensity. At this stage the available evidence for
any effects on rainfall intensity is not conclusive. This
Manual does not recommend any increase in design rainfall
intensities due to the Greenhouse Effect.
Nevertheless, designers of major urban stormwater
drainage systems should consider the possibility of climate
change due to greenhouse effect. Sensitivity testing can
be performed in critical cases.
Designs should be
sufficiently robust and incorporate safety margins to allow
for this possibility.

U&an Stormwater Management Manual

DESIGN

RAINFALL INTENSITZES

13.2.1 Definitions
The specification of a rainfdl event as a "design storrn"is
common engineering practice. The related concepts of
frequency and average recurrence interval (ARI) were
discussed in Chapter 11.
Although the design storm must reflect required levels of
protection, the local climate, and catchment conditions, it
need not be scientifically rigorous. It is more important to
define the storm and the range of applicability so as to
ensure safe, economical and standardised design.
Two types of design storm are recognised: synthetic and
actual (historic) storms. Synthesis and generalisation of a
large number of actual storms is used to derive the former.
The latter are events which have occurred in the past, and
which may have well documented impacts on the drainage
system. However, it is the usual praace in urban
stormwater drainage to use synthetic design storms and
most of this Chapter concentrates on these storms.

Design sror,n duration is an important parameter that

defines the rainfall depth or intensity for a glven frequency,
and therefore affects the resulting runoff peak and volume.
Current practice is to select the design storrn duration as
equal to or longer than the time of concentration for the
catchment (or some mmimum value when the t!me of
concentration is short). Intense rainfalls of snort durations
usually occur within longer-durat~onstorms rather than as
isolated events. The theoretically correct mctice is to
compute drscharge for several design storms ~41th
different
durations, and then base the desrgn on che "cmcal" storm
wh~chproduces the maximum discharge. '.ioXvveverthe
"critical" storm duration determined in this waj may not be
the most critical for storage design.
Recommended'
practice for catchments containing storage is to compute
the design flood hydrograph for several storms with
different durations equal to or longer than the time of
concentration for the catchment, and to use tne one which
produces the most severe effect on the pond slze and
discharge for design. This method is further discussed in
Chapter 14.

13.2.2

Rainfall Intensity-Duration-Frequency (IDF)

Relationships

The total storm rainfall depth at a point, for a given rainfall

duration and ARI, is a function of the local climate.
Rainfall depths can be further processed and converted
into rainfall intensities (intensity = depthJduration), which
are then presented in IDF curves. Such curves are
particularly useful in stormwater drainage design because
many computational procedures require rainfall input in the
form of average rainfall intensity.

13-1

The
vambles, frequeniy, intensity and duration, are
aii r-elated to each other. Tine data are normally presented
as curves dispiayinr; two of the variables, such as intensity
and duration, for a range of frequencies. These data are
then used as the input in most stormwater design
processes.

13.2.3 Areal Reduction Factor

- - ......24 hours;

It is important to understand that IDF curves give the

rainfall intensity a t a point
Storm spatial characteristics are important for larger
catchments. I n general, the larger the catchment and the
shorter the rainfall duration, the less uniformly the rainfall
is distributed over the catchment. For any specified ARI
and duration, the average rainfall depth over an area is
less than the point rainfall depth.
The, ratio of the areal average rainfall with a specified
)duration and A R I to the point rainfall with the same
duration and ARI is termed the areal reduction factor.
Areal reduction factors are applied to design point rainfall
intensities only, to account for the fact that it is not likely
that rainfall will occur at the same intensity over the entire
area of a storm (the principle of design storms assumes
that the design storm is centred over the catchment). The
areal reduction is expressed as a factor less than 1.0. For
large catchments, the des~gn rainfall is caiculated wtth
Equation 13.1:

where F, is the areal reduction factor, I, is the average

rainfall over the catchment, and I, is the point rainfall
intensity.
Suggested values of areal reduction factor FAfor Pentnsular
Malaysia are gtven in HP No.1-1982. These values are
reproduced in Table 13.1 below for catchment areas of up
to 2GG km2. The values are plotted in Figure 13.1.
Intermediate values can be interpolated from this Figure.

Table 13.1
Catchment
Area

Values of Areal Reduction Factors (FA)

Storm Duration (hours)

0.40

- -

-6

-s-

0.20
10

---0.5 hour

100
Catchment Area (km2)

Figure 13.1

hours

*3 hour
hours
---

1000

GraphicalAreal ReductionFactors

No areal reduction factor is to be used for catchment arsas

of up to 10 km2. The majority of urban drainage areas wtll
fall into this category.
Areal reduction factors should not be applied to real ralnfall
data, such as recorded daily rainfalls. Instead an attempt
should be made to obtain and use all available data from
other rain gauges in the catchment.
Stcrm d~rectronand movement can have marked effects,
pa~icularlyIn areas w ~ t hpredommatmg weather patterns,
and are particularly relevant to the case of operation
ar; or conti01 of a large system of stormwater draicage
ner;$orks. However, for urban dra~nage~t IS customary :o
assume that deslgn storms are Gat~onary.

13.2.4 I D F Curves for Selected Cities and Towns

The publication " Hydrolog~cal Data - RainfaN and
Evd,ooratiOn Records for Malaysia " (199 1 ) and
"Hydrological Procedure No. 26 "by the Department of
Irrtgation and Drainage (DID), have maximum ramfall
intensity-duration-frequency curves for 26 and 16 urban
areas in Peninsular Malaysia and East Malaysia (Sabah and
Sarawak), respectively. These curves will cover the needs
of the majority of users of this Manual.

Users need to be aware of the limitations of these IDF

curves:
The curves have not been revised since 1991. The
patterns should be reviewed using the additional data
that is now available.
The period of data from which the curves was derived
was very short, in some cases only 7 years. Few of
the stations had more than 20 years of data. This
means that there is a large potential error ~n
extrapolating to long ARI such as 100 years.

-13-2

-Urban Stormwater Management Manual

-.

,.ye Ic;;lier :;::;;: sf the dg;atjcns

anajypd was 15
mlnutes. DID should expedite the installation of
a!rjltai piuvlometers to capture data from short storm
.-Auul,u,
down to 5 minutes duration.

a to d are fittlng constants iependent on ARI.

C.

The limits of rainfall ARI were between 2 years and

100 years.
The curves were not in a convenient form for use in
modern computer models.
There was no guidance given for urban areas outside
the 42 centres listed.

I t is recommended that the curves should be updated by

DID to incorporate additional data and extend the
coverage as outlined above.
13.2.5 IDF Curves for Other Urban Areas
IDF curves are calculated from local pluviometer data.
Recognising that the precipitation data used to derive the
above were subject to some interpolation and smoothing,
it is desirable to develop IDF curves directly from local
rain-gauge records if these records are sufficiently long
and reliable. The analyses involve the following steps:
Data Series (identification)

ii
Data Tests

il
Distrbution 1dentificaL;on

C
Estimation of D~stributionParameters
Selection of Distrbutlon

I!
Quantile Estlmat~onat chosen ARI
The required analyses are hlghly specialised and would ke
outside the scope of interest of most users of this Manual.
Local authorities are advised to find out from the DID co
the availability of IDF curves or coefficients for the~r
respective areas, or to obtain local pluviometer data for
those wishing to conduct their own analysis.

Four coefficients are considered in Equation 13.2 to keep

the calculation simple for a yeasonable degree of accuracy.
Higher degree of polynomial can be used to get more
accurate values of rainfall intensity. The Equation can be
used for deriving rainfall intensity values for a given
duration and ARI, once the values of coefficients a to dare
known. The equation is in a more suitable form for most
spreadsheet of computer calculation procedures.
The curves in "Hydrological Data" (1991) are valid for
durations between 15 minutes and 72 hours. Extrapolation
of the curve beyond these limits introduces possible errors,
and is not recommended. Also, Equation 13.2 should not
be used outside these limits. Alternative procedures for
deriving IDF values for short durations are given in
Section 13.2.7.
The possible uncertainty range of the IDF figures derived
in accordance with this Manual is likely to be up to rt 20/0.
Among the sources of error noted are: problems of
extrapolation to long ARIs, use of local rather than
generalised analysis, and problems with the accuracy of
short-duration intensity records. The error is likely to be
highest for the durations shorter than 30 minutes and
longer than 15 hours, and for ARI longer than 50 years.
For particularly critical applications it may be appropriate to
conduct sensitiv~tytests for the effects of design rainfall
errors.
Table 13 2 glves v a l ~ e s of the fitted coe%e.its
In
Equation 13 2 for Kuala i u ~ p u r fcr
, rainfail ARIs between
2 years and 100 years and durat~ons~ l t h l n30 to 1000
mlnutes (see F~gure13.2 for the graphs) Appendrx 13 A
gwes derived values of the coeffic~entsIn Equatlon 13.2 for
the 26 and 10 urbar centres in Peninsular and East
Malaysla, respectively
Due to irregular shape of the
curves, coeffiaents for 6 other urban centres In East
Malaysia are not sutable to be used In Equatlon 13 2. IDF
values for these 6 stat~onsshould be taken from their
respectwe curves ava~lableIn HP-26 (1983)
Table 13.2

Coefficients of the Fitted IDF Equation for

Kuala Lumpur

13.2.6 Polynomial Approximation of IDF Curves

Polynomial expressions in the form of Equation 13.2 have
been fitted to the published IDF curves for the 35 main
citiesftowns in Malaysia.

where,

= the average rainfall intensty (mmfhr) for ARI and

duration t
R = average return interval (years)

R ~ t

Urban Stmwater Management Manual

(data period 1953 - 1983); \/a!!dity: 3C c t 5 1000 minutes

13-3

2 Yr

. . - - -.1 y r ARI
10C
Duration (minutes)
Figure 13.2

IDF Cuwes for Kuala Lurnwr

Equatlon 13 3 should be used for durations less than 30

minL:?s
For durat~onsbetween 1 5 and 30 mlnbtes, che
resu:s should be checked against the pubhshed IDF
c u n s The relatiorship 1s valid for anv ARI wthin :he
rance of 2 to 100 years

13.2.7 IDF Values for Short Duration Storms

I t 1 8 recommended that Equat~on13 2 be used to derive
design ralnfall intensities for durations down to a lower
limit of 30 minutes. This value corresponds to the original
range of durations used in deriv~ngthe curves.

The ,:aiue of FD is obtained from Table 13.3 as a function

of 'i,,,, the 2-year AX1 24-hour rainfall depth Values of
'P.,. for Pen~nsular Maiaysia are given in Figwe 13.3
I n t s - ~ e d i a t evalues should be ~nterpoiated.

Estimation of rainfail intensit~esfor durations between 5

and 30 minutes involves extrapolation beyond the range of
the data used In deriving the curve fitt~ngcoefficients. The
e e c o m m e n d e d method of extending the data is based on
HP No.1-1982, which gives a rainfall depth-duraticn
plotting
graph f x durations between 15 minutes and 3
hours. T h ~ sgraphical procedure was converted into an
equation and extended as described below. An additional
adjustment for storm intensity was included based on the
method used in "PNG Flood Estimation Manual" (SMEC,
1990), for tropical climates similar to Malaysia. This
adjustment uses the 2 year, 24-hour rainfall depth 'p24h as
a parameter.

Note that Equatior! 13.3 !s !n terms of r3!n.f3!! depth, not

I f intensity is required, such as for roof
intensit\/.
dra~nage,the depth P, (mm) IS converted to an intensiP{ i
(rnm,'hr) by div~dingby the duration d i n hours:

The design rainfall depth P, for a short duration d

(mmutes) is given by,

Table 13.3
I

j Duration

5
10
15
- 20
30

2P2qh(mm)

(minutes)
where PjOr P60 are the 30-minute and 60-minute duration
rainfall depths respectively, obtained from the published
design curves. Fo is the adjustment factor for storm
duration

Values of FD for Equation 13.3

1 100

1 2.08
1.28
0.80
0.47
0.00-

West Coast
120
150

:, 180

1.85
1.13
0.72
0.42
O.O@

1.40
0.86
0.54
0.32
0.00

1.62
0.99
0.62
0.36
0.00

I
East Coast
1
All
1.39
1.03
0.74
0.48
0.00

1
i

13-4

Urban Stormwater Management Manual

13.3

DESIGN RAINFALL TEMPORAL

PATTERNS

i3.3.1 Purpose

13.2.8 IDF Values for Frequent Storms

Water quality studies, in particular, require data on IDF

values for relatively small, frequent storms. These 9orms
are of interest because on an annual basis, up to 90% of
the total pollutant load is carried in storms of up to
3 month ARI. Chapter 4 recommends that the water
quality design storm be that with a 3 month ARI. The
typical IDF curves given in Appendix 13.A have a lower
limit of 2 years ARI and therefore cannot be used directly.
The following preliminary equations are recommended f ~ r
calculating the 1, 3, 6-month and 1 year AR! rainfail
intensities in the design storm, for all durations:

The temporal distribut~on of rainfall within the design

storm is an important factor that affeds the runoff volume,
and the magnitude and timing of the peak discharge.
Design rainfall temporal patterns are used to represent the
typical variation of rainfall intensities during a typical storm
burst.
Standardisation of temporal patterns allows
standard design procedures to be adopted in flow
calculation.

I t is important to emphasise that these temporal patterns

are intended for use in design storms. They should no: be
confused with the real rainfall variability in historical
storms.
Realistic estimates of temporal distributions are best
obtained by analysis of !ocal ramfall data From recarding
gauge networks. Such an analysis may have to be done
for several widely varying storm durations to cover various
types of storms and to produce distributions for various
design problems. Different distributions may apply to
different climatic regions of the country.
Temporal patterns shouid be chosen so that the resulting
rnnzff hydrqiaphs a x consistent with observed
hydrographs. Therefore the form of the temporal pattern
and the method of runoff computation are closely interlinked. The statistical basis of this approach 18 discussed in
".4usistr2/iar?.!?a~fA'
.3,7C .?;.ncff' (P.R&R, 1987).

,;Jhere, ".".r~.
- , '-'zI-,, "i, and 'I, are the required 1, 3, 6nionth and !-year ARI rainfall ~ntensitiesfor any duraticn
D, and '1, :s th2 2-year ARI rainfall Intensity for the s a w
duratior! D, obta~nedfrom IDF curves.
Users should be aware of the limitations of these Equations
13.5a to 13.5d. They were derived by fitting a distributlcn
to the 1-hour duratjon rainfalls, and extrapolating the
distribution to frequent ARIs. This method IS subject to
considerable uncertainty. These preliminary equations
were derived using Ipoh rainfall data. Further research 18
required to confirm the relationships, particularly in other
parts of Malaysia where different climatic influences apply.
13.2.9 IDF Values for Rare Storms

Further research is required in order to allow design rainfall

mformation to be given for storms with ARI greater than
100 years.
This Manual does not cover the design of major structures
such as dams or bridges, for which a special hydrologic
analysis is required.

Urban Stormwater Management Manuai

A range of methods to distribute rainfall have been

suggested ~nthe literaturs:
1

Aderage temporal pacerns developed from local mintrainfall data measured In short trme intervals (15
mlnutes cr less)

Sim~le ~deallsed rairfall d~strlbutron fitted to local

storm data by the method of moments.

Temporal patterns from local IDF relat~onsh~ps.

The second method IS not recommended, as the idealised

patterns are not representative of real storm patterns.
Triangular patterns, for example, give unrealistically high
peak intensities.
The third approach for distributing rainfall within a design
storm makes use of the local IDF relationship for the
design ARI. This approach is based on the assumption
that the maximum rainfall for any duration less than or
equal to the total storm duration should have the same
ARI. For example, a 10 year ARI three-hour design storm
of th~stype would contain the 10 year ARI rainfall depths
for all durations from the shortest time interval considered
(perhaps 5 minutes) up to three hours. These rainfalls are
generally skewed.

Deslgn Rainfall

Figure 13.3

13-6

Values of ' P for~use with Table 13.3

(source: HP 1, 1982)

Urban Stomwater Management Manual

4Ithcijgh there are some theoretical sbjections to the third

approach, on the grounds that it combines peaks from
diffsent historical storms, it 1s nevertheless conservative
and very convenient for design.
it is therefore
recommended.
This distribution can be derived from the local IDF curves
and the analysis of skewness of actual storms. The design
temporal patterns presented in this Chapter have been
derived on this basis.
13.3.2 Present Malaysian Practice

The 1982 update of Hydrological Procedure No. 1 gave

recommendations on temporal patterns to be adopted for
design storms in Peninsular Malaysia. Patterns were
prepared for six standard durations: 0.5, 3, 6, 12, 24 and
72 hours.

stormwater drainage design. I f a rarniaii iuraticr!

longer than 6 hours is required, the aser should
consult HP No.1-1982.
No temporal pattern data is available in HP No. 26 for
Sabah and Sarawak (1983). For preliminary studies,
the patterns for the East Coast of Malaysia, in
Appendix l3.B, could be adopted for Sabah and
Sarawak. Because the climatic conditions are more
comparable to the East Coast than the West Coast. A
further study to derive temporal patterns suitable for
use in Sabah and Sarawak should be undertaken by
specialist hydrologists.
There are too few data points found in each Standard
Duration temporal pattern. This causes a systematic
bias which will under-estimate the peak rainfall as
shown in Figure 13.4. AR&R-1987 estimates that this
error will underestimate the true peak by as much as
10%.

Nine rainfall stations located in different parts of Peninsular

Malaysia were used in this analysis. The data covered nine
years from July 1970 to June 1979.
The procedure used was to extract the rainfall pattern of
the annual maximum stormbursts at each station. This is
Method 1 of the previous sub-section. By comparing the
results, a set of representative patterns were derived. I t
was found that different patterns applied to the West
Coast and East Coast, except for the 0.5 hour duration,
and therefore different regional patterns were
recommended.

This procedure used in HP No.1-1982 is less

comprehensive than that used in, for example, AR&R
(1987), because there was a much smaller amount of data
available.
18 years have elapsed since Hydrological Procedure
No. 1 was updated. The patterns should be reviewed
using the additional data that is now available.
The range of Standard Durations used in the
Procedure is insufficient for the full range of design
conditions. Design calculations need to be made for
periods as short as 5 minutes, in the case of Roof
Drainage (see Chapter 23). The gap between each
Standard Duration in HP No.1-1982 is too wide.
Temporal patterns should be internally consistent
(AR&R, 1987). There are inconsistencies between the
HP No.1-1982 patterns.

Durations longer than 6 hours are not covered in this

Manual as they are unlikely to be required in urban

Urban Stomwater Managem&

Manual

Indicated Peak

cicn
sc

2
.-c

The recommended patterns were presented in HP No. 1 in

graphical form (Figures D.8 to 0.13 of HP No.1). These
graphs are difficult to read and are subject to misinterpretation. Therefore, the data has been converted to
tables in this Manual, as shown in Appendix 13.8.
13.3.3 Review of Standard Temporal Patterns

Instantaneous Peak
Intensity

Time

Figure 13.4

Example of the Under-estimation of a

Hydrograph by Discretisation

13.3.4 Temporal Patterns for Standard Durations

The recommended patterns in this Manual are based on

those from AR&R for durations of one hour or less and
from HP No. 1 (1982) for longer durations.
The recommended patterns follow Method 3 mentioned in
Section 13.3.1. Checks were made to ensure that the
patterns comply with the requirements in Section 13.3.3.
Where necessary, the ordinates of the patterns were
adjusted to meet these requirements.
Most of the
adjustments made were relatively minor.
The standard durations recommended in this Manual for
urban stormwater studies are listed in Table 13.4. The
interim temporal patterns to be used for these standard
durations are given in Appendix 13.8.

13-7

Design Rainfall

23.4.3 Data Quality and Acceptance

- daily iainfa!l record should be examined for quality,

-L
8 lie

3?i! in particular to identify any instances of missing

records.

Table 13.4 Standard Durations for Urban

Stormwater Drainage

Standard Duration
(minutes)

Number of
Time Intervals

Time Interval
(minutes)

10

Occasional missing daily records may be acceptable,

depending on the purpose for which the data is used. I n
general, for water quality studies the amount of missing
record should not exceed 3% i.e. 10 days per year. For
flood studies, particular care is required because it is often
found that the missing record is the actual record of most
interest, i.e. a large storm or flood event.
Some older daily rainfall records are still in English
(Imperial) units.

13.4.4 Long-Duration Rainfalls

Iote that minutes are used in this Table, for consistency

lith the units in Equation 13.2.

.3.3.5 Temporal Patterns for Other Durations

;or other durations, the temporal pattern for the nearest
standard duration should be adopted. It is NOTcorrect to
average the temporal patterns for different durations.

13.4

RAINFALL TIME SERIES

13.4.1 Introduction
Calculations for stormwater quality may involve the time
series of runoff, which In turn is related to the volume of
.airifall.
jaily rainfall gauges are widespread thraughout Malaysia,
I comparison to the smaller number of pluviometers.
aily rainfall records are also of longer duratior? than
luviometer data. Both of these attributes make daily data
aluable for statistical studies. Daily rainfall data is
ormally readily available at or close to any location of
rterest for urban stormwater studies.

13.4.2 Sources of Rainfall Data

I n Malaysia, rainfall data is collected by several
departments and authorities including the Meteorological
Service and DID. Other agencies such as water and
sewerage agencies, may collect data for their own
purposes.
The quality of the designs and analyses depends to a large
degree, on the quality of the rainfall data used. Therefore
every effort should be made to search for and obtain data
from the data collection agencies.

Some of the design guidelines given in this Manual require

calculation of long-duration rainfall totals, such as the 75
percentile 5-day total. Daily records should be used for
this purpose.
Example 13.C.2 in Appendix 13.C is a worked example of
the calculation of 5-day rainfall totals for Ipoh, Perak. The
75 percentile 5-day rainfall is used in the design procedure
for sediment basins, in Chapter 39.

13.4.5 Adjustment of Daily Rainfalls

Daily rainfall data is read with fixed observation times each
day. Daily records do not necessarily reflect the rainfall
figures for a 24 hour storm duration because the storm
period is unlikely to commence at exactly the time of the
daily reading.
I f conversion is necessary from daily rainfall totals to 24haur totals, it should be done using Equat~on13.6:

where, I . is the 24-hour storm rainfall for a given ARI,ID is

the daily rainfall, and Cis a conversion factor with values
of 1.16 (East Coast) and 1.12 (West Coast).
This equation is unchanged from the 1982 edition of
HP No.1.
The conversion factor C is obtained from
analyses of rainfall records. The values of Crecommended
above are consistent with those in SMEC (1990) and Miller
et al (1973).

13.4.6 Continuous Simulation

Continuous simulation models are important for water
quality studies. Most of these models are designed for use
with daily rainfall data.

13-8

Urban Stormwater Management Manual

s~mulatrony n o d sf several months to ach~evestabiiib.

Usually !t 6 conuenienc io rnr! the models fcr a 12-mcnth
simulatioii. This ensures that seasonal variations, if any,
are taken into account.
The data requirement for such a simulation is a full year,
or more, of continuous daily rainfall records. Even data
with a small number of missing records can be accepted
(see Section 13.4.3). The data should be prepared in the
form of a list or spreadsheet column of daily rainfall totals,
in rnillirnetres. If, possible, data should be obtained in
digital form to minimise data entry and 'the possibility of
transcription errors.

standards for water quality control works, as dixussec .

Chapter 4. The result is likely to be similar rn other part;
of the West Coast of the Peninsula. Results in sttier a m
of the country may be different due to different rainfall
characteristics. Therefore, a 6milar calculation should be
performed at other locations.

.E

2
m

80%
0

13.4.7 Role of Small, Frequent Storms

The land area and cost of a treatment facility such as a

pond is approximately in proportion with its volume.
Increasing the design standard from (say) 3 month to
6 month ARI will result in significant size/cost increase for
only a minimal increase in pollutant removal and this can
be expressed in terms of the law o f diminishing returns
that defines the point at which increasing the design
standard is no longer cost-effective. It is generally found
that for water quality control, most benefit is derived from
the treatment of small frequent storms and that it is not
cost-effective to design facilities for large, rarer events.
The prelimmary calculations presented later (Chapter 15)
using rainfall data for Ipoh give a guideline to determme a
suitable cho~ceof design storm for sizing the water quai!?]
treatment measures. The results of the calculations are
plotted in Figure 13.5. From Figure 13.5 it can be seen
that, for Ipoh (Perak):
On a long-term average basis,
93% of total rainfall occurs in storms
equal to or smai'ler than 3 month ARI

Urban Stomwater Management Manual

7o0/o

+--

'I
I

-m
ij

'

90%

Like most decisions on design standards, the selection of a

suitable design standard for water quality control works
involves performance and economic considerations. I t
requires a trade-off between the benefits of providing a
higher level of protection, performance and the size and
cost of works needed to provide that protection.

looO/o

60%

u
50% .
0.01

Figure 13.5

13.5

0.1
I
Event A R I (years)

10

Amount of Total Rainfall which Occurs in

Storms Less than a Given ARI (Ipoh Data)

HISTORICAL STORMS

A discussion of rainfall data would not be complete without

mention of historical storms. Historical storm data is used
in calibration of models, as well as for the checking of past
flood occurrences.

I n an urban drainage situation, it is relatively rare to have

good historical rainfall data available close to the study
area or catchment. Nevertheless, every effort should be
made to obtain such data. Data sources are discussed in
Section 13.4.2.
Goyen and O'Loughlin (1999) have shown the importance
of locating rainfall gauges close to or preferably within the
study area if accurate calibration is to be achieved. I f
detailed studies are being undertaken and good calibration
data is required, the density of rain gauges should be at
least 1 per square kilometre.

13-9

Design Rainfall

APPENDIX 13.A

F I n E D COEFFICIENTS FOR IDF CURVES FOR 35 URBAN CENTRES

Pulau Ptnang

Penang

Perak

Ipoh

Perak

Bagan Serai

Perak

Teluk Intan

Perak

Kuala Kangsar

;
----I

20
50

5.7854
6.5736

1 0.1175

'

-0.2903

: 0.0044 ;
-0.0482 : 0.00002 '

1 -0.1244
'

Perak

Setiawan

Selangor

'

--

- -

(Continued)

U&n

Stormwater Management Manual

13-11

-aGie i 3 . A i

Coefic~encsfor the iDF Equations for the Different Major Cities and Towns

State

~ocation

1953-1983

Maiacca

Malacca

19;l-1990

--

1
Neger~Semb~lan

Negeri Sembilan

Kuala Pilah

Johor

Kluang

1970-1990

Seremban

:
'

2
5
10
20
SO
100
2
5
10
20
50
100
2
5
10
20

1
1
j

5.3255
5.1086
4.9696
4.9781
4.8047
5.0064
3.7091
4.3987
4.9930
5.0856
4.8506
5.3796
5.2565
5.4663
6.1240
6.3733

--

Kuala Lumpur

(32 5 r 5 1000 m~r.:!

Coefficients of :he :CF ?OIL~CC

at E ~ ~ a t ~ a n s

Data Per~od
1

Federal Terrtory

~ rMaiaysla
!

---

'

0.1806
0.5037
0.6796
0.7533
0.9399
0.8709
1.1622
0.7725
0.4661
0.5048
0.7398
0.4628
0.0719
0.0586
-0.2191
-0.2451

--

-0.1322
0.0047
0.0112
; -0.2155
-0.2584 ; 0.0147
-0.2796
0.0166
-0.3218 j 0.0197
1 -0.3070
0.0186
/ -0.3289 0.0176
/ -0.2381 j 0.0112 ,
1 -0.1740 / 0.0069 ,
' -0.1875 0.0082 .
-0.2388
0.0117
,
-0.1826
0.0081
-0.1306
0.0065
-0.1269
0.0062
-0.0820
0.0039
-0.0888
0.0051
,

Johor

Mersing

--

Johor

Batu Pahat

Johor

Johor Bahru

Johor

Segamat

1i

1970-1983

I
-

.
-

--

-- --

continued)

-.

13-12

Urban Stormwater Management Manudl

iocat~on

Stare

A N

Data Pemd

(year) ,
2

!
I

,I

Pahang

I
I

Raub

/
Pahang

Cameron Highland

1951-1990

1
I

1966-1983

Pahang

Kuantan

1951-1990

i
I

Pahang

Temerloh

1971-1983

Terengganu

Kuala Terengganu

1951-1983

Kelantan

a
4.3716
4.5461

I 5.4226

5.2525
4.8654
5.1818
4.9396
4.6471
4.3258
4.8178
5.3234
5.0166
5.1899
4.7566
4.3754
4.8517
5.0350
5.2158
4.6023
5.3044
4.5881
4.4378
4.4823
4.5261

0,

0.3725

-0.1274

d
0.0026

0.4017 i -0.1348 j 0.0036 ,

-0.1521 / -0.0063 ! -0.0056 1
0.0125 I -0.0371
-0.0035
0.3420
-0.1058
0.0012
' 0.2173
-0.0834
0.0001
0.2645
-0.1638
0.0082
, 0.4968
! 0.7684 -0.2549 0.0134
0.5093 i -0.2022 j 0,0100 ,
0.2213
-0.1402 i 0.0059 ,
0.4675 1 -0.1887 I 0.0089
0.2562
-0.1612 j 0.0096
0.6589
-0.2529 ! 0.0167
0.9634 I -0.3068
0.0198
0.7649
-0.2697
0.0176
0.7267 , -0.2589
0.0167
0.6752 I -0.2450 i 0.0155
0.4622
-0.1729
0.0066
0.0115
-0.0590
-0.0019
0.5465
-0.1646 ' 0.0049
0.71 18
-0.1960
0.0068
0.8403
-0.2288
0.0095 _0.7210
-0.1988
0.0071

i
'

'
1

'

'

---

+
-

5
10
20
50
100
2
5
10
20
50
100
2
5
10
20
50
100
2

10
20
50
-- 100

Kuala Dungun

Kelantan

/
I
I
I

1970-1983

Terengganu

Coeffk~eiitrof the iGF Polynom~alEquations

Kota Bharu

1951-1990

Gua Musang

(Continued)

Urban Stormwater Management Manual

13-13

-! able 13.P.T

Coefficients for the IDF Equations for the Different Major Cities and -OWES in Maiaysii! (30 i t 5 1000 min)

- .-

Data Penod

Locat~on

State

Sabah

Kota Kmabalu

Sabah

Sandakan

1957-1980

I
1

1
I

Sabah

Tawau

1966-1978

I
!

Sabah

Kuamut

1969-1980

Sarawak

S~manggang

1963-1980

Sarawak

S~bu

1962-1980

--

c
,
d
2
5.1968
0.0414
-0.0712 1 -0.0002
5
5.6093 , -0.1034 j -0.0359 1 -0.0027
lo
-0.0050
5.9468 i -0.2595 1 -0.0012
I
20
0.0026
5.2150 j 0.3033 I -0.1164
50
5.1922
0.0027
0.3652 / -0.1224
3.7427
2
1.2253 / -0.3396
0.0191 ,
4.9246 1 0.5151 1 -0.1886
0.0095
5
10
5.2728 1 0.3693 / -0.1624
0.0083 I
0.0133
4.9397 ! 0.6675 1 -0.2292
20
5.0022
0.0123 1
0.6587 ' -0.2195
50
2
i 4.1091 i 0.6758 i -0.2122 0.0093
5
1 3.1066
1.7041 j -0.4717 / 0.0298 10
4.1419
1.1244 I -0.3517 / 0.0220
20
4.4639
1.0439
-0.3427 1 0.0220
i
2
4.1878 ' 0.9320
-0.3115 / 0.0183 ,
i
5
3.7522
-0.4086 / 0.0249
1.3976
10
!
4.1594
1.2539
-0.3837 j 0.0236
20
1 3.8422
1.5659
-0.4505 / 0.0282
,
50
5.6274
0.3053
-0.1644 i 0.0079/ 100 I 6.3202 -0.0778 -0.0849 ! 0.0026
0.7773
-0.2644 / 0.0144
j
2
! 4.3333
-0.1985 1 0.0100
i 4.9834
0.4624
5
! 5.6753
0.0623
-0.1097 j 0.0038
10
I
20
1
5.9006
-0.0189 ' -0.0922 i 0.0027
1
2
3.0879
1.6430
-0.4472 ; 0.0262
1
5
3.4519
1.4161
-0.3754 , 0.0200_
10
3.6423
1.3388
-0.3509 ' 0.0177
20
3.3170
1.5906
-0.3955
0.0202_2
5.2707-- -- 0.1314
-0.0976
0.0025
-- -- -- -5
-0.0919
0.0031
-5.5722
-------0.0563
6.1060
-0.2520
-0.0253
5.0012
-- 10
-- - .
20
6.0081
-2.Ll73
-0.0574 I O.OOX_-0.0244
' -0.0008 ,
0
.
2
5
8
4
6.2652
50---______
- --3.2235
1.2714
-0.3268
0.0164 2
-0.0700
-0.00320.2745
5
j 4.5416
10
4.5184
0.2886
-0.0600
-0.0045
20
,
5.0785
-0.0820
0.0296 1 -0.01102- 5.1719-,
0.1558 _ -0.1093
0.0043 __
-0.1455 1 0.0068
5
4.8825
0.3871
10
I 5.1635
0.2268 . -0.1039 / 0.0039
20 i 5
-0.0968
0.0035
i 50 1 5.2780 0.2240
4.9302
0.2564 1 -0.1240 ' 0.0038
j
2
-0.0021 '
5.8216 / -0.2152 1 -0.0276
1
5
10
6.1841 ' -0.3856 1 0.0114
-0.0048
20
1 6.1591 -0.3188 j 0.0021 -0.00444
50
1 6.3582 -0.3823 / 0.0170 1 -0.0054
I
'

1957-1980

Crzefficients of :he :DF Polynom~alEquations

AP.!

(year)

'

---

Sarawak

B~ntulu

1953-1980

Sarawak
----

--

1964-1874

Kapt
----

.-

Sarawak

1951-1980

Kuch~ng

---

Sarawak

MI~I

i
I

13-14

I
I

II

1953-1980

Urban Stormwater Management Manual

Design Ra~nfall

APPENDIX i 3 . B DESIGN TEMPORAL PATTERNS

Tabie 13.B1
No. of
Time
Periods
2
3
6
12
8
6
6

Duration
(min)
10
15
30
60
120
180
360

Temporal Patterns - West Coast of Peninsular Malaysia

Fraction of Rainfall in Each Time Period

0.570
0.320
0.160
0.039
0.030
0.060
0.320

0.430
0.500
0.250
0.070
0.119
0.220
0.410

0.180
0.330
q.168
0.310
0.340
0.110

0.090
0.120
0.208
0.220
0.080

0.110
0.232
0.090
0.120
0.050

0.060
0.101
0.119
0.040
0.030

0.089
0.094

Time Period

U&n

3
4
Time Period

Stormwater Management Manual

0.017

Time Period

120 minute Duration

0.5-

Time Period

180 minute Duration

0.028

30 minute Duration

60 minute Duration

0.031

Time Period

Time Period

0.048

15 min Duration

0.057
0.030

360 minute Duration

Time Period

13-15

Design Rainfall

-1

Table 13.82

Temporal Patterns - E a n Coast nf seninsular Maiaysia

"

Fraction of Ramfall jn Each Time Period

(min)

Periods

10 min Duration

15 min Duration

Time Period

Time Period

9 1 0 1 1 1 2

Time Period

II

Time Period

Time Period

360 mlnute Duratlon

. --- .

--

05

120 minute Duratm

180 mlnute Duration

Time Period

30 minute Duration

Time Period

( # these patterns can also be used in Sabah and Sarawak, until local studies are carried out)
-13-16

-Urban Stormwater Management i%nua/

Design Rainfall

APPENDIX 13.C
13.C.1

Table :3.C:

WORKED EXAMPLE

Exampie of Caiculation of Daily and 5-day

Rainfall Totals Rainfall Station 45: 111,
Politeknik Ungku Ornar at Ipoh, Perak

Calculation of 5 minute Duration Rainfalls

Problem: Calculate the 5 minute duration, 20 year ARI

rainfall for use in a roof design in Kuala Lumpur.

L D a t e

Daily

1994 *
1996
Day ~ o t a l l Daily ( 5 Day Total

IS

Solution: Five minutes duration is shorter than the

period of validity of Equation 13.2. Therefore, refer to
the procedure for other durations in Section 13.2.7.
From Equation 13.2, for 20 year AM and 't= 30 minutes
20~30= 142.4 mm/hr and the corresponding
rainfall depth is
= 71.2 mm.

we obtain

20~'0

Similarly, Z0~60= 91.0 mmlhr and the corresponding

rainfall depth is
= 91.3 mm.
From Figure 5 in HPl-1982, P' for Kuala Lumpur is read
as approximately 100 mm. The corresponding FDfactor
from Table 13.3 is 2.08.

j (Intermedizte Data from 1511 to 19/f2 are not Shown)

Substituting these values in Equation 13.3,

i
j
Canvert this depth
Equation 13.4:

to

rainfall

intensity

using

20/12
21/12
22/12

0.0
0.5

25/12
/ 7.5
i
26/12
j 7.0
i- 27/12
0.0
' 3 / 1 2 ' 0.0
p-29/12 ~ - 7 0 . 0
!-- 30112
0.0
I
31/12

,r

0.0
0.5
0.5

j
I

1
j

0
1

0.5
8.0
14.5
14.5
14.5
7.0
0.0

7 -

0.0

9.5
0.0

0.0
0.0
0.5
2.5

0.0

1
j

I
1

/
!
!2351.0 1

43.5
1.5

15

12.0
21.5
21.5
30.5
20.0
20.0
11.0
13.5

3.0
3.0
46.5
47.5

L-__i-._.

13.C.2

Use of Daily Rainfall Data

/
Problem: Use daiiy rainfall records to calculate the 5-day
Compute the 25
rainfall totals for Ipoh, Perak,
percentile, 5G percentile and 75 percentiie 5-day rainfali
totals. The 75 percentile 5-day total is required for
design of a wet sediment basin, in accordance with
Chapter 39.
Solution: The results are presented in Table 13.C1.
Some of the intermediate lines of data have been
omitted. Calculations are carried out using statistical
functions in a spreadsheet such as EXCEL.

Total
1783.0 I
Missing days
9
j
25 percentile;X____-_

perce~tilei

? -. iqdicates Missing Dara

HYDROLOGICAL PROCEDURE NO. 4

MAGNITUDE AND FREQUENCY

OF FLOODS IN PENINSULAR MALAYSIA
(REVISED AND UPDATED)

JABATAN PENGAIRAN DAN SALIRAN

KEMENTERIAN PERTANIAN MALAYSIA

Hydrological Procedure No. 4

MAGNITUDE AND FREQUENCY

OF FLOODS
IN PENINSULAR MALAYSIA
(REVISED AND UPDATED)

1987

BAHAGlAN PARIT DAN TALIAIR

KEMENTERIAN PERTANIAN, MALAYSIA

Hydrological Procedure No. 4

MGGIMTUDE AND FREQUENCY

OF FLOODS
IN PENINSULAR MALAYSIA
(REVISED AND UPDATED)

1987

Contributor : Ong Chee Yan

Price: $lo/=

Drainage and Irrigation Department

Ministry of Agriculture
Malaysia.
First published in 1 974
Revised and updated in 1987

SYNOPSIS
This procedure is a revised and updated version o f the Drainage and Irrigation Department Hydrological
Procedure No.4 (1974) - "Magnitude and Frequency of Floods in Peninsular Malaysia". The Hydrological
Procedure No. 4 whlch was first published in 1974 was developed based on hydrological data up t o year
1970 and regional analysis was used to estimate design floods for Peninsular Malaysia.
This revised and updated version also estimates floods using the technique of regional ana'lysis. However,
an additional 10 years or more hydrological data ( u p to year 1982) was used t o establish the new flood
frequency regions for Peninsular Malaysia.

The regional analysis carried out in this procedure generally consists of the two major parts - (i) Development of a set of regional dimensionless flood frequency curves and (ii) Development of a set of regional
regression equations relating mean annual flood to the catchment characteristics (catchment area and mean
annual catchment rainfall).
Hence two maps are included in this ~ r o c e d u r e- Map 1 identifies the various flood frequency regions
(FF regions) in Peninsular Malaysia and Map 1 identified the various mean annual flood (MAF regions)
in Peninsular Malaysia. By knowing the flood frequency region and mean annual flood region a river basin
of interest belongs to, the design floods of the basin can be estimated using the regional flood frequency
curve and the regional M A F equation.
This procedure will be revised and updated again when an additional ten years of hydrological data is
available.

iii

CONTENTS
Page

INTRODUCTION
1.1

Regional Flood Frequency Analysis

1.2

Frequency Distribution Used

- Gwnbel Type I

DEVELOPMENT OF PROCEDURE
2.1

Selection of Catchments

2.2

Extraction of Annual Flood Data

2.3

Frequency Analysis of Individual Station Flood Data

2.4

Derivation of Regional Dimensionless Flood Frequency Curves

2.5

Derivation of Regional M A F (Mean Annual Flood) Equations

APPLICATION OF PROCEDURE
3.1

Method of Application

3.2

Worked Examples

ACCURACY O F PROCEDURE
4.1

Comparison with Observed Data

4.2

Comparison with W 4 (1974)

4.2.1 Comparison of Results using H P q 1988) and HP4( 1974)

RELIABILITY OF PROCEDURE
REFERENCES
APPENDIX I

List of Catchments and Catchment

Characteristics used in Procedure

APPENDIX II - Results of Individual Station Frequency

Analysis

MAP 1

Flood Frequency Regions of Peninsular Malaysia

MAP 2. - Mean Annual Flood Regions of Peninsular Malaysia

In the planning and design of water resources projects, engineers and planners are often interested to
determine the magnitude and frequency of floods that will occur at the project areas. An estimate of the
magnitude of a aood of a certain recurrence interval (commonly known as the "design flood") that is
likely to occur is fundamentd to ensure that economic engineering design with adequate standards of
safety can be achieved.
In Malaysia, streamflow records from gauged rivers offer a fairly accurate means of estimatig design floods
through the application of various statistical methods. However, not every river in Malaysia is gauged and
moreover, gauged rivers are only gauged at certain strategic points of the rivers. If a project is located in
an ungauged catchment with no streamflow records, then the design floods for the catchment will have
to be estimated by other flood estimation techniques.
Hydrologists have developed numerous techniques for estimating design floods, the main ones being the
rational method, unit hydrograph method, conceptual and statistical rainfall-runoff models and regional
frequency analysis. This procedure describes the use of regional flood frequency analyses to estimate
design floods for Peninsular Malaysia.
1.1

Regional Flood Frequency Analysis

The regional approach to flood frequency analysis has been widely used in many countries, in United
Kingdom (NERC, 1975), in United States (Riggs, 1973) and in New Zealand (Beable et ai, 1982). Regionalization or regional analysis is concerned with the extension of records from gauged sites of close proximity
to cover that of a region: it provides a means of applying information from gauged sites in one region to
ungauged sites in the same region. In regional frequency analysis, individual frequency curves from gauged
sites are averaged to form a regional curve which is postulated to apply to all catchments in the region.
In this study, the regional flood frequency analysis method used by the Natural Environmental Research
Council (NERC, 1975) is adopted. Basically, the method involves the development of two components:
(i) A set of dimensionless regional frequency curves relating QT/MAF to T where QT is the peak discharge of T-years recurrence interval, MAF is the mean annual flood or peak discharge and T is the
recurrence interval in years.
(ii) A set of regional regression equations relating the mean annual peak discharge to the catchment

characteristics of catchment area and mean annual catchment rainfall.

1.2

Frequency Distribution Used - Gumbel Type 1

A probability distribution commonly used in flood frequency analysis is the Gumbel Type 1 distribution.
This distribution has been adopted for regional flood frequency analysis in Unted Kingdom (NERC, 1975).
New Zealand (Beable et al, 1982) and other countries throughout the world. The Gumbel Type I distribution was adopted for flood frequency analysis in this procedure.
The return period or recurrence interval of any ranked flood in a Gumble Type 1 distribution is skewed
towards the mode of the theoretical distribution. The theoretical fit is then determined by the method of
moments and is in the form of the following equation (Haan, 1977):

where XT is the magnitude of the event having a return period of T years.

X
v-

is the arithmetic mean value of the magnitudes of the events.

is the standard deviation from the mean.
is the Chow Frequency Factor for the Extreme Value Type I distribution.

The data were also fitted using Log-Pearson 111 probability distribution. This distribution is recommended
by the United States Water Resource Council (USWRC,1967) for flood frequency studies in the U.S.A.
However, the Log-Pearson I11 distribution was not adopted for this procedure because:
(i) the curvature of this distribution varies greatly from one data set -to another making it difficult to
obtain a regional curve.
(ii) there is a high degree of skewness in the sample distributions.

(iii) 20%of the sample data failed the Smirnov-Kolmogorovgoodness-of-fit test.

2.

DEVELOPMENT OF THE PROCEDURE

The method used in developing this procedure is summarized below:

(a)

Selection of catchments.

(b)

Extraction of annual flood flow (annual peak discharge) data.

(c)

Frequency analysis of individual station flow data.

(d)

Derivation of regional dimensionless flood frequency curves.

(e)

Development of regional mean annual peak discharge (MAF) regression equations.

2 1.

Selection of Catchments

Streamflow records from al3 river stations operated by the Drainage and Irrigation Department were investigated for the study. A total of 61 stations with lengths of records vary from 8 to 36 years in Peninsular
Malaysia were selected for the analysis. The stations selected are listed in Appendix I.
The flow records from each station were assessed and selected based on the following criteria:
(i) There must be at least eight years of good quality data.
(ii) The catchment has not changed significantly'over the period of records because of urbanisation,
agricultural development or industrial development.
(iii) There is no substantial regulation of flow upstresm of the station due to r e s e ~ o storage
r
or diversion

of flow from the river.

(iv) The area of the catchment is greater than 20 square kilometres.
(v) There is no tidal influence at the station.
(vi) The catchment is predominantly rural.

2.2 Extraction of Annual Flood Data

The maxin~umpeak discharge in each year of record is extracted from D.1.D Streamflow Record Publ~cations (Pre-1960 to 1980). Each data sample is thoroughiy investigated for incompleted data (missing record
during any year). If records were missing for a particular year then they are examined to determine if the
missing records occur during the dry season (when the annual peak discharge is unlikely to occur). I f it is
suspected that the missing records includes the annual peak discharge then that particular year's data is
treated as missing.
2.3

Frequency Analysis of Individual Station Flood Data

The annual flood peaks that were collected for each station were reduced to the dimensionless form of
QJMAF. The MAF is d e f i e d as the arithmetic mean of the annual flood series:
MAF =
where Qi
n

Qi ............................................
(2)

= annual flood peak for the i th year

= number of records in years (number of annual flood peaks in the series)

The plotting positions of each sample is determined based on the following Gurnbel criteria (Haan, 1977):
(a)

The plotting position must be such that all observations can be plotted.

(b)

The plotting position should lie between the observed frequencies of (m - l)/n and mln where m
is the rank of the observation beginning with m=l for the largest value and n is the number of years
of records or the number of observations.

(c)

The return period of a value equal to or larger than the largest observation and the return period of
a value equal to or smallest observation should converge toward n.

(d)

The observations should be equally spaced on the frequency scale.

(e)

The plotting position should have an intuitive meaning, be analytically simple, and be easy to use.

The Weibull plotting position formula meets all the criteria stated above. Therefore, the pjotting positions
(recurrence interval in years) of each sample are calculated using the Weibull formula:

where T
n
m

= plotting position of the peak discharges in years

= length of record in years

= rank of the peak discharge in the series

The dimensionless frequency curve for each sample is obtained by plotting ratios of Q,/MAF against the
recurrence interval of QT using the Gumbel distribution. The Srnirnov-Kolmogorov Goodness-of-Fit Test
was used to test each distribution's fit to the data. Only those distribution which fit the data at 95% confidence limits were accepted.
A computer program was developed to fit the Cumbel distribution to the dimensionless annual flood
series by method-of-moments. The program also calculates the 95% confidence h i t s and test out the
goodness-of-fit. The program outputs the QT/MAF values where QT is Qi for return period T. The results
of' frequency analysis cdrried out on individual station's data are presented in Appendix 11.

2.4

Derivation of Regional Dimensionless Flood Frequency Curve

Stations which exhibited similar dimensionless frequency distribution were indentified and grouped into
various groups. The grouping of stations into regions is done by superimposing the dimensionless curves
together and examining the similarity of the curves. If a curve from one station lies closely to a curve
from another station then the catchments of these two stations are grouped together under one region.
Consideration is also given to catchments which are located closely to one another. Catchments in close
proximity with one another are more lkely to be classified under the same region.
Factors lrke climate, topography and hydrological characteristics which influence the flood flows in a river
basin are also taken into consideration before finalking the regional flood frequency boundaries. The
delineation of the flood frequency regions is guided by the Mean Annual Rainfail Maps (D.I.D, 1973,
Hydrological Region Maps (Goh, 1974), Average Annual Water Resources Maps for Peninsular Malaysia
(Teh, 1982) and topographical maps of Peninsular Malaysia published by Survey Department, Malaysia.
After the flood frequency regions has been established, the regional curve for each region is derived by
averaging the dimensionless curves of stations belonging to that region. The regional curve is then the
representative flood frequency curve for all rivers in that region.
A total of 6 flood frequency regions were established for Peninsular Malaysia. These regions are shown
in Map 1. and the regional curves belonging to each region are shown in Figure 1.
2.5

Derivation of Regional MAF (Mean Annual Flood) Equations

The magnitude of the M A F of a river basin is affected by both the physiographical, meteorological and
catchment characteristics of the basin. Catchment area, mean annual catchment rainfall, mean channel
slope, mean channel length and drainage pattern are some of the easily defined characteristics that could
affect the catchment's MAF.
In a study on the degree of correlation of catchment characteristics with the MAF (Nash and Shaw, 1965),
it was found that a combination of the catchment area and the mean annual catchment rainfall exhibits
the hlghest correlation with the MAF. Table 1 shows the correlation coefficient for different combination
of catchment characteristics with the MAF derived from the Nash and Shaw Study.

S'O

Edd

b.dd

Sdd

S'E

Correlation Coefficients of Catchment Characteristics with the MAF derived by Bash and Shaw

Table 1 :

Catchment Characteristics
1)
ii)
iv)

ARS
AR
AS
RS

v)

vi>

R
S

iii)

vii)

A
R
S

Coefficient of Correlation Squared

catchment area
.
mean annual catchment rainfall
mean channel slope

In this procedure, the catchment area and mean annual catchment rainfall are the catchment characteristics
chosen for the derivation of the MAF equations. These two parameters are easily available. The catchment
areas of streamflow stations used in the procedure are obtained from the D.1.D Hydrological Stations
Inventory (D.I.D, 1987). The mean annual catchment rainfall for catchments in Peninsular Malaysia were
abstracted from the Peninsular Malaysia Mean Annual Rainfall lsohyetal Map (1950-1975) (D.I.D, 1976)
by planirnetering of isohyets within each catchment.
The relationship existing between catchment characteristics and its MAF is assumed to be in the form of:
MAF = c A' R' ............................................ (4)
Where

MAF is the mean annual flood

A is the catchment area
R is the mean annual catchment rainfall

c,a,b are catchment characteristics constants to be estimated

The MAF equation is reduced to a multiple h e a r regression by transforming Equation (4) into its logarithmic form:

Log MAF = log c + a log A + b log R ............................................ ( 5 )

The residual r, from the equation (5) is defined as:
r

= log (MAF obs.)

- log (MAF est.) ............................................ ( 6 )

where
r is the log residual of the MAF equation
MAF obs. is the observed MAF
MAF est.. is the estimated MAF
The MAF regions are established using the same method adopted in D.1.D Hydrological Procedure No. 12
- "Magnitude and Frequency of Low Flows in Peninsular Malaysia". (Toong, 1985). Grouping of catchments into regions is done by separating the residuals into positive and negative residuals. Catchments with
positive residuals formed one region and catchments with negative residuals formed another region. Refmement is carried out by repeating the grouping in each region until an ideal number of MAF regions
is attained. Six final MAF regions are established and these regions are shown in Map 2. The MAF equations with the catchment characteristic constants and the correlation coefficients squared derived for each
region are Jisted in Table 2.

Table 2 : Regional Mean Annual Flood (MAF) Equations

Catchment Characteristics Constants
Region

MAF Equation

MAF 1

MAF = 0.6582 A

MAF 2

MAF = 0.9630 A

MAF 3

MAF = 0.1 192 A

MAF 4

MAF = 0.1048 A

MAF 5

MAF

MAF 6

MAF = 0.4783 A

0.0140 A

0.7901

0.6541
0.61 75
0.7177
0.79Srl
0.9066

Multiple Coefficient
of
Correlation Squared

0.1980

0.8093

R
R

R
R

3.0571
3.0224
5.0354

0.9463

Catchment Area in km.

Mean Annual Catchment Rainfall in metres.

NOTE : R is measured in metres.

APPLICATION OF PROCEDURE

3.

in the development of this procedure, many constraints were set by the nature of the hydrological data
used in deriving the regional flood frequency curves and the MAF equations. Therefore in the application
of this procedure, the catchment of interest should also satisfy the following criteria:
(i) The catchment must not be significantly regulated (by reservoir, diversion, etc).

(ii) The catchment must not be influenced by tidal effects.

(iii) The catchment area must be greater then 20 square kilometres.

(iv) The catchment is predominantly rural.
3.1

Method of Application

The method of application of this procedure to estimate design floods for an ungauged catchment involves
the following steps:
Step 1

Determine the catchment area A in square kilometres.

Step 2

Estimate the mean annual catchment rainfd R in metres (1000 x rnillimetres).

i)

The catchment mean annual rainfall can be estimated from available rainfall records
of D.I.D. rainfall stations within or near the catchment.

ii)

If no rainfall records are available, R can be estimated from the 1: 1,000,000

Mean Annual Rainfall lsohyetal Map (1950- 1975) for Peninsular Malaysia (D.l.D,
1976).

Note : The unit for R used in the MAF regression analysis is in metres.
Step 3

Determine the MAF region of the catchment from Map 2.

Step 4

Compute the MAF from the appropriare regional MAF equation.

Step 5

Determine the flood frequency (FF region) region of the catchment from Map 1.

Step-6

Obtain the dimensionless ordinates &/MAF from the appropriate regional flood frequency curves for the return periods required.

Step.7

Determine QT for the various return periods by multiplying the Q,/MAF factor by the
MAF obtain in Step 4.

Worked Examples

3.2

Example I

Determine the 30-year design discharge for an ungauged site on Sg. Batang Kali located
3 ~ Long. 101' 38' E.
at ~ a t . 23'N,

Step 1

Catchment area A 88 km2.

Step 2

Mean annual catchment rainfall R = 2.S50m (2550mm).

Step 3

From Map 2, the site is located in mean annual flood region MAF 3.

Step 4

The MAF equation for region MAF 3 is

MAF=0.1192 A 0 ' 6 1 7 s R
MAF = 0.1 192 (88

0.6175

'

) (~.5503~0571

MAF = 33.10 m3/s.

Step 5

From Map I , the site is located in flood frequency region FF2.

Step 6

The QT/MAF value for the 30-year return period of region FF2 is obtained from the
flood frequency curve in Figure I :QJ0/MAF = 1.94.

Step 7

Q30 = Q30/MAF x MAF

QjO = 1.94 x 33.10
3

Q30 = 64.21 m 1s.

Example 2 -

Results :

Derive the mean annual flood and flood frequency curve for Sg. Langat at Kajang (DID
Stn. no. 2917442) with a catchment area of 380 km and a mean annual catchment
rainfall of 2675mm (2.6751~1).

From Map 1 and Map 2 , the site is located in mean annual flood region M A F 3 and flood
frequency region FF3. Following the steps in section 3.1. the results obtained from regional
analysis are presented in Table 3. For comparison purposes, the results of the single station
analysis are also presented in Table 3.

Table 3 : Results of design peak discharges derived from regional analysis and single station analysis for
stn. no. 2917442
Method
of
Analysis

Q, for T-year recurrence interval (rn 1s)

MAF (m3/s)

T=2

T=5

T=IO

T=20

T=jO

T=100

Regional

94.59

88.91

127.20

153.24

177.83

210.00

233.64

Single Station

99.61

94.62

131.49

155.39

178.30

208.18

23 1 .I0

4.

ACCURACY OF PROCEDURE

Three d~fferentcomparisons are made to assess the accuracy of the procedure. The first comparison gives an
indication of how the 10-year design peak discharges estimated from regional analysis (using this procedure)
vary with .the 10-year peak discharges recorded by river stations throughout Peninsular Malaysia. The
second comparison shows the differences between design peak discharges derived from this procedure HP4
(1987) and design peak discharges derived from the old procedure, HPq1974). The last comparison gauges
the accuracy of regional analysis used in the procedure as compared to other methods of design flood
estimation used in flood studies for rivers in Peninsular Malaysia.

4.1

Comparison with Observed Data

In this comparison, the observed 10-year peak discharges from all the sixty one river stations and the
10-year peak discharges estimated using regional analysis in this procedure are plotted in the form of a
scatter diagram in Figure 2. It can be seen that no major discrepancies exist between the peak discharges
derived by regional analysis and that derived from single station frequency analysis. However, the scatter
diagram shows that 54.1 percent of the flood estimates made using this procedure tends to be overestimated. 77.0 percent of the 10-year peak discharges estimated from thls procedure are within the range
of 0.67 to 1.50 times the recorded values.
The percentage breakdown of the 10-year peak discharges for all the river stations derived from regional
analysis as compared to the 10-year peak discharges derived from single station frequency analysis is presented in Figure 3.

Figure

S c a t t e r d i a g r a m c o m p a r i n g Qlo v a l u e s o b t a i n e d f r o m this
p r o c e d u r e H P 4 (19871, Qpro a n d [rto v a l u e s o b t a i n e d f r o m
s i n g l e s t a t i o n f r e q u e n c y analysis of observed d a t a ( Qob,).

R a t i o o f Predicted QIO (regional malysis ) /observed

Qlo

Figure 3 : Frequency diagram showing the percentage breakdown of the ratios of Qlo values from regional
analysis t o Q1, values from observed records.

4.2

Comparison with HP4( 1974)

This procedure, HP;Y 1987) is developed totally independent of the old procedure, the first edition of the
D.I.D Hydrological Procedure No. 4 "Magnitude and Frequency of Floods in Peninsular Malaysia" (Heiler
and Chew, 1974). The old procedure was developed based on streamflow data from year 1948 to 1970.
Since then. an additional ten years or more data has been collected and made available for analysis. Hence,
this procedure utilizes a longer period of data (up to year 1982) in its analysis. Also, a different approach
in the method of' analysis has been adopted t o improve the accuracy and performance of the procedure.
The major difference of this procedure HPq1987) compared t o the old procedure HPq1974) are summarized below:
(i) Additional Streamflow Data Used

An additional ten to twelve years of streamflow data (up to year 1982) is used to derive the dimensionless flood frequency curves and the mean annual flood regression equations in this procedure.
(ii) Two Regional Maps
In this procedure, two regional maps are developed: one for the flood frequency regions (Map 1)
and the other for the mean annual flood regions (Map 2). The flood frequency regions are demarcated
for areas in which the set of dimensionless flood frequency curves (see Figure 1) applies whereas
the mean annual flood regions are demarcated for areas which share the same mean annual flood
regression equations coefficients (see Table 2).

(iii) Additional Characteristic of Mean Annual Catchment Rainfall used t o derive t h e Mean
Annual Flood Regression (MAF) Equations
Other than catchment area, an additional characteristic, the mean annual catchment rainfall is used
to derive the regional mean annual flood regression equations in this procedure. Unlke the old
procedure (HP4 1974) where only catchment areas is correlated with mean annual floods to derive
the regression equations coefficients, this procedure used both the characteristics of catchment area
and catchment mean annual rainfall to derive the mean annual flood regression equationtcoefficients.
(iv) Streamflow d a t a from different gauging stations used
Streamflow data from gauging stations used in this procedure are not all from the same stations
used in the old procedure, HP4 (1974). Streamflow data from stations that do not fit into the Gumbel
Type I Distribution and data from stations that exhibited great changes since 1970 (usually from
catchments of non-homogeneous nature) are not included in the anaiysis of this procedure.
4.2.1 Comparison o f Results using

HP4( 1987) a n d HP4(1974)

The mean design peak discharge estimated for three ungauged catchments at Sg. Batang Kali, Sg. Setiu
and Sg. Lipis using this procedure HPq1987) and the old procedure HPq1974) are listed below:

(i) Site: Sg. Batang Kal at Lat. 3" 23' N, Long. 101' 38' E
Catchment area A: 88 km2 (34 mile2)
Mean annual catchment rainfall R = 2.550m (2550mm)
HP4(1987): Flood Frequency Region FF2
Mean Annual Flood Region MAF 5
MAF = 0.1 192 (88'6'75) (2.5503.05
= 33.10 m3/s.

HP4( 1974): Flood Frequency Region F 4

Q2.33

= lo4 A

0.693

= 104 (34.693,

= 1197.70 ft3/sec

= 33.94 m3/s.

Table 4 :

Cornparison for results obtained from HPq1987) and HPq1974) for Sg. Batang Kali at Lat.
3" 23' N, Long. 101" 38' E
1

Hydrological
Procedure

Used

Region

Qr for T-year recurrence interval (m3/s)

FF

M AF

2.33

10

20

50

100

HPq1987)

FF2

MAF3

33.10

43.36

51.31

62.3 1

59.58

77.45

HP4(1974)

F4

33.94

39.03

43.44

46.84

5 1.S9

55.32

(ii) Site: Sg. Setiu at Lat. 5" 31' N,Long. 102' 44' E
Catchment Area A: 161 km2(62 mile2)

Mean annual catchment rainfall R = 3490 mm

HP4 (1987): Flood Frequency Region FF5

Mean Annual Flood Region MAF6
MAF= 0.4783 ~0.9066~0.9463

= 0.4783 ( 1 6 1 O . ~ ~(3.49$.9463)
~~)
= 156.35 m3/s.

HP4(1974): Flood Frequency Region F 4 ,

0.883
Q2.33 = 218 A
= 218(62.883)
= 8339.47 ft3/s
= 236.33 m3js.
Table 5 :

Comparison of results obtained from HPq1987) andHP.111974) for Sg. Setiu a i Lat. 5" 31'N,
Long. 102' 44'E
Region

Hydrological
Procedure
Used

FF

MAF

2.33

10

20

50

100

HP4( 1987)

FF5

MAF6

156.35

233.91

312.70

383.06

469.05

537.84

HP4( 1974)

F8

236.33

271.78

302.50

326.14

359.22

385.22

QT

for T-year recurrence interval (m3/s)

(iii) Site: Sg. Lipis at Lat. 4" 00' N, Long. 101" 40' E
Catchment area A: 130 ~ r (50
n mile2)
~
Mean annual catchment rainfall R = 2.200m (2200mm)
H . 4 (1987): Flood Frequency Region FF 1
Mean Annual Flood Region MAF5
MAF = 0.0140 (130-79s4) (2.2Oos. O 35
= 35.63 m3/s.

HP4 (1 974): Flood Frequency Region F10

= 1250A0 . 3 8 6 0

Qa.j3

= 1 2 5 0 ( 5 0 ~1. ~ ~ ~ ~
= 5658.64 ft3/sec
= 160.30 rn3/s.

Table 6 :

Comparison of results obtained from HP4.1987) and HPq1974) for Sg. Lipis at L.at.4" 00' N,
Long. 101" 40' E

Hydrological
Region
Procedure .
Used
FF
MAF

2.33

HPq1987)

FFI

MAF5

35.63

HPq1974)

F10

160.30

5.

QT

for T-year recurrence interval (m3/s)

10

20

50

100

45.96

54.16

62,35

72,69

80.52

203.58

242.05

275.72

323.87

354.26

RELIABILITY OF PROCEDURE

In statistical studies, error analyses are usually carried out to determine the reliability of data and methods
used. Hence, in this procedure, it is also essential that the statistical reliability of the estimated design
peak discharges be considered. Due to the complex nzture of the errors involved in this procedure, no
theoretical expression for the standard error is derived. For users to subjectively evaluate the accuracy of
any design floods estimated using this procedure, the various sources and causes of errors are discussed
below:

(i)

.Lower peak discharge record from stick gauges

The present automatic recorder stations (used in this procedure) that were established prior to 1960 were
all operated by stick gauges before being upgraded to the automatic recorder system. 23% of streamflow
stations presently used in this procedure are still being operated ~olelyby stick gauges.

For stick gauges, the water level is read manually twice a day (8 a.m. and 8 p.m. daily). During flood
times, the peak water-level could happen at any time between 8 a.m. and 8 p.m. causing the peak level
to be missed and not recorded. A lower peak water-ievel would be recorded. This error will eventually
result in the underestimation of the design flood estimated using this procedure.
(ii)

Poor quaIity rating curve and peak discharge records

Rwer cross-sections frequently change in urbanised catchments where rivers have been canalised and
widened. Other changes in river cross-section also happened, especially in rivers that are. subseptible to
scouring and silting. When a river has changed its cross-section, then the records will be rendered nonhomogenous.
The other factor that causes inacurrate peak discharge data is the extrapolation of rating curves. For certain
flood events, the peak water-levels recorded are beyond the range of the stage-discharge rating curves.
Hence, the rating curves would have to be extrapolated in order that the peak water-levels recorded can
be converted t o peak discharge value. .
Other factors like errors in data collection, errors in data analysis, non-functioning of water-level recorders
and inaccurate gauging measurements also attribute to the poor quality of peak discharge records of a
streamflow station.
(iii)

Different length of records

It is recommended in the U.S.G.S "Flood Frequency Analysis" (Dalyrmple, 1960) that all period of records
be adjusted to a common base period. The common base period, derived from the longest available records.
time are different in magnitude and distribution
is necessary because storms and floods of one p e r d
to those of another period. Hence, for different length of records, missing years may be filled in by corrolation techniques. The estimated peaks are not used directly but serve the purpose of allowing the correct
recurrence interval to be assigned to the recorded peak discharge data in order that the records can be
compared or combined.
In this procedure, no adjustment of the length of records to a common base period was carried out because
the variation in the lengths of records of streamflow stations used in this procedure is not high enough to
justify for the adjustment of a common base period. The average length of record for the streamtlow
stations used is 23.2 years, the shortest period being 8 ycrtrs and the longest period is 36 years. 36'3 o f the
length of records i s between 20 to 30 years. Hence the error attributed by the different length of recvrds
is relatively small.

REFERENCES
Beable, M.E and Mckerchar, A.1 (1982) "Regional Flood Estimation in New Zealand", q a t e r and
Soil Technical Publication No. 20.,Water and Soil Division, .Ministry
.
of Works and Development,
Wellington, New Zealand.
Department of Engineering Hydrology, University College, Galway Ireland "Flood Frequency Estimation for Ungauged Catchments", University of Ireland, Galway, Ireland, U.K.
Dalrymple, T (1960) "Flood Frequency Analyses", U.S.G.S Water Supply paper 1543-A, U.S
Department of The Interior, Washington D.C, U.S.A.
Drainage and irrigation Department, Malaysia (1976) "Hydrological Data - Rainfall Records for
Peninsular Malaysia 1970 - 1975", Ministry of Agriculture, Malaysia.
Goh, K.S (1974) "Hydrological Regions of Peninsular Malaysia" Water Resources Publication No.
2 . , Drainage and Irrigation Department, Ministry of Agriculture and Rural Development, Kuala
Lumpur, Malaysia.
Haan, C (1977) "Statistical Methods in Hydrology", The Iowa State University Press, Ames, Iowa,
U.S.A.
Heiler, T.D and Chew, H.H (1974) "Magnitude and Frequency of Floods in Peninsular Malaysia",
Drainage and Irrigation Degartment Hydrological Procedure No. 4, Ministry of Agriculture and
Fisheries, Kuala Lumpur, Malaysia.
Nash, J.E and Shaw, B.L (1965) "Flood Frequency as a Function of Catchment Characteristics",
k v e r Flood Hydrology Proceedings of Symposium, Institution of Civil Engineers, London, United
Kingdom.
Natural Environmental Research Counci1 (1975) "Flood Studies Report, Volume I
Studies", N.E.R.C, United Kingdom.

- Hydrological

Shahin, M.A (1980) "Statistical Analysis in Hydrology. Vol I and Vol II", International Institute
for Hydraulic and Environmental Engineering, Delft, Netherlands.
Teh, S.K and Kelsom, A (1982) "Average Annual and Monthly Surface Water Resources of Peninsular
Malaysia"' Water Resources Publication No. 12, Drainage and Irrigation Department, Kuala Lumpur,
Malaysia.
Toong, A.T (1985) "Magnitude and Frequency of Low Flows in Peninsular Malaysia", Drainage and
Irrigation Department Hydrological Procedure So. 12., Ministry of Agriculture, Kuala Lumpur,
Malaysia.
United States Water Resources Council (1976) "Guidelines for determining FIood Flow Frequency",
Bulletin No. 17 of the Hydrology Committee, U.S Water Resources Council, Washington D.C, U.S.A.
United States Water Resources Council ( 1 967) "A Uniform Technique for determining Flood Flow
Requency", Bulletin No. 15. of the Hydrology Committee, U.S Water Resources Council, Washington
D.C. U.S.A.

APPENDIX I
List of Catchments and Catchment Characteristics

Station
No.

Catchment

Sg Johor at Rantau Panjang

Sg Sembrong at Brizay Br~dge
Sg Bekok at Jln. Yong Peng Air Hitarn
Sg Kcsang at Chin Chin
Sg Lenggor at Bt. 4 2 Kluang Mersing
Sg Melaka at Pantai Belimbing
Sg Linggi at Sua Betong
Sg Pedas at Kg. Pilin
Sg Gemencheh at Jln. Gemas Rompin
Sg Muar at Buluh Kasap
Sg Muar at Bt. 5 7 Iln. Gemas Rompin
Sg Linggi a t Rahang
Sg Muar at Kuah Pilah
Sg Langat at Dengkil
S_e Langat at Kajang
Sg Semenylh at Semenyih
Sg Triang at Juntai
Sg Combak at Jln. Tun Razak
Sg Batu at Sentul
Sg Triang at Jln. Keretapi
Sg Selanpor a t Rantau Panjang
Sg Sernantan .4t Jam. Keretapi
Sg Pahang at Ternerloh
Sp Bentong at Jam. K. Marong
Sg Bernam at Tanjung Malim
Sg Bernam at Jam. SKC
Sg Trolak at Trolak
Sp Bil a t Jln. Tg.Malim - Slim
Sg Slim at Pekan Slim
Sg Sungkai at Sungkai
Sg Bidor a t Bt. 9 Jln. Anson
Sg Bidor at Bt. 18 Jln. Anson
Sg Lipis at Benta
Sg Batang Padang at T. Keramat
Sg Batang Padang at Tapah
Sg Gedong at Bidor
Sg J e h i at Jam. Bunggor
Sg Cherul a t Ban Ho
SF Ternbeling at Merting
Sg Kemaman at Ran. Panjang
Sg Kampar at Kg. Lanjut
Sg Kinta at Bt. Gajah
Sg Kinta at Kellas
Sg Perak at Jam. Iskandar
Sg Dungun a t Jam. Jerangau
Sg Plus at Kg. Lintang
Sg Ara a t Bt. 20 Jh.Taiping
Sg Krian at Dusun Rimau
Sg Ijok at T,iti Ijok
Sg Terengganu at Kg. Tanggol
Sg Krian at Selama
Galas a t Dabong
Sg Kulirn at Ara Kuda
Sg Muda a t Victoria Estate
Sg Muda at Batu Pekitka

~g

5405421
5505412
5506413

Length
Catchment Observed
Cdrchrnent Mean Annual
MAE
of
Rainthll
(13331s)
Records Area ( k m 2 )
(Years)
mm

Predicted
MAF
(17131s)

136.76
42.45
59.32
151.03
94.59
57.68
88.86
42.99
49.25
184.26
217.50
418.71
1489.43
85.80
66.21
201.73
34.33
24.16
ll6.!)9
85.61
112.70
105.87
333.02
123.07
116.34
51.45
1053.27
366.75
652.05
485.83
107.91
179.64
71.99
620.12
956.33
183.31
59.37
143.83
71.36
1925.59
131.27
3684.33
37.54
545.64
468.11

Region
FF

MAF

I'F4
FF3
FF4
FF3
FF3
FF3
FF3
FF3
FF3
FFI
FF2
FF1
FFI
FFI
FF2
FF2
FF2
FI-2
FF?
FF2
FF2
FF2
FF1
FF2
FF2
FF2
FF1
FF5
FFI
FF5
FF2
FF2
FF2
FF2
FF5
FF2
FF2
FF2
FF2
FF5
FF2
FF6
FF2
FF2
FF2

MAF4
MAF4
MAF4
MAF3
MAF3
MAF3
MAF3
MAF3
MAF3
MAFS
MAF3
MAFS
MAFJ
MAPS
MAF3
MAF3
MAFI
MAF2
MAF?
MAF2
MAF?
MAF2
MAFS
MAF2
MAFZ
MAFZ
MAFS
MAF6
MAFS
MAF6
MAF2
MAF2
MAF2
MAF2
MAF6
MAF2
MAFZ
MAFl
MAF2
MAF6
MAFl
MAF6
MAFl
MAFl
MAFl

5506416
550641 7
5721 442
5806414
6022421
6204421

Sg Sedim at Merbau Pulas

Sg Karangan at Titi Karangan
Sg Kelantan at Jam.
Guillermard
Sg Muda at Jeniang
Sg Kemasin at Peringat
Sg Padang Terap at Lengkuas

26
10

440
83

2835
3100

101.97
24.75

99.22
27.04

FF2
FF2

MhFl
MAFl

25
36
20
22

1 1900
1710
48
1270

2430
218.5
2875
1830

5236.86
310.26
49.69
185.69

5486.45
275.44
43.43
210.23

FF6
FF2
FF5
FF2

MAF6
MAFl
MAF6
MAFl

HYDROLIGAL PROCEDURES PUBLISHED

Price
H.P. No . 1 -

Estimation of the Design Rainstorm in Peninsular Malaysia (Revised and

updated, 1982) . .
..
..
..
..
..
..
..

..

MS 10.00

H.P. No . 2 -

Water Quality Sampling for Surface Water (1973)

..

..

H.P. No . 3 -

A General Purpose Event Water Level Recorder ( 1 973) . . . .

..

M$ 3.00

H.P. No . 4 -

Magnitude and Frequency of Floods in Peninsular Malaysia (1 974)

..

M$ 6.00

H.P. No . 5 -

Rational Method of Flood Estimation for Rural Catchments in Peninsuiar

Malaysia (1974) . . . . .
..
.
..
..
..
..

M$ 3.00

'

M$ 3.00

..

..

..

M% 3.00

..

..

..

MS 5.00

Field Installation and Maintenance of Capricorder 1599 (1971) . .

..

M% 5.00

H.P. No . 9 -

Field 111stallationand Maintenance of Capricorder 1598 (1974) . .

..

M$ 5.00

H.P. No . I0 -

Stage-Discharge Curves (1977)

..

MS 5.00

H.P. No . l l -

Design Flood Hydrograph Estimation for Rural Catchments in Peninsular

Malaysia ( 1 976) . .
..
..
..
..
..
..
..

M$ 6.00

Magnitude and Frequency of Low Flows in Peninsular Malaysia (Revised

and updated 1985)
..
..
..
..
..
..
..

MS20.00

The Estimation of Storage-Draft Rate Characteristics for Rivers in Peninsular Malaysla (1976)
..
..
..
..
..
..
..

MS 5.00

Graphical Recorders - Instruction for Chfirt Changing and Annotation

(1976)
..
..
..
..
..
..
..
..
..

M$ 5.00

H.P. ?io . 15 -

River Discharge Measurement by Current Meter (1 976) . . . .

..

MS 5.00

H.P. 30. 16 -

Flood Estimation for Urban Areas in Peninsular Malaysia ( 1 976)

MS 5.00

H.P. No . I7 -

Estimating Potential Evapotranspiration using The Penman Procedures

(1977)
..
..
..
..
..
..
..
..
..

H.P. No . 6 -

Hydrological Station Numbering System (1 974) . .

H.P. No . 7 -

Hydrological Station Registers (1974)

H.P.No . 8

1 P Yo . 1

H.P. No . 13 H.P. No . 14 -

..

..

..

..

..

..

MS 5.00

H.P. No . 18 -

Hydrological Design of Agriculture Drainage Systems (1 977)

..

..

M$ 5.00

H.P. No . 19 -

The Determination of Suspended Sediment Discharge (1977)

..

..

M$ 5.00

H.P. No . 2 0 -

Hydrological Aspects Related to Agricultural Planning and Irrigation

Design (1 977) . .
..
..
..
..
..
..
..

M$10.00

H.P. No . 21 -

Evaporation Data Collection using U.S. Class "A" Aluminium Pan (1981)

MS 5.00

H.P. No..22 -

River Water Quality Sampling ( I 9 8 1)

H.P. No . 23. -

Operation and Maintenance of Cableway Installations (1982)

H.P. No . 24 -

Establishment of AgrohydroIogical Stations (1982)

..

..

M$ 5.00

..

..

MS 5.00

..

..

..

M$ 3.00

..

..

..

M$ 5.00

H.P.No . 26 - Estimation of Design Rainstorm in Sabah and Sarawak (1983) . .

..

M$10.00

..

H.P.No . 25 - Standard Stick Gauge for River Station (1982)

..

..

..

HYDROLOGICAL PROCEDURE NO. 5

RATIONAL METHOD OF FLOOD

ESTIMATION FOR RURAL CATCHMENTS
IN PENINSULAR MALAYSIA

JABATAN PENGAIRAN DAN SALIRAN

KEMENTERIAN PERTANIAN MALAYSIA

Hydrological Procedure No. 5

RATIONAL METHOD
OF

FLOOD ESTIMATION FOR RURAL

CATCHMENTS IN PENINSULAR
MALAYSIA
(REVISED AND UPDATED)
1989

Bahagian Pengairan dan Saliran

Kementerian Pertanian
Malaysia

RATIONAL METHOD

FLOOD ESTIMATION FOR RURAL

CATCHMENTS IN PENINSULAR
MALAYSIA
(REVISED AND UPDATED)

CONTRIBUTORS:
AZMl MD. JAFRl
ZAHARI OTHMAN

Price:

DRAINAGE AND IRRIGATION DIVISION

MINISTRY OF AGRICULTURE
MALAYSIA

First Published in 1974

Revised and Updated in 1989

CONTENTS
Page

...

...

...

...

1.

INTRODUCTION

2.

THE RATIONAL METHOD AS A FLOOD ESTIMATION PROCEDURE

...

...

...

...

...

2.1

General ...

2.2

Essential Features of the Rational Method

3. THE INVESTIGATIONS

...

...

...

...

...
...

...

...

...

...

...

...

...

...

...

...

...

...

...

...

...

...

...

...

3.1

General ...

...

...

...

...

...

...

3.2

Methodology of the investigation ...

...

...

...

...

...

...

...

...

...

...

...

...

...

...

...

3.2.3 Estimation of the Average Intensity of the Design Storm

...

...

...

...

...

3.2.1 Design Sequence

...

...

...

3.2.2 Estimation of Time of Concentration

3.2.4 Estimation of Runoff Coefficient

...

...

...

...

3.2.5 Regional Runoff Coefficient C based in Flood Frequency Regions

4.

...

...

...

...

4.1

Comparison w~thDID HP5 (1972) . . . . . . . ...

...

...

...

4.2

Comparison with the Regional Flood Frequency Analysis

...

...

...

...

ACCURACY OF THE PROCEDURE

...

...

5.

LIMITATIONS OF THE PROCEDURE

...

...

6.

USE OF THE PROCEDURE

. . . . . .

...

...

...

...

...

REFERENCES

...

...

...

...

...

...

...

APPENDIX A .
Multiplying Factor to take Account of Catchment Development
APPENDIX 8 .
Check on Estimation of Tc Using Empirical Relationships

...

The Rational Method of flood estimation is widely used throughout the world. Recent studies have shown
that it is most useful if regarded as a statistical link between the frequency distributions of the design rainstorm
and the design flood. In this procedure, the statistical approach was adopted and used to prepare a flood
estimation for small rural catchments up to 100 square kilometres in Peninsular Malaysia. With the establishment of a few representative and experimental catchments and the increase in record length of some of
the catchments used in the previous study, more data are available for reviewing and bpdating the old prccedure. In this study, records from 20 small rural catchments with 5 years or more of continuous data were
analysed to provide new design data. Since the methodology adopted in the new edition was basically similar
to the previous edition, the presentation and arrangement are kept as similar as possible for the convenience
of the users.

1 . INTRODUCTION

This procedure is the result of a study into the applicability of the Rational Method of flood estimation for
small rural catchments in Peninsular Malaysia.
The use of the Rational Method for urban environment has worked reasonably well in many countries.
For rural catchments the use of Rational Method has received much criticism. Few overseas researchers
who have studied the method as a deterministic model and tested it with observed data tqund that the method
gave low accuracy when individual storms and resulting peak discharges were considered. However, studies
by French et al. (1 974) who examined the validity of the method have shown that statistically the method
served the purpose of engineering practice where peak discharges of a given frequency are linked with the
rainfall intensities of the same frequency.
Since the annual total expenditure on many small hydraulic structures such as bridges, culverts, diversion works and so on amounted from a hClndred thousand dollars to tens of million, the need to have a procedure to guide the practitioners to arrive at a more reasonable design of such structures is both urgent
and important. For this purpose DID HP NOS (Heiler. 1974) has been published and used as a basis for the
design of above structures by most practitioners.

2 . THE RATIONAL METHOD AS A FLOOD ESTIMATION PROCEDURE

2.1 General
The Statistical Rational Method for peak discharges can be written as:

where Q, is the peak discharge of a design flood in m3/s with return period T years selected using any
recommended design return periods applicable in Malaysia.
i T is the average intensity of the design rainstorm of duration normally taken as be~ngequal to
time of concentration, Tc and of return period T years in mmlhr.
A is the catchment area in km2.
C-is a dimensionless runoff coefficient normally considered to be a function of the catchment and
design storm characteristics and of return period T years.
and Tc, the so-called time of concentration referred above is defined as being the time taken for
a drop of water to travel from the most remote part of the catchment to the outlet design point.

2 . 2 Essential features of the Rational Method

The Rational Method is generally considered to be one of the best available flood estimation procedures
for small urban and rural catchment areas. However, much confusion has resulted in the past and is still
common today with regard to the principles underlying the method.
The Rational Method has been criticised because of its inability to reproduce particular flood events
when actual rainfall is used as the input. Such criticism implies an assumption that the method is a "deterministic" one to represent the physical operation on rainfall-runoff process. This is not the intended use of
the method. It has long been recognised the most realistic way to use the Rational Method is to consider
it as a "statistical" link between the frequency distributions of rainfall and runoff. As such it provides a mean
of estimating the design flood of a certain return period, and of a duration equal to the time of concentration.
Rewriting equation (1) in a slightly different form, the following equation results:

where C, is now taken to be dimensionless statistical link between the frequency of peak discharge,
q in m3/sper km2and mean intensity of the design storm ,i in mmlhr. Subsequently, for a particular catchment having adequate flood and intensity data, design values of runoff coeficients, CT can be derived by
the use of the formula

where q, is a peak runoff rate of return period T years derived from a frequency analysis of observed flood
and i, is the mean design storm intensity of duration equal to Tc derived from a frequency of storms of durat ion equal to Tc.
Using the relation in the above equa'tion a consistent increase in C, with increasing return period for
20 small rural catchments in Peninsular Malaysia as shown in Table 1 was found. This conforms with the
investigation by French et al. (1974) on 37 rural catchments in New South Wales, Australia.
3 . THE INVESTIGATION
3.1 General
It is the intention of this section to outline the general methodology employed for the interest of the general
user, rather than to describe in detail the developmen: of the procedure. The frequency analysis of the annual maximum flood data from the catchments are no: covered. Interested readers should refer to the revised and updated editions of DID Hydrological Procedure No. 1 and No. 4 for details

3 2 Methodology of the investigation

3.2.1 Design Sequence
In using the Rational Method for flood estimation the usual design sequence is as follows:
(a) estimate the critical duration of the design storm (made equal to TJ.
(b) compute the various values of mean intensity (i,) for duration = T,
(c) estimate the values of C, from storm and flood frequency regions,
for various values of Tp
(d) compute the estimates of peak discharge (aT)
using equation (1).
3.2.2 Estimation of time of concentration
An essential part of the Rational Method is the estimation of the time of concentration, Ti previously defined in para 2.1. While this is an unrealistic physical concept in a natural catchment, there is little doubt that
some "characteristic time" exists that is critical for particular catchment. This cannot be defined precisely.
and probably varies from season to season and from storm to storm. Various practical methods were proposed by previous researchers to estimate time of concentration. At the initial stage of the study, average
time lags between centroids of rainfall hyetographs to the peak of corresponding hydrographs as proposed
by Schaake et al. (1967) was attempted to estimate T-. Due to inadequate rainfall data for several catchments used in the study, derivation of Tc by this meihod was abandoned. The minimum time of rise of
a flood hydrograph that reflects total catchment contribution as proposed by Pilgrim et al. (1974) was selected
for this study. The minimum time of rise against duration of storm was studied for 20 small rural catchments
in Peninsular Malaysia. A typical plot of the relationship is shown in Figure 1 where the typical time of rise
was adopted as the time of rise for that particular catchment. In order to establish a relationship between
the time of concentration and the physiographic characteristics of the catchments, a multiple linear regression analysis was carried out by presupposing a relation of the form

= K A a S b .................... (4)
is the time of concentration in hours
is the length of the main stream in kilometres.
is the slope in percent from the main stream catchment boundry intersection to the design
point, measured along the main stream.
k, a and b are constants.

TJL
where T,
L
S

Transforming equation (4) into logarithmic form the equation which resulted from the regression
analysis was

with a multiple coefficient of correlation = 0.87 between the observed and estimated values of T-. The
graphical solution to this equation is presented in Figure 2.
3.2.3 Estimation of the average intensity of the design storm, i,
The method of estimating the design rainstorm contained in the revised and updated 010 Hydrological Procedure No. 1 (Fadhlillah et al. 1982) has been used through out this investigation for computing the
characteristics of the design storm for eaih of the study catchments. The only input for using this procedure
is the duration of the storm, which is made equal to Tc found from equation (4), and the geographical location of the design point. Note that the design intensity should be adjusted to take account of the reduction
in storm intensity with catchment area according lo Table 6, page 12 of the above procedure or Figure 6
page 10 of DID Water Resources Publication No 17.
3.2.4 Estimation of runoff coefficient, C
The runoff coefficient C in the Rational Method which is often regarded as a simple parameter is complex
and affected by various factors and processes. Several factors affecting runoff coefficients C are infiitrat~on
losses, variation of rainfall intensities, catchment storage, antecedent wetness and physical characteristic
of the drainage area. Various approaches have been made available to present design runoff coefficient,
C in tabular selection tables, graphical relations and simple recommended values which can be found in
various reference books. Most of these approaches have been based on engineering judgement and
experience rather than values derived from observed flood data. In this procedure, the approach used was
to derive C for various return periods from frequency analysis of observed flood data and design rainfall
intensities for 20 small rural catchments in Peninsular Malaysia.
3.2.5 Regional Runoff coefficient, C based on Flood Frequency Regions
Mean values of runoff coefficient for a return period of 10 years, C,, were computed for each region. For
the application of this procedure, Peninsular Malaysia was divided into 4 regions (Figure 3). These are Flood
Frequency Regions based on DID HP No. 4 (Ong, 1987) where runoff coefficients C within the same region
would not be much affected by the frequency distribution of flood peaks and the flood producing rainstorms.
Mean values of C,IC,, where C, are runoff coefficients of return period T years were computed. A family
of curves representing different regions was established as shown in Figure 4. Knowing the region in which
a particular project lies and using the appropriate regional runoff coefficient curve, the factor C-IC., IS obtained for any particular return period and hence C, can be obtained.
4. ACCURACY OF THE PROCEDURE
Three methods were employed to assess the accuracy of the new procedure. One method was the scatter
diagram (as shown in Figure 5) which compared the 10 years design peak discharges estimated using the
new procedure with the 10 years peak discharges of single station frequency analysis of observed data.
Another method was the comparison between the design peak discharges derived using the new procedure and the design peak discharges derived using the previous procedure (DID HP5, 1974). The third
comparison checked the accuracy of the procedure by comparing it with the regional flood frequency analysis
(DID HP4, 1987).
4.1 Comparison with HP5 (1972)
Table 1 compared the results obtained from DID HP5 (1988) and DID HP5 (1974) for two single stations whlch
are Sg. Durian Tunggal at Batu 11, Air Resam and Sg. Chalok at Jambatan Chalok. Although the values
estimated using HP5 (1974) gave higher value, the accuracy of the new procedure (HP5, 1988) should have
been improved for the following reasons:
(i)

Additional streamflow and rainfall data up to the latest record (1986) have been used wherever
possible.

Table 1 - Details of Study Catchments

-

TEM STATION
NO.

STATION NAME

CATCHMENT MAINSTREAM SLOPE AT

SOURCE
LENGTH
AREA

KIM"
Parit Madirono
Sg. Permandi at Bt. 27 J.B./Mersing
Sg. Mupor at Bt. 32 Jalan J.B.
Sg. Kahang at Ulu Kahang
Sg. Durian Tunggal at Bt. 1 1 Air Resam

Sg. Mantau at Kg. Mantau

Sg. Kepis at Jam. Kayu Lama
Sg. Pontian at Jam. Balak
Sg. Lui at Kg. Lui
Sg. Batu at Kg. Sg. Tua

Sg. Serai at Kg. Unchang

Sg. Tekam, Site A
Sg. Tekam, Site C
Sg. Gedong at Bidor
Sg. Cendriang at Bt. 32 Jln. Tapah
Sg.Chalok at Jam. Chalok
Sg. Lanas at Air Lanas
Sg. Kemasin at Peringat
Sg. Pelarit at Titi Konkrit
Sg. Buloh at Batu Tangkup

KM

1.96
6.44
6.44
21.6
14.48
6.4
11.0
11.9
15.2
14.48
12.07
0.99
1.34
24.0
10.1
7.0
15.2
17.4
10.8
6.17

0.50
0.90
0.87
3.30
1.50

3.40
1.49
0.47
4.03
7.98
0.20
0.013
0.80
4.99
2.07
0.83
3.80
0.06
1.52
0.60

DERIVED RUNOFF COEFFICIENTS(C)

Tc
(hrs.)

2.62
4.23
4.32
7.87
6.44
3.37
6.46
7.41
5.29
4.40
9.57
4.97
2.08
7.12
3.70
4.82
5.18
19.79

6.96
6.O4

HYDROLOGICAL PROCEDURES PUBLISHED

H.P No.

Title

Price

Estimation of the Design Rainstorm in Peninsular Malaysia

$1 0.00
(Revised and updated, 1982)
$ 3.00
Water Quality Sampling for Surface Water (1973) .. .. .. .. .. .. .. .. .. .. ..-.. .. ..
A General Purpose Event Water Level Recorder ~ a ~ r i c o r d e r $ 5.00
Mode1 1598 (1973)
$ 6.00
Magnitude and Frequency of Floods in Peninsular Malaysia (1974)
$ 3.00
Rational Method of Flood Estimation for Rural Catchments
in Peninsular Malaysia (1 974)
$ 3.00
Hydrological Station Numbering System (1974) .. .. .. .. .. .. .. .. .. .. .. .. .. .. ..
Hydrological Station Registeis (1974) .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. ..
$ 5.00
$ 5.00
Field Installation and Maintenance of Capricorder 1599 (1974)
$ 5.00
Field Installation and Maintenance of Capricorder 1598
Digital Event Water Level Recorder (1974)
$ 5.00
Stage-Discharge Curves (1977) .. .. .. . . .. .. .. . . . . . . .. . . .. .. .. . . . . .. .. .. .. . . . .
$ 5.00
Design Flood Hydrograph Estimation for Rural Catchments
in Peninsular Malaysia (1974)
$ 5.00
Magnitude and Frequency of Low Flows in Peninsular Malaysia (1976)
$ 5.00
The Estimation of Storage-Draft Rate Characteristies for
Rivers in Peninsular Malaysia (1976)
$ 5.00
Graphical Recorders Instructions for Chart Changing and
Annotation (1976)
$ 5.00
River Discharge Measurement by Current Meter (1976) .. .. .. .. . . .. .. .. .. .. ..
Flood Estimation for Urban Areas in Peninsular Malaysia (1976)
$ 5.00
$ 5.00
Estimating Potential Evapotranspiration Using the Penman
Procedure (1977)
$ 5.00
Hydrological Design of Agriculture Drainage Systems (1977)
$ 5.00
The Determination of Suspended Sediment Discharge (1977)
$10.00
Hydrological Aspects of Agriculture Plan'ning and
Irrigation Design (1977)
$ 5.CO
Evaporation Data Collection Using U.S. Class "A"
Aluminium Pan (1981)
River Water Quality Sampling (1 981) .. .. . . . . .. .. .. . . .. .. .. .. .. . . .. .. .. . . .. ..
$ 5.00
$ 5.00
Operation and Maintenance of Cableways Installation (1981)
$ 3.00
Establishment of Agro-hydrological Stations (1982) .. .. .. .. .. .. .. .. .. .. .. .. ..
$ 5.00
Standard Stick Gauge for River Station (1982) .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. ..
$10.00
Estimation of Design Rainstorm in Sabah and Sarawak (1983)

(ii) The rainfall regions have been divided into four as compared to only two in the previous study
(HP5, 1974). Further subdivision of rainfall regions is not possible due to the lack of small catchment data.
(iii) Streamflow stations other than the stations used in the previous procedure (HP5, 1974) have been
included. Most of the stations in this study (HP5, 1988) are recorder stations. These stations provided
continuous recorded data which should be more accurate in term of reliability than manual stations.
(iv) Most of the streamflow stations in this study have more gentle slopes compared to the stations used
in the previous study. This will be more representative of most of the practical river catchments in
Malaysia and the steeper slopes in the previous study may have contributed to higher flows.
Table 2
-

- Comparison of

- --

Q,, value for

Sg. Durian
Tunggal at Bt. 11
Air Resam
(m3/s)

Method

--

Results by Various Methods

O,, value for

!3g. Serai at
Kg. Unchang

(m3/s)

Ratronal Method (1974)

Rat~onalMethod (1 988)
Frequency Analys~sOf
Observed Data
Regtonal Flood Frequency
Analysis HP4, 1987

4.2 Comparison with the regional flood frequency analysis DID (HP4, 1987)
The comparison of DID HP5 (1988) to DID HP4 (1987) in Table 2 showed that DID HP4 (1987) estimated
a higher value. It should be noted that DID HP4 (1 987) used the extension of records and other informations
from gauged sites of close proximity to cover a region to get a regional mean annual flood equation.

5. LIMITATIONS OF THE PROCEDURE

From the theoretical basis of the Rational Method, two important factors are neglected. These are the effects of channel storage and temporal and spatial variations of rainfall intensities. Due to such limitations
and also that the procedure was derived utilizing data from rural catchments with areas ranging from 0.3
- 100 km2,the use of the procedure to estimate the runoff for larger areas is not recommended. Mult~plying factors to take account of catchment development (Appendix A) should serve as a general guide to arrive
at a reasonable estimate although there has been no study to substantiate this recommendation. The procedure may not be reliable to estimate the runoff for,areas with steeper slope. For general guidence the
limit is approximately from 0.01 - 8 percent; the slope values of which this procedure was developed. Like
any other flood estimation procedures, the design flood obtained from this procedure should be checked
with other available procedures and the decision to adopt the estimated design values should be complemented
by a sound engineering judgement.

6. USE OF THE PROCEDURE

6.1 Component of the procedure
The following items are required to use this flood estimation procedure:
(i) Figure 3,showing the general disposition of four regions proposed for application of Figure 3,
(ii) Figure 2, being a graphical solution to the Tc formula presented as equation (5),
(iii)' Figure 4, showing the relationship of mean frequency factor C,./C, for different regions and
(iv) . D.I.0 Hydrological Procedure No. 1 (1982) "Estimation of the Design Rainfall in Peninsular Malaysia".

2
4
Duration of

6
Storm

10

Figure 1 -Relationship
B e t w e e n Time
Rise and Duration of S t o r m
Sungai

12

(hours)

Gedong

at

Bidor.

of Hydrograph
For

510

20
30
40
Ca.tchment

50
60
Area I

70

urn2

80

90

100

110

F i g u r e 2 - T i m e of C o n c e n t r a t i o n G r a p h
G r a p h i c a l S o l u t i o n of Tc = 1 . 2 8 6 L
~ 0 . 2 2 3 SO-263

120

130

140

Legend

Discharge

Figure 3

Station

used

in

the

study

L o c a t i o n of study c a t c h m e n t s
R e g i o n a l A p p l i c a bil i t y o f t h e
procedure.

and

2
Re turn

Figure 4

Period

5
10
20
Years ( Normal Probability

Relation of mean frequency

for d i f f e r e n t regions.

SO
Scale

factor

Cy /CtO

q,o f r o m

Figure 5

procedure

i cumec I

km2 1

-Scatter
d i a g r a m c o m p a r i n g q,, v a l u e s o b t a i n e d
f r o m this p r o c e d u r e HP 5(1988) qpro a n d q,o
values obtained from frequency analysis of
observed

d a t a , qobs

6.2 Design Sequence

The detailed design sequence for using this procedure is as follows:
Step 1 For t$e design situation being considered select a design return period (T,).
Step 2 Estimate the critical storm duration, T, from catchment characteristics from Figure 2.
Step 3 Compute the depth of design storm o i duration, Tc for following return period T,, 2, 10. 20
years (Note that estimates for return periods of 2 and 20 years are required to make confidence
limit estimates of the design rainstorm).
Step 4 Compute the confidence limits of X(T,, T,) estimate using X(2, T,) and X(20, T,) values, (See
pp. 4-5 in DID Hydrological Procedure No. 1 1982).
Step 5 Estimate the value of C from Figure 4.
Step 6 Compute the estimate of design discharge QTtogether with confidence limits of the estimates,
using equation (1).
Step 7 Adjust Q, estimate according to future land use development using factor (F) shown in A p
~ e n d i xA.
6.3 Worked examples
Example 1:
A flood estimate is required for a point latitude 4'00' N and longitude 102"001E for a catchment possessing
the following' characteristics:
Area
slope
Length of mainstream
Development from jungle
Solutions:
Step 1
Step 2

= 25.90 sq.km.
= 3%

= 6.44 krn
= 40%

From any recommended hydrological procedure, choose

T = 10 years
From
=

Tc

1.286 x L
A0 223 x SO263
1.286 x 6.44
25.90223 x 30 263

= 3 hrs.

Step 3

From DID HP No. 1, find

X(2,3) = 78 mm
X(l O,3) = 122 mm
X(20,3) = 140 mm

Step 4

Confidence interval = 0.43 D

D = X(20) - X(2) = 140 - 78mm = 6 2 m m
0.43D = 0.43 x 62 = 26 mm

Step 5

i,

= X(10) = 122 +- 26:7 = 40.7 a 8.9 mmlhr

3

Step 6

From the curve frequency factor C&,,

for different regions, find for region 4

C,, = 0.38
Step 7

Q,, =

0.278 x C,, x i,, x A

24.35 m3/s

= 111.36
Step 8

From Appendix A multiplying factor for development

F = 1.05
Q,, = (1 17.36 x 1.05) ,+ (24.35 X 1.05)
= 116.92 + 25.57 m3/s

Example 2:
A flood estimate is required for a point latitude 4O00' N and longitude 102 "00' E for a catchment possessing
the following characteristics:
*
Area
Slope
Length of mainstream
Development for jungle
Solutions:
Step 1

= 6.44 km
= 40%

T = 10 years

Step 2

From Tc

=
=

Step 3

= 25.90 sq. km
= 1 O/O

1.286 x 6.44

25,902'23x 10263

4.0 hrs.

From DID HP No. I , find

X(2,4) = 80 mm
X(10,4) = 98 mm
X(20,4) = 105 mm

Step 4

Confidence Interval = 0.43 D

D = X(20) - X(2) = 105 - 80 = 25 mm
0 .43D = 0.43 x 25 = 10.7 mm

Step 5

i,, = X(1O) = 98 2 10.7

4
4
= 24.5 r 2.7 mmlhr

Step 6

From the curve frequency factor C,/C,,

for different regions, find for region 4

C,, = 0.38
Step 7

= 0.278 x C,, x i., x A

= 0.278 x 0.38 x (24.5 _t 2.7) x 25.9
= 67.03 + 7.39 m3/s

Cj,,

Example 3:
Obtain a flood estimate for a culvert on a main trunk road at location latitude 5O00' N and longitude 10300'
E on a catchment possesing the following characteristics:

= 5.18sq.km

Area
Slope
Length of mainstream
Development for natural vegetation
Solutions:
Step 1
Step 2

= 5O h
= 2.41 km
= zero

From any recommended hydrological procedure

choose T = 20 years
From Tc

1.286 x L
A0 233 x S
O 263

1.406 hrs.

Step 3

From DID HP No. 1, find

X(2, 1.406) = 75 mm
X(10, 1.406) = 118 mrn
X(20, 1.406) = 135 mm

Step 4

Confidence Interval = 0.43D

D = X(20)
X(2) = 135 75 = 60 mm
0.43 D = 0.43 x 60 = 25.80 mm

Step 5

i,, = X(10)

i2, = X(20)

= 118 +. 25.80 = 83.0

=

1.4
135 +. 25.80
1.4

= 96.0

,+

18.4 mmlhr

,+

18.4 mmlhr

Step 6

From the curve of frequency factors C,/C,

C,, = 0.38
C, = 0.429

Step 7

Q,, = C,, x
Q,

for different regions, find for regions 4

i,, x A
= 0.278 x 0.38 x (83 + 18.4) x 5.18
= 45.42 rt 10.06 m31s
= 0.278 X C,, x ,i x A
= 0.278 X 0.429 x (96.0 +- 18.4) x 5.18
= 59.30 -+ 11.36 m3is

Chee Yan, Ong "Magnitude and Frequency of Floods in Peninsular Malaysia" Min~stryof Agriculture Malaysia. Kuala Lumpur, 1987
French, et al. "Experimental Examination ot the Rational Method tor Small Rural Catchments" Civil Eng Transactions, Institution
o f Engineer, Australia, Month Vol. CE 16, 1974. pp. 95-102.
Heiler. T.D "Rational Method of flood Estimation for Rural Catchments in Peninsular Malays~a"Ministry of Agrrculture Malaysra,
Kuala Lumpur. 1974.
institution of Engineers. Australia "Austral~anRainfall and Runoff". 1977.
Mahrnood. Fadhlillah "Estimation of the Design Rainstorm, D 1 D Hydrological Procedure No. 1 ~ i n r s t r ~ o f ~ ~ r i c u~alaysra,
lture
Kuala Lumpur, 1982.
Schaake. John C "Experimental Examination of the Rational Method" Journal of the Hydraulic Divrsion, ASCE. Nov. 1967,
pp. 353-359.
"

APPENDIX A
MULTIPLYING FACTORS TO TAKE ACCOUNT OF
CATCHMENT DEVELOPMENT
Multiplying Q, estimates from
undeveloped area by factor
shown (F)

Development to Agriculture
from jungle in percent
0-25
25-50
50-75
75- 100

..........................................
..........................................

..........................................
..........................................

1 .OO
1.05
1.1 5

1.20

Source: DID HP5 i1974) page 7.

APPENDIX B
CHECK ON ESTIMATION OF Tc USING EMPIRICAL RELATIONSHIPS
-

STATION
N 0.

NOTE 1

Catchment
Area
k m2

Measured T, (in hours)

Time of using Bransby
Rise in
Williams
hour
Formula

T, (in hours)
using D.I.D.
Formula
(1974)

T, (in hours)
using D.I.D.
Formula
(988)
I

The Bransby-Williams Formula is widely used in Australia and Papua New Guinea for estimating
T, from catchment characteristics, and was the most successful of the overseas relationships tried in matching observed T, of Malaysian catchments. It is included in this appendix
for comparison purposes.

NOTE .2 - Bransby-Williams (Australian Rainfall and Runoff, 1977)

Tc =-

58.5L

A'''

(minutes)

so'*

L = Length of mainstream in km
A = Area in km2
S = Siope in mlkm.

HYDROLOGICAL PROCEDURE NO. 5

RATIONAL METHOD OF FLOOD

ESTIMATION FOR RURAL CATCHMENTS
IN PENINSULAR MALAYSIA

JABATAN PENGAIRAN DAN SALIRAN

KEMENTERIAN PERTANIAN MALAYSIA

Hydrological Procedure No. 5

RATIONAL METHOD
OF

FLOOD ESTIMATION FOR RURAL

CATCHMENTS IN PENINSULAR
MALAYSIA
(REVISED AND UPDATED)
1989

Bahaglan Pengairan dan Sallran

Kementerian Pertanian
Malaysla

RATIONAL METHOD

FLOOD ESTIMATION FOR RURAL

CATCHMENTS IN PENINSULAR
MALAYSIA
(REVISED AND UPDATED)

CONTRIBUTORS:
AZMI MD. JAFRl
ZAHARI OTHMAN

Price:

DRAINAGE AND IRRIGATION DIVISION

MINISTRY OF AGRICULTURE
MALAYSIA

First Published in 1974

Revised and Updated in 1989

CONTENTS
Page
iii
1,

...

INTRODUCTION

...

...

...

...

...

...

...

2. THE RATIONAL METHOD AS A FLOOD ESTIMATION PROCEDURE

...

...

...

2.1

General ...

2.2

Essential Features of the Rational Method

3. THE INVESTIGATIONS

...

..,

...
...

...

,..

...

...

...

...

...

...

...

...

...

...

...

...

...

...

...

...

3.1

General

...

...

...

...

...

...

3.2

Methodology of the investigation ...

...

...

...

...

...

...

...

...

...

...

...

., .

...

...

...

3.2.3 Estimation of the Average Intensity of the Design Storm

...

...

...

...

...

...

3.2.1 Design Sequence

...

...

...

3.2.2 Estimation of Time of Concentration

3.2.4 Estimation of Runoff Coefficient

...

...

..

...

3.2.5 Regional Runoff Coefficient C based in Flood Frequency Regions

...

...

...

...

...

4.1

Comparison with DID HP5 (1972) . . . . . .

...

...

...

.,.

4.2

Comparison with the Regional Flood Frequency Analysis

...

...

4. ACCURACY OF THE PROCEDURE

...

5.

LIMITATIONS OF THE PROCEDURE

...

...

...

...

...

...

6.

USE OF THE PROCEDURE . . . . . .

...

...

...

...

...

...

...

...

...

...

...

,..

...

REFERENCES

...

...

...

APPENDIX A - Multiplying Factor to take Account of Catchment Development

APPENDIX B - Check on Estimation of Tc Using Empirical Relationships

...

SYPNOSIS
The Rational Method of flood estimation is widely used throughout the world. Recent studies have shown
that it is most useful if regarded as a statistical link between the frequency distributions of the design rainstorm
and the design flood. In this procedure, the statistical approach was adopted and used to prepare a flood
estimation for small rural catchments up to 100 square kilometres in Peninsular Malaysia. With the establishment of a few representative and experimental catchments and the increase in record length of some of
the catchments used in the previous study, more data are available for reviewmg and hpdating the old p r ~
cedure. In this study, records from 20 small rural catchments with 5 years or more of continuous data were
analysed to provide new design data. Since the methodology adopted in the new edition was basically similar
to the previous edition, the presentation and arrangement are kept as similar as possible for the convenience
of the users.

1 . INTRODUCTION
This procedure is the result of a study into the applicability of the Rational Method of flood estimation for
small rural catchments in Peninsular Malaysia.

The use of the Rational Method for urban environment has worked reasonably well in many countries.
For rural catchments the use of Rational Method has received much criticism. Few overseas researchers
who have studied the method as a deterministic model and tested it with observed data found that the method
gave low accuracy when individual storms and resulting peak discharges were cons~dered.However, studies
by French et at. (1974) who examined the validity of the method have shown that statistically the method
served the purpose of engineering practice where peak discharges of a given frequency are linked with the
rainfall intensities of the same frequency.
Since the annual total expenditure on many small hydraulic structures such as bridges, culverts, diversion works and so on amounted from a hondred thousand dollars to tens of million, the need to have a procedure to guide the practitioners to arrive at a more reasonable design of such structures is both urgent
and important. For this purpose DID HP N0.5 (Heiler, 1974) has been published and used as a basis for the
design of above structures by most practitioners.

2. THE RATIONAL METHOD AS A FLOOD ESTIMATION PROCEDURE

2.1 General
The Statistical Rational Method for peak discharges can be written as:

where Q, is the peak discharge of a design flood in m3/s with return period T years selected using any
recommended design return periods applicable in Malaysia.
i- is the average intensity of the design rainstorm of duration normally taken as being equal to
time of concentration, T, and of return period T years in mmlhr.
A is the catchment area in km2.
C: is a dimensionless runoff coefficient normally considered to be a funct~onof the catchment and
design storm characteristics and of return period T years.
and Tc, the so-called time of concentration referred above is defined as being the time taken for
a drop of water to travel from the most remote part of the catchment to the outlet design point.

2.2 Essential features of the Rational Method

The Rational Method is generally considered to be one of the best available flood estimation procedures
for small urban and rural catchment areas. However, much confusion has resulted in the past and is still
common today with regard to the principles underlying the method.
The Rational Method has been criticised because of its inability to reproduce particular flood events
when actual rainfall is used as the input. Such criticism implies an assumption that the method is a "deterministic" one to represent the physical operation on rainfall-runoff process. This is not the intended use of
the method. It has long been recognised the most realistic way to use the Rational Method is to consider
it as a "statistical" link between the frequency distributions of rainfall and runoff. As such it provides a mean
of estimating the design flood of a certain return period, and of a duration equal to the time of concentration.
Rewriting equation (1) in a slightly different form, the following equation results:

where CTis now taken to be dimensionless statistical link between the frequency of peak discharge.
q in m3/s per km2and mean intensity of the design storm i, in mmlhr. Subsequently, for a particular catchment having adequate flood and intensity data, design values of runoff coeficients, C, can be derived by
the use of the formula

where q, is a peak runoff rate of return period T years derived from a frequency analysis of observed flood
and i, is the mean design storm intensity of duration equal to T, derived from a frequency of storms of durat ion equal to T,.
Using the relation in the above equation a consistent increase in C, with increasing return period for
20 small rural catchments in Peninsular Malaysia as shown in Table 1 was found. This conforms with the
investigation by French et al. (1974) on 37 rural catchments in New South Wales, Australia.
3. THE INVESTIGATION
3.1 General
It is the intention of this section to outline the general methodology employed for the interest of the general
user, rather than to describe in detail the development of the procedure. The frequency analysis of the annual maximum flood data from the catchments are not covered. Interested readers should refer to the revised and updated editions of DID Hydrological Procedure No, 1 and No. 4 for details.

3.2 Methodology of the investigation

3.2.1 Design Sequence
In using the Rational Method for flood estimation the usual design sequence is as follows:
(a)
(b)
(c)
(d)

estimate the critical duration of the design storm (made equal to TJ,
compute the various values of mean intensity (i,) for duration = T:
estimate the values of C, from storm and flood frequency regions,
compute the estimates of peak discharge (Q,) for various values of Tc
using equation (1).

3.2.2 Estimation of time of concentration

An essential part of the Rational Method is the estimation of the time of concentration, T, previously defined in para 2.1. While this is an unrealistic physical concept in a natural catchment, there is little doubt that
some "characteristic time" exists that is critical for particular catchment. This cannot be defined precisely,
and probably varies from season to season and from storm to storm. Various practical methods were proposed by previous researchers to estimate time of concentration. At the initial stage of the study, average
time lags between centroids of rainfall hyetographs to the peak of corresponding hydrographs as proposed
by Schaake et al. (1967) was attempted to estimate T-. Due to inadequate rainfall data for several catchments used in the study. derivation of T, by this me<hod was abandoned. The minimum time of rise of
a flood hydrograph that reflects total catchment contribution as proposed by Pilgrim et al. (1974) was selected
for this study. The minimum time of rise against duration of storm was studied for 20 small rural catchments
in Peninsular Malaysia. A typical plot of the relationship is shown in Figure 1 where the typical time of rise
was adopted as the time of rise for that particular catchment. In order to establish a relationship between
the time of concentration and the physiographic characteristics of the catchments, a multiple linear regression analysis was carried out by presupposing a relation of the form

= KAaSb .................... (4)

is the time of concentration in hours
is the length of the main stream in kilometres.
is the slope in percent from the main stream catchment boundry intersection to the design
point, measured along the main stream.
k, a and b are constants.

T,/L
where T,
L
S

Transforming equation (4) into logarithmic form the equation which resulted from the regression
analysis was

with a multiple coefficient of correlation = 0.87 between the observed and estimated values of Tc. The
graphical solution to this equation is presented in Figure 2.
3.2.3 Estimation of the average intensity of the design storm, i,

The method of estimating the design rainstorm contained in the revised and updated DID Hydrological Procedure No. 1 (Fadhlillah et al. 1982) has been used through out this investigation for computing the
characteristics of the design storm for eakh of the study catchments. The only input for using this procedure
is the duration of the storm, which is made equal to T+ found from equation (4), and the geographical location of the design point. Note that the design intensity-should be adjusted to take account of the reduction
in storm intensity with catchment area according to Table 6, page 12 of the above procedure or Figure 6
page 10 of DID Water Resources Publication No 17.
3.2.4 Estimation of runoff coefficient, C
The runoff coefficient C in the Rational Method which is often regarded as a simple parameter is complex
and affected by various factors and processes. Several factors affecting runoff coefficientsC are infiltration
losses, variation of rainfall intensities, catchment storage, antecedent wetness and physical characteristic
of the drainage area. Various approaches have been made available to present design runoff coefficient,
C in tabular selection tables, graphical relations and simple recommended values which can be found in
various reference books. Most of these approaches have been based on engineering judgement and
experience rather than values derived from observed flood data. In this procedure, the approach used was
to derive C for various return periods from frequency analysis of observed flood data and design rainfall
intensities for 20 small rural catchments in Peninsular Malaysia.

3.2.5 Regional Runoff coefficient, C based on Flood Frequency Regions

Mean values of runoff coefficient for a return period of 10 years, C,,were computed for each region. For
the application of this procedure, Peninsular Malaysia was divided into 4 regions (Figure 3). These are Flood
Frequency Regions based on DID HP No. 4 (Ong, 1987) where runoff coefficients C within the same region
would not be much affected by the frequency distribution of flood peaks and the flood producing rainstorms.
Mean values of C,IC,, where C, are runoff coefficients of return period T years were computed. A family
of curves representing different regions was established as shown in Figure 4. Knowing the region in which
a particular project lies and using the appropriate regional runoff coefficient curve, the factor (2-iC., is obtained for any particular return period and hence C, can be obtained.
4 ACCURACYOFTHEPROCEDURE

Three methods were employed to assess the accuracy of the new procedure. One method was the scatter
diagram (as shown in Figure 5) which compared the 10 years design peak discharges estimated using the
new procedure with the 10 years peak discharges of single station frequency analysis of observed data.
Another method was the comparison between the design peak discharges derived using the new procedure and the design peak discharges derived using the previous procedure (DID HP5, 1974). The third
comparison checked the accuracy of the procedure by comparing it with the regional flood frequency analysis
(DID HP4, 1987).
4.1 Comparison with HP5 (1972)

Table 1 compared the results obtained from DID HP5 (1988) and DID HP5 (1974) for two single stations which
are Sg. Durian Tunggal at Batu 11, Air Resam and Sg. Chalok at Jambatan Chalok. Although the values
estimated using HP5 (1974) gave higher value, the accuracy of the new procedure (HPS, 1988) should have
been improved for the following reasons:
(i)

Additional streamflow and rainfall data up to the latest record (1986) have been used wherever
possible.

Table 1

- Details of Study Catchments

TEM STATION
NO.

STATION NAME

CATCHMENT MAINSTREAM SLOPE AT

LENGTH
SOURCE
AREA

KM*
Parit Madirono
Sg. Permandi at Bt. 27 J.B./Mersing
Sg. Mupor at Bt. 32 Jalan J.B.
Sg. Kahang at Ulu Kahang
Sg. Durian Tunggal at Bt. 11 Air Resarn
Sg. Mantau at Kg. Mantau

KM

Yo

1.96

0.50

6.44

0.90

6.44

0.87

21.6

3.30

14.48

1.50

6.4

3.40

Sg. Kepis at Jam. Kayu Lama

Sg. Pontian at Jam. Balak
Sg. Lui at Kg. Lui

11.o

1.49

11.9

0.47

15.2

4.03

Sg. Batu at Kg. Sg. Tua

Sg. Serai at Kg. Unchang
Sg. Tekam, Site A

14.48

7.98

12.07

0.20

Sg. Tekam, Site C

Sg. Gedong at Bidor
Sg. Cendriang at Bt. 32 Jln. Tapah

0.99

0.01 3

1.34

0.80

24.0

4.99

10.1

2.07

Sg. Chalok at Jam. Chalok

7.0

0.83

Sg. Lanas at Air Lanas

Sg. Kemasin at Peringat
Sg. Pelarit at Titi Konkrit
Sg. Buloh at Batu Tangkup

15.2

3.80

17.4

0.06

10.8

1.52

6.17

0.60

DERIVED RUNOFF COEFFICIENTS(C)

The rainfall regions have been divided into four as compared to only two in the previous study
(HPS, 1974). Further subdivision of rainfall regions is not possible due to the lack of small catchment data.
(iii) Streamflow stations other than the stations used in the previous procedure (HPS, 1974) have been
included. Most of the stations in this study (HPS, 1988) are recorder stations. These stations provided
continuous recorded data which should be more accurate in term of reliability than manual stations.
(iv) Most of the streamflow stations in this study have more gentle slopes compared to the stations used
in the previous study. This will be more representative of most of the practical river catchments in
Malaysia and the steeper slopes in the previous study may have contributed to higher flows.
(ii)

Table 2

- Comparison of Results by Various Methods

Q,, value for
Sg. Ourian
Tunggal at Bt. 11
Air Resam
( rn3/s)

Method

Q,, value for

Sg. Serai at
Kg. Unchang

Rational Method (1 974)

Rational Method (1 988)
Frequency Analysis Of
Observed Data
Regional Flood Frequency
Analysis HP4, 1987

4 . 2 Comparison with the regional flood frequency analysis DID (HP4, 1987)

The comparison of DID HP5 (1988) to DID HP4 (1987) in Table 2 showed that DID HP4 (1987) estimated
a higher value. It should be noted that DID HP4 (1987) used the extension of records and other informations
from gauged sites of close proximity to cover a region to get a regional mean annual flood equation.

5. LIMITATIONS OF THE PROCEDURE

From the theoretical basis of the Rational Method, two important factors are neglected. These are the effects of channel storage and temporal and spatial variations of rainfall intensities. Due to such limitations
and also that the procedure was derived utilizing data from rural catchments with areas ranging from 0.3
- 100 km2,the use of the procedure to estimate the runoff for larger areas is not recommended. Multiplying factors to take account of catchment development (Appendix A) should serve as a general guide to arrive
at a reasonable estimate although there has been no study to substantiate this recommendation. The procedure may not be reliable to estimate the runoff for areas with steeper slope. For general guidence the
8 percent; the slope values of which this procedure was developed. Like
limit is approximately from 0.01
any other flood estimation procedures, the design flood obtained from this procedure should be checked
with other available procedures and the decision to adopt the estimated design values should be complemented
by a sound engineering judgement.

6. USE OF THE PROCEDURE

6.1 Component of the procedure
The following items are required to use this flood estimation procedure:
(i)
(ii)
(iii)'
(iv)

Figure 3, showing the general disposition of four regions proposed for application of Figure 3,
Figure 2, being a graphical solution to the Tc formula presented as equation (S),
Figure 4, showing the relationship of mean frequency factor CJC,, for different regions and
D.1.D Hydrological Procedure No. 1 (1982) "Estimation of the Design Rainfall in Peninsular Malaysia".

Duration o f

Storm

(hours)

Figure 1 -Relationship
B e t w e e n Time
R i s e a n d D u r a t i o n of S t o r m
Sungai

Gedong

at

Bidor.

of H y d r o g r a p h
For

510

20
30
40
Catchment

50
60 70 80
A r e a ( ~ r 1n ~

90

NO

110

F i g u r e 2 - T i m e of C o n c e n t r a t i o n G r a p h
G r a p h i c a l Solution of Tc = 1.286 L
~0'223

so. 263

120

130

140

Terenggan

*.

EGION

Oischarge

Figure 3

Station

used

in

the

study

L o c a t i o n of study c a t c h m e n t s
Regional Applicability of t h e
procedure.

and

Region

10 y e a r r u n o f f

10

20

SO

R e turn Period Y Years ( Normal Probability Scale 1

Figure 4

R e l a t i o n of m e a n f r e q u e n c y
for d i f f e r e n t r e g i o n s .

factor

Cy/CtO

0.5
q 10

from

1.0
procedure

2- 0

1.5
(

cumec I

km2

2.5

3.0

Figure 5 -Scatter
d i a g r a m c o m p a r i n g q,, v a i u e s obtained
f r o m this procedure H P 5(1988) qpro a n d q,,
v a l u e s o b t a i n e d f r o m frequency a n a l y s i s o f

observed

d a t a , q,

6.2 Design Sequence

The detailed design sequence for using this procedure is as follows:
Step 1 For the design situation being considered select a design return period (TI).
Step 2 Estimate the critical storm duration, Tc from catchment characteristics from Figure 2.
Step 3 Compute the depth of design storm of duration, Tc for following return period T,, 2, 10, 20
years (Note that estimates for return periods of 2 and 20 years are required to make confidence
limit estimates of the design rainstorm).
Step 4 Compute the confidence limits of X(Tl, TJ estimate using X(2, T,) and X(20, T,) values, (See
pp. 4-5 in DID Hydrological Procedure No. 1 1982).
Step 5 Estimate the value of C from Figure 4.
Step 6 Compute the estimate of design discharge QT together with confidence limits of the estimates,
using equation (1).
Step 7 Adjust Q, estimate according to future land use development using factor (F) shown in A p
pendix A.
6.3 Worked examples
Example 1
A flood estimate is required for a point latitude 4'00' N and longitude 102"OO' E for a catchment possessing
the following' character~stics:

= 25.90 sq.km.
Area
Slope
= 3%
Length of mainstream
= 6.44 km
Development from jungle = 40%
Solutions:
Step 1
Step 2

From any recommended hydrological procedure, choose

T = 10 years
From
=
=

1.286 X 6.44
25.90 223 x 33 263
3 hrs.

Step 3

From DID HP No. 1, find

X(2,3) = 78 mm
X(l O,3) = 122 mm
X(20,3) = 140 mm

Step 4

Confidence interval = 0.43 D

D = X(20) - X(2) = 140 - 78 mm = 62 mm
0.43D = 0.43 x 62 = 26 mm

Step 5
Step 6

From the curve frequency factor C,/C,,

C,, = 0.38

Step 7

Q,,

= 0.278 x C,, x i l o x A

= 111.36
Step 8

for different regions, find for region 4

24.35 m31s

From Appendix A multiplying factor for development

F = 1.05

Q,, = (1 11.36 x 1.05)

(24.35 X 1.05)
= 116.92 rt 25.57 m3/s

Example 2:
A flood estimate is required for a point latitude 4 '00' N and longitude 102 '00' E for a catchment possessing
the following characteristics:
m
Area
Slope
Length of mainstream
Development for jungle
Solutions:
Step 1

= 40%

T = 10 years

Step 2

From Tc
= 1.286 x L
~ 0 . 2 2 3x S
O 263
= 1.286 x 6.44
25,90223 x 10 263
=

Step 3

= 25.90 sq. km
= 1 O/O
= 6.44 km

4.0 hrs.

From DID HP No. 1, find

X(2,4) = 80 mm
X(10,4) = 98 mm
X(20,4) = 105 mm

Step 4

Confidence Interval = 0.43 D

D = X(20) - X(2) = 105 - 80 = 25 mm
0 .43D = 0.43 x 25 = 10.7 mm

Step 5

i,, = X(10) = 98 2 10.7

4
4

Step 6

From the curve frequency factor C&,,

C,, = 0.38

Step 7

Q,, = 0.278 x C,, x i., x A

= 0.278 x 0.38 x (24.5 _t 2.7) X 25.9
= 67.03 rt 7.39 m31s

for different regions, find for region 4

Example 3:
Obtain a flood estimate for a culvert on a main trunk road at location latitude 5'00' N and longitude 10300'
E on a catchment possesing the following characteristics:

= 5.18 sq.km

Area
Slope
Length of mainstream
Development for natural vegetation
Solutions:
Step 1
Step 2

= 5 O/O
= 2.41 km
= zero

From any recommended hydrological procedure

choose T = 20 years
From Tc

= 1.406 hrs.

Step 3

From DID HP No. 1, find

X(2, 1.406) = 75 mm
X(10, 1.406) = 118 mm
X(20, 1.406) = 135 mm

Step 4

Confidence Interval = 0.430

D = X(20) - X(2) = 135 - 75 = 60 mm
0.43 D = 0.43 x 60 = 25.80 mm

Step 5

i., = X(10) = 118

_+

25.80 = 83.0 2 18.4 mmlhr

1.4
,i

= X(20) = 135

+ 25.80 = 96.0
1 .4

18.4 mmlhr

Step 6

From the curve of frequency factors C,IC,, for different regions, find for regions 4
C,, = 0.38
c, = 0.429

Step 7

Q,,

Q,

= C,, x i,, x A
= 0.278 x 0.38 x (83

k 18.4) x 5.18
= 45.42 + 10.06 m31s
= 0.278 x C
, x ,i x A
= 0.278 x 0.429 x (96.0 2 18.4) x 5.18

= 59.30

11.36 m 3 k

REFERENCES
Chee Yan, Ong "Magnitude and Frequency of Floods in Peninsular Malaysia" Ministry of Agriculture Malaysia, Kuala Lumpur, 1987
French, et al. "Experimental Examination of the Rational Method for Small Rural Catchments" Civll Eng Transactions, Institution
of Engineer, Australia, Month Vol. CE 16. 1974, pp. 95-102.
3. Heiler. T.D "Rational Method of flood Estimation for Rural Catchments in Peninsular Malaysia" Ministry of Agriculture Malaysra.
Kuala Lumpur, 1974.
4
Institution of Engineers. Australia "Austrahan Rainfall and Runoff", 1977.
re
5 . Mahmood, Fadhlillah "Estimation of the Design Rainstorm, 0.I.D Hydrological Procedure No. 1' ' ~ i n i s t r k o~f ~ r i c u l t uMalaysia.
Kuala Lumpur. 1982.
6. Schaake, John C "Exprimental Examination of the Rat~onalMethod" Journal of the Hydraulic D~vrsron,ASCE. NOv. 1967,
pp 353-359.
1.

2.

APPENDIX A
MULTIPLYING FACTORS TO TAKE ACCOUNT OF
CATCHMENT DEVELOPMENT
Multiplying Q, estimates from
undeveloped area by factor
shown (F)

Development to Agriculture
from jungle in percent

Source: DID HP5 (1974) page 7

APPENDlX B
CHECK ON ESTIMATION OF Tc USING EMPIRICAL RELATlONSHlPS
STATION
N0.

NOTE 1

Catchment
Area
km2

Measured T, (in hours)

Time of using Bransby
Williams
Rise in
hour
Formula

T, (in hours)

using D.I.D.
Formula
(1974)

T, (in hours)
using D.I.D.
Formula
(1988)

The Bransby-Williams Formula is widety used in Australia and Papua New Guinea for estimating

Tc from catchment characteristics, and was the most successful of the overseas relationships tried in matching observed T, of Malaysian catchments. It is included in this appendix
for comparison purposes.
NOTE .2 - Bransby-Williams (Australian Rainfall and Runoff, 1977)
Tc =-

58.5 L

A'.'

(minutes)

so'*

L = Length of mainstream in km
A = Area in km2
S = Slope in mlkm.

HYDROLOGICAL PROCEDURES PUBLISHED

H.P No.

Title
Estimation of the Design Rainstorm in Peninsular Malaysia
(Revised and updated, 1982)
Water Quality Sampling for Surface Water (1973) .. .. .. .. .. .. .. .. .. .. .:
.. ..
A General Purpose Event Water Level Recorder ~ a p r i c o r d e r
M ode1 1598 (1973)
Magnitude and Frequency of Floods in Peninsular Malaysia (1974)
Rational Method of Flood Estimation for Rural Catchments
in Peninsular Malaysia (1 974)
Hydrological Station Numbering System (1974) .. .. . . .. .. . . . . .. .. . . .. . . . . .. ..
Hydrological Station Registeis (1974) .. .. .. . . .. .. . . .. .. . . . . . . . . . . .. .. . . . . .. . .
Field Installation and Maintenance of Capricorder 1599 (1974)
Field Installation and Maintenance of Capricorder 1598
Digital Event Water Level Recorder (1974)
Stage-Discharge Curves (1977) .. .. .. .. . . .. .. . . . . . . .. . . . . .. . . .. . . .. .. .. . . .. ..
.Design Flood Hydrograph Estimation for Rural Catchments
i n Peninsular Malaysia (1 974)
Magnitude and Frequency of Low Flows in Peninsular Malaysia (1976)
The Estimation of Storage-Draft Rate Characteristies for
Rivers in Peninsular Malaysia (1976)
Graphical Recorders Instructions for Chart Changing and
Annotation (1976)
River Discharge Measurement by Current Meter (1976) .. .. .. .. . . .. .. .. .. .. ..
Flood Estimation for Urban Areas in Peninsular Malaysia (1976)
Estimating Potential Evapotranspiration Using the Penman
P rocedu re (1977)
Hydrological Design of Agriculture Drainage Systems (1977)
The Determination of Suspended Sediment Discharge (1 977)
Hydrological Aspects of Agriculture Planning and
I rrigation Design (1977)
Evaporation Data Collection Using U.S. Class "A"
Aluminium Pan (1981)
River Water Quality Sampling (1 981) .. .. . . . . . . .. . . . . .. .. .. .. . . . . . . . . .. . . . . . .
Operation and Maintenance of Cableways Installation (1981)
Establishment of Agro-hydrological Stations (1982) .. .. .. .. .. .. .. .. .. .. .. .. ..
Standard Stick Gauge for River Station (1982) .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. ..
Estimation of Design Rainstorm in Sabah and Sarawak (1983)

Price

HYDROLOGICAL PROCEDURE NO. 11

DESIGN FLOOD HYDROGRAPH

IN PENINSULAR MALAYSIA

JABATAN PENGAIRAN DAN SALIRAN

KEMENTERIAN PERTANIAN MALAYSIA

HYDROLOGICALPROCEDURE No. 11

DESIGN FLOOD HYDROGRAPH

ESTIMATION FOR RURAL CATCHMENTS
IN PENINSULAR MALAYSIA

Jabatan Pengairan dan Saliran

Kementerian Pertanian Malaysia

HYDROLOGICAL PROCEDURE NO. 1 1

DESIGN FLOOD HYDROGRAPH

ESTIMATION FOR RURAL CATCHMENTS
IN PENINSULAR MALAYSIA

Contributors:
M.A.W. Taylor and Toh Yuan Kiat

DRAINAGE AND IRRIGATION DEPARTMENT

MINISTRY OF AGRICULTURE MALAYSIA

ACKNOWLEDGEMENTS
The work reported on here has been carried out by the Hydrology Branch of the
rainage and Irrigation Department (D.1.D) of the Ministry of Agrihlture and Rural
evelopment, in Association with the Engineering Export Association of New Zealand
c. (ENEX). The basic data used in this study has been collected by Federal and State
Wcers of the D.I.D., and by officers of the National Electricity Board (N.E.B). The use

f data supplied by N.E.B. is gratefully acknowledged.

First Printed in 1980

Reprinted in 1994
Price: RM6.00
Published by and available from:
Publications Unit, Ministry of Agriculture,
Wisma Tani, Jalan Sultan Salahuddin,
50624 Kuala Lumpur.
Tel: 03-2982011
1994

CONTENTS
Page
...

...

...

...

......

...

...

...

...
...
...
3. DEVELOPMENT OF PROCEDURE
...
... ...
...
...
...
...
...
3.1 General
...
3.2 The Design Storm
...
...
...
...
...
...
... ... ... ... ...
3.2.1 Recurrence Interval
...
...
...
3.2.2 Point Rainfall Depth and Frequency
...
...
...
3.2.3 Areal Distribution ...
...
.,.
...
...
...
3.2.4 Temporal Distribution
...
,. .
...
...
...
...
...
3.2.5 Duration
... ...
...
...
3-2.6 Design Storm Recommendations . . . . . .
...
...
...
3.3 The Estimation of Runoff from Rainfall
...
...
...
...
3.3.1 Rainfall - Runoff Analysis
...
...
...
3.3.2 The Rainfall - Runoff Relationship
...
...
...
...
3.4 The Time Distribution of Runoff
...
...
...
...
...
...
3.4.1 General
...
...
...
...
3.4.2 The Dimensionless Hydrograph ...
...
...
...
...
3.4.3 The Triangular Hydrograph
...
...
...
3.4.4 Triangular Hydrograph Geometry
...
...
3.5 The Estimation of Lg for Ungauged Catchment
3.6 The Estimation of D,, Tband T, for Ungauged Catchments
...
...
...
...
...
3.7 The Design Baseflow
...

...
...

1. INTRODUCTION

...

...

...

2. SPECIFICATION FOR PROCEDURE

...

...
..,

...

...
...
...

...
...

...
...

...
...
...
...
...
...
...
...

...

...

...

...

...

..

...

...

5.1 General
... ... ... ... ... .., .-.
5.2 Test One Comparison with Recorded Hydrographs
...
5.3 Test Two - Comparison with Flood Frequency

...
...

...

...

...

...

...

...

...

...

...

...

...

4. SUMMARY OF PROCEDURE

. . . . . .

5. ACCURACY O F PROCEDURE ...

...

...

6. LIMITATIONS O F PROCEDURE

7. THE TRIANGULAR UNIT HYDROGRAPH

...

...

...

...

...

,..

...

...

...

..,

...

...

...

...

...

...

- WORKED EXAMPLES ...

...

...

..,
...

...

...

References
List of Symbols

APPENDIX
A

...

APPENDIX
3 - DETAILS O F STUDY CATCHMENTS

...

APPENDIX
C - DATA USED TO DERIVE RAINFALL - RUNOFF
... ... ... ... ...
RELATIONSHIP ...
APPENDIX
D - DETAILS OF TEST CATCHMENTS

...

...

SYNOPSIS
Thc design ofmany water control structures quires a reliable and realistic &matt of the design
floodhydrograph. When no streamttow .records are available, as is often the case, the design flood hydrograph may be derived from a design storm.

This procedure presents a dettRninistic method of estimating the design flood hydrograph for ungauged Nllal catchments in Peninsular Malaysia. The procedure is based on the development of three
components: a design storm, a rainfall-runoffrelationship, and a triangular hydrograph. The procedure
was testcd on 12 gauged catchments and gave average results. The limitations of the proctdun an:discussed and a number of worked examples illustrating the use of the procedure are given.

DESIGN FLOOD HYDROGRAPH ESTIMATION

FOR RURAL CATCEMENTS IN PENINSULAR MALAYSIA
A probkm commonly encountered in the engineering ficld is the determination of the design flood. The
design flood may be defined as the flood adopted for the design of a water control structure after wmideration of the hydrologic and economic factors.
Design flood estimation using established methodology is relatively simple when records of streamflow and rainfall are available for the catchment conamed. The difficulties arise when no such records
are available in which case the designer is faced with two alttrnatives:

(i) To instrument the catchment for the period required to collect the hydrologicai data necessary to
&rive the design flood.
(ii) To estimate the design flood using a flood estimation procedure.
The former approach is generally time consuming and therefore expensive and is generally only
warranted on projects involving major capital expenditure. The latter approach is undoubtedly subject
to a greater degree of uncertainty, but nevertheless has to be used in the absence of any hydrological
data. Design flood estimates made using a flood estimation procedure should therefore be interprctad
sensibly within the limitations of the method, and where possible checked using any alternative flood
estimation methods available. Aitllren (1973) has reviewed existing methods of flood estimation in
common usage.

Two procedures for estimating tlhe design flood from ungauged rural catchments have r e a d y baen
adopted for use in Peninsular Malaysia -the Rational Method (Heiler I974 h) and the Regional Flood
Frequency Method (HeSles and Chew 1974). Both procedures have been compiled fiom flood studies on
Malaysian catchments.
The Rational Method and the Regional Hood Frequency Method provide a means of estimating the
design flood peak only. Although this is often suiiicient, the design of many engineering works requires a
consideration of storage upstream of the structure e.g. dam spillways, culverts with upstream ponding
etc. The complete desii flood hydrograph is therefore nwesaq for the determination of the dcsign
discharge at the point of interest.
This procedure complements th8etwo procedures mentioned previously in providing a method of
estimating the total flood hydrograph for ungauged w a 1 catchments. The procedure is not appticabla
to urban catchments.
2. SPECIFICATION FOR PROCEDURE

The nquircmcnts cons id^ mcwary for this procedure were that it should:
ti) Estimate the peak flow, the volume and time distribution of runoff for various rcmmnca
intervals.
(ii) Account for the significantdifferences in the catchment characteristics that affect floods.

(iii) Utilize catchment data that can be readily determined fiom topographical maps.
(iv) Be simple and nlatively fast to apply.

3. DEVELOPMENT OF PROCEDURE

Most synthetic procedures for estimating design flood hydrographs arc deterministic in that the design flood is &rived from a hypothetical design storm. A review of some of the more widely used proccdures for estimating design flood hydrographs has been made by Cordery et a1 (1970). Three basic
steps are common to this methodology of flood estimation:
(i) The specification of the design storm of which the important characteristics are usually the
recurrence interval, the total rainfall volume, the areal distribution of rainfall over the catchment, the temporal distribution of rainfall, and the duration of rainfall.
(ii) The estimation of the runoff volume resulting from the design storm.
(iii) The estimation of the time distribution of runoff from the catchment.
These three main components were studied in the developmelit of this procedure.
Over recent years there have been numerous and diverse techniques developed for estimating all of the
above components. However it is considered that the main limitation in the development of a design
flood hydrograph estimation procedure lies in the avaiIability of rainfall and streamflow data, rather
than any inherent limitations in the techniques used to develop the procedure. In this respect the problem is that there are very few major floods for which reliable rainfall and streamflow data are available,
particularly on small catchments. Any relationships developed are therefore based on data from relatively small storms, and hence the flood estimates are made from extrapolated relationships.
The techniques used in the development of this procedure have been adopted primarily to retain a
degree of simplicity commensurate with the data available.

3.2 The Design Storm

As with most deterministic flood estimation procedures it is assumed that the recurrence interval of
the design flood equals the design storm recurrence interval.
The choice of design recurrence interval reflects the severity of the potential damage in the event of
the design flood being exceeded. On large schemes the design recurrence interval is usually based on a
cost benefit analysis. On smaller schemes it is too difficult to quantify the costs pertaining to flood
damage and the benefits derived from flood alleviation, and the design recurrence interval is usually
selected somewhat arbitrarily. The design flood estimation procedure developed in this investigation is
intended to apply to smaller schemes where the latter approach holds.
Heiler and Tan (1974) have recommended design recurrence intervals for different types of water
control structures in Malaysia. Although these recommendations are tentative they serve as a useful
guide to the selection of design recurrence intervals for small schemes.
In cases where there is considerable risk of major damage and loss of life in the event of the design
flood being exceeded, it is usual practice to calculate the upper limit of the flood regime. This upper
limit is the probable maximum flood and it is derived from the probabk maximum storm.The probable maximum storm represents the upper limit of the design storm, for which the recurrence interval
is not defined.
The techniques for estimating the probabk maximum storm are beyond the scope of this procedure
since it is considered that the d e w of security involved warrants a fairly detailed hydrologkd d y s i s
specific to the project under consideration.

3.2.2 Point Rainfall Depth and hqpencr

A comprehensive depthduration-frequency study of storm rainfall throughout Peninsular Malaysia
has been compiled in the fonn of a hydrological design procedure by Heiler (1973). This procedure can
be used to estimate the depth of rainfall for a specified duration and frequency for any point location
in Peninsular Malaysia. In the procedure the short duration rainfall refers to intense bursts of rain
within a storm and not the total rainfall resulting from the storm.
The procedure is based on relatively short term data (5-13 years) from 59 recording raingauges and
32 daily raingauges distributed throughout Peninsular Malaysia. It is considered that the procedure
gives a reasonably reliable estimate of point rainfall depths, although it should be realized that local
variations are likely to occur due to the relatively thin population of available raingauges used to
develop the procadure.

3.2.3 Areal Distribotioe

During a storm, rainfall is usually distributed unevenly over the catchment area with rainfall depths
demasing with distance from the storm center. Large variations in rainfall depth can occur over
relatively short distances, particularly in areas where thunderstorm activity is common. This characteristic of rainfall in Malaysia has been noted by Dale (1959), although no studies on the areal patterns
of storm rainfall for small catchments in Malaysia have been found in the literature. A uniformly
distributed areal pattern has been adopted for this procedure using an areal reduction factor to reduce
the point rainfall estimate to an areal average.
In Table 8 (pg. 13) of his procedure, ~eilergives factors for the reduction of point rainfall to areal
mean rainfall for a range of raitlfalt durations and catchment areas. These factors are based on the U.S.
Weather Bureau recommendations, sine only a limited study of Malaysian storms has been done by
the D.I.D. (1970).
During the study reported on here, 92 storms on 18 different catchments were analysed to compute
the ratio of catchment mean rainfall to the maximum recorded point rainfall. The catchment mean
rainfall was calculated by the Thiessen weighting method. Because most of the study catchments were
equipped with only 1 recording raingauge, only 24 hour durations could be investigated using records
from the daily raingauges on the catchment.
The areal reduction factors for the study data are shown in Fig. 1 together with the U.S.Weather
Bureau recommendations. The plotted points define a typical rainfall depth-area trend with the con-

'

U S W o C u r t - 2 L ~ 0

6 k.0.m
- 1h~O..*n

,t

-fh
u
m

0
0

0
0

&
I

,"

0
0

0
0

0
0

FIO.1- WUAnON OFAREALBDUCTK]N

F#CTOR WITH CATCHMENT AREA

k.dmk,

siderable scatter in the points reflecting different storm rainfall characteristics and recurrence intervals.
The U.S. Weather Bureau curve for 24-hour duration rainfall forms an upper envelope containing most
of the study data. it is considered that this curve represents the likely upper limit of the variation of the
areal reduction factor with catchment area for 24-hour rainfall typical of the more severe flood-producing storms. The variation of the aeral reduction factor with catchment area for short duration rainfall as recommended by the U.S.Weather Bureau is also shown in Fig. 1. At present there is insufficient
data to deny the validity of these factors in Malaysia and they are considered satisfaciory for use in this
procedure.

3.2.4 Temporal Distributioo

The temporal distribution of the design storm is usually adopted from a study of the temporal distribution of recorded storms, as that pattein which produces the maximum discharge from the catchment
(Chow 1964 pg. 25-29). The D.I.D. (unpublished data) has compiled the temporal pattern for a large
number of storms of varying durations recorded at a number of rainfall stations in Malaysia. The
results indicate that the median temporal pattern is when 80 % of the total rainfall falls in the first 50 %
of the storm duration. A design storm temporal pattern of 80% of the rainfall in the first 60% of the
storm duration has been recommended for general flood estimation in Singapore by Chang (1969170).
He derived this pattern as the mean pattern from 377 storms recorded in Singapore. No relationship
was found between the temporal pattern and the duration and depth of rainfall.
The temporal pattern indicated by the D.I.D. data for Malaysia is not necessarily typical of the
storms producing large floods, since it is based on point sampling of small and large storms whose
potential flood-producing characteristics are unknown. The pattern also refers to the total rainfall
during the storm, not the intense bursts of rain within the storm on which the procedure by Heiler is
based.
It is shown later (section 5.2) that quite large errors are possible in predicting the volume of rainfall
excess from the storm rainfall. In view of this limitation and the lack of data on the temporal pattern of
large flood-producing storms, the refinement of a varying design storm temporal pattern was considered
to be not justified. The flood estimation procedure developed in this investigation therefore does not
include the temporal pattern as a design storm parameter.

3.2.5 Duration
The design storm duration is usually adopted as that duration which gives the maximum discharge.
This critical duration is found by trial and error by calculating the design flood for a range of storm
durations. A similar practice has been adopted in this procedure.

3.2.6 Design Storm Recommendations

The following procedure is recommended for estimating the design storm:
(i) The design recurrence interval is selected on the basis of the guidelines suggested by Heiler and
Tan ("Hydrological Design Return Periods", Provisional Hydrological Procedure, Drainage and
Irrigation Department, Malaysia 1974).

(ii) The design storm depth for the required recurrence interval is calculated from the procedure by
Heiler ("Estimation of the Design Rainstorm", D.I.D. Hydrological Procedure No. 1, Ministry
of Agriculture and Fisheries, Malaysia).
(iii) The areal reduction factors recommended by Heiler are used to convert point rainfall to catchment mean rainfall.
(iv) The temporal pattern of the design storm is not considered.

(v) The design storm duration is taken as that duration which gives the highest peak discharge.

FIG. 2

H't'ORO(XAR4 SEPARATION

3.3 The Estimation of Runoff from Rainfall

3.3.1 Rainfall-Runoff Analysis

The purpose of this section is to briefly describe thegeneral principles of rainfall-runoffanalysis. Since
there are considerable differences in the interpretation of common hydrological terms, it is import~dt
to define the meaning of the terms as used in this study.
Referring to Fig. 2 the discharge hydrograph is a graph of instantaneous discharge a t the catchment
outlet versus time. The area under the hydrograph over any specified time interval is runoff. The discharge hydrograph can be separated into two components direct runoff and baseflow. Direct runoff
refers to that volume of the runoff which reaches the catchment outlet soon after the causative rainfall.
The residual flow is baseflow.

Since direct runoff and basflow are empirical concepts, the method of baseflow separation is largely
empirical . In this study the baseffow was separated by drawing a straight line from the start of rise of
the hydrograph to an arbitrary point on the recession of the hydrograph. The arbitrary point was determined as the point after which the recession curve plotted as a straight line on log-artihmetic paper. The
log-arithmetic plot was done for 1 or 2 hydrographs for a catchment, and the remaining hydrographs
were separated by inspection, keeping the length of the baseflow separation line more or less constant.
The rainfall hyetograph (Fig. 2) is a plot, in discrete form, of the rainfall intensity over the catchment
versus time. The point rainfall intensity was extracted at hourly intervals from the recording raingauge
data and then adjusted linearly so that the area under the hyetograph equalled the catchment mean
rainfall over the same period. As mentioned previously the catchment mean rainfall was determined as
the Thiessen weighted mean of the 24 hour rainfall totals recorded by the daily raingauges.
The rainfall hyetograph can also be divided into several empirical components. The 4 index or loss
rate is defined as the rainfall intensity above which the volume of rainfall or rainfall excess equals the
volume of direct runoff. The volume of rainfall below this intensity goes to catchment recharge.

TIME

-A

FIG. 3 ANALMS OF URECT RUNOFF HYWOGRAPH

In flood studies the portion of the discharge hydrograph of major interest is the direct runoff hydrograph, and the portion of the rainfall hyetograph of interest is the storm rainfall (Fig. 3). The storm
rainfall is that portion of the rainfall hyetograph for which the intensity exceeds the (b index. It is
assumed that rainfall of intensity less than the 4 index does not contribute to direct runoff.
The time distribution of direct runoff can be measured objectively by the lag parameter Lg.As shown
in Fig. 3, Lgis the time from half the duration of rainfall excess to half the volume of direct runoff.
The concepts described above are largely empirical and obviously over-simplify the very complex
relationship between rainfall and runoff. Nevertheless they are fundamental to hydrological analysis
and are considered satisfactory for the purposes of this study.

3.3.2 The Rainfali-Runoff Relatiomhip

In practice there are two methods of deriving the volume of runoff from the volume of rainfall. The
first method is the loss rate approach in which an initial loss prior to the onset of direct runoff and a
continuing loss during the storm are abstracted from the design storm. The selection of a continuing
loss or 4 index for the design storm is usudly the most important factor in this approach.
In the second approach a rainfall-runoff relationship is developed from which the volume of runoff
can be estimated for any design storm volume. This approach was adopted for this procedure since it is
felt that the use of loss rates tends to obscure the difficulties associated with estimating volumes of runoff
for design storms which are usually outside the range of the observed data.

It is important for the rainfd-runoff relationship to be compatible with the &sign storm as estimated
from the procadwe described in section 3.2. Tbis procedure is based on a frequency analysis of intense
bursts of rain within a storm. Since only tbe intense rainfall is assumed to contribute to direct moff, it
is necessary to use storm rainfail as defined previously and not total rainfd in the development of the
rainfall-runoff relationship.

In this study 175 storms from 38 catchments in Peninsular Malaysia were analysed. For each storm
the volume of direct tunoff, the volume of storm rainfall and the 4 index were computed using the
methods described in section 3.3.1. Of the data andysed 97 stonns from 19 catchments were used to
develop the rainfail-runoff relationship. These data are listed in Appendix C. On the remaining catchments the sparse areal coverage of raingauges did not pcrmit reliable estimates of the storm rainfall and
these data were therefore not used.

The dincct runoff plotted against storm rainfall is shown in Fig. 4. The scatter of the points is to be
expected since the volume of runoff varies with other factors in addition to rainfail amount, such as the
catchment moisture status prior to the storm, the surface cover, the soil type and the intensity of rain-

fall. To account for some of these additional fsctors, multiple correlation methods have been used as
for example by L i ~ k et
y a1 (1958 pg. 173-178).
In this study an attempt was made to include the catchment antecedent moisture status in the rainfallauno~rclationlip.A 5-day antecedent rainfall indcx and a W o w index were tried Although the
latter indcx accounted for some of the scatter in Fig. 4 the results were not conclusive enough to justify
including an index of catchment antecedent moisture status in the rainfall-runoff relationship.

For practical purposes it is useful to express the rainfall-moff relationship as a mathematical

equation rather than graphically. Several forms of equation have been reportad by Boughton (1966)

who used a hyperbolic tangent function and Chow (1964 pg. 21-28) who describes an empirical equation
developed by the U.S. Soil Conservation Service and of the form:

where Q = direct runoff (ins.)

P, = total rainfall during the storm minus the initial loss (ins.)
I = potential infiltration (ins.)

The value of I is dependent on the soil type, ground cover and the antecedant moisture conditions.
A similar form of equation has been adopted for this procedure except that the storm rainfa11 P is
used in place of Pe,and a single value of 1 is used. The choice of equation was somewhat arbitrary since
there were insufficientdata to indicate the suitability of any one type of equation. However the above
equation is attractive in that the proportion of estimated runoff relative to rainfall increases as the storm
rainfall increases. This is logical since the catchment recharge is high in the early stages of a storm and
dyreases to a more or less constant rate as the storm duration increases.
In Fig. 4 the equation was fitted to the observed data by eye giving emphasis to the relatively few
points representing the targer floods analysed. The fitted curve does not match the observed data for the
smaller storms, and for storm rainfall below 3 inches the linear relationship shown in Fig. 4 is recommended. The equations for estimating direct runoffQ frorr: . ..In rainfall P are :
for storm rainfall less than 3 inches,

= 0.33P

for storm rainfall greater than 3 inches,

P2

Q=-

(P+6)
Because of the scarcity of data for major storms, the uncertainties associated with predicting runoff
volumes from rainfall are clearly evident in the relationship shown in Fig. 4. Comparison with other
rainfall-runoff studies in Malaysia by Goh (19721, Tan (1967) and Charlton (1964) indicate that the
relationship developed is reasonable within the range of data available.
A comparison with overseas data is available from the review made by Pilgrim (1966) of loss rates for
catchments in the U S A . , Australia and New Zealand. Frequency diagrams of loss rates for each
country are shown in Fig. 5, together with a frequency diagram of loss rates computed in this study.
Since the methods used to derive the loss rates are empirical and therefore to some extent subjective,
only general comparisons can be drawn from Fig. 5. However it is obvious that loss rates for Malaysian
catchments are substantially higher than loss rates observed in the U.S.,Australia and New Zealand.
In terms of the rainfall-runoff relationship this means that the volume of direct runoff as a proportion
of storm rainfall is considerably less for Malaysian catchments than for catchments in the U.S.,Australia and New Zealand.
3.4 The Time Distribution of Runoff

There are several methods of distributing the direct runoff volume with time of which the best known is
probably the unit hydrograph. Numerous studies of the unit hydrograph have been made and despite
some criticism of its theoretical validity, the unit hydrograph is widely used in design flood estimation.
One advantage of the unit hydrograph is that it can be used to distribute runoff from storms of varying
temporal pattern. The unit hydrograph also preserves the curvilinear shape characteristic of observed
hydrographs. A practical disadvantage of the unit hydrograph is that it is fairly tedious to apply.
In this procedure a triangular distribution of direct runoff is adopted. The triangular hydrograph is
very simple to apply and reproduces the hydrograph shape sufficiently accurately for design purposes. It
is developed from the dimcnsiodess hydrograph described in the following section.

The basic assumption underlying the dimensionlesshydrograph method as described in the U.S. Bureau
of Reclamation Manual ("Design of small Dams" pg. 39-41) is that the shape of the direct runoff
hydrograph reflects the influence of all the catchment characteristics on the flood-producing rainfall.
Conversion of the direct runoff hydrograph to a dimensionless form eliminates most of the effccts of the
catchment charac~ristics
and enables the comparison of the essential features of the direct runoff hydrograph for different catchments. The ordinate and abscissa of the dimensionless hydrograph are defined
as:

dimensionless abscissa =

tx100
-(L,x Dl21

(3.4)

where q(t) = discharge otdinata of the d i d runoff

hydrog&ph (cusacs)
t = time (hrs)

La = catchment lag (hrs)

D = duration of storm rainfall (hrs)
A = catchment a m (sq. mls.)
Q = direct runoff (ins.)
Equations (3.3) and (3.4) describe a curvilinear dimensionless hydrograph. Because the dimensionless
form 3argdy eliminates the e f l ' of catchment characteristics, the dimensionless hydrographs for
catchments having similar flood-producing features show certain similarities. A representative dimensionless shape can be derived which is characteristicof any group of hydrologically similar catchments.
It is assumed that this dimensionless hydrograph can be applied to similar ungauged catchments.

F& practical purposes the important shap parameters of tbc direct runatfbydrograph am the peak
flow snd some measure of the time distribution of which L, is a convenient index. The direct runoff
hydrograph can thus be representad by a triangular hydrograph as shown in Fig. 6.

The p a k discharge of the triangular hydrograph is derived from equation (3.3) as

,-

Dp~ . ~641)
4 YQ
qD = -(jgf
~j
where qp = peak discharge of the triangular hydrograph (cusecs).

D,

(3.5'1

peak ordinate of the dimensioniess hydrograph that is

characteris& of the catchment.

Catchment lag time is a function of the catchment characteristics and Dpis typically constant for any
group of hydrologically similar catchments. Therefore if D, is known and Lgcan be estimated for an
ungauged catchment, the peak discharge can be estimated and the triangular hydrograph constructed
for any storm. The assumption made here is that the lag time is constant for a given catchment and is
independent of the storm characteristics.

3.4.4 TrianguIar Hydrograpb Geometry

For a givm runoff volume the shape of the triangular hydrograph is fully defined by the two parameters
q, and Lg.However for case of constructing the triangular hydrograph, the base time Tband the time
of rise Tp of the triangular hydrograph can be derived as foliows :

Starting with equation (3.5) put (Lg+D/2) 3 C

DPxQxAx64O
q, =

The volume of the triangular hydrograph (Fig.6) b

FIG. 7

- TRIANGULAR HYDROGRAPH GEOMETRY

~ubstitntin~
for q, fmm equation (3.6) gives
+~T.X%~QXAX&

640xAxC

With the basc of the triangular hyhgrapsl-h d thm arc the two possibilities as shown in Fig.7 that
Tpis either less than C [case (i)] or goaraer Lhan C [case (u)].

Substituting for Q from equation (3.7) gives

2 x (Tb-qa
Tr . ~ i
Tb
uow Tp Tb-Tr

Equations (3.9) and (3.10) are only valid if Tpis lcss than C.
Theref6t.efrom equation (3.10)
(ab-20-Tb8)<mb
Tbg-3rnb-k X a > O
(T~-~C)X~~-C)>O
So Tb must be greater than 2C.
From equation (3.8)

So when the dimensionless peak is lcss than I, Ta is grcater than 2C and Tp is less than C. This cornsponds to the normally observed shape of the direct runoff hydrograph where the recession limb is longer
than the rising limb.
@)

Case (ii)

If Tb <2C then from equation (3.8)

This implies a hydrograph shape w h e n the recessionlimb is shorter than the tising limb. The quation
for T, is derived from Fig. 7 as follows:

f x Q= f X

cG

QP

Substituting for Q from equation (3.7) gives

Tp I:

2 0

TB

The triangular hydrograph can be easily constructed from the three parameters qp, Tb and Tp as
estimated from equations (3.5). (3.8) and (3.10) or (3.1 1) depending on the value of Dp. Assuming that
the design storm is fully ddined, the only unknowns in these equations are L, and Do. Methods for
estimating these parameters for ungauged catchments arc described in the following sections. It should
be noted that for a given value of Dpthe ratio of Tp to Tb is amstant.

Valucll of catchment lag were derivedfw 38catchments in R t h d a r Malaysia.For each eatchmtat a

number of om (1-11) wcmanalysd by the methods described in scdhn 3.3.1, resultingina r q
of LI
for each catcheat.

Tbis variation in lag time for Australia catchments har been studied by Askew (1970). who found
that the lag time decraassd as tbs man di8charge rate of the flood hydrograph inmassd Tha variation
in catchmnt kg times for diff'at storms in Malaysia has beem discussad in the Pahang River Basin
Study (1974). Tht physical explanation proposed k that subsurface flow is a major compoaeat of the
flood hydmgraph in humid tropical onas where most of the rain that contributes to direct runoff tntm
the soil. Subsurfaa flow has a bighly.variabk: response to rainfall &pending mainly on the antecadtnt
wctaass of tile various soil layers.

The variation in catchmart lag dues derived in this study is considered to k partly due to the
variable subsurface rnponse and partly due to a partid area storm effect. Depending on the spatial
distributionof rainfall and antbccdent soil m o h , oaiy part of the catchment may contribute to direct
runoff during a storm. Lag times derived from partial area storms rrflect the lag characteristics of the
anoff-gmmting part of the catcbmt only. The median of the lag times derived for each catchment
was taken to be tbe npnsentativcr value for that catchment. The median lag values for the study catchments are given in Appendix B.
Catchmat lag is a quantitative measure of thc infiuencc of catchment storage in modifying the shape
of the rainfall excess hyetograph. It is to be expected therefore that catchment lag should be related to
those physical chamcbristics that determine the storage behaviour of the catchment. A number of
studies relating some measure of catchment lag to catchment characteristics haws been carried out
previously. Examples arc by Say& (1938). Taylor and Schwan ( 1952), McSparran (1 968), Cordcry and
Webb (1974). 'lie relationship adopted for this procedure is of a similar form to that proposed by
Linsley et a1 (1958 pg.
- 207) as

L.

LL
a

ctX(-)

"

(3.12)

where L = main stream length from the outlet to the c a t c h ~boundary

t
(miks)
Lc = main stream length from the outlet to the catchment centroid (miles)
S = weighted mean stream slope (fk/mf.)
G, n arc constants.

The product L L is a measure of the size and shape of the catchment, and S is a masure of catchment
topography

The 38 catchments studied were arranged into three rcpr~~~lltative

hydrological groups. The group
were selected partly on the basis of similar topographical characttristics and partly on a regional basis.
For each group, values of Ls were plotted against LL&S as shown in Fig. 8. Initially the th:parameters
Ct and n were computed by b t squares, but it was coasidared that for design flood estimation more
emphasis should be given to the lower range of lag times observed in each group. Therefore the values of
Ct and n adopted for design flood estimation were determined graphically. The values am giwn in
Tdde 1.
TAW! I
VALUES

OF

AND n FOR EQUA'RON (3.12)

Equation (3.12) provides an objective means of computing the lag timc of an uageueed catcfuncnt.
Tbe paromcns L, Le and S are readily dctermincd from a topographical map. Thc difficulty is to
idant@ which hydrological group the
catchmeat conforms with. A brief qualitative dcscrip
tion of each p u p is given in Table 1. However it is recommended that the s c l d w of the group be
based on a comparison of the topographical similarities of the ungaugcd catchment with those of the
study catcbnrants. This can be done by inspection from a topographical map. It is felt that catchment
slope is probably the most important topographical characteristic d'ting catchment lag time. The
stmarn slope factor S does give some indication of the catchment slope particularly on hsmall catchments, but it is not sufficiently repmntative to provide an objective means of Jassifying the thm
hydrological groups.

Tht lag rebtionships derived in this study are significantlydifferent from the relationships dcvdopcd
for comparative catchments ovcrsurs. Linsley et. al. (1958 pg. 207) reports values for C+of 0.35 to 1.2
and n equal to 0.38 for some U.S.catchments. Thcr longer lag times observed on Malaysian cakhmmts
is again indicative of the high proportion of subsurf' flow in the f
i
d hydrographs. The travel time of
runoff from where it is depositd as rainfall to where it reaches the stregm channel is longer for subd a m flow than for overland flow. This delaying effect is most significant on small catchmentswhere
the travd timc of ruaoff in the stream channel is only a small proportion of the total lag time. As the
catchment size increases the travel timc through the soil becamcs less signi6cant than the trawl time
in the straam channd.

As with La,the values of Dp derived from different storms on the catchment varied, and* median
valw waa adopted as being representative of the catChmcIlt. The median Dp values for t& study catch-

meats arc given in Appendix B. The differences obstrvcd in the Dp values bctwm catchments wera
consistent with the three hydrological group defined previously. Catchments with steep s l o w tended
to have higher values of Dp than the flat catchments. The distribution of Dp values within each hydrological group is shown in Fig. 9. For dcsign flood estimation the median values of D, for each p u p arc
used (Table 2).
TASLE
2

VALUES OF Dp. Tb AND Tp

Group 1
Group 2
Group 3

1.06

1 . N

0.94C0

O.so*

0.89

2 . X

0.87C

0.39

0.75

2.67C

0.58C

0.22

* Adopted for design flood estimation.

Having fixed the value of Dp, the additional shape parameters TI,and Tp were estimated using
equations (3.8), (3.10) and (3.11). For hydrological Group l whem Dpexceeds 1.O, equation (3.11) gives
Tp greater than Tb/2. This is contrary to observed hydrograph shapes and arises from the triangular
approximation of the curvilinear hydrograph for short, peaky hydrographs. For design purposes it is
recommended that Tp be made equal to Tb/2 for catchments conforming with Group 1.
The values of D,, Tb and Tp for the threa hydrological groups arc shown in Table 2. Thcst values may
be used for design flood estimation fbr ungauged catchments in Peninsular Malaysia.

The baseflow component of the hydrograph reflects the antecedent rainfall activity over the catchment.
It is therefore very difficult to predict the statistical variation of basetlow prior to major floods. Fortunately bascflow is usually a relatively small quantity compared to direct runoff. It is rccommcndcd
that 5 cusecs per square mtle is a satisfactory f w e for design bamflow on catchment less than 200
square miles in area. This figure represents moderately wet antecedent conditions as observed on the
study catchments.
!?:

4.

SUMMARY OF PROCEDURE

The steps for estimating the design flood hydrograph are as foUows:
1. From a topographical map, compare the topography of the catchment with similar catchments
studied in this investigation and select the appropriate hydrological group. Compute L. LC,A and S
for the catchment.

The measurement of L and A is quite simple. To measure LCthe location of the catchment centroid
must be known. This is determined by suspending a cutout of the catchment from three different
points, and finding the intersection of plumb lines from each point. Lois then the stream length from
the point of interest to the intersectionof thu perpendicular from the antroid to the stream alignment.

The stream slope S is measured as the weighted sum of the incrcmcntal slopes between successive
stream contours.

wherc 1, = incramental stream length

St = incmnenta! s l o p

The number of increments m should be sufhbnt to &fine tho stream pro&.

2. Calculate L, for the catchment using equation (3.12)' with n equal to 0.35, and Ct for the group from
Table 1.
3. Caknrlab the design storm for the catchment using D.I.D. Hydrological Procedure No. 1 (Heiler
1973). The dasign storm should be calculated for a range of durations. Experience suggests that the
critical duration giving the bighest peak discharge is often similar to the catchment lag time.

4. Calculate Q from equation (3.1) or (3.2).

5. Cakthte e,from equation (3.5). If the total hydrograph is required calculate T b and Tpfrom the
appropriate values in Table 2.
6. Add the design baseflow component of 5 cusccs per square mile.

7. If the design strucnvc involves storage upstntam of the structure, the inflowhydrographs for varying
starm durations should be routed through the storage, and the critical outflow hydrograph detcrmi& by trial and error.
Some workcd examples illustrating the use of this procedure arc given in Appendix A.
5.

ACCURACY OF PROCEDURE

5.1 Gaerrl

Two tests werc made to assess the accuracy of the procedure. The first test compared recorded flood
hydrographs with hydrographs estimated from recorded rainfall. This test indicated how closely the
triangular hydrograph represents the time distribution of runoff. The second test compared estimates of
the 20-year flood with the 20-year flood determined from a frequency analysis of recorded peak discharges. This test indicated how well the procedure estimates statistical floods.None of the test catchments werc used to develop the procedure.
5.2. Test one -Cornpuban wit& Recorded Hydrographs

Only one catchment equipped with rainfall and streamflow recorders was availabIefor this test. Details
of the test catchment (station number 3118447) are given in Appendix D. Twa annual maximum flood

events w e n analysed to &termhe the direct runoff hydrograph and the v o l and
~ duration of raintall
excess. The triangular hydrographs wen then estimated from the recorded rainfall excess using cqution
(3.5) and (3.12) to estimate LIand qp mpectively. Values of Tp, Tband D p m saladbdfromTable 2
assuming that the catchment is typical of Group 1.

Fig. 10shows the recorded and estimated direct runoff hydrographs The first flood is caused by long
duration and relatively low intensity rainfall, and the second floodis caused by short duration and
relatively high intensity rainfall. For both floods the estimated peak discharges compare reasonably
well with the observed peak discharges. The triangular hydrographs tcptcsent the rising limb of the
observed hydrographs reasonably well, but not the latter part of the recessions. However the tail end of
the inflow hydrograph is seldom significant in the design of storage structures and for practical purposes
the triangular hydrograph appears to be satisfactory.
For the same two storms the volume of runoff was estimated from the recorded storm rainfall uxiag
equations (3.1) and (3.2). A comparison with tho recorded runoff volumes as shown in Table f dft.
monstrates the possible inaccuracies in estimating volumes of runoff from stom raidall.
OBSERVED A N D ESTIMATED RUNOFF VOLUMES
-

Date of
Storm

Storm

Rainfall (ins)

Direct Runoff cis)

Observed

Estimated

Percent

Diflennce

Although the analysis of only 2 storms is not suffcitnt to draw any firm conclusions, the test does indicate that the estimation of the runoff volume is likely to ba om of the largest sources of inaccuracy in
estimating the design flood.

5.3 Test Two - Cornprison witb Flood Frequency

Frequency analysis of historical flood resords is generally recognised as one of the most reliable methods
of determining the magnitude of statistical floods. A frequency analysis of annual maximum discharges
was done for 12 test catchments. None of the test catchments were used in the development of the flood
estimation procedure because, apart from station 31 18447, they were not equipped with rainfall and
streamflow recorders.
For each catchment, the annual maximum discharge series was fitted to a Type I extreme value
distribution by least squares. The 20-year peak discharge determined from the frequency distribution
was then compared to the 20-year flood calculated using the flood estimation proaxlure. The results are
listed in Appendix D and shown graphically in Fig. 11.
Fig. I1 indicates that the procedure tends to somewhat over-estimate the 20-year peak discharge as
determined from the frequency analysis. A possible mitigating factor is that the peak discharge data
used in the frequency analysis are based on 12 -hourly staff gauge readings for most of the test catchments.Since the actual flood peak may occur between 12-hourly readings, it is possible that the 20-year
floods estimated from the frequency analysis are somewhat low. However there were no major discrepancies apparent in the flood frequency curves for the 12 test catchments and it is considered that
they provide a reasonably reliable basis for the test. Fig. I 1 shows that 84 percent of the flood estimates
made using the procedure are withrn the range 213 to 13 of the flood frequency estimates, which is good
enough .to justify the use of the procedure for design flood estimation.
6.

LIMITATIONS OF PROCEDURE

This procedure has been prepared primarily to assist engineers in the selection of sensible and realistic
design floods in cases where hydrological data for the catchment is sparse or nonexistent. To this end it
is important that the limitations of the flood estimation methodology are clearly understood. The main
limiting assumption inherent in the method is that the flood of T-year recurrence interval is actually
caused by the storm of the same recurrence interval. There are three main reasons why the validity of
this assumption is questionable.
Firstly, the prior moisture status of the catchment affects both the volume and time distribution of
runoff. Generally speaking the proportion of direct runoff relative to rainfa11 is greater when the catchment antecedent moisture status is high. In the derivation of the rainfalf-runoff relationship for this
procedure a single curve representing average conditions has been drawn through the observed data.
The variation in the time distribution of runoff is evident in the catchment lag variation observed on the
study catchments. The physical causes of this non-linearity have been discussed in section 3.5. In this
procedure the median lag values were taken as being representative of catchment lag time. Although
there is little theoretical justification for using median values of catchment lag and average proportions
of runoff to rainfall, it is a practical alternative to introducing the statistical variation of catchment
moisture status as an additional variable. This latter approach has been explored briefly by Nash (1958).
The second reason is due to the areal variability of catchment rainfall during a storm. This has
several effects. It contributes to the variatibnin lag times observed for different storms on a catchment,
and also makes the assumption of uniform areal distribution of the design storm invalid. These effects
become more restrictive as the catchment size increases.
The third reason why the T-year flood may not be caused by the T-year storm is because of the
possibility that the peak discharge may result from a complex storm due to successive bursts of rain in
the catchment with the progression of runoff. Chow (1964 pg.^Z5-3l) states that this effect is usually
typical of thunderstorm situations over large catchments. On the West Coast of Malaysia, where large
areal variations in thunderstorm activity can occur over very short distances, this possibility is very real,
even on small catchments. On the East Coast most of the major flooding occurs during the northeast
monsoon when the rainfall tends to be more widespread. Major flood results from complex storms
were observed on a number of the catchments used in this study. A good example of this type of event is
the flood recorded at station number 31 18447 on the 4th J a n ~ j f l l This
.
flood, which was analysed
in section 5.2, was caused by successive bursts of low intensity rainfall occuring
- over several days, The

recurrence interval of this ftood as estimated from the frequency analysis is about 20 years. By comparison the %year design storm calculated using the procadure is a short duration, high intensity
storm. This clearly emphasizes the difficulties associated with estimating statistical floods from thoo-

nticat design storms.

For the reasons mentioned above, it is mommended that the flood estimation proasdun summarid
in section 4 should:
a. Not be used on catchments larger than 200 square miles.
b. Not be used as a design basis when serious consequences such as major damagc and loss of life
would result from the deign Aood being exoteded. In this case the probable maximum storm
should be used in preferenct to the design storm, with some consideration being given to the
possible combination of anal and temporal pattern giving the highest peak discharge. The
temporal and areal pattern of the probable maximum storm should be determinod from an analysis
of extreme flood producing storms in the area. The probable maximum precipitation may be
estimated from meteorological data using the standard techniques given by'Weism (1970).
To handle a varying storm temporal pattern the unit hydrograph must be uscd. In casts when nb
streamflow data arc available the triangular unit hydrograph may be estimated as shown in the following
section.
7. THE TRIANGULAR UNIT HYDROGRAPH

In some cases the unit hydrograph may be required to estimate the design flood from a storm of varying temporal pattern. The results presented in this study may be uscd to derive triangular unit hydrographs for ungauged rural catchmcnts. The procedure is as follows:
I. Sckct the appropriate hydrological group and compute Lgfrom equation (3.1 2) as before.

2. From equation (3.5) put Q = I and D = U,

where qu = peak discharge of the triangular unit hydrograph
U = period of the triangular unit hydrograph.
3. Culculate T b and Tpfrom the expressions in Table 2 as before. In this case Tb and Tprefer to the
base time and time to F& of the triangular unit hydrograph respectively.

A m w , A.P. (1973):

"Flrrod Ejtimation for t,kban and Rural Catchments", Australian Road Research, Val. 5, No. 3.

Asmw, A.J. (1970): " Variation in Lag T i mfor Natural Catchn~ents",Roc. A.S.C.E., Vol. 96,NO. HY2.
BOUGHTON,
W.C. (199): "A Mathernutical Model for Reluting Rune$ to Rainfall with Daily Data", Th8 h t i t u t i 0 ~of
Engineers Australra, Civ. Eng. Trans.. April.

~ NKOON(1969170): "Temporal Pattern of Design Storm for Singapore", Journal. Institution of Engineers
Singapore, pg. 9-12.
CHARLTON,
F.G.(1964): ''Stamiarct Catchmertllr in the Estimation of Flood Flows". The Journal ofTropical Geogputphy.
Vol. IS.
~ A N O
K

"Handbook of Applied Hydrology", McGraw-Hill. New York.

C~RDERY,I AND PILGRIM,
D.H. (1970): "Design Hydrograph Methou3 of Flood Estimation for S m d Rural Catchments".
Tbe Institution of Engineers. Australia, Civ. Eng. Trans.. October.
A Design Method", The Institution
C O R ~ & R Y1.AND WEBB,S.N.(1974): FlwdEstirndion in Eastern New South Wales
of Engineers Australia, Civ. Eng. Trans., Vot. CE 16 ,No. 1.
CHOW.V.T. (1964). Ed.:

"The Rainfall of Malaya Part I", Jour. of Tropical Geography, V O ~13.

. pg.26.
DRAINAGE
AND IRRIGATIONDEPARTMENT.
(1970): "Spaiial Distribution of Rainfofl of Large Area Storms in S e l a w " ,
Rcsearch Station lnfonnation Paper No. 4/70.
GOHKIMCHUAN
(1972): " A Comparurive Study of the Rainfall-Rumfl Characteristics of a Developed d a F
d
Catchment". M.A. Thesis. University of Malaya.
HEILXR,
T.D. (1973): "Estinratbn of the Design Rainstorm". D.I.D. Hydrological Procedure No. I. Ministry of Agriculture and Fisheries. Kuala Lumpur, Malaysia.
HEILER,
T.D. (1974): "Rational Metkod of Flood Esrimation for Rural Catchments in Peninsular Malaysia", D.I.D.
Hydrological Procedure No. 5, Ministry of Agriculture and Fisheries. Kuala Lumpur. Malaysia.
HEILER,
T.D. AND CHEW
HAI HONG(1974): "M4gnitude and Freqwrrcy of Floods in Peninsular Malaysia", D.I.D.
Hydrological Procedure No. 4. Ministry of Agriculture and Wsheries. Kuala Lumpur, Malaysia.
HEILER,T.D.AND TANHOETIM(1,974): "Hydrological Design Return PeriodF". Provisional Hydrological Procedure,
Drainage and Irrigation Department. Malaysia.
HEND~RSON.
F.M.(1966): "Opw Channel Flow", pg. 356-359, The MacMillan Co.. New YO&.
-LEY.
R.K.et a1(1958): "Hydrology for Engineers". McGraw-Hill. New York.
MCSPARXAN,
I.E. (1968) : "Design Hydrographs for Pennsylvania Watersheds", Proc. A.S.C.E.. Vol. 94, No. HY4, July.
NASH,
J.E. (1958): "Dzterminhg Runofffrom RainfaII", Proc. Instit. of Civ. Eng. Vol. 10, June.
PAHANO
RIVER
BASIN
STUDY.(Aug. 1974): VoI. 3, pg. 59, Prepared by AUSTEC for the Australian and Malaysian
Governments under the Colambo Plan.
PILGRIM,
D.H.(1966): "Storm Loss Rates for Regions with Limited Data", Proc. A.S.C.E., Vol. 92, No. HY2. March.
SNYDER,
F.F. (1938) : "Synthetic Unit Hydrograph", Trans. Am. Geophys. Union. Vol. 19, Part I.
TAYMR,
A.B. AND SCHWARZ, H.E. (1952): "Unit-Hydrograph Lag and Peak Flow Related to Basin Characteristics".
Trans. Am. Geophys. Union. Vol. 33, April.
TANHOETIM(1967): "Standard Carchment for tile Derivation of Rainfall-RunoffRelation Required for Flood Estimation
of Small Carchmerrrsin Maloya". U.N. Water Resources Series No. 30.
UNITEDSTATES
DEPARTMENT
OF THEINTERIOR,BUREAU
OF RECLAMATION
(1968): "Design ofsmall Dam", Oxford and
1BH Publishing Co., New Delhi.
WEJSNER,
C.J. (1970): "Hydrometeoro~og.y".Chapman and Hall, London.

DALE,
W.L. (1959):

LIST OF SYMBOLS
Catchment area (sq. mls.)
Lag time plus half the duration of rainfall excess (hrs)
Constant in equation (3.12)
Duration of rainfall excess (Ihrs)
Fkak ordinate of dimensionless hydrograph
Potential infiltration index
Main stream length from outlet to catchment boundary (mls)
Length of main stream from outlet to catchment centroid (mls.)
Catchment lag (hrs.)
Incremental stream length between successive contours (mls.)
Constant in equation (3.1 2)
Storm rainfall (ins.)
Total rainfall minus initial loss (ins.)
Volumc of direct runoff or rainfall excess (ins.)
Peak discharge of triangular hydrograph (CUSCCS)
Discharge ordinate of direct runoff hydrograph (cusecs)
P a k discharge of triangular unit hydrograph (cusacs)
Weighted mean slope of main stream (ft. per milc)
Incremental stream slope between successive contours (ft, per mile).
Recurrence interval (yrs.)
Base time of triangular hydrograph (hrs.)
Time to peak of triangular hydrograph (hrs.)
Recession time of triangular hydrograph (hrs.)

Time
Period of the triangular unit hydrograph (hrs.)

APPEINom A
WORKED EXAMPLES

Calculate the:design flood for a catchment having the following characteristics:

Area (A) = 20 sq. miles.
Stream length (L) I
.10 miles.
Length to centroid (L)= 5 miles

Stream slop (S) = 400 R.lmik:

btchment Location 4"N, 100"E

In this casa only the @ dischargtis required for a B y c a r tecurrencc interval. The catchment is very
stcep and covered in jungle i.a a Group 1 catchment.

= I .

Gatdate catchment lag fromaquation(3.12)

(-)

10x5

Ls = 2.0 x 4400
= 2.8 hn.

0.35

SrBP2
From D.LD. Hydrological Procedure No. I, calculate the design storm for durations of 2,3 and 6 hours.
P = 3.8 ins.
2-hr storm
P = 4.2 ins.
3-hr storm
P
= 4.7 ins.
6-hr storm
STEP3

Calculate the direct runoff volume from equation (3.2)

3.82
2-hr storm Q = (3.8 6) = !.Sins.
+

4.2"

3-hr storm Q = -- = 1.7 ins.

(4.2 + 6)

6hr

4.72
+ 6 ) = 2.1 ins.

Q = (4;7

STEP4

Calculate the 20-year peak discharge from equation (3.5). For a Group 1 catchment Dp = 1.06.
1.06x20x64ox 1.5

2-hr storm qp = -(2.8 + 1)

- = 5356 cusecs

3-hr storm qp =

1.06x2ox&10x1.7
(2.8+1.5)-

6 h r storm q, =

1.66~20~640~2.1
= 4913 cusecs
(2.8 + 3)

- 5364 cusecs
-

Example 2
It is proposed to construct a small dam across a river draining a 7 square mile catchment. Assuming a
50-yearrecurrence interval, calculate the design flood for the dam spillway given that:
Catchment is typical of Group 2
Catchment location 2"40'N, 103'40'E
Stream slope (S) = 45.5 ft/mile
Stream length (L) = 4.4 miles
Ltn,gthto Centroid (LC) = 2.0 miles
Dam s t o w area = 200 acres
Width of spillway (W) = 60 ft.
The spillway is an uncontrolled averflow type with discharge characteristics of:
discharge = 3.97 x W x H1.5
Where H is the spillway head (ft). It can be assumed that the dam storage area does not change significantly with H.

The approach here is to compute the inflow hydrographs for a range of storm durations and calculate
the outflow hydrographs at the spillway by storage routing. The outflow hydrograph with the highest
peak discharge is the design flood.
STEP

Calculate catchment lag from equation (3.12)

STEP 2

From D.I.D. Hydrological Procedure No. I, calculate the design storm for durations of 3, 6,12,24 and
48 hours.
P = 8.1 ins.
3-hr storm
P = 10.1 ins.
6-hr storm
P = 12.5 ins.
12-hr storm
P = 17.8 ins.
24-hr storm
P = 23.4 ins.
48-hr storm
STEP 3

Calculate the direct runoff volume from quation (3.2)

8.12
3-hr storm Q = -- 4.7 ins.
(8.1 6)

10.12

6-hr storm Q = --=6.3 ins.

(10.1 f 6)
12-59
= 8.4 ins.
12-hr storm Q =
(12.5+6)
17.82
= 13.3 ins.
24-hr storm Q =
(17.8 6)
23.4a
48-hr storm Q = (23.4 6) = 18.6 ins.

4
Calculate the triangular hydrograph for each storm using equation (3.5) to estimate qp, and the values
of Tp and TI,
as given in Table 2.
0 . 8 9 ~ 7 ~ 6 4 0 ~ 4=. 7
_3176CUStCS
3-hr storm qp =;
(4.4+ 1.5)

STEP

0.89~7~640~6.3
= 3394 cusecs
(4.4 f 3,O)
= 6.4 hrs
T, 0.87 x (4.4 +3)
= 16.6 hrs
Tb= 2.24 x (4.4+ 3)
0 . 8 9 ~ 7x640x8.4
r;: 3220 cusecs
12-hr storm q,, =
(4.4 +6.0)
9.0 hrs
TD= 0.87~(4.4+6.0)
Tb= 2.24 ~(4.4+6.0) = 23.3 hrs
0 . 8 9 ~ 7x640x13.3
24-hr storm q, = ----------3
: 3234 cusecs
(4.4 c 12)
r 14.3 h r ~
Tp = 0.87 x (4.4 + 12)
= 36.7 hrs
Tb= 2.24 x (4.4 + 12)
0.98 ~ 7 x 6 1 18.6
0 ~
=. 261 1 cusecs
48-hr storm qp =
(4.4 +24)24.7 hrs
Tp = 0.87 x (4.4+24)
= 63.6 hrs.
TI,= 2.24 x (4.4+24)
6-hr storm q, =

---

=E

---

Add a baseflow component of 7 x 5 = 35 cusecs to each hydrograph. The inflow hydrographs a n shown
in Fig, 12.

FIG.12

- !NFLOW

HVOROGRAPHS FOR E X A m E 2

STEP5

Route the inflow hydrographs for each storm duration through the dam storage. Any standard routing
technique may be used. The method used in this example is given by Henderson (1966). For brevity the
calculations are not given here, but the routed outflow hydrographs are shown in Fig. 13.
Comments

Tha design floodfor the particular spillway is 2900 cusecs.

The outflow hydrograph having the highest peak discharge is caused by a 24-hr storm, whereas the
intlow hydrograph having the highest peak discharge is caused by a 6hr storm. This illustrates the
point that for structures having a relatively large storage capacity, the short intense storms are
usually less critical than longer duration storms producing large runoff volumes.

FtG. 13 OWFLOW HYDROGRAPHS FOR EXAMPLE 2

The probable maximum flood is required for the design of a river bund protecting a small town. The
probable maximum flood is to be estimated using the unit hydrograph and a probable maximum storm.
CalmLrt the I-hr triangular unit hydrograph for the catchment given that:
Catchment area (A)
= 180 sq.mi1es.
= 22 miles.
Stream Length (L)
!Length to Centroid (LC)= 12 miles.
Stream slope (S)
= 150 ft./miles.
Assume the catchment is typical of Group 2.
STEP 1

Calculate catchment lag from equation (3.12)

22x12
(rn)
0.3s

LI = 4 X

= 11.7 hrs.
STEP 2

Calculate the peak discharge of the triangular unit hydrograph using equation (3.5)

0.89~180~640~1
(1 1.7q-0.5)
= 8404cusecs
Calculate Tp and Tb from the values given in Table 2
TR=+ 0.87 X (1 1.7 +0.5) s 10.6 hrs
Tb = 2.24 ~(11.7+O.5) = 27.3 hrs
q,,

"

D ~ A f L SOF STUDY CATCHMENTS

Station

Number

OLD
GROUP

CATCHMEMS

Station Name

D.I.D.

4423
4426
4436
8418
8427
8429
8467

Sg. Bidor @ Bidor

so. Batan0 fidoa8 @ Tapah
Sg. Gcdong @ Bidor
Sg. Kinta @ Taqiong Rambutan
fig. Selpneor @ km7 Frasas Hill Road
Sg. Kerlmg @ 6000 Ulu Kuala
Sg. Gombak Q kmZ1 GombPk
Sg. Batam @ Robinson Falls Intake
Sg. Perting @ Beatong
Sg. Benw @ Janda Baii
Sg. L h g @ Pimco Estate Road

2421

Sg. Kulim @ Ara Kuda

D.I.D.

3454
3456
3459

3463

GROUP 2

CATCHMENTS

NEW

Sg. Jawi @ Jan. Jalan Raya

Sg. Sungkai @ Sungkai
Sg. Kampar @ Kg. W u t
Sg. Raia @ Ladang Kita K e l h
Sg. Bmam 8 Tanjong Malim
Sg. Selangor @ Rasa
Sg. Klang @ Lebob Pasar, Kuala Lumpur
Sg. Gombak @ Jln. Pckcliling
Sg. Batu @ Scntul
Sg. Sanenyih @ Semeayih
Sg. Lui @ Kg. Lui
Sg. Mantau @ Mantau
Sg. Bcntong @ Jam. Kuala Marong
Sg. Telom @ U/S Telom Intake

I*

..

*I

J.K.R.
L.L.N.
J.K.R.

L.L.N.
**
**

I*

**

**
*I

..
*s

I*

v*

**
*.

**
**

**
.s

L.L.N.

Station Name

OLD

GROUP 3

CATCHMENTS

3466

NEW

operating

Authority

(Sq.mls.)

L
(mls.)

- -

---

LC

(mls.)

(ftlml.)

Median Value
of

LC(hrs.)
Sg. Pari @ Jalan Siiibin. Ipoh
Sg. Dmmsip.'8 @ S u h g
Sg. Durian Tunggal@ Mik 1I Air Resam
Sg. Kesaag $ Chin Chi
Sg. Sembroag @ Jambatan Brizay
Sg. Sayong @ Hilir Sayong
Se. Sayow @ Layang Layew
Se. Sebol@Jambatan
Sg. Permandi @ Batu 27. Johor
Bahru/Mersrog
Sg. Mupor @ p t u 32, Johor

Bahru/M-w
!3g. Le-r
@ Batu 42. Kluang Me*
Sg. Kanasin @ Peringat

D.I.D.
J.K.R.
D.I.D.

..

--

NO.of storms
Wysed

DD

DATA USED TO DERlVE RAlNFALL-RUNOFF RELATlONSHIP

Station-Number

OLD

NEW

(Sq. rnls.)

Date of
Peak
Discharge
27- 5-73
20- 4-73
10-11-72
30-10-72
23- 10-72
13- 9-72
18- 9-71

8-11-70
21-10-71
22-10-73
27- 5-73.
19-11-72
23-10-69
28-1 2-70
23-10-69
28-10-69
19- 5-68
29-1 l 6 7 *
2519116-

2-69
7-68
5-68
4-66
1012-65
23- 11-72
13-10-72
6 1 1-71
5- 1-71*

21- 5-65
IS- 6 6 7
29- 5-68
7-10-69
:26-11-70
30-10-65
12- 7-68.
5- 1-71
17-11-72
14- 9-64
18- 6-67
17- 9-68
20- 1-64'
31- 3-69
29-12-68
7-12-73
17- 6-67
17- 9-68
29- 3-68
18-11-68'
-

Complex Storm.

P
(ins.)

(ins.)

(hrs.)

+(ms./hr.)
Index

Areal

No. of
Rain

Factor

Gauges

Reduction

Station N u m b

OLD

NEW

(Sq.mb.)

-of
Peak
Discharge

Q
i s .

D
(bn.)

In&
$m./hr.)

Anal
Reduction

Factor

No.of
Rain

GWm

DETAILS OF TEST CATCHMENTS

Station Number

OLD

20 Yr. Flood Estimate

Station Name

NEW

Sg. W i m @ Merbau Pulas

Sg. Trolak @ Trolak

Sg. Bil @
Tg.IMalim-Slim Road
Sg. Ara @ Mile 20 Taiping-Ijok Road
Sg. ljok @ Titi ljok
Sg. Bidor (3 Mile 18 Anson-Karnpar

Operating
A
Authority ( (Sq.mls.)

D.I.D.
D.I.D.

..
I.

1.

Road

9.

Sg. Kinta @ Ipoh

99

Sg. Langat @ Kajang

Sg. Langat @ Ulu Langat Mile 20
Sg. Muar @ Kuala Pilah
Sg. Linggi @ Sua Bentong
Sg. Linggi @ Rahang

1*

J.K.R.
D.I.D.
I.
9

(mls)

LC
(mls.)

S
Catchment
(ftlml.) Group

Procedure

Flood
Max. Flood No. Years
Frequency on rccord
Rccord

(cusew)

1-(

1-(

PROCEDURES PREVIOUSLY PUBLISHED

No . 1 -- Estimation of the Design Rainstorm (1973)

...

...

No . 2 -- Water Quality Sampling for Surface Water (1973)

...

...
...

$5.00

...

...
...
...
...
...

...

...

$3.00

...
... ...
... ...

...

$5.00

...
...

$5.00

..,
...

...
...

No . 3 -- A General Purpose Event Water-Level Recarder Capricordcr Model 1598 (1973)

No . 4 -- Magnitude and frequency of floods in Peninsular Malaysia (1974)

.
No. 6 -- Hydrological station numbering system (1974) ...

No 5 -- Rational method of flood estimation for rural catchments (1974)

No. 7 -- Hydrological Station Registers (1974)

..

...

...

... ... ...

No . 8 - Field Installation and Maintenance of Capricordcr 1599 (1 974)

...
...
...
...

...
... ... ...
Recorder Digital Event Water Level (1974) ...
... ... ... ...
No. 10 -- Stage-Discharge Curves (1975) ...
No . 9 -- Field Installation and Maintenance of CapricordGr 1598

.a

...

...
... ...

...

58.00

$3.00

$6.00

$3.00

$5.00

$5.00

HYDROLOGICAL PROCEDURE NO. 16

FLOOD ESTIMATION
FOR URBAN AREAS
IN PENINSULAR MALAYSIA

JABATAN PENGAIRAN DAN SALIRAN

KEMENTERIAN PERTANIAN MALAYSIA

HYDROLOGICAL PROCEDURE NO. 16

FLOOD ESTIMATION
FOR URBAN AREAS
IN PENINSULAR MALAYSIA

Bahagian Partt dan Taliair

Kementwian Pertanian

ACKNOWLEDGEMENTS

The work reported here has been carried out by the Urban Drainage Unit of the
Drainage and Irrigation Department in association with Sinclair Knight & Partners
Pty Ltd Consulting Engineers.

(Reprint: 1980)

Published by and available from:

Publications Unit, Ministry of Agriculture

Jalan Swettenham, Kuala Lumpur.

FLOOD ESTIMATION
FOR URBAN AREAS
IN PENINSULAR MALAYSIA

Contributors:

T. J. FRICKE & K. V. LEWIS

DRAINAGE AND lRR IGATION DIVISION

MINISTRY OF AGRICULTURE AND RURAL DEVELOPMENT

MALAYSIA

85/-

MINIsrRY OF AGRICULTURE
MALAYSIA

TABLE OF CONTENTS

2.

REQUIREMENTS OF THE PROCEDURE

3.

DEVELOPMENT OF THE PROCEDURE

3.1 General

3.2 Runoff Coefficient, C

3.3 Rainfall Intensity, I
3.4 Time of Concentration, t,

3.5 Storage Coefficient, C,

3.6 Comparison of Procedures

3.7 Design Hydrograph

4.

CONCLUSIONS

5.

USE

OF THE PROCEDURE

5.1 Problem

5.2 Solution

REFERENCES
APPENDIX A
COMPARlSON OF IRUNOFF COEFFICIENTS

FROM VARIOUS SOURCES

1.

INTRODUCTION

In December 1975, Planning and Design Procedure N o 1 titled, "Urban Drainage Design
Standards and Procedures for Peninsular Malaysia", (11, was published by the Drainage
and lrrigation Division of the Ministry of Agriculture and Rural Development, Malaysia.

All the Standards and most o f the Procedures contained in the Publication were based
on recognised current overseas practice in urban drainage design and were established
following an extensive review o f overseas and local literature. The Flood Estimation
Procedure however, was developed by the Urban Drainage Unit o f the Drainage and
lrrigation Department.
This Paper develops the Flood Estimation Procedure recommended for use in urban
areas in Peninsular Malaysia and serves as a technical support volume for the earlier
publication. Results of the developed procedure are compared with other flood
estimation techniques in sample catchments in the Kuala Lumpur Conurbation.

2.

REQUIREMENTS OF THE PROCEDURE

It is required that the flood estimation procedure satisfy the following conditions:

1.

Estimate design peak discharges for urban catchments in Peninsular Malaysia

for various:
Return Periods
Degrees of urban dlevelopment
Catchment areas u p t o 20 square miles.

2.

Where necessary produce a hydrograph corresponding t o the peak discharge.

3.

Utilise input data which can be readily determined from topographic and land
use maps.

4.

Be capable o f simple hand calculation.

Of all the flood estimation procedures reviewed, the Rational Method provided the
most. suitable means o f achieving the above requirements. It was realised however
that this procedure overestimates the peak discharge for large catchments (over 200
acres), because it does not take i n t o account areal and temporal variation in storm
rainfall, and detention storage present in surface depressions, gutters, and channels.
In the development of the recommended flood estimation procedure it was not
considered eitber necessary or possible (due to lack of local data) to reduce the
peak design discharge for the areal and temporal variation in the storm rainfall,
and this' has been neglected in the procedure. It was however considered necessary
to account for detention storage effects of the larger catchments, particularly
because most drainage systems are open channel with considerable storage capacity.

The following seCtion o f this Paper develops the recommended procedure, a Modified
Form of the Rational Method, by briefly discussing the important parameters in the
Standard Rational Method atnd developing a simple coefficient which when applied t o
the Standard Rational Method can account for the effects o f detention storage.

3.

DEVELOPMENT OF THE PROCEDURE

3.1

General

The Standard Rational Method is usually expressed in terms of the following

equation:

where Q is the peak discharge in cusecs, C is the runoff coefficient depending o n

the characteristics o f the drainage area, i is the uniform rate o f rainfall intensity in
inches per hour for a duration equal to the time of concentration, and A is the
drainage area in acres. The time o f concentration, tc is defined as the time which
would be required for the surface runoff from the most remote part of the catchment t o reach the point being considered and is the sum o f the overland flow time,
to and the time o f flow in the drain, td,

T o account for channel storage, an additional coefficient, Cs, has been added t o
obtain the Modified Form o f the Rational Method as follows:

A discussion o f the important parameters in the Standard Rational Method: the runoff
coefficient, C, the rainfall intensity, I, and the time o f concentration t, together with
the development o f the storage coefficient, C, follows.

3.2

Runoff Coefficient, C

The choice o f the value of C is the most intangible aspect in the use of the Rational
Method, Taken literally, it represents the multiplier o f a 100 percent runoff peak
(assuming n o infiltration or storage] required t o obtain the design peak. This coefficient
has t o account for the various climatic conditions and physiographic characteristics o f
the. catchment.

Table 3.1 Runoff Coefficients for Urban Centres

-

- --

- -

Land Use
Business:

Runoff Coefficient

City Areas Fully built-up

and Shophouses
Industrial: Fully built-up
Residential: -

4 housedacre

- 8 housedacre

- 12 houses/acre
12 housedacre

Pavement
Parks (normally flat in urban areas)
Rubber
Jungle (normally steep in urban areas)
Mining land

For fully built-up city and industrial areas most sources were in close agreement and
values o f 0.9 and 0.8 were adopted respectively. For pavements and parks, standard
values as recommended in mast o f the literature have been adopted. Useful guides for
values of C for rubber and jungle were obtained from Heiler (4) and AUSTEC (51,
and the recommended values are in close agreement with these sources. For residential
areas the literature showed a wide variation in values, due mainly t o the different types
of residential development. T o ascertain lower limits, the percent imperviousness o f
various types o f development were analysed. The work was carried out on subdivision
maps of Kuala Lumpur and the results are shown in able 3.2.

Table 3.2

Percent Imperviousness as a Function of Land Use

Development

Housing Density
(housedacre)

imperviousness
(percent)

Terraced houses
Semi-detached houses
Detached bungalows

To determine recommended values of C, an allowance for the pervious areas was added
to the above figures based on a qualitative assessment of the runoff coefficients
presently in use.
It will be noted that no variation in C is recommended for changes in rainfall intensity
as is done in Australian Rainfall and Runoff, (6). This is because the operation of

varying C in urban drainage design is tedious and cumbersome and is not considered
warranted.

3.3

Rainfall Intensity, I

The method of estimating the design rainstorm (and hence rainfall intensities) is contained
in DID Hydrological Procedure No 1, (7), and relationships for major urban areas in
Peninsular Malaysia are published in the Urban Drainage Design Standards and Procedures
for Peninsular Malaysia, (1). The Rainfall Intensity-Duration-Frequency Relationship for
Kuala Cumpur, Figure 3.1 is included in this Paper as an example and will be used later
in this Paper in testing the procedure.
To determine the appropriate rainfall intensity it is necessary to estimate the duration of
rainfall, which in the Rational Method is the time of concentration, t,.

3.4

Time of Concentration, t,

A review of the literature revealed many empirical formulas for estimating the time of
cbncentration. Most however are based on local conditions. For urban areas in
Peninsular Malaysia it is recommended that the time of concentration be estimated
from the sum of the overland tiime, t,and the time of flow in the drain, k.

10

20

30

Storm
Figure 3 -

Rainfall Intensity

60

40

Duration

60

Duration

708090XX>

Frequency

2M)

minutes
Relationship

Kuala Lumpur

300

It is recommended that the Rantz extension (8) of the Wright McLaughlin chart (9)
be used for overland flow time (t,,) estimation. This requires a knowledge of the
runoff coefficient, C and the overland flow length and slope. For areas for which
these parameters cannot be determined such as an area yet to be mapped out, an
overland flow time of 10-15 minutes can be assumed. The Rantz chart is shown
in Figure 3.2.

The time of flow in the drain (td) can be determined from normal hydraulic formulas,
given the channel cross section, length, roughness and slope. In areas yet to be mapped
out t d can be determined by dividing the estimated drain length by 10 ft/sec for proposed
lined drains or by the average velocities indicated in Table 3.3. (Note that these velocities
are for natural streams.)

3.5

Storage Coefficient, C,

To simulate basin storage, iB Rational Method hydrograph formed by the time-area diagram
of a catchment can be routed through a hypothetical basin reservoir a t the design point.
In addition, by adopting a conservative approach to the Muskingum Routing Equation,
it can be assumed that the relationship between basin storage and the discharge is of
the form:

S = KO

(3.5)

where S is the storage volume, 0 is the discharge from the storage area, and K is a
storage delay time constant.
The determination of the proper value of K is the uncertainty in the procedure.
(b)

Determination of Storage Delay T h e Constant, K

Aitken
Aitken, (111, analysed six urban catchments in Australia with essentially the same method as described above and adopted a K value equal to 0.3%. This
procedure he termed the Clarke Model A. He compared the hydrograph obtained in
his six catchments with tho Rational Method hydrographs (using the time-area diagram),
and with the Road Research Laboratory Method hydrographs, (12).

/il

Aitken found that the Rational Method over-estimated the observed peak discharge
whilst the Clarke Model A and the RRL Methods both gave good agreement with
the observed values. Aitken preferred the Clarke Model A to the R R L Method
however because the routing was linear, and therefore simple, and because the need
to compute a non-linear storage relationship by calculating volumes of storage in
pipes was obviated.

Figure 3.2

Design Chart for Estimah'on of Overland ZTme of FIow

Table 3.3

Approximate Stream Velocities (10)

Average Slope of Channel

(percent)

Average Velocity

(feet/second)

A further simplification was introduced by taking the ratio of the peak discharges
estimated by the Clarke Model A to those estimated by the Rational Method. He
found that the reduction in peak discharge over that given by the Rational Method
depended on the shape of the time-area diagram. The relationship shown in
a storage coefficient - as the ratio of the
Figure 3.3 was arrived at, by taking C,
Clarke Model A discharges to the Rational Method discharges.

Time

Figure 3.3

of

Concentration

Storage Coefficient Curve

ftll- minutes

- By Aitken (K =

C, reached a minimum value of 0.8 for either large catchments or those with straight
line time-area diagrams. Smaller catchments with lower values of t, tended to have
"S" shaped time-area diagrams and as a consequence higher values of C,. In fact it
can be shown by a finite method, that the minimum value of C, is directly related to
the ratio of the storage delay time constant chosen, to the time of concentration.
The minimum values are achieved, either with inflow hydrographs with straight line
timearea diagrams or from large catchments. The minimum values of Cs from a
finite method analysis are shown in Table 3.4.

Table 3.4
Storage Delay Time
Constant { K)
.(proportion of )
.q

Minimum Values
of. Storage
Coefficients (CJ

Minimum Values of Storage Coefficient, Cs

1.0

.50

.90

54

.80

.57

.70.

.60

.50

.40

.30

.20

.I0

.61

.65

.68

.74

.80

.86

-93

Linsley, Kohler and Pmlhus

Linsley, Kohler and Paulhus (LKP) ( 13) described
(ii)
a lag and route procedure t o simulate basin storage, where the lag time was taken as
the time of flow in the drain (GI. The Muskingum Method w i t h K = lag and
x = 0 (ie. a concentrated linear storage) was used for storage routing and this was
claimed to be conservative.

The use o f K = td for large catchments is closely equivalent t o K = t, as the effect

of overland flow time (to) diminishes. LKP used what was essentially a Rational
Method triangular hydrograph as the inflow to the catchment storage which is equivalent t o assuming a straight line time-area diagram. Table 3.4 has indicated that for
K = 1.0t,, the attenuation o f the hydrograph formed b y a straight line time-area
diagram is C, = 0.50 which is close t o the peak reduction achieved b y LKP in their
method which used K = l.Otd.
A significant advantage in using K as a proportion o f the time in the drain is that as
the catchment gets larger, t d approaches t, and the storage coefficient approaches a
minimum. Conversely as the catchment gets smaller, q is a small proportion o f the
total time o f concentration and the storage coefficient approaches 1.0, (see Table 3.4).
This is desirable as it was intended that the procedure should approach the Standard
Rational Method for the smaller catchments b u t also t o account for basin storage f o r
the larger catchments.
Conclusions - The work b y Aitken was carried o u t in Australian catchments
(iii)
where the storage was related t o the capacity o f the drainage pipes. These were
compared t o the R R L method which also relates the storage t o the capacity of the
pipes. In Malaysian conditions the storage capacity in open channels is considered
greater than in closed pipe systems generally because average velocities are lower
thereby requiring larger cra;s-section areas (and hence larger volumes) for similar
discharges. It is therefore considered that the value o f K = 0.3% which gives a
minimum storage coefficient o f C, = 0.8 is t o o conservative. It is also considered
that the minimum storage coefficient o f C, = 0.5 achieved b y the LKP Method
with K = I.&,
is t o o low t o use without substantive data. It was therefore
decided t o take the middle course and adopt a minimum value of C, = 0.67, which
from Table 3.4 is equivalent t o having a storage delay time constant, K = 0.5%. It
was further decided that t o ensure the storage coefficient would approach 1.0 for
the smaller catchments and 0.67 for the larger catchments that the storage delay
time constant should be sat equal t o 0.5td, and this value has been adopted for
further study.

(4

Simpiifiii Fonn of the Stonrge Caeffcient, Cs

One requirement of the flood estimation procedure was that it be capable o f simple
hand calculation. It was therefore necessary to develop a simple formula for C,
that would eliminate the tedious flood routing. It was further necessary that the
formula would develop a minimum value o f 0.67 for large catchments and approach
1.0 for small catchments.

A guide to the form of th~eequation was found in the procedure used by the City
of Philadelphia, Chow (14). It was found in the Philadelphia lntet Method that a
factor, F,' could be applied to a version of the Rational Method to account for
detention storage.

In the abwe equation T is the time from the beginning of intense rainfall to the
end of the period of maximum rainfall intensity, L is the length of the drain from
the inlet to the design point, and V is the mean velocity of flow in the drain.
The approach in this procedure was to let T equal the time of concentration, &, and
.8L/V be equivalent to the time in the drain, td. The storage coefficient equation
then reduces to:

which for larger catchments where td approaches t, is equivalent to C, = 0.67 and for
small catchments where
is very small compared to t,, the value of C, approaches
1.0. This equation therefore satisfies the two required end conditions and is also simple
to use.

3.6

Comparison of Prowdures

To test the validity of using a storage coefficient, C, in lieu of the more elaborate
routing procedure, a comparison was made between the two procedures for four
catchments in the Kuala Lumpur Conurbation. The results were also compared to
those obtained by the Standard Rational Method. The details of the various catchments investigated are shown in Table 3.5, and the results of the investigation are
shown in Table 3.6.

Details o f the Catchments Investigated

Table 3.5
Catchment
Area, A (acres)
Runoff Coefficient, C

Design Return Period, T (years)

Drain Length, L (feet)
Average Drain Velocity, V (ftlsec)
Drain Time, td (min)
Overland Time, to (min)

Time of Concentration, t, (min)

Raitifall Intensity, I (incheslhr)

Table 3.6 Comparison of Results

Design Discharge (cusecs) f o r Catchments

A
B
C
D

Procedure
Rational Method
(Q = CIA)

Routing Procedure
with K = .std
Modified Rational Method

(0= C,CIA) where C, =

180

2%

554

1314

2215

2%

t,,

It can be seen from the above Table that the values obtained for the Modified Rational
Method are in close agreement with those obtained by the more elaborate routing
procedure. It was therefore concluded that the Modified Rational Method with
C, = 2tC/(2t, + td) is acceptable for use.

3.7

Design Hydrograph

In the occasional situation where a flood hydrograph is required, such as for detention
storage, a Modified Rational Method hydrograph can be developed based on the
following assumptions:

1.

The peak discharge is equal to C,CIA.

2.

The area under the modified hydrograph (volume of water) is the same as
for the Standard Rational Method hydrograph.
By setting b = base of the modified hydrograph, then

by substituting 2tc/(2t, + $1 for Cc then

= 2t,

+ t,

The problem then remains to proportion the base of the hydrograph (b) between the
rising and recession limbs. To be conservative it was decided to set the rising limb
equivalent t o the time of concentration, t, leaving the recession limb equivalent to
t, + td. This has been done and the results are shown in Figure 3.4.

Time

Figure 3.4 Modified Ration Method Hydrograph

A flood estimation procedure has been formulated which embodies the straight-forward
concepts of the Rational Method. The procedclre incorporates an additional parameter
which makes an allowance for the channel storage effects which are significant in large
urban areas in Peninsular Malaysia. The procedure is essentially empirical in nature and
does not lay claim to being a conceptual model of the runoff process.
The procedure is considered to be applicable t o the drainage design of urban catchment
areas of up to 20 square miles.

USE OF THE PROCEDURE

5.

The use of the procedure is illustrated in the following worked example.

5.1

Problem

A 5 year design discharge is required for a culvert in a catchment in Kuala Lumpur with
the following ct.laracteristics.

Area (A)

650 acres

Average Runoff Coefficient (C)

0.65

Overland Travel Distance

(I,1

500 feet

Overland Slope (S,1

5 percent

1
Overland Runoff Coefficient KO

0.3

Stream Length to Design Point

(L)

5 percent

Average Stream Slope (S)

5.2

3000 feet

Solution

The Five Year Design Discharge Q5 = Cs C I5 A

(4
From

- Overland Time, to
Figure 3.2, with L, = 500
Step I

feet, S, = 5 percent, C, = 0.3, to = 22 minutes

Step 2 - Drain Time, td

(b
With L = 3000 feet and Velocity = 4 feethecond (from +able 3.3 for S = 5 percent),
t d = (3000) / ( 4 x 60) minutes = 12.5 minutes.

(d

Step 3 - Time of C~mcentration,tc

From equation 3.2, t, = 22 + 12.5 = 34.5 minutes.

fd)

Step 4 - Storage Cwffient, C,

From equation 3.4, C, = (69/81.5) = 0.85

fej

Step 5

- Rainfill

Intensity, 1

With t , = 34.5 minutes from Figure 3.1, l5 = 4.2 inchedhour

ffl

Step 6 - Design antye, Q5

CIS = (0.85) 10.65) 14.2) (650) cusecs,

Q5

= 1508 cusecs.

REFERENCES
1.

Lewis K. V.
Cassell P.A. and
Fricke T.J.

"Urban Drainage Design Standards and

Procedures for Peninsular Malaysia",
Kementerian Pertanian dan Pernbangunan
Luar Bandar, Bahagian Parit dan Taliair,
1975.

2.

Aitken. A.P.

"Hydrologic Design in Urban Areas

A
Review", Australian Water Resources Council,
Technical Paper No. 5, 1973

3.

Schaeke J.C. et/ai

"Experimental Examination o f the Rational

Method", Pr. ASCE, Hydraulics Division,
Vol. HY6, November, 1967.

4.

Heiler T.D. and

Chew Hai Hong

"Rational Method of Flood Estimation for

Rural Catchments i n Peninsular Malaysia",
Hydrological Procedure No. 5, Kementerian
Pertanian dan Perikanan, Malaysia, Bahagian
Parit dan Tatiair, 1974.

5.

AUSTEC

"Pahang River Basin Study," Vol 5, Gov. of

Malaysia, Gov. o f Australia, August 1974.

6.

"Australian Rainfall and Runoff", Institution

of Engineers, Australia, 1st Report o f the
Stormwater Standards Committee, 1958.

7.

Heiler T. D.

"Estimation o f the Design Rainstorm",

Hydrological Procedure N o 1, Kementerian
Pertanian dan Perikanan, Malaysia, Bahagian
Parit dan Taliair, Dsc. 1973.

8.

Rantz S.E.

"Suggested Criteria for Hydrologic Design of

Storm Drainage Facilities in the San Francisco
Bay Region, California", U.S. Department o f
the Interior, Geological Survey, Water Resources
Division, November 1971.

9.

Wright-McLaughlin Engineers

"Urban Storm Drainage Criteria Manual",

Volumes 1 and 2, Denver Regional Council
of Governments, U.S. Department of Commerce,
March 1969.
"Design of Small Damsr', United States
Department o f the Interior, 1965.

11.

Aitken A.P.

"The Application of Storage Routing Methods

to Urban Hydrology", Journal of Institution
of Engineers, Australia, Jan./Feb. 1968.

12.

Watkins L.H.

"The Design of Urban Sewer Systems", Road

Research Technical Paper No. 55, A.M.S.O.
London 1962.

13.

Linsley R.K.
Kohler M.A. and
Paulhus J.L.W.

"ti ydrology for Engineers" McGraw Hill 1958.

14.

Chow V.T.

"'Handbook of Applied Hydrology", McGraw

Hill 1964.
'Use of the Rational Method in Urban Storm
Drainage Design", Drainage and Irrigation
Department, Research Station Information
Paper No. 2/70, August 1972.

16.

Proctor and Redfern

International Ltd.

"Kuala Lumpur Drainage Study", Gov. of

Canada, Gov. of Malaysia, July 1969.
"Design and Construction of Sanitary and
Storm Sewers", ASCE-Manuals and Reports
on Engineering Practice-No. 37.

APPENDIX A
COMPARISON OF RUNOFF COEFFICIENTS FROM VARIOUS SOURCES.
This Appendix contains tables o f runoff coefficients that were used to varying degrees in
determining the recommended values as shown in fable 4.1 of the text.

Tablo A-1

C Values Used By JPT (16)

Type of Drainage Area

Coefficient C

Business Areas
Residential Areas
Light Industrial Areas
Unimproved Areas
Streets
Lawns:
Sandy soil, flat 2%
Sandy soil, av. 2

- 7%

Sandy soil, steep 7%

Heavy soil, flat 2%

Heavy soil, au. 2

- 7%

Heavy soil, steep 7%

Table A-2

C Values Used By PKNS

Type of Drainage A m

Coefficient. C

Highly built-up areas

0.8

Terrace lots

0.7

Detached bungalow lots

0.5

Flat open areas

0.2

- 0.9
- 0.75
- 0.6

- 0.3

Table A-3

C Values Recommended By Proctor and Redfern (16)

Land Use

llnmined Land

Flat
0.10

steep

0.14

0.19

3 HousedAcre

0.17

0.22

4 Houses/Acre

0.20

0.25

5 HousedAcre

0.22

0.27

2 HousesIAcre
2 HousesIAcre

6 HousesIAcre

7 HwsedAcre
8 Houses/Acre

12 HousedAcre
15 HousedAcre
Multiple Family
Flats or Apartments
Shophouses
Commercial
Industrial (Light)
Institutional
Open Space
Rubber
Jungle
Fully Paved Areas
Tin Mining Land

0.15

C Values Used By The Ministry'of Environment Singapore

Table A-4

Characteristic of the
catchment when fully developed

Aversge
dOp6

less than
1 in 100

Jopa
1 in 20
to

1 in 100

Avenge
dope

grouter
than
1 in 20

Roofs connected direct t o channel or drain

Asphalt and dense pavements
City areas fully and closely built
Densely built residential areas
Residential districts not densely built-up
Rural areas with fish ponds and vegetable
gardens

Table A-5

C Valuer Recommended By AUSTEC 15)

Ultimate Land Use

Runoff Coefficient

Determined by Capacity o f
outlet from Padi Scheme

Impervious Areas - City areas

both fully and solidly built u p
Detached Houses in LJrban Areas
Rubber, Oil Palm, Secondary
Jungle, Average Grassed or
Medium-Timbered Land
Primary Jungle o r Forest
Flat Grass Areas

Padi Areas

Table A--6

C Values Recommended by ASCE (17)

Description of Area
Business
Downtown
Neighbourhood
Residential
Single-family
Multi-units, detached
Multi-units, attached
Residential (suburban)
Apartment
Industrial
Light
Heavy

Parks, Cemeteries
Playgrounds
Railroad yard
Unimproved
Character o f Surface
Pavement
Asphaltic and Concrete
Brick
Roofs
Lawns, sandy soil
Flat, 2 percent
Average, 2 t o 7 percent
Steep, 7 percent
Lawns, heavy soil
Flat, 2 percent
Average, 2 t o 7 percent
Steep, 7 percent

Runoff Coefficient

PROCEDURES PREVIOUSLY PUBLISHED

. Estimation of the Dasign Rainstorm (1973) ............. $8.00

No. 2 . Water Quality Sampling for Surface Water (1973)......... 83.00

No 1

Na . 3

. A General Purpose Event Water-Level Recorder Capricorder

Model 1598 (1973)

.............................. $5.00

. . Magnitude and Frequency of Floods in Peninsular Maiaysia

(1974) ........................................$6.00
No.5 . Retional Method of Flood Estimation for Rural Catchments
(1974) ....................................... $3.00
No. 6 . Hydrological Station Numbering System (1974) . . . . . . . . . . $3.00
No. 7
Hydrological Station Registers (1974) ................. $5.00
No 4

No 8

Field Installation and Maintenance of Capricorder 1599 (1974) $5.00

Field Installation and Maintenance of Capricorder 1598 Digital

$5.00
Event Water Level Recorder (1974)

No 9

...................
No. 10 . Stage-Discharge Curves (1975) ....................... $5.00
No. 11 . Design Flood Hydrograph Estimation for Rural Catchments
in Peninsular Malaysia (1975) ....................... $5.00
..

No 12 . Magnitude and Frequency of Low Flows in Peninsular

Malaysia (1975)

.................................85.00

No 13 . The Estimation of Storage-Draft Rate Characteristics for

River in Peninsular Malaysia ( 1975)

................... $5.00

NO '14 . Grolphical Recorders . Instruction for Chart Changing and

Annotation ( 1976)

.............................. $5.00

No 15 . River Discharge Moasurernent by Current Meter (1976)

...... 85.00

DKETAK DI JABATAN PERCET.AKAN NEGARA

OLEH MOKHTAR BIN H1. SHAMSUDDIN, K.M.N., PJ.J.,
PEMANGKU KETUA PENGARAH PERCETAKAN NEGARA,
KUALA LUMPUR
1960